CD1. European Economy. Basic editions. 44/1990. One ...

ISSN 0379-0991

EUROPEANECONOMY

COMMISSION OF THE EUROPEAN COMMUNITIESDIRECTORATE-GENERAL FOR ECONOMIC AND FINANCIAL AFFAIRS

**** ** ***, ****

One market, one money

An evaluation of the potentialbenefits and costs of forming

an economic and monetary union

No 44 October 1990

European Economy appears four times a year. It contains importantreports and communications from the Commission to the Council and tothe Parliament on the economic situation and developments. In addition,European Economy presents reports and studies on problems concerningeconomic policy.

Two supplements accompany the main periodical:

— Series A—'Economic trends' appears monthly except in August anddescribes with the aid of tables and graphs the most recent trends ofindustrial production, consumer prices, unemployment, the balanceof trade, exchange rates, and other indicators. This supplementalso presents the Commission staff's macroeconomic forecasts andCommission communications to the Council on economic policy.

— Series B—'Business and consumer survey results' gives the mainresults of opinion surveys of industrial chief executives (orders,stocks, production outlook, etc.) and of consumers (economic andfinancial situation and outlook, etc.) in the Community, and otherbusiness cycle indicators. It also appears monthly, with the exceptionof August.

Subscription terms are shown on the back and the addresses of the salesoffices are shown on page 3 of the cover.

Unless otherwise indicated the texts are published under the responsibilityof the Directorate-General for Economic and Financial Affairs of theCommission of the European Communities, rue de la Loi 200, 1049Brussels, to which enquiries other than those related to sales and subscrip-tions should be addressed.

Commission of the European Communities

Directorate-General for Economic and Financial Affairs

October 1990 Number 44

sduquenoy

Reproduction is subject to acknowledgement of the source.

Printed in Belgium, 1990

Catalogue number: CB-AR-90-044-EN-C

sduquenoy

sduquenoy

One market, one moneyAn evaluation of the potential benefits

and costs of forming an economicand monetary union

Study of the

Directorate-General for Economic and Financial Affairs

sduquenoy

sduquenoy

Abbreviations and symbols usedCountriesB BelgiumDK DenmarkD Federal Republic of GermanyGR GreeceE SpainF FranceIRL IrelandI ItalyL LuxembourgNL The NetherlandsP PortugalUK United KingdomEUR 9 European Community excluding Greece, Spain and PortugalEUR 10 European Community excluding Spain and PortugalEUR 12 European Community, 12 Member States

CurrenciesECU European currency unitBFR Belgian francDKR Danish kroneDM DeutschmarkDR Greek drachmaESC Portuguese escudoFF French francHFL Dutch guilderIRL Irish pound (punt)LFR Luxembourg francLIT Italian liraPTA Spanish pesetaUKL Pound sterlingUSD US dollarSFR Swiss francYEN Japanese yenCAD Canadian dollarOS Austrian schilling

Other abbreviationsACP African, Caribbean and Pacific countries having signed the Lome ConventionECSC European Coal and Steel CommunityEDF European Development FundEIB European Investment BankEMCF European Monetary Cooperation FundEMS European Monetary SystemERDF European Regional Development FundEuratom European Atomic Energy CommunityEurostat Statistical Office of the European CommunitiesGDP (GNP) Gross domestic (national) productGFCF Gross fixed capita) formationLDCs Less-developed countriesMio MillionMrd 1 000 millionNCI New Community InstrumentOCTs Overseas countries and territoriesOECD Organization for Economic Cooperation and DevelopmentOPEC Organization of Petroleum Exporting CountriesPPS Purchasing power standardSMEs Small and medium-sized enterprisesSOEC Statistical Office of the European Communitiestoe Tonne of oil equivalent: Not available

sduquenoy

Preface

The renewed move to EMU became operational on I July1990 when Stage I of the process proposed in the DelorsCommittee Report was begun. In December 1990 the EuropeanCommunity will begin an intergovernmental conference tonegotiate new Treaty provisions that will lay the constitutionalfoundations of economic and monetary union (EMU). TheEuropean Council has agreed that the new Treaty provisionsshould be drawn up, ratified by all Member States and avail-able for use before I January 1993. It has also decided that asecond intergovernmental conference should work at the sametime on further Treaty provisions to adapt the institutions ofthe Community to the requirements of the political union forefficient and democratic decision-making.

The Community has thus established its agenda with clarityand precision. It is an agenda of historic importance.

While the content of the agenda is both economic and political,the whole process will stand or fall on the basis of the functionalqualities of the economic and monetary union.

This is why the Commission decided to prepare a thorougheconomic appraisal of the likely economic effects — costs aswell as benefits — of the move to EMU. In so doing it wasasking its own staff in the Directorate-General for Economicand Financial Affairs to build a bridge between the politicalnegotiators on the one hand, and the community of academiceconomists on the other, and to stimulate a two-way processof motivation of economic research and supply of economicadvice. The findings of the research have been summarized inthe proposal on economic and monetary union submitted inAugust 1990 to the Council and Parliament as the Com-mission's contribution to the work of the forthcoming inter-governmental conference. The systemic proposals containedin that document are thus supported by detailed economicanalysis.

The present volume is the analysis of the Commission's staff.Another volume of papers, to be published later in EuropeanEconomy, will offer a cross-section of academic analysis ofprincipal features of the EMU challenge.

With the 1992 programme to complete the internal market, theCommission has acquired some experience of the interactionbetween policy analysis and policy implementation. In 'Theeconomics of 1992', published in European Economy inMarch 1988, the Commission presented its economic analysisunderlying the Cecchini Report on the potential gains fromcompleting the internal market. That study helped the Com-mission clarify its priorities in implementing the 1992 pro-gramme, and also to establish the credibility of 1992 in theeyes of business opinion.

Although methodologically quite different, the present studyon EMU is in some respects a sequel to this earlier work on1992. It appears at a particularly important time. While notpretending to resolve all the uncertainties, the study offers tothose concerned with preparation of the EMU Treaty pro-visions a handbook on the economic implications of theirpossible choices. The main actors in the economy — businesses,trade unions, governments — are offered a reference studywhich should help them prepare their strategies to exploit tothe full the potential benefits of EMU and minimize its possiblecosts.

While the past few years have been extremely positive for theCommunity, both in terms of its actual economic trends andits systemic developments, the onset of the Gulf crisis in thesummer of 1990 announces a more difficult period ahead.Already the oil price increase is approaching in magnitudethose experienced in the 1970s. How does this affect the EMUproject? The present study has much to say on this. While itmakes no attempt to discuss short-run questions, a pervasivetheme is how best the European economy may be equipped toabsorb major external economic shocks in order to minimizethe cost of their impact. The major conclusion is that a stronglyunified European economic and monetary system can indeedlower these costs, compared to a poorly coordinated or dispar-ate response.

Stage I of the movement towards EMU provides the frame-work within which a resolute and well-coordinated Communityresponse to the current tensions in the international economycan be arrived at. To complete the membership of the exchangerate mechanism of the EMS has always been regarded as avital element of the programme for Stage I. It is thereforeencouraging that Spain (on 20 June 1989) and the UnitedKingdom (on 8 October 1990) have now joined the eightoriginal members of the exchange rate mechanism. The twonewcomers are due, moreover, to move from the broad to thenarrow fluctuation margins during Stage I after a transitionalperiod. This extension of participation in the exchange ratemechanism means that over 90 % of the Community economyis now managed under the influence of a continuously operatingand powerful mechanism of monetary policy coordination.This is a good augury for the further convergence sought forthe remainder of Stage I, in parallel with the negotiations ofthe forthcoming intergovernmental conference.

Jacques DelorsPresident

Henning ChristophersenVice-President

Brussels, 10 October 1990

The present study was directed by Michael Emerson, Director for the economic evaluationof Community policies, with the support of a task force of economists of the Directorate-General for Economic and Financial Affairs. Two economic advisers of the Directorate-General, Daniel Gros and Jean Pisani-Ferry, made major contributions. The task forcewas based on the division for integration economics, headed by Horst Reichenbach,with Alexander Italianer (head of sector), Stefan Lehner and Marc Vanheukelen. Theeconometric modelling work was undertaken by Jean Pisani-Ferry, Alexander Italianerand Andries Brandsma. The research on transaction costs was undertaken by MarcVanheukelen. Valuable contributions were made also by the directorate for nationaleconomies, notably its director, Jorge Braga de Macedo with Joan Pearce and JiirgenKroger; by Antonio Cabral and Pedro Santos of the directorate for the evaluation ofCommunity policies; and by Christian Ghymers of the monetary directorate.

The work of the task force benefited from the guidance of a steering group consisting ofthe Director-General — successively Antonio Costa and Giovanni Ravasio — HeinrichMatthes, Michael Emerson, Jean-Fran?ois Pons, Jorge Braga de Macedo, Horst Reichen-bach and Herve Carre.

The task force received valuable support from Rudy Druine, Rod Meiklejohn, MarcNelissen, Rui Pericao, Jose Secades; Verena Barwig, Anna-Maria Durr, Brigitte Devereux,Kathrine Kaad Jacobsen, and especially Carine Collard who coordinated the preparationof the final text.

Helpful advice is acknowledged from several economists outside the Commission, includ-ing Michel Aglietta, Richard Baldwin, Peter Bofinger, Anton Brender, Ralph Bryant,Jean-Michel Charpin, Alex Cukierman, Andrew Hughes-Hallet, Peter Kenen, WillemMolle, Manfred Neumann, Richard Portes, Andre Sapir, Niels Thygesen, Frederik vander Ploeg, Paul Van Rompuy and Charles Wyplosz. ••

Certain organizations have also provided valuable assistance by way of data or surveyresults. These included several central banking institutions; the Banca d'ltalia, the Banquede France, the Deutsche Bundesbank, the Institut Belgo-Luxembourgeois de Change, theBank of International Settlements; and a considerable number of private financialinstitutions, as well as the Association for the Monetary Union of Europe and the Bureaueuropeen des consommateurs.

Appreciation is also expressed to the International Monetary Fund for permission tomake use of the Multimod model, and in particular to Paul Masson and StephanSymansky of the staff for helpful support and suggestions.

sduquenoy

sduquenoy

sduquenoy

Summary of contents

Preface 5

Executive summary 9

Part A — Synthesis and economic principles 15

Chapter 1. Synthesis 17Chapter 2. The economics of EMU 31

Part B — The main benefits and costs 61

Chapter 3. Efficiency gains 63Chapter 4. Benefits of stable prices 87Chapter 5. Implications for public finance 100Chapter 6. Adjusting without the nominal exchange rate 136Chapter 7. External dimensions 178

Part C — The impact through time and space 201

Chapter 8. Transitional issues 203Chapter 9. Spatial aspects 212Chapter 10. National perspectives on the costs and benefits of EMU 235

Annexes 249

Annex A Exchange transaction costs 251Annex B Germany and the Netherlands: the case of a de facto monet-

ary union 269Annex C European and German economic and monetary union: a

comparison 280Annex D Shocks and adjustment in EMU: simulations with the Quest

model 284Annex E Exchange rate regimes in the EC: simulations with the

Multimod model 303

Detailed contents 335

List of tables and graphs 343

sduquenoy

Executive summary

Executive summary

Summarized in terms of the three major objectives of econ-omic policy, the likely impact of EMU is:

(i) Microeconomic efficiency: sure advantages, as a sin-gle currency and economic union complement the sin-gle market and add to its impact. One market needsone money. Economic analysis supports the percep-tions of industrialists that the benefits could substan-tially reinforce the gains being obtained from 1992 —see Graph 1.1.

(ii) Macroeconomic stability: sure advantages as regardsbetter overall price stability (i.e. both very low in-

flation on average, and low variability) assuming thatthe issues of institutional central bank design arehandled well, and probably some gain also in terms ofthe stability of the real economy (lesser fluctuationsin output and employment) — see Graph 1.2.

(Hi) Equity as between countries and regions: opportunitiesand risks for all regions, and no a priori balance ofrelative advantage for the original or newer MemberStates. The least-favoured regions have a real oppor-tunity for rapid catch-up. EMU, like 1992, is a posi-tive-sum game.

The present study evaluates the benefits and costs of formingan economic and monetary union (EMU) in the EuropeanCommunity.

Economic principles

Defining EMU and its alternatives. Monetary union canconsist of either a fixed exchange rate regime or a singlecurrency. While both are possible, a single currency is foundto offer a better benefit-cost result in economic terms, andso is the main focus of the analysis.

Economic union consists of a single market for goods, ser-vices, capital and labour, complemented by common policiesand coordination in several structural, micro- and macro-economic domains. An efficient economic union requiresmuch less centralization of policy competences than monet-ary union.

For the purpose of comparison with a future EMU, thusdefined, the point of departure is assumed to be a Com-munity which has completed the internal market accordingto the 1992 programme, combined with the European Mon-etary System in which all Member States take part(1992 +EMS).

If the Community were not to move ahead to EMU, would'1992 +EMS' be a stable alternative? This is not sure, sincecomplete capital liberalization requires virtually a unifiedmonetary policy if exchange rates are to be stable. Otheralternatives to EMU might therefore be 1992 with less thana completed EMS, or a completed EMS with less than a

fully integrated single market. The net benefits of EMUwould be correspondingly greater in relation to these alterna-tives.

The economic mechanisms in play. The macroeconomic per-formance of the Community is at stake. The impacts ofEMU will go to the heart of the determinants of the rateof inflation, growth, public finance trends and even themanagement of the world economy.

Because of these different types of impact (e.g. both growthand stability) an aggregate, quantified estimate of the poten-tial impact of EMU is not feasible. However many individualeffects can be assessed empirically and this is done through-out the study. For example, savings in monetary transactioncosts with EMU would be comparable to savings in frontiercosts with 1992. In both cases these relatively small directgains are no more than a small part of the indirect andfurther dynamic gains.

This view is supported in surveys of industrialists' opinions,undertaken by independent research organizations, fromwhich it appears that the addition of a single currency tothe single market more than doubles the number who wouldexpect a very positive impact on the European businessclimate.

The larger part of the potential economic gains would notbe the automatic results of the institutional changes. The fullgains would require the concerted commitment of nationalgovernments, employers and employees as well as the Com-munity itself to what amounts to a change of economicsystem. This is because the systemic changes deliver onlypart of their economic benefits directly; in a larger measurethe benefits would flow indirectly from policy changes in-

sduquenoy

Executive summary

Two views of EMUGRAPH 1.1: A business perception of the microeconomic impact of EMU

Opinions on the prospects for the business climate become very much more positive when a single currency complementsthe single market.

1992 plus% 1992 % single currency

100-.

80-

60-

40-

20-

Positivc Littleor none

Negative

Source Business survey undertaken for the Commission by Ernst & Young.

Negative

GRAPH 1.2: An economist's perception of the macroeconomic impact of EMU

Compared to a floating exchange-rate regime, EMU improves greatly on the stability of inflation and real economic activity;it also improves on the EMS especially as regards the stability of real economic activity.

120

100-

80-

60-

Qploat

EMS (of mid-1980s)

EMS (evolving)

(3 EMU (fixed exchange rates)

EMU (single currency)

80 100 120

Output variability

Soun-e Simulations of IMF Multimod model undertaken b> the Commission staff.

10

sduquenoy

Executive summary

duced by the new institutions and rules, and changed behav-iour in the private economy (Graph 1.3 illustrates this point).

In order to mobilize expectations and to begin to inducethese indirect effects in anticipation of EMU, it is vital thatthe intention of the Community to establish the definitiveEMU within a relevant time-horizon be credible. While theprecise means of making such commitments credible is one ofthe tasks of the forthcoming intergovernmental conference(rather than the present study), it is possible in some degreeto model how the economy will respond to a fully credibleEMU. Simulations presented in this study suggest that theeconomy would be not only more efficient and less inflation-ary, but also subject to less variability of prices and outputlevels.

The main benefits and costs

The main chapters of the study organize the analysis andfindings along the following lines:

(i) Efficiency and growth. Elimination of exchange rateuncertainty and transaction costs, and further refine-ments to the single market are sure to yield gains inefficiency. Through improving the risk-adjusted rate ofreturn on capital and the business climate more gener-ally there are good chances that a credible commitmentto achieving EMU in the not-too-distant future willhelp further strengthen the trend of investment andgrowth.

(ii) Price stability. This is a generally accepted objective,and beneficial economically in its own right. The prob-lem is that of attaining price stability at least cost, andthen maintaining it. The Community has the oppor-tunity of being able to build its monetary union on thebasis of the reputation for monetary stability of itsleast inflationary Member States. Given the paramountimportance of credibility and expectations in winningthe continuous fight against inflation at least cost, thisis a great advantage.

(iii) Public finance. A new framework of incentives andconstraints will condition national budgetary policies,for which the key-words will be autonomy (to respondto country-specific problems), discipline (to avoid ex-cessive deficits) and coordination (to assure an appro-priate, overall policy-mix in the Community). EMU willalso bring valuable gains for many countries' nationalbudgets through reductions in interest rates, as inflation

and exchange risk premiums are eliminated. These ben-efits will very probably outweigh the loss of seignioragerevenue to be experienced by some countries.

(iv) Adjusting to economic shocks. The main potential costof EMU is that represented by the loss of monetaryand exchange rate policy as an instrument of economicadjustment at the national level. This loss should notbe exaggerated since exchange rate changes by the Com-munity in relation to the rest of the world will remainpossible, whereas within the EMS the nominal exchangerate instrument is already largely abandoned, and EMUwill reduce the incidence of country-specific shocks.Relative real labour costs will still be able to change;budgetary policies at national and Community levelswill also absorb shocks and aid adjustment, and theexternal current account constraint will disappear.

Moreover, model simulations suggest that with EMU,compared to other regimes, the Community would havebeen able to absorb the major economic shocks of thelast two decades with less disturbance in terms of therate of inflation and, to some extent also, the level ofreal activity. This is of renewed relevance, given thatthe Gulf crisis of summer 1990 once again subjects theCommunity to a potentially damaging economic shock.

(v) The international system. With the ecu becoming a majorinternational currency, there will be advantages for theCommunity as banks and enterprises conduct moreof their international business in their own currency;moreover the monetary authorities will be able to econ-omize in external reserves and achieve some inter-national seigniorage gains. EMU will also mean thatthe Community will be better placed, through its unity,to secure its interests in international coordination pro-cesses and negotiate for a balanced multipolar system.

The impact through time and space

(vi) Transitional costs and benefits. The costs of the tran-sition to EMU (in disinflating, reducing budget deficits),for the countries not yet fully prepared, would be greatlyreduced by the setting of clear political commitments tothe definitive EMU at a not-too-distant time in thefuture. If economic agents (public authorities, com-panies, trade unions, individuals) perceive these com-mitments to be credible, they will anticipate EMU intheir economic strategies and behaviour. Such a process

11

Executive summary

GRAPH 1 .3 : Simplified schema of the effects of EMU

Systemicchanges

(economic andmonetary union,

internationalsystem)

Policychanges

(economicand

monetarypolicy)

Behaviouralchanges ,

(by economic agents:business, trade unions.

financial markets)

Finaleconomicimpacts

(microeconomicefficiency,

macrocconomicstability,

equity betweencountries

and regions)

Note: See Graph 1.4 for greater detail.

has already been at work to advantage with the 1992programme, and a similar strategy could be equallybeneficial for EMU, if not more so.Since most of the costs of moving to EMU arise in thepreparatory stages of the process (as set out in theDelors Committee report), whereas some of the import-ant benefits (elimination of exchange rate uncertaintyand transaction costs) arise only in the definitive regimewith a single currency, there is a clear economic casefor a relatively short duration for the transitional period.

(vii) Impact on the regions. As regards the regional distri-bution of the impact, which is relevant to the objectiveof longer-term convergence of economic performance,there are no a priori grounds for predicting the patternof relative gains and losses. There are risks and oppor-tunities of different types affecting both categories ofregions. Policies are already at work to reduce lo-

cational disadvantages of the least favoured and geo-graphically peripheral regions. However, the key to thecatching-up process lies in obtaining synergies betweenCommunity and national efforts to upgrade the leastfavoured regional economies. The fixing of clear policyobjectives, such as for the single market and EMU, arealso highly relevant here for mobilizing such efforts.

(viii) Convergence. About half the Community (D, F, B, NL,L, DK, IRL) could proceed now to EMU with littledifficulty, notably given their advanced degree of con-vergence in terms of inflation and cost trends. Threeother countries (I, E, UK) have some adjustments tomake still, but these are surely feasible within a fewyears. The two remaining countries (GR, P) have largeradjustments to make, but these countries too could,with political will, set their sights on participation inthe full EMU, at the same date as the rest of theCommunity.

12

sduquenoy

sduquenoy

sduquenoy

Executive summary

Key points relevant to the forthcoming intergovernmentalconference, which will prepare Treaty provisions f or EAfU.are:

(i) the case for a single currency as the first-best formof monetary union is supported on economic grounds,since only in this way can exchange uncertainty andtransaction costs be completely eliminated and thebenefits of full credibility of the union fully secured;

(H) the case for the central bank to be granted a clearmandate to secure price stability and institutionalindependence is also supported, and indeed necessaryto secure the full potential gains of a single currency;

(Hi) the case for centralized powers over budgetary policyis much weaker, but the future regime will still haveto conciliate the need for a common standard ofdiscipline over deficits and debt with that for flexibleresponse at the national level to country-specificshocks;

(iv) the transitional adjustment of private and public sec-tors to the disciplines of EMU will be made morerapidly and at less cost if the final objectives forEMU are the subject of credible commitments by thepublic authorities for completing the monetary unionat a not-too-distant time in the future.

Further conditions for a successful functioning of EMUare:

(i) an effective economic policy coordination functionwill be required, built presumably around the powersof the Council of Ministers for Economics and Fi-nance;

(it) this coordination function will have to concern theoverall macroeconomic policy mix of the Community,and thus require concertation with EuroFed and aunified Community participation in internationaleconomic and monetary cooperation;

(Hi) Member States will be largely responsible for manag-ing national budget deficits at the low levels consist-ent with a high standard of monetary stability, sinceCommunity rules in this area should not establish aprecise centralized control;

(iv) the costs of absorbing country-specific economicshocks will be minimized if labour costs adjust rela-tively flexibly;

(v) the catch-up process of the least-developed countriesand regions depends critically on securing synergiesbetween on the one hand national development andstabilization efforts, and on the other hand the Com-munity's policies for the single market and EMU.

13

sduquenoy

sduquenoy

sduquenoy

Part A

Synthesis and economic principles

15

sduquenoy

sduquenoy

Chapter 1. Synthesis

Chapter 1

IS

The present study is an evaluation by the staff of the Com-mission of the potential economic impact of forming aneconomic and monetary union (EMU) in the EuropeanCommunity. It provides an analytical background to theproposals which the Commission addressed to the Counciland Parliament in August 1990 entitled 'Economic and mon-etary union'.

1.1. What alternative regimes are to becompared?

The processes of completing the single market by 1992 andnow of achieving EMU are continuous ones. To make ananalysis of their benefits and costs requires that the pointsof comparison be clear, which is not itself such a simplematter.

The end-point of the process must be defined, as also thealternative regime which would otherwise be assumed toprevail at that time.

As regards the end-point, the Delors Committee reportidentifies the definitive monetary regime as consisting eitherof a fixed exchange rate system or a single currency system.There are significant economic and political differences be-tween the two, with the single currency emerging as the first-best economic system. The latter is therefore the principalreference for the end-point, although the comparison withthe fixed exchange rate alternative is thoroughly discussed(Chapters 2, 8).

Economic union is a less clear-cut concept than monetaryunion. A completed internal market for goods, services,labour and capital is a large and necessary feature of eco-nomic union. Beyond that, however, there is a wide spectrumof conceivable possibilities. These range between addingrather little to 1992 to examples of federal economic unionssuch as in the United States, with a very large federalbudget and associated responsibilities for public expenditurefunctions and taxation.

The approach of the present study is to follow the principleof 'subsidiarity' that is increasingly advocated by the Com-munity institutions. According to this principle only thoseeconomic policy functions which can be more efficientlydischarged at the Community level are transferred from thenational level. The list of policies that qualify on these

grounds is likely to evolve gradually over time as increasinginterdependence makes various policies more difficult tomanage efficiently at the national level. Indeed, the increasein Community competences assumed for the purpose of thepresent study, to be added to 1992 and so to constitute theeconomic union, are basically evolutionary in nature (seeChapters 2, 3 and 5).

As regards the alternative regime, as a first approximationthis might be the status quo in the first half of 1990, justbefore the beginning of Stage I of the EMU process, but, aswill be seen, this is not entirely satisfactory.

The status quo of the first half of 1990 sees the legislativeprogramme for completing the internal market as a littleover half adopted. There were already, however, signs thatthe economy has been anticipating 1992, both at the level ofbusiness decisions and national economic policies. Moreoverthe economic impact of 1992 has earlier been the subject ofdetailed evaluations in the Cecchini report and associatedliterature (see 'The economics of 1992' in European EconomyNo 35 of March 1988). For these reasons the status quomay be more suitably defined as assuming completion of the1992 programme.

On the monetary side the status quo of 1990 sees eightcountries adhering to the narrow-band variant of the ex-change rate mechanism (ERM) of the European MonetarySystem (EMS). Two countries (Spain and the UnitedKingdom) joined the broader band of the ERM recently (on20 June 1989 and 8 October 1990 respectively), and twocountries (Greece and Portugal), remain outside the ERM.However, all Member States are committed to joining thenarrow band of the ERM as well as completing the liberaliza-tion of capital movements. While the timing of these remain-ing actions to complete the EMS is not fixed, it is agreedthat this will take place in Stage I of the EMU process whichbegan on 1 July 1990.

The present study concentrates therefore on comparing thenet advantage of moving from a Community that had com-pleted the single market and membership of the EMS(1992 + EMS) to an EMU comprising a monetary unionwith a single currency and an economic union possessing aminimum of competences.

In one important respect, however, this approach may wellerr on the side of underestimating the net advantages ofmoving to EMU. This is because the advantages of1992+EMS may not all be sustainable without the clearprospect of further movement to the definitive EMU. Theeconomic advantages of 1992 are certainly not fully achieva-ble without a single currency, especially in the field of finan-cial market integration. In addition the EMS in its present

17

sduquenoy

Part A — Synthesis and economic principles

stage of development may not be compatible with completecapital market liberalization as required by 1992. The statusquo may not, therefore, be a stable one. If the move to EMUwere not to take place, it is quite likely that either the EMSwould become a less stable arrangement or capital marketliberalization would not be fully achieved or maintained.For these reasons it can be argued that the alternativeregime for the benefit-cost analysis should be something lessadvanced in terms of integration than either the full 1992programme or the present EMS or both (these alternativesare discussed more thoroughly in Chapter 2). In essence, thealternative to progress is regress, rather than the status quo.If this argument is accepted, it logically strengthens theeconomic case for moving ahead to EMU.

Intermediate stages of monetary integration which are in-tended to be transitional are not substantially analysed, sincetheir overall economic consequences are hard to evaluate,for reasons commented on below (Chapter 8). This is trueboth of the Stage II of the Delors Committee report, andthe alternative Stage II proposed by the British Government.

1.2. The conceptual framework

EMU will have a very pervasive impact on the workings ofthe economy. Many different mechanisms will come intoplay and interact. It is therefore particularly important toestablish a clearly structured conceptual framework withinwhich to discuss the detail.

The impact of EMU may be viewed schematically (in asimplified form in Graph 1.3 above) as a chain of cause andeffect that passes through successive phases: first systemicchanges, these leading to actual policy changes, and then onto behavioural changes in the economy, before emergenceof the final economic impacts in terms of the objectives ofpublic policy. Thus, more precisely:

(i) the definitive EMU represents initially a set of systemicchanges, or changes to the economic constitution. Thisinvolves the new single currency (the ecu) in its domesticand international role, the new central bank (EuroFed),the completed single market (1992) and a changed rolefor national budgets (budget rules);

(ii) the policy changes occur in both the monetary andeconomic branches. The introduction of a single cur-rency and the EuroFed institution has a profound im-pact on the management of monetary policy. For exam-ple, the commitment of monetary policy to price stab-ility should acquire decisively strengthened qualities ofunity and credibility. Economic policy-makers will also

find themselves operating within the constraints of newdisciplines and responsibilities as regards budgetarypolicy;

(iii) the behavioural changes of economic agents concernenterprises, trade unions, households and financial mar-kets. They come to recognize that the systemic changesnot only affect them directly through the single marketand single currency, but also indirectly through thepolicy-makers (monetary and budgetary policies, forexample, become less accommodating with respect tonon-competitive costs and prices);

(iv) the final economic impacts may be identified in termsof the three conventional objectives of economic policy:microeconomic efficiency in resource allocation andeconomic growth, macroeconomic stability with regardboth to inflation and output and employment, andequity with respect to the distribution of impacts be-tween countries and regions. The systemic, policy andbehavioural changes all culminate in impacts that fallunder one or more of these objectives.

However, as was suggested in Graph 1.3, some of the impactsof the systemic changes deliver benefits directly, such aswhen the single currency eliminates transaction costs, with-out passing indirectly through policy and behaviouralchanges. Similarly for policy changes, there is a mix of directand indirect impacts.

Moreover there are some important interactions betweenthese final impacts. For example, lower inflation contributesto higher efficiency; and faster economic growth contributesto the catch-up of backward regions.

This schema of analysis, in its simplified form, is not yetsufficiently precise to identify the economic mechanisms thatactually deliver the final benefits and costs. Greater detail istherefore given in Graph 1.4, in which the mechanisms areidentified and these form the core structure to the study asa whole. Each of these mechanisms is discussed individuallylater in this chapter, and they are also summarized as acheck-list in Box 1 at the end of this chapter.

Sixteen mechanisms may appear complicated, but this rep-resents the reality that the impacts of EMU will be bothcomplex and heterogeneous in nature. There will be costs aswell as benefits, direct as well as indirect effects, automaticas well as conditional effects. Compared to the economicimpact of 1992, which essentially improves the efficiency ofthe economy, EMU will in addition have an importantimpact in lowering the average rate of inflation, as well asin lowering the variability of both the level of output andinflation. All of these impacts correspond to distinct objec-tives of economic policy. However reliable quantification is

18

sduquenoy

Chapter I. Synthesis

only possible for some of them. For this reason, quite apartfrom the problems of weighting together heterogeneous im-pacts, no overall quantitative estimate of the potential netbenefits of EMU is presented. This difference with the Com-mission's earlier study on the impact of 1992 does not meanthat EMU is any less far-reaching in its implications, on thecontrary. The economic methodology used to assess thesevarious types of impact is explained in Chapter 2, withfurther detail in supporting chapters and annexes.

While the individual mechanisms at work may be numerousand complex, the overall picture becomes clear when it isobserved that they all converge in their effects upon one ormore of the three classic objectives of economic policy:efficiency, stability and equity.

Thus:(i) Efficiency: the addition of a single currency to a single

market will perfect the resource allocation function ofthe price mechanism at the level of the Community asa whole. Without a completely transparent and surerule of the law of one price for tradable goods andservices, which only a single currency can provide, thesingle market cannot be expected to yield its full benefits— static and dynamic. With 'one market and onemoney', the Community economy could confidently beexpected to improve further its underlying economicperformance. This already appears in the views of indus-trialists recorded in surveys, according to which theaddition of a single currency to the single market wouldmore than double the number of enterprises expecting

GRAPH 1.4: DetaOed schema of the effects of EMU

Systemic changes, Policy changes Behavioural changes Final economic impacts

/*"~ Microeconomic "Monetary union

Ecu as singlecurrency

Common, indepen-dent Central Bank

V J

Economic union

Single marketRules for nationalbudgetsShock -absorbingfunctions of budgets ,

f International ^\system

Ecu as international

Tn- polar monetarysystem

•

*»

~t

Monetary policy

Credible.stability-oriented

- monetary policyLoss of nationalexchange rateEffective

V international policy J

Economic policy

Disciplined_ national budgets

Coordination ofbudgetary policyPublic sector

L efficiency i

f Behaviour of ̂ \economic agents

Wages and prices

Interest rates

Trade and capitalL flows j

- ' — *

' — fc

efficienc>

Static and dynamicgams in efficiency

-and growth

Internationa]financial gains^ J

f ^ A acroeconomic^Nstability

InflationAverage rate

1" Variability

OutputAverage level

i Variability j

/"" Equity between "\countries and regions

Catch-up of

Recovery of areashit by shocks

19


a 'very positive1 impact on the business climate in theCommunity — as was shown in Graph 1.1.

(ii) Stability: governments aim to secure a favourable com-bination of two main macroeconomic equilibria — theprice level and real economic activity (output and em-ployment). More precisely, the objectives are to achievelow inflation and unemployment levels in terms of theirequilibrium rates, and also minimum variability in bothcases. EMU, with a well designed institutional structurebuilding on an inherited reputation for price stability,is a highly probable means of securing the price stabilityobjective. The growth effects of extra induced invest-ment could also, possibly, reduce the underlying unem-ployment rate. The prospects of securing a reducedvariability of inflation and output may be less self-evident, but evidence from new methods of model simu-lation suggest that a favourable outcome can be securedhere too, as was shown in Graph 1.2. These simulationsreplicate the typical economic shocks experienced in thelast two decades such as the oil shocks. They suggestthat under EMU the Community would have absorbedthese shocks at less cost than was the case under lessunified monetary regimes. (Correspondingly, the newoil shock of 1990 could be better handled if EMU werealready in place.)

(iii) Equity: the main focus here is between regions andcountries whose initial conditions differ markedly, forexample advanced versus less-developed. There are ar-guments that go in off-setting directions, for exampleeconomy of scale advantages of the centre versus lowerlabour costs and faster potential productivity gains inthe least favoured regions. There are no grounds forexpecting the net balance of relative advantage to fallsystematically in one direction or the other. Communitypolicy instruments are addressed to making good weak-nesses in basic resource endowment where possible. Butthe major determinant of the desired catch-up processwill be whether the public authorities and economicagents of the regions and countries concerned can securesynergies between Community strategies for 1992 andEMU and domestic development efforts.

1.3. The main benefits and costs

The 16 mechanisms already announced are now explainedsufficiently to indicate their most important features, with afull account given in supporting chapters.

These mechanisms are presented in groups (five in all, corre-sponding to Chapters 3 to 7). The first two, concerningmicroeconomic efficiency and price stability and monetary

policy, introduce the most important direct effects of eco-nomic and monetary union.

The next two groups, concerning public finance and theproblem of adjusting the economy without changes in thenominal exchange rate, introduce important indirect impactsfor both the stability and efficiency of the economy, andalso concern conditions for a successful functioning of EMU.

The final category concerns impacts on the internationaleconomic and monetary system. This has implications formacroeconomic stability and, at the microeconomic level,for business and banking through the role of the ecu as aninternational currency.

1.3.1. Concerning efficiency and growth

EMU will result in an amplification of the type of economicbenefits that follow from the 1992 programme. Indeed onlya single currency allows the full potential benefits of a singlemarket to be achieved. For this reason it is no coincidencethat they are historically observed to go together. The eco-nomic literature also now points out how the static efficiencygains from market and monetary integration should alsotranslate into dynamic gains in terms of the rate of economicgrowth for a medium-term oeriod at least.

Exit exchange rate variability and uncertainty(mechanism 1)

The EMS has observed a considerable reduction in nominalexchange rate variability between the ERM participants, butthis is not the case for currencies outside the ERM. In recentexperience (1987-89) the average monthly variability of indi-vidual EC currencies against all other EC currencies hasbeen 0,7% for ERM currencies and 1,9% for non-ERMcurrencies. These are therefore the degrees of variability thatEMU could eliminate.

The reduction of real exchange rate variability has been lesspronounced. However, as argued further below, some realflexibility (i.e. in relative labour costs and competitiveness)is actually desirable.

Exchange rate uncertainty is as damaging to the economyas variability if not more so. Uncertainty, as revealed byexchange risk premiums in interest rate differentials, canpersist well beyond the actual stabilization of the exchangerate. This is why moves from the present EMS to monetaryunion would be highly beneficial to the business environ-ment, even if the recent stability of EMS exchange rates hasbeen impressive.

20

sduquenoy


There is evidence that foreign direct investment respondspositively to exchange rate stability. Recent surveys showthat exchange rate stability is generally highly valued byindustrialists. Econometric evidence on the impact on tradeis weak, but the impact through the efficiency and volumeof investment seems likely to be more important.

Exit transaction costs (mechanism 2)

These typically range from the very small for inter-banktransactions (0,1 %), to the more substantial for commercialtransactions (0,3 % for an ECU 100 000 transfer, 0,5 % forECU 10000), to the very costly for small bank transfers(12% for ECU 100). The cost of exchanging bank-notesbetween major and minor currencies can be an extremelyhigh 25%. In addition companies incur in-house costs,which case studies indicate can amount to around 4 % ofthe profits on sales in other EC markets.

Overall transaction costs can be conservatively estimated toamount to around 1/2 % of GDP (ECU 13 to 19 billion peryear) for the Community as a whole, which is to be comparedto the 1/4 % of GDP estimated in the Cecchini report forborder costs for the circulation of goods. Both appear to beonly a small part of the overall market-segmenting influ-ences. On the monetary side the combination of risk factorsand transaction costs has a particularly damaging impacton efficient resource allocation.

Transaction costs are particularly high (1 % of GDP) forsome of the Member States which have small and very openeconomies or weak currencies that are little used inter-nationally.

Apart from foreign exchange transaction costs, cross-bordertransactions bear additional costs (bank charges and com-missions, delays) due to the present inefficiency of inter-national payments systems. Only the combination of 1992and a single currency can offer the prospect of reducingthese costs to a minimum, such as within the United States.

In a fixed exchange rate system the authorities could conceiv-ably impose a 'par-clearing' practice on banks, as was thecase at one stage in the United States: (i.e. no bid-offermargins and no special exchange transaction costs would bepermitted). But this would still presumably require monetaryreforms, so as to give all currencies of the system the samepar value, which practically amounts to moving to a singlecurrency.

Building on 1992 (mechanism 3)

The 1992 programme will achieve the essential break-through for market integration. However, there are several

respects in which microeconomic policies of the Communityshould be able to secure further economic benefits beyondthose following from implementation of the 1985 WhitePaper. These include deeper liberalization of energy andtransport markets than provided for in the White Paper.European infrastructural investments for cross-bordertraffic, including high-speed trains and better air trafficcontrol, will complement market-opening policies. Compe-tition policies for reducing subsidies and controlling undesir-able concentrations have a large agenda for action ahead.Increasing integration will call for some tax harmonizationinitiatives to avoid inefficient market distortions. The R&Dpolicies of the Community could be further developed toprofit from economy of scale advantages for many projects.The current expansion of the structural Funds, in the courseof being doubled between 1988 and 1992, will have to beevaluated and adapted to the needs of economic union.

From static to dynamic gains (mechanism 4)

The gains from 1992 and monetary integration have tradi-tonally been analysed in terms of step improvements ineconomic efficiency and welfare, i.e. a once and for all gain,even if this is achieved only progressively over a period ofyears. Actual experience of how the 1992 process seems toaffect the economy and recent developments in the economicliterature have drawn attention to how a major impetus,such as 1992 or EMU or both together, may also trigger anincrease in the underlying rate of growth of the economy.Two features of this analysis may be highlighted.

First, if the productivity of factors of production is improved(as through the 1992 programme) or if the risk-adjusted rateof return is increased (as through EMU in diminishingexchange rate risk), then investment will increase and socarry with it the rate of economic growth at least in themedium run. Estimates show that if EMU reduced the riskpremium in the required rate of return by a moderate amount(0,5 percentage points) there could be a substantial growthoffered over the long run, accumulating to perhaps 5 % ofGDP.

Second, analysis of the European unemployment problemhas recently given attention to the 'hysteresis1 effect, accord-ing to which alternative equilibrium rates of unemploymentare possible, with the actual one heavily influenced by recenthistory or inertial factors. According to this interpretationan impetus that creates more optimistic expectations in theeconomy may effectively shift the economy onto a higherperformance trajectory, with a dynamic adjustment from oneequilibrium unemployment rate to a second more favourableone. More optimistic expectations,- initiated by a credible

21


policy deemed favourable to the business environment, maythus become self-fulfilling.

Although such arguments may still be speculative, they bearcomparison with the rise of business confidence and invest-ment in the Community since the time when the 1992 pro-gramme became credible. Given the very positive businessopinions on EMU reported above, it is quite conceivablethat the addition of EMU to 1992 could indeed carry theeconomy onto a still stronger trend performance.

1.3.2. Concerning price stability

Since it is proposed that the Community's future centralbank (EuroFed) be given the statutory duty to secure pricestability as its priority objective, it is important to be clear ofthe economic justification for this choice, and why monetaryunion could achieve it at least cost.

The benefits of price stability (mechanism 5)

Taking the macroeconomic record of the industrializedcountries in terms of inflation, growth, unemployment andincome per capita levels over medium- to long-run periods,there is no evidence that higher inflation can be exploited asa means of securing the other economic policy objectivesmentioned. The old theory that there is a trade-off betweenhigh inflation and low unemployment is now unsupportedas a matter of theory or empirical analysis, except for short-run periods. On the contrary there is some (statisticallyweak) positive relationship between low inflation and highreal economic performance. The post-war macroeconomicexperience of the industrialized world suggests that, on aver-age, high inflation countries have a higher unemploymentrate and a lower income per capita.

To identify more precisely why this should be so, economicanalysis makes the distinction between anticipated and unan-ticipated inflation.

Anticipated inflation imposes costs on the holders of money,and on producers in the need to change price lists. It does,however, give quasi-tax revenues to the government (seig-niorage), which is only an advantage to the extent govern-ments cannot collect taxes efficiently.

Unanticipated inflation may allow governments to benefitin the short run by depreciating the public debt in realterms; but in the long run this becomes a disadvantage sincemarkets impose an inflation risk premium on governmentswhich do not earn a reputation for stability-oriented policy.

High inflation rates are also more variable and uncertain,and cause more relative price variability. This makes theprice mechanism less efficient in its resource allocation func-tion. At the macroeconomic level it results in 'stop and go'policies, which make efficient business strategies extremelydifficult.

Institutional factors conducive to price stability(mechanism 6)

While any government and central bank can in principlesecure price stability, there are reasons to expect this objec-tive to be achieved with a higher probability and minimalcost by an independent central bank which is given thisobjective as its statutory duty.

Political science recognizes, in this context, the importanceof the different incentive structures of central banks andgovernments. Governments typically have electoral cycleswhich may at times make it attractive to exploit short-term gains against longer-term costs. Monetary policy isvulnerable in this situation, unless the central bank haspolitical independence, board members with long and securetenure, and statutes establishing an explicit duty to givepriority to price stability.

In practice these factors appear responsible for the observedcorrelation between institutional independence of the centralbank and price stability, with Germany and Switzerlandrepresenting the most positive examples.

Thus there are sound reasons for the design of EuroFedto follow these examples, more than some average centralbanking constitution.

The costs of disinflation (mechanism 7)

The reduction of inflation is usually resisted because it in-volves transitional costs. These arise for two reasons: wageand price rigidity and the difficulty of securing credibilityfor the stabilization objective.

These two factors are related, and both can be stronglyinfluenced by institutional strategies. This has already beenobserved in the processes of convergence in the EMS, withthe linkage of the exchange rate of countries whose stabiliza-tion strategies were initially weak in credibility to those withstrong credibility. Political commitment to a central bankingsystem, which itself has a reputation for price stability, isthus a key to minimizing the transitional costs of disinflation.Expectations are at the heart of the inflation process. Thisis why disinflation costs could be reduced by agreement in

22

sduquenoy


the Community on a suitable institutional design of Euro-Fed, and clear commitments to its full entry into operationat a fixed date.

Such a date should be just long enough to allow for necessaryadjustments, but not so distant as to appear irrelevant toeconomic agents.

1.3.3. Concerning public finance

The public finances of Member States will be deeply affectedby the move to EMU. This will involve direct benefits, aswell as some costs, for the budgets of some countries in thetransitional period. But more important will be the recastingof the role of national budget policies in the monetary union,and the intensified competitive pressure in due course onnational public expenditure and tax systems.

Revision of the framework for macroeconomic budgetarypolicy (mechanism 8)

Loss of the monetary policy and exchange rate instrumentat the national level will place new demands on budgetarypolicy at the national level for stabilization and adjustmentpurposes in the case of country-specific disturbances. Thisrequires flexibility and autonomy, at least within a normalrange of sustainable public deficit and debt levels.

However, a high standard of monetary stability implies thatunsustainable public deficits and debt cannot be monetizedanymore. Budgetary discipline, in order to avoid excessivelyhigh deficits, will need to be intensified, and the Communityis discussing proposals to this effect. Measures to this effectwould be justified because it is not self-evident whethermonetary union will itself have the automatic effect of en-hancing budget discipline at the national level. While moneti-zation will cease to be possible, financial market integrationwill make available a larger source of savings to nationalauthorities and the external current account will disappearas a direct financial constraint.

Increased economic interdependence between MemberStates will also warrant more intensified coordination ofbudgetary policies and mutual surveillance, especially inrelation to the overall monetary-budgetary policy-mix andthe Community's external current account.

Direct impacts on expenditures and revenues(mechanism 9)

In EMU the profits from the issue of currency will naturallybe collected at the Community level, and these seigniorage

revenues will in some way be returned to the national econ-omies (central bank profits are usually transferred to thebudget), and to this extent there will be no loss of seigniorage.Countries with the highest inflation rates may suffer- netseigniorage losses, but not greater than 1/2 to 1 % of GDP,given that the 1992 programme will already induce throughcompetitive pressures a reduction in the very high reserverequirements imposed in some countries by the central bank.

On the other hand countries with high interest rates willbenefit as these come down to the rates prevailing incountries with the lowest inflation rates. In part these re-ductions will only be in the nominal interest rate, offset bya lower erosion by inflation of the real value of the publicdebt. Even this, however, is not without significance, sincea lower nominal interest rate would help reduce the apparentbudget deficit automatically, and imply less need to cut realpublic expenditure programmes that might seem necessary.However, real interest rates could also be reduced wherethese carry at present a significant premium reflecting marketperceptions of expected depreciation and exchange-rate risk.

This margin is in several countries 2 % or more. While thesepremiums are transitional in nature, they are typically largecompared to the likely net loss of seigniorage in the case ofCommunity countries which are already embarked on thetransition towards monetary stability. (Seigniorage is onlyreally important for countries that have permanently veryhigh inflation rates and poorly developed tax systems).

Indirect impact on expenditures and taxation(mechanism 10)

EMU will intensify competitive pressures on certain publicexpenditures and taxes on mobile factors. As a general spurto public sector efficiency these pressures are to be welcomed,for example to improve education systems and economicinfrastructure, and avoid social security systems imposingunnecessarily heavy taxes on employment. However carewill also need to be taken to avoid under-provision of publicgoods whose benefits spread across frontiers, or the erosionof taxation on tax bases that can easily migrate. This is theeconomic justification for acts of harmonization or establish-ment of minimum standards of public goods or tax rates inappropriate cases.

1.3.4. Concerning adjustment withoutexchange rate changes

The principal risk in forming a monetary union is thatinvolved in losing the possibility to change the nominalexchange rate. This section evaluates that risk, and the

23


alternative instruments for adjusting economies in circum-stances when nominal exchange rate changes might other-wise be made. The chief uses of exchange rate changesare to improve labour cost competitiveness (e.g. devalue toprotect employment) or control inflation better (e.g. revalueto reduce prices). The inflation issue has already been dis-cussed above, and so the present section focuses on the usesfor the exchange rate for purposes of preserving employmentand regional balance in the case of adverse economic shocks.

Loss of the nominal exchange rate instrument(mechanism 11)

The loss of the nominal exchange rate instrument impliedby EMU is the most important economic cost involved, butthis should not be exaggerated for several reasons. First, thenominal exchange rate of the ecu, in relation to the rest ofthe world, will still be variable. Secondly, participants in theERM have already forgone nominal exchange rate changesto a high degree, and even those currencies outside the ERMpursue 'firm' exchange rate policies to avoid risks of higherinflation. Thirdly, changes in nominal exchange rates havebeen no ultimate guide to real exchange rate changes (i.e.relative labour costs and competitiveness) between ECcountries in the last decade; a correlation is only observedfor an initial period. They therefore chiefly serve only tooffset differences in inflation rates. Exchange rate changescan help for an initial period to protect employment fromthe effects of an adverse economic shock, but at the cost ofpoorer results in terms of inflation and also, in the mediumterm, dissipation of the employment gains too. (Simulationresults are given in Chapter 6 and Annex D.)

Real adjustments in competitiveness (mechanism 12)

Real exchange rate changes remain possible in EMU, sincesome important prices can change still between regions (e.g.those of housing and commercial property) as well as wagecosts. Some federations (e.g. Canada) have seen as large realadjustments between provinces as the EC national economiesobserved between themselves in the last decade. While ad-justments to adverse shocks need to remain possible, theevidence of the last two decades within the EC shows thattrend changes in real exchange rates have not explaineddifferences in growth rates at all. Persistent real devaluationstrategies do not seem to buy faster growth, at least not inWestern Europe.

EMU will reduce country-specific shocks (mechanism 13)

The case for exchange rate changes relates mainly to theincidence of country-specific shocks that cause losses of

demand for a country's typical products, or of competitive-ness for its producers. While the possible occurrence ofcountry-specific shocks cannot be eliminated, they are likelyto become less probable for three reasons. First, integrationas a result of 1992 and EMU leads to changes in industrialstructures in the direction of deeper 'mtra-industry' tradeand investment relations, which means that most countriesbecome involved in both exporting and importing the prod-ucts of many industries. Old-style comparative advantage,in which countries specialize their production in distinctcommodities, becomes less important. As a result sector-specific shocks become to a lesser degree country-specific intheir impact, Secondly, a credible monetary union will affectthe behaviour of wage-bargainers. They will be more carefulabout risking becoming uncompetitive, given that devalu-ation will not be an option. Thirdly, EMU will eliminate animportant category of country-specific shocks which orig-inate in exchange rate movements themselves and imper-fectly coordinated monetary policy. Recent methods ofmodel simulation suggest that EMU could as a result signifi-cantly reduce the variability of inflation and also to someextent output (see Chapter 6 and Annex E). These resultsapply also to the costs of asymmetric responses by MemberStates to common economic shocks, of which oil price risesare the major example (recurring in 1990 for the third timein recent history).

Financial flows will be available to absorb shocks(mechanism 14)

A major effect of EMU is that balance of payments con-straints will disappear in the way they are experienced ininternational relations. Private markets will finance all viableborrowers, and savings and investment balances will nolonger be constraints at the national level. National budgetswill, as mentioned, retain their capacity to respond to na-tional and regional shocks through the mechanisms of socialsecurity and other policies. The Community's structuralpolicies have a complementary role, and may be furtherdeveloped in response to needs.

1.3.5. Concerning the international system

The primary aim of EMU is to strengthen the integrationof the Community and improve its economic performance.However, due to its weight in the world economy, EMU willnecessarily have far-reaching implications for the inter-national economic and monetary system. Care will have tobe taken to avoid some potential problems, and to make thechanges a positive-sum game for the international com-munity as a whole.

24

sduquenoy


The ecu as a major international currency (mechanism 15)

As the single currency of the Community, the ecu will com-pete as an international vehicular currency with the dollarand yen on an equal footing. The ecu would no doubtincrease its international market share as a numeraire fortrade and contract thus saving around 0,05 % of GDP tothe Community in transaction costs; also as a means ofpayment and store of value in the denomination of assetsand liabilities, perhaps to the extent of a conservative 5 %of the total world portfolio. Community enterprises andbanks would have the advantage of doing more of theirbusiness in their own currency. The economy would be lesssubject to short-term variability in its terms of trade as aresult. Seigniorage gains from the use of ecu banknotesin other countries would develop, gradually accumulatingperhaps to around USD 35 billion (as a one-time gain, notannually). EMU would also be beneficial to those partnercountries, mainly in the rest of Europe presumably, whichwere to choose to peg their currencies to the ecu. Inter-national reserve holdings of ecu-denominated assets outsidethe Community will probably increase, but since these willbear market interest rates they represent little direct gain.The Community itself would be able to economize in externalreserve assets, perhaps by as much as one half, or USD 200billion. The various portfolio shifts into ecus could conceiv-ably cause an unintended exchange rate appreciation, butthis would not necessarily happen since borrowers and len-ders might increase their demand for ecus in parallel, andthe movement could be a gradual one over many years.

International cooperation (mechanism 16)

EMU would effectively unify and strengthen the Com-munity's presence in international forums. As the Com-munity and its Member States would speak with one voice,a reduction in the number of players (e.g. from G7 to G4)should facilitate the coordination process. The Communitycould intervene more effectively to encourage developmentsof the world policy-mix that would be advantageous to itsinterests. In terms of systemic developments EMU could bea decisive building block for establishing a balanced tri-polar monetary regime.

1.4. The impact through time and space

While the ultimate impact of these many consequences ofmoving to EMU is highly likely to be exceedingly beneficialthere are further dimensions to be evaluated. For the Com-munity as a whole there is the question how the flow of

benefits and costs will develop over time through the stagesproposed in the Delors Committee report. This will influencethe optimal speed of the move towards the single currency.For each Member State there is the question how it inparticular stands to be affected, given differences in eco-nomic structure at the outset. A preoccupation of severalcountries is how the economically least-favoured regions arelikely to be affected. (These issues are examined inChapters 8 to 10).

1.4.1. Transitional issues

The optimal speed of the move to the definitive EMU willdepend on three inter-related factors: first, how benefits andcosts of EMU develop stage by stage, secondly, whether theintermediate stages will face risks of instability, and thirdly,the progress made by Member States in the convergence oftheir economic performance. The overall conclusion pointsto the advantages of relatively rapid moves through to thesingle currency, but taking sufficient time to foster greaterconvergence for several countries. There is a strong economiccase for setting fixed and credible time-tables, to spur conver-gence and encourage anticipatory adjustments on the partof governments and businesses.

Benefits and costs by stages

Exchange rate variability within the ERM has already beenreduced by three-quarters of what preceded establishmentof the EMS, and this will presumably be the experience alsoof new ERM participating currencies as its membership isgeneralized in Stage I. Correspondingly, a large part of thecosts of adjustment to exchange rate stability have alreadybeen borne, or will be so in Stage 1. However, exchange rateuncertainty, and associated interest rate costs may remain agood deal longer.

The benefits from achieving fully credible monetary stability(internal and external) can only be achieved by a singlecurrency, since exchange rate fixity can be affirmed but notproved. A single currency would be for all enterprises, tradeunions and individuals sign of a genuine regime change, andwould help ensure that they adapt to the required wage andprice discipline. In addition transaction costs will only beeliminated with a single currency. International advantageswill also accrue especially with a single currency.

The costs of doing without exchange rate changes as anadjustment instrument will gradually diminish over time aseconomic integration becomes deeper.

25

sduquenoy


Overall the adjustment costs of EMU mostly arise early inthe transition, and the benefits are harvested to a consider-able extent only at the end of the process, as is illustratedby Table 1.1 for the 16 mechanisms discussed above. Thesingle currency is clearly the first-best solution economically.

Possible instability in the transition

The complete openness and increasing integration of finan-cial markets resulting from the 1992 programme will increasethe potential for currency substitution. Speculative attacks

on individual currencies could be of enormous power iffinancial markets are given any reason to doubt the commit-ment of the authorities to defend the fluctuation margins ofthe ERM. The transition will also have to manage smoothlythe change of monetary policy leadership from one basedon Germany to that of the independent EuroFed. Thesefactors favour a short Stage II while EuroFed becomesestablished but does not have full responsibility. It alsopoints to the advantages of the single currency as the defini-tive regime, in preference to fixed exchange rates, in termsof clarity, credibility and therefore stability.

Table 1.1Economic mechanisms generating benefits and costs, by stages of EMU, as in the Delors Committee Report

Stage I Stage Ilia(fixed

exchange rates)

Stage I lib(single

currency)

Efficiency and growth

1. Exchange-rate variability and uncertainty2. Exchange transaction costs3. Extending 1992 to economic union4. Dynamic gains

Price stability

5. Price discipline6. Institutions conducive to stability-oriented monetary policy7. Transitional costs of disinflation

Public finance

8. Autonomy, discipline, coordination9. Lower interest-rate costs (less seigniorage losses)10. Public sector efficiency

Adjustment without exchange-rate changes

11. Loss of nominal exchange-rate instrument12. Adjustment of real wage levels13. Lesser country-specific shocks14. Removal of external constraints

International system

15. Ecu as international currency16. Improved international coordination

Note: + benefit, - costs. • insignificant or uncertain. The comparisons are between the stages and a baseline case that assumes completion of the single maiket and membership of the EuiopeanMonetary System's exchange-rate mechanism.

26

sduquenoy


Convergence requirements

Monetary union requires virtually complete convergence ofinflation. At the limit, with perfect markets it imposes thelaw of one price on all tradable goods and services. Monetaryunions nevertheless often sustain persistent divergences inrelative wage levels, in line with productivity and competitiv-ity differences. In the case of countries catching up in levelof productivity and incomes, there will as a result be somemargin (possibly around 1 to 2% per year) for consumerprices on average to rise a little faster than in the Communityas a whole as the prices of non-tradable services convergeon the levels found in high income countries.

Budget balances and current account balances can divergeeven more substantially, as the Dutch example of highbudget deficits and current account surpluses and the reverseBritish example suggest. The essential requirements forbudgetary policy is sustainability of the public debt withoutrecource to monetary financing.

It is therefore important that a reasonable degree of conver-gence of price inflation, and reduction of excessive budgetdeficits are achieved before a locking of the exchange rates.However, the convergence of price and cost performanceand budgetary policies will be strongly influenced by thecredibility of commitments to a given future monetary re-gime, so long as the time-horizon is not too long. For thisreason commitments to the later stages of EMU should notbe held up until the ultimately required degree of conver-gence is already achieved.

Of the existing ERM countries, Germany, France, Belgium,Luxembourg, the Netherlands, Denmark and Ireland arealready sufficiently convergent for monetary union as re-gards price inflation. The other large Member States —Italy, Spain and the United Kingdom — are not so farbehind and adequate convergence should be attainablewithin a few years. Even if Portugal and Greece have abigger adjustment to make, these countries could, with amajor effort of political will, sufficiently catch up on the restof the Community so as to join the definitive EMU at theoutset.

1.4.2. Regional impact

Neither economic theory nor the current experiences of theleast favoured and geographically peripheral regions of theCommunity point to a bias in the sense that these regionsmight systematically profit either more or less from EMUthan the average. While the economic centre of the Com-munity benefits from economy of scale advantages, it is not

evident that these relative advantages are destined to grow.The least favoured regions still have other advantages. Whilethe latter may fear the economic power of the large corpor-ations of the centre, the trade unions of the centre fear thecompetition of low cost locations in the periphery. Com-panies are quite willing to relocate to the periphery, accord-ing to business surveys, if competitive advantage within theinternal market is offered.

The Community's policies are already addressed to reducingthe locational disadvantages and weaknesses in resourceendowment of the western and southern periphery. This isdone both in the 1992 programme and the structural Funds(for example in the transport and telecommunications sec-tors). The structural Funds also aim to improve basic re-source endowment of weaker regions through manpowertraining and investment.

The diverse experiences of Ireland, Spain, Portugal, Greeceand the Italian south point to the possibilities of a successfulcatching-up process, but also to this being far from inevitableand to be conditioned essentially by the extent of the commit-ment of the countries and regions concerned to generalizedmodernization. Such modernization has to extend verydeeply, and to acquire socio-political as well as purely eco-nomic dimensions. The Community of 1992 and EMUclearly offers a framework for such efforts, which may am-ount to a thorough regime change for society as well as theeconomy. This will be even more emphatically illustrated inthe case of the Community's newest region, East Germany.

1.4.3. Benefits and costs by country

As regards the benefits of achieving price stability, EMUclearly offers most at this point to the five Member States(Italy, Spain, United Kingdom, Portugal, Greece) with in-flation rates above the 2 to 3% standard encompassing theother seven countries. The adjustment costs of disinflationshould be lessened as these countries enter into commitmentsto join progressively more ambitious stages of the EMUprocess (indeed, Italy and Spain have both given examplesof this in the last year in respectively moving into the narrowand broad bands of the ERM).

Germany is apprehensive over the risk that EuroFed's mon-etary policy might not keep to the highest stability standard.While this issue should be resolved in deciding the statutes ofEuroFed, a broader perspective suggests positive advantagealso for Germany. In the absence of the move to EMU it isunlikely that Germany's partners would be so committed toprice stability, and the risks of importing inflation from

27


them would be correspondingly greater given also that thetheoretical option of a floating exchange rate is now excludedin practice.

While the extent of the remaining exchange rate variabilitystill to be reduced is closely related to present ERM commit-ments, significant costs of exchange rate uncertainty (asincorporated in interest rate premiums) are borne by mostERM countries. In particular, all ERM countries exceptGermany and the Netherlands stand to gain from lower realinterest rates in passing to a fully credible EMU. Non-ERMcountries would also secure such advantages at a comparablestage in the stabilization process. These interest rate savingswill benefit public budgets as well as private investment.

The elimination of transaction costs in passing to a singlecurrency will especially benefit the countries with poorlydeveloped financial markets and traditionally weak curren-cies. These costs are particularly heavy for the tourist sectorsof the southern Member States.

All Member States should benefit from the improved busi-ness climate created by EMU, and by the stronger inter-national presence of the ecu and the Community.

1.5. The Community as an advantageousmonetary area

One way of drawing together the arguments set out aboveis to consider how they relate to the traditional argumentsof the economic literature concerning the 'optimal currencyarea'. The present study has made use of the main insights ofthis theory. However, there are further relevant arguments,especially in the case of the Community, that need to bebrought into account in order to guide the overall andpolicy-relevant judgement, which is whether EMU couldindeed be advantageous when all benefits and costs areconsidered together. (The optimal currency area theory wasoriginally developed to define the optimal geographic juris-diction of a given money; the Community more nearly facesthe inverse task of defining the optimum economic andmonetary competences of a given geographic jurisdiction.)

The two original insights of work on the optimal currencyarea were:

(i) labour and capital should be sufficiently mobile acrossthe regions of the union to permit adjustment to region-specific economic shocks to take place without too seri-ous costs, especially in terms of unemployment (seeChapter 6);

(ii) monetary union would be beneficial in eliminating trans-action costs in exchanging currencies (Chapter 3).

The formation of currency areas would depend then on theactual trade-off between these two factors. Since in theCommunity labour mobility is still limited, the net balanceof advantage might on these grounds alone be uncertain.

Subsequent developments to the theory added two moreconsiderations that already make the benefit-cost balancemore favourable in the case of the Community:

(i) if the States concerned are individually small and open,then exchange rate changes will cause significant disturb-ances to price stability (Chapter 4);

(ii) if an economy has a diversified industrial structure, withmore intra-industry trade than narrow specializationsin particular commodities, then the likely incidence ofeconomic shocks warranting exchange rate changeswould be less (Chapter 6).

Further branches of economic analysis and empirical factsparticularly relevant to the Community push the probablebenefit-cost ratio much further still in favour of EMU in theCommunity:

(i) the analysis of inflation and efficient price stabilizationstrategies has come to attach increased importance toexpectations, the credibility of policy commitments, andthe reputation of institutions. When these analyticaldevelopments are combined with the actual possibilityin the Community of building the monetary unionaround the stability standard established by its leastinflationary members, an important argument for EMUis added (Chapter 4);

(ii) moreover recent methods of macroeconomic model si-mulation, incorporating the effect of forward-lookingexpectations, provide evidence that the overall stabilityproperties of the economy, of both price and to someextent real activity levels, would be improved underEMU (Chapter 6);

(iii) related considerations arise in political economy litera-ture on problems of 'government failure' in budgetaryas well as monetary policy. EMU offers also an oppor-tunity to correct problems of this nature experienced bysome Member States with a new institutional structureand set of incentives and constraints for national autho-rities. In the extreme cases this amounts effectively to a'regime change', which is being achieved over a periodof years in the newer Member States, and more suddenly

28


now in the case of East Germany (Chapters 9 and 10and Annex C);

(iv) recent developments in the theory of economic growthsuggest that the static efficiency gains of 1992 and EMUcould well translate also into dynamic gains in the rateof growth (Chapter 3);

(v) finally, the Community is a special case in that its largesize means that formation of EMU will not leave theinternational system unaffected (as would, for example,a Benelux EMU), but offers further advantages throughcausing changes in that system (Chapter 7).

The overall results of the benefit-cost analysis of the Com-munity facing the prospect of EMU may therefore be sum-marized as follows. Old economic theory guiding judgmenton the formation of monetary unions might have suggested,for the Community, that the trade-off between the twoarguments then considered relevant were of uncertain advan-tage. New economic theory and particular features of theCommunity's actual structure and situation add sevenfurther arguments, all pointing to benefits for EMU in theCommunity. On these grounds the economic case becomesstrongly advantageous. Political union objectives mayfurther be added. But the case can stand powerfully oneconomic criteria alone.

29

sduquenoy


Box 1.1: Check-list of economic impacts of EMU

Impacts relating to efficiency and growth

\. EMU would completely eliminate nominal exchange ratevariability which in recent years has averaged 0,7% per monthfor ERM currencies and 1,9% for non-ERM currencies (eachcurrency against all other EC currencies). It would also eliminateuncertainty (which interest rate premiums show to be consider-able even where actual exchange rates have been stable).

2. Only a single currency completely eliminates the transactioncosts of exchanging currencies. These costs are non-trivial formany businesses, and can be sizeable for small transactionsbetween persons and for tourists. They amount to at least 1A%of GDP per year (ECU 13-19 billion) for the EC as a whole,and up to 1 % for smaller Member States.

3. Going beyond the single market measures of the 1985 WhitePaper, additional economic advantages can be secured by furthermeasures in such fields as energy, transport, competition,R&D, environmental and taxation policies.

4. The combination of the 1992 programme and EMU couldwell translate into not only considerable 'static' once and for allefficiency gains, but also 'dynamic1 gains (i.e. a higher sustain-able rate of economic growth). Estimates show that a moderatereduction in the riskiness of investment (e.g. exchange rateuncertainty) could have a substantial growth effect in the long

Impacts relating to price stability

5. Price stability is itself advantageous for efficient resourcesallocation. While difficult to measure, this is borne out in macro-and microeconomic evidence.

6. EuroFed will be most likely to secure price stability if itsstatutes establish this as its priority duty and grant it politicalindependence to fulfil this duty. The actual price stability per-formance of independent central banks is clearly positive.

7. The reduction of inflation to a common very low rate couldbe achieved at minimal transitional cost through clear politicalcommitment to a EuroFed of this design, to become fully oper-ational at a not too distant date.

Impacts relating to public finance

8. The role of national budgetary policies will be substantiallyrevised, with new needs for autonomy to permit flexibility com-

bined with enhanced discipline over excessive deficits and coordi-nation to ensure an appropriate policy-mix for the Community.

9. The most inflationary countries will lose seigniorage revenuesup to 1% of GDP but against this there will often be interestrate savings of larger size, notably where these comprise asignificant exchange rate risk premium.

10. Competitive pressures should increase the efficiency of pub-lic expenditures and taxation, but in some cases the Communitymay have to establish minimum tax rates and cooperate in theprovision of public goods to avoid inefficient outcomes.

Adjustment without exchange rate changes

11. The main cost of EM U is the loss of the national monetaryand exchange rate instrument. This cost should not be exagger-ated, since it has already largely been renounced within theERM, while changes will remain possible for the Communityas a whole in relation to the rest of the world.

12. Changes in real exchange rates (competitiveness) remainpossible and desirable within EMUs, and examples from feder-ations show that this is not just a theoretical possibility. This iswhy wage and price flexibility is a necessary condition of success.

13. EMU will reduce the incidence of country-specific shocksthat warrant real exchange rate changes, as a result of changesin industrial structure and wage bargaining. Also shocks result-ing from exchange rate instability and uncoordinated monetarypolicy will be eliminated, and so the variability of output andinflation reduced.

14. Financial flows will be available to absorb economic shocks.Monetary union will remove external financial constraints, andnational and Community budgets will also help absorb shocks.

Impacts on the international system

15. As the Community's single currency, the ecu will developinto a major international currency, resulting in several kindsof financial advantage for the Community's economy: less trans-action costs in international trade, more ecu-dominated financialissues managed by European banks, smaller needs for externalcurrency reserves, seigniorage gains on foreign holdings of ecunotes.

16. EMU should facilitate international coordination, and givemore weight to the Community in encouraging developmentsof the world policy-mix favourable to its interest. It should alsofacilitate establishment of a balanced tri-polar regime.

30

sduquenoy

Chapter 2. The economics of EMU

Chapter 2

The economics of EMUThis chapter presents the main economic assumptions and thetheoretical and methodological tools upon which the study isbased. Section 2.1 is devoted to a discussion of the content ofEMU and of the alternatives against which its effects areassessed, whereas Section 2.2 presents the underpinnings andmethods of the cost/benefit approach.

The main conclusions are (he following.

Economic union will increase the degree of integration ofcapital, goods and labour markets across the Community, butnot to the degree of integration reached by national economies.This will differentiate EMU from existing economic and mon-etary unions consisting of federal States, where labour mobilityacross States is higher than will be the case in the Community.

Already with irrevocably fixed exchange rates, monetary unionimplies complete equalization of nominal interest rates onequivalent assets, almost perfect substitutability of nationalcurrencies when these remain in a fixed exchange rate systemwith monetary policy in the hands of a single institution. Asingle currency does not increase the macroeconomic con-straints imposed by EMU. It offers, however, a number ofadditional advantages. This study concentrates, therefore, onthe benefits and costs of a full EMU with a single currency.

The experience of monetary unions suggests that in EMUinflation convergence will not need to be perfect. Moreover,as the catching-up of less advanced Member States proceeds,their real exchange rate should appreciate through a slightlyhigher inflation rate, due to the catch-up in the relative priceof non-tradable products. Rough calculations suggest thesedifferentials might in some cases exceed 1 % and even 2 %per annum. The same applies to wages and labour costs as awhole, for which differentials may be even larger.

The gains from EMU are evaluated with respect to a '1992+ Stage I' baseline in which realignments would still be

possible, but in practice only in the case of shocks. It isassumed that Stage I would remain as asymmetric as thepresent EMS, and would also lead to further convergence ininflation rates. Since it cannot be taken for granted that StageI could remain stable indefinitely, consideration has also to begiven to the gains from EMU with respect to alternativeexchange rate regimes.

There is no ready-to-use theory for assessing the costs andbenefits of EMU. Despite its early insights, the 'theory ofoptimum currency areas' provides a too narrow and somewhatoutdated framework of analysis. Recent developments in both

micro- and macroeconomics have not yet led to a unifiedtheory of monetary unions. However they provide buildingblocks for a comprehensive analysis of EMU.Four major categories of permanent effects can be expectedfrom EMU:(i) microeconomic efficiency gains, which arise from the

elimination of exchange rate uncertainty and transactioncosts, and lead to a permanent increase in output;

(ii) macroeconomic stability effects, which arise both fromthe elimination of infra-Community exchange rates andfrom policy discipline in the monetary and fiscal fields,and impact on the variability of output, prices, and othermacroeconomic variables;

(in) regional equity effects, which concern the distribution ofcosts and benefits of EMU among Member States andregions;

(iv) external effects, through a wider international role ofthe ecu, tighter international policy coordination, andpossibly changes in the international monetary regime.

In addition, important macroeconomic effects should arise inthe transition towards EMU:(i) Both the lack of a unified theory and the diversity of

effects involved imply that an attempt to make an overallquantitative assessment of EMU would be meaningless.However, partial quantifications are provided when poss-ible in this report.

(ii) In comparison to alternative benchmark exchange rateregimes, i.e. financial market autarky and free float,EMU can be expected to yield significant benefits. Thus,in so far as the risks of instability in Stage I could leadthe system to revert to some mix of capital controls andcrawling peg as in the early EMS, the net benefit of EMUwould only be greater.

2.1. EMU and alternatives

The basic definition of EMU is institutional and legal. Sec-tion 2.1.1 first discusses its economic content, which isnecessary for an assessment of its costs and benefits. Section2.1.2 then discusses the economic content of the referencecases against which the costs and benefits of EMU areassessed, namely a'Stage I + 1992'baseline, and alternativeexchange rate regimes.

2.1.1. The economic content of EMU

The Madrid meeting of the European Council of June 1989endorsed the approach to EMU proposed by the report of

31


the Committee for the Study of Economic and MonetaryUnion (the Delors Report), and as decided by the EuropeanCouncil the first stage of the realization of EMU began on1 July 1990.

The economic definition of EMU given by the Delors Reportis now widely accepted. For monetary union, it uses as adefinition the three conditions stated in the Werner Report(1970), namely:(i) total and irreversible convertibility of currencies;(ii) complete liberalization of capital transactions and full

integration of banking and financial markets;(iii) elimination of margins of fluctuation and irrevocable

locking of exchange rate parities.

As mentioned in the Report, the first two conditions havealready been met, or will be with the completion of theinternal market programme.' Therefore, the decisive steptowards full monetary union appears to be the irrevocablelocking of exchange rates, which implies the creation of asingle central bank (EuroFed). As discussed in more detailbelow, the adoption of a single currency would then behighly desirable. Since membership in the exchange ratemechanism (ERM) of the EMS is presently narrower thanCommunity membership, it would be important that all itsMember States participate in the ERM beforehand.

The definition of economic union is less clear-cut since itinvolves measures relating to different fields. It is also, bynature, more open-ended than monetary union. The DelorsReport describes it in terms of four basic elements:(i) the single market within which persons, goods, services

and capital can move freely;(ii) competition policy and other measures aimed at

strengthening market mechanisms;(iii) common policies aimed at structural change and re-

gional development;(iv) macroeconomic policy coordination, including binding

rules for budgetary policies.

The precise contents and implications of these measures andpolicies are discussed below and analysed in various sectionsof this study. Table 2.1 below gives a synthetic presentationof the features of EMU compared to those of the singlemarket programme.

Capital markets liberalization is effective in most Member States. OnlySpain, Portugal and Greece still maintain exchange controls and thesewill be progressively removed by 1992. In some cases the phasing-outmay be extended until 1995.

Table 2.1Major features of 1992 and EMU

•1992' Stage IEMU

FinalEMU

Monetary union

ConvertibilityFree capiial movementsIrrevocable parities/single currencyAll Member States in the ERM/EMU

Economic union

XX

X

XXXX

Single marketCompetition policyRegional and structural policiesMacroeconomic coordination

XPP

XPPP

XEEE

P = panially. E = enhanced.

General principles

The above presentation makes clear that the overall concep-tion of EMU aims at a certain balance in the progresstowards monetary union on the one hand, and economicunion on the other. This approach is called parallelism andwas stressed by the European Council meeting of Madrid.Other principles to be followed in the design of EMU aresubsidiarity and the necessity to allow for specific situations.

Subsidiarity is an important criterion for assigning tasks tothe different levels of government in a multi-level govern-ment system. In its most general form, it states that tasksshould be assigned to the lowest level of government, unlesswelfare gains can be reaped by assigning it to a higher level.2It is therefore a principle which aims at the decentralizationof government functions as long as this is justified on ef-ficiency grounds. In the Community context, the applicationof this principle should ensure that a policy function isassigned to the Community level only when it can beperformed in a more efficient way at that level than bynational or local governments.

Application of the principle of subsidiarity in the economicfield can be based on the two familiar criteria of cross-country spill-overs and economies of scale as discussed inBox 2.1.

2 This definition follows Van Rompuy, Abrahams and Heremans (1990),to which the reader can refer for a detailed discussion of economicfederalism.

32


Box 2.1: The economic meaning of the principle of subsidiarity:assigning tasks to the Community on efficiency grounds

Two economic criteria can be used as necessary conditions forassigning on efficiency grounds a particular policy function tothe Community:

(i) assignment of a policy function at the national level is inef-ficient because of the existence of cross-country spill-oversgiving rise to externalities; since national governments donot take fully into account the consequences of their actionson the rest of the Community, they are bound to takesuboptimal decisions;

(ii) the management of a policy function involves indivisibilitiesand/or economies of scale, which imply that productivityand effectiveness are improved when it is performed at ahigher level.

For both criteria it is essential that externalities or economiesof scale are significant at the Community level. Environmentaleffects (e.g. acid rain) provide classic cases of externalities;other examples can also be found in macroeconomic policy.Community-wide economies of scale are apparent in certainR&D investments (e.g. space programmes).For the assignment to the Community level to be an adequateresponse, it is however necessary that two additional conditionsare met:(a) this assignment is demonstrated to yield net benefits after

administrative costs and the balance of government versusmarket failures are taken into account, and

(b) ad hoc coordination among national governments is notsufficient to correct for inefficiencies.

Other motives of assignment of tasks to the Community levelcan stem from distributional or citizenship considerations, whichare not discussed here.

The necessity of taking into account the diversity of specificsituations refers to the fact that all Member States are notat present in the same economic condition, given for exampledifferent inflation rates or levels of development. This issueis taken up where necessary in the analytical chapters below,and on a country-by-country basis in Chapters 9 and 10.

The remainder of this section analyses the economic contentof EMU and presents the related assumptions upon whichthis study is based.

Economic union

The economic consequences of 1992 have been evaluated indetail in the study The economics of 1992 (Commission ofthe EC, 1988). Table 2.1 summarizes the elements of eco-nomic union which were not, or only partially, part of thesingle market programme: mainly regional and structuralpolicy on the one hand, and macroeconomic coordinationon the other.

An important issue for the analysis of EMU is the extentof the integration effects that can be expected from thecompletion of the internal market. Although persons, goodsand services, and capital will be allowed to move freelyacross frontiers, this does not mean that integration will beperfect from the outset. Effective integration can be expectedto lag somewhat behind de jure integration. However, adistinction has to be made in this respect between the mo-bility of capital, goods, and persons.

1. Capital mobility across countries can be expected to bealmost perfect once exchange controls are removed in thenarrow sense that arbitrage equalizes interest rates corrected

for the forward exchange discount. This condition, knownas covered interest rate parity, means that it becomes theor-etically equivalent to borrow in the home currency or toborrow in a foreign currency and to hedge against exchangerate variations. As pointed out in Chapter 6, covered interestrate parity already holds for countries which have removedexchange controls.

Nevertheless, this only means that agents throughout theCommunity have equal access to lending or borrowing inthe same currency whichever currency they choose. Butcovered interest rate parity does not imply that neithernominal nor real interest rates are equalized across currenciesnor that capital circulates across the EC in the same way aswithin the United States. Without monetary union, agentswould still form expectations of exchange rate changes and/or ask for risk premiums when they lend in a foreign cur-rency, with the consequence that real interest rates differacross the Community and that cross-country financing islimited by the exchange rate risk. The importance of thisremaining barrier is a matter of theoretical and empiricaldiscussion which is taken up in Chapters 3, 5 and 6, but itsexistence is indisputable.

Since monetary union would remove this last barrier, it canbe assumed that perfect capital mobility would be achieved.Government bonds of the same maturity and risk wouldbecome perfect substitutes across the Community. However,domestic and foreign stocks would still be imperfect substi-tutes due to a different degree of information on domesticand foreign firms. To that extent, some segmentation ofcapital markets would remain even in EMU.3

3 See, for example, Artus (1990).

33

sduquenoy

sduquenoy

sduquenoy

sduquenoy


2. Integration of the markets for goods and services can beexpected to increase in the course of the progress towardsthe single market. Nevertheless, this will be a lengthy processbecause obstacles to perfect integration not only arise fromlegal or technical barriers but also from habits of consumersand the behaviour of firms, especially in the service sectorwhere so-called 'non-tradables' account for a large part ofthe production. Even for typical manufactured tradables,integration is far from complete since goods produced intwo different Member States are imperfect substitutes andconsumption structures are still biased towards homegoods.4 This imperfect integration is apparent at the macro-economic level as the marginal propensity to spend on homecountry goods is much higher than the marginal propensityto import in most Member States (exceptions being thesmallest, very open economies).

Imperfect integration has important consequences for thedesign and properties of an economic and monetary union.5

First, goods market integration cannot be expected to arbi-trage away differences in price levels as would be the casefor homogeneous commodities, which means that changesin real exchange rates (i.e. price competitiveness) are stillpossible within the Community whatever the exchange rateregime. As documented below, the experience of existingmonetary unions indicates that, although nominal exchangerates are fixed, real exchange rates can be affected by differ-entiated movements in domestic prices. Moreover, and sec-ondly, such real exchange rate changes remain a necessarycomponent of the national adjustment to policy or non-policy shocks of a domestic or external origin. Thirdly,country-specific shocks can arise in product markets (e.g. afall in the demand for goods from a given country) or theeffects of global shocks can differ across countries. Theseissues are taken up in Chapter 6.

3. Labour mobility within the Community can also be ex-pected to increase but is bound to remain limited, except forspecific skills, due to cultural and linguistic barriers. Evenwithin Member States or within the United States, mobilityis actually far from perfect as exemplified by differencesin regional unemployment rates. Therefore, at least in thedecades ahead labour market integration will remain limitedand, except for certain well-established migration flows,labour mobility between countries cannot be expected to act

Markets for automobiles are a good example.The importance of imperfect integration of goods markets for theanalysis of exchange rate regimes has been stressed by Baldwin andKrugman (1987) and Krugman (1989) in response to the 'global mone-tarist' approach of McKinnon. For an attempt to measure macro-economically this integration in the Community, see Aglietta, Coudertand Delessy (1990).

as a significant equilibrating mechanism within EMU. Thisissue is further developed in Chapter 6.

A reasonable assumption for EMU is therefore that thedegree of integration of markets will be high for capital, stilllimited but increasing for goods and services, and low forlabour. This is an important characteristic which will, atleast in the years existing ahead, differentiate the Communityfrom economic and monetary unions such as the federalStates, whose degree of integration is higher.

Another major difference between EMU and existing monet-ary unions in federal States will be the degree of centraliza-tion of public finance. In modern federal States like theUnited iStates, Canada, Australia or Germany, federalgovernment expenditures account for at least half of totalgovernment expenditures. 6 This gives to the federal budget,first, the role of reducing automatically regional incomedifferentials (through progressive income taxation and socialtransfers), and second, enough weight to be used in aggregatemacroeconomic management. Since the Community budgetonly amounts to 2 % of total EC government expenditures,neither its interregional nor its global function can be com-pared to that of federal budgets. In so far as shocks affectincomes of Member States in an asymmetric way, otheradjustment mechanisms will have to take the place of acentral budget as an automatic stabilizer. To the extent thataggregate fiscal policy measures are required, most if not allof this policy will have to be implemented through coordi-nation among Member States.

Monetary union with irrevocably fixed exchange rates

When a monetary union has the form of irrevocably fixedexchange rates, the logical consequences of this definitionhave to be made clear in order to avoid misunderstandings.'Irrevocability' of exchange rates has to be taken literally ifmonetary union is to mean something else than a merehardening of the EMS.

Three propositions can be made in order to characterize themacroeconomic effects of a monetary union:(i) Irrevocably fixed exchange rates imply nominal interest

rates equalization. With perfect capital mobility, nomi-nal interest rates on assets of the same nature, maturityand specific risk can only differ because of (a) expecteddepreciation and (b) exchange risk premiums. Both canonly arise to the extent that their irrevocability is notfully credible in the eyes of the markets. Thus, trulyirrevocable exchange rates imply the same interest rates

The minimum is 45 % for Canada. See Chapter 6, Section 6.7.

34

sduquenoy


across the union.7 This is a direct consequence of theincompatibility of fixed exchange rates, full capital mo-bility and autonomy of monetary policies alreadyunderlined by the Padoa-Schioppa Report (1987).

(ii) Irrevocably fixed exchange rates require that nationalcurrencies become almost perfect substitutes. Sinceinterest rates are equalized and conversion rates arefixed, the only differences between holding currency x ory are that (a) the acceptability of x and y for transactionsdiffer on their respective territories (in particular due tolegal tender provisions), and (b) conversion of x into yentails transaction costs as long as par-clearing is notestablished. From a macroeconomic point of view, ir-revocably fixed exchange rates are equivalent to a singlecurrency except that the rate of conversion is not onefor one.

(iii) Irrevocably fixed exchange rates imply that monetarypolicy is put under the control of a single institution.Exchange rates can only be deemed irrevocable if allofficial monetary institutions within the Communityguarantee without limits the conversion of any currencyof the system into another at the given rate. Indeedwithout this multilateral guarantee there would be nodifference between monetary union and the unilateralcommitments to exchange rate fixity that are alreadymade within the ERM, albeit whose credibility is hardto achieve and always subject to changes in the judg-ment of the markets. But this requires giving the re-sponsibility for monetary policy to a single institution,since in the absence of this centralization any creationof money by a national central bank would impactthroughput the union without any possibility for an-other Member State to shelter from its consequences.Without a central institution, the system would there-fore incorporate a strong incentive for countries to 'freeride' at the expense of their neighbours, i.e. to expandmoney supply and seigniorage gains excessively at homewithout bearing the full associated inflation costs.8Even the 'anchor country' of the ERM would lose itsmonetary autonomy since irrevocably fixed exchangerates would remove all asymmetries within the system.

The macroeconomic implications of a monetary union withirrevocably fixed exchange rates are therefore the same asthose of a single currency. In spite of the persistence ofnational monetary symbols whose denominations and valuesdiffer, these in fact become different images of the same

currency. In consequence, the discipline imposed upon mon-etary policy is exactly the same in both cases: there is noroom for decentralization and differences in the conduct ofmonetary policy.9

The above propositions are based on the assumption thatall agents believe in the irrevocability of exchange rates.However, the nature of the commitments which could leadagents to rule out any possibility of parity changes is not easyto define. Exchange rate credibility can either be achievedthrough the building of a reputation or through commit-ments whose breach would be very costly in political and/or economic terms.

Experience shows that reputation-building is a lengthy pro-cess, and that even long-lasting exchange rate stability doesnot necessarily lead to the complete disappearance of bondrate differentials. Two relevant experiences are those of theNetherlands, which since 1983 has maintained a fixed centralrate within the ERM with respect to the Deutschmark, andthat of Austria, which at least since 1981 has pegged itscurrency to the DM.10 Although both countries benefitedfrom particularly good starting conditions because of theirsize, their economic links with Germany, their inflationrecord and the autonomy of their central banks, the conver-gence of bond rates took in both cases several years to beachieved. Moreover, the option of a permanent and unilat-eral Deutschmark peg is not really available for larger Mem-ber States like France, Italy, Spain and the United Kingdomfor obvious economic and political reasons.

Credibility is a direct function of the strength of the commit-ment to exchange rate fixity and of the difficulty of secedingfrom the union: tying one's hand, in the words of Giavazziand Pagano (1988), yields benefits to the country whichmakes this commitment. In an EMU, an externality is alsoat work since credibility is a property of the system whichis shared by all its members. Commitment can never beabsolute, since whatever its technicalities monetary union isthe result of a treaty among sovereign States which canalways be renounced. However the institutional setting of amonetary union affects its credibility. As long as currenciesas well as national money and bond markets remain distinct— which would be the case with irrevocably fixed rates butnot the single currency — it is still possible to leave thesystem at a moderate economic cost. Moreover, since theexistence of separate markets for bonds denominated in

Strictly speaking, this is only true if currencies are exchanged at parsince in the presence of transaction costs, interest rate arbitrage is notperfect, and short-term interest rate differentials can therefore appear.This is shown for example by Casella (1990) and Krugman (1990).

9 This only refers to macroeconomic policy. Other functions of centralbanks, like supervision, may be more decentralized, but are not discussedhere.

10 On the Dutch experience, see Annex B. The Austrian experience isanalysed by Genberg (1989) and Hochreiter and Tornqvist (1990).

35


different currencies provides a measure of credibility throughyield differentials for assets of the same category and risk,the irrevocability of the exchange rates can always bequestioned by market forces.

Monetary union with a single currency

Since a strong and binding commitment to exchange ratefixity is a necessary condition of the success of EMU, thequestion arises how institutional devices could enhance thecredibility of this commitment. A natural candidate is theadoption of a single currency which would replace existingnational currencies. " Starting with the Werner report, the

choice of a single currency has been considered an alternativedefinition of monetary union, and it has been argued abovethat both definitions are macroeconomically equivalent, pro-vided fixed exchange rates are indeed irrevocable. Howevera single currency has a number of additional advantageswhich are discussed throughout this report and summarizedin Box 2.2. The only cost it entails, apart from to the costof introducing it, is in effect a benefit: it makes exit fromthe union very difficult. An early move to a single currencywould therefore increase the net benefit of EMU.

The nature of this single currency has to be discussed brieflyin order to avoid misunderstandings. Two major questionsare, first its relation to existing national currencies, andsecond its relation to the present ecu.

Box 2.2: Six differences between a single currency and irrevocablyfixed exchange rates1. Transaction costs: a single currency would eliminate all thecosts arising to firms and individuals from conversions fromone Community currency into another (Chapter 3).2. Transparency of prices: as goods and services would be pricedin the same currency, this would further strengthen the pro-competitive effect of the single market (Chapter 3).3. Economies of scale: a single currency would lead markets forthe same categories of financial instruments to merge, yieldingbenefits in terms of market depth and efficiency (Chapter 3).

4. Credibility: a single currency gives from the outset maximumcredibility to monetary union (Chapter 4).

5. Visibility: a single currency would be for all Communityagents a visible sign of the creation of EMU and would makethose agents more conscious of the associated wage and pricediscipline (Chapter 6).

6. External benefits: only with a single currency can the EMUlead to a recasting of international currencies and to a morebalanced international monetary regime (Chapter 7).

As to the relation to existing national currencies, the intro-duction of a single currency (the ecu) would necessitate ineach country a change in the unit of account. The new unitof account could be used for all bank notes and coin inthe Community, which would no doubt continue to beardifferent national symbols (as in the Belgian and Luxem-bourg francs, or English and Scottish pounds). Alternatively

notes and coin could continue to bear their traditional namesin each country, but would be exchanged one to one forecus. Therefore, 1 franc would be equivalent to 1 lira or 1mark, which would be different names for the ecu. A furtheralternative is that notes and coins could be denominated inecu on one side and the national currency name on the other,so long as the same quantity was indicated. There would beno significant economic differences between these alterna-tives.

The term 'common currency' is ambiguous: it sometimes refers to asingle currency, which would replace the national currencies, and some-times to a parallel currency, which would be issued in addition to

êxisting ones. It is therefore not used throughout this report. See alsobelow on the hard ecu proposal.

As to the relation between the present private ecu and futureecus, they would be different in that the new ecu would bea full currency instead of a mere financial instrument. Itssupply would be under 7the control of a single institutioninstead of being the result of separate policies governing the


value of the national currencies constituting the basket ecu.In spite of this difference, however, the choice of the sameunit of account could ensure that at the outset of the monet-ary union both would have the same value. There would beno change in ecu denominated contracts.

This study concentrates on the benefits and costs of a fullEMU with a single currency. However, specific differencesbetween the single currency and a fixed exchange rate systemare discussed throughout the text.

Macroeconomic convergence in EMU

Since EMU will be characterized on the one hand by imper-fect economic integration and on the other by a singlemonetary policy, the issue arises of what degree of price andmore generally macroeconomic convergence is required forthe stability of EMU.

In existing federal States, inflation can frequently diverge inthe short run by a few percentage points. Table 2.2 presentsaggregate evidence drawn from the Canadian experience.Due to different local conditions, consumer price inflationvaries across cities, while GDP deflators exhibit large spreadsacross provinces because of different industrial structures(some provinces being heavily specialized in the productionof oil and raw materials). In fact, real exchange rate varia-bility is sometimes even higher within Canada than acrossERM countries. l2

Table 2.2Indicators of city and regional inflation divergence within Canada

Annual rates I971QI-1979Q4 1980Q1-1987Q4

Relative city-consumer prices(8 cities/Toronto)(i) average standard deviation 1,1 1,6

(ii) maximum standard deviation 2,2 2,2

Relative provincial GDP deflators(9 provinces/Ontario)(i) average 8,6 4,1

(ii) maximum 33,6 11,3

Source: adapted from Poloz (1990).Note: based on standard deviations of relative price levels with respect to Toronto for cily-CPIs (8 cities), and lo the Province of Ontario for provincial GDP deflators (9 provinces).

In the long run, however, price inflation does not in generaldiverge by more than a fraction of one percentage point peryear across States or regions. In Europe the monetary unionsbetween Ireland and the United Kingdom, and Belgium andLuxembourg, for example, led to almost complete conver-gence in inflation rates over the long run. From 1950 until1978 (when the Irish pound ceased to be linked to the UKpound and joined the EMS), the average difference in annualconsumer price inflation between Ireland and the UK wasless than 0,4%. Similarly, between Luxembourg andBelgium the average difference over the period 1950-88 wasabout 0,3 % per annum.

However, even these small differences in rates of growth ofconsumer prices can lead to considerable differences in pricelevels. For the Ireland/UK case the cumulative differencewas about 10 % (over 28 years) and for the Belgian/Luxem-bourg case it was about 15% (over 38 years). BetweenGermany and the Netherlands there was even a cumulativedifference of about 20% for the 19 years (from 1950 to1969) during which the exchange rate was fixed. However,this lack of convergence, which became stronger towards theend of the period, might have led to the cut in the link after1969 (Graph 2.1).

12 See Poloz (1990) for details.

GRAPH 2J : Evolution of relative consumer prices in threemonetary unions, 1950-88

£;»..0

•8

0 , 7 . . . . . . . .1950 1955 1960 1965 1970 1975

B/Ljij>-

•-"""•-i .-,,«-*" '

» UK/IRL

D/NL

1980 1985

+ Cicrmanj NetherlandsI nited Kingdom lrti<injBi-ljaum 1 mi'mKiurji

Ban- pv-nud jvt-idgt' I l.ino w i t h symbol* represent periods wi thconstant nominal exchange rdii— Lines without symbols rcpicsenl periodswith variable ev-hangc raits

37


Long-run convergence in inflation is the result of arbitrage intradable goods whose price, although not equalized, cannotpermanently diverge. There is, however, no equivalent mech-anism to foster convergence in the price of non-tradablegoods. Movements of factors of production, i.e. capital andlabour, impose some limits to divergences in the price ofnon-tradables, but experience shows that this process is veryslow. Real estate prices, for example, can diverge consider-ably inside a monetary union if local growth conditionsdiffer. Convergence does not have to be perfect therefore interms of consumer price inflation and certainly not in termsof the level of consumer prices.13

Inflation differentials can even arise from equilibratingmechanisms, when the starting point is one of divergence.A well-known example of such mechanism is the real ap-preciation of the exchange rate generally experienced bycatching-up countries.14 Although such figures are sur-rounded with considerable uncertainty, comparison of pricelevels indicate that in 1989 consumer prices were about 40 %below the Community average in Portugal and some 20 %below average in Spain.15 As catching-up proceeds, this gapwill narrow through higher inflation in these countries thanin the rest of the EC. Assuming, for example, full catching-up and equalization of price levels in 20 years, this wouldmean that recorded consumer price inflation would eachyear be 2,5 percentage points higher in Portugal, and about1 % higher in Spain than in the whole Community. It mustbe stressed, of course, that such differentials would only bewarranted to the extent that they went hand-in-hand withhigher productivity gains and growth.

Labour costs also require convergence. But in this area thereare even more reasons for national differences in the shortand long run since differences in productivity growth ratesshould be approximately reflected in the growth of realwages. Actually, convergence in labour costs without conver-gence in productivity could only lead to regional pockets of

Even prices of tradables can exhibit wide divergence in level terms. Arecent survey by Runzheimer Mitchell Europe shows that both beforeand after tax retail prices of typical standardized manufactured goodsof the same brand name often vary by 100% or more across theCommunity. Arguably, price transparency within EMU could reducethese variations, but some divergence would surely remain. See TheFinancial Times, 9 July 1990.The standard rationale behind this phenomenon is that as catching-upproceeds, the relative price of non-tradables, for which productivitygains are lower, increases with respect to that of tradables. Assumingpurchasing power parity in the traded-goods sector, this leads to realexchange rate appreciation. The analysis of mechanisms of this typegoes back to Ricardo. For a modern restatement, see Balassa (1964).Source: Eurostat, Indicedesprixalaconsommation, supplement 1, 1990.

unemployment and stagnation. It is therefore of paramountimportance that regional labour costs remain in line withproductivity differentials.

The experience of the three European monetary unions al-ready examined can again be used to illustrate this. Therelative wage levels displayed in Graph 2.2 suggest that realwages could diverge by as much as one percentage point peryear over long periods of time. This is almost three limeshigher than the corresponding figure for differences in con-sumer price inflation mentioned above. For example, wagesin Ireland have increased by about 40 percentage pointsmore than in the UK during the period 1950-78, an averagedifference of more than 1% p.a. over this 28 years period.The difference in the degree of convergence of wages andconsumer prices can be seen by comparing Graphs 2.1 and2.2, showing that wages typically diverged more than con-sumer prices in three cases of intra-European monetaryunions.

GRAPH 2.2: Evolution of relative wage levels in three monet-ary unions. 1950-88

1,5 •

1.4 •

1,3

1,2 •

1 , 1 ,

1 •

0,9

0,8

UK./IRL

1950 1955 I960 1965 1970 1915 1980 1985

O Germany, Netherlands+ United Kingdom/IrelandI . Belgium Luxembourg

Base period average = 1. Lines with symbols represent periods withconstant nominal exchange rates. Lines without symbols represent periodswith variable exchange rates.

Since EMU will comprise a group of countries with largeinitial differences in labour productivity levels, nominal wageincreases should diverge as long as productivity grows atdifferent rates. Present GDP per employed person (a roughmeasure of labour productivity) is estimated to be about

38

sduquenoy


50% of the Community average in Portugal.16 If this gapwere closed in 20 years, real wages could rise annually by3,5 % more in Portugal than in the Community as a whole.German reunification provides an even more striking exam-ple : if East German labour productivity were before reunifi-cation about 50 % of the West German level, a catch up inabout 10 years could see wage increases by 7% more eachyear in East Germany.

The need for convergence in public budget imbalances ismore difficult to assess since in existing federal States na-tional or regional deficits (or surpluses) do not in generalconverge to a common value. Moreover, Member Statesshould retain a large autonomy as regards short-run budget-ary policy. However, excessive deficits that lead to explodingpublic debts are not compatible with EMU, essentially be-cause the policy of price stability of the EuroFed might bejeopardized if individual member countries run up excessivepublic debts or deficits. Coordination of fiscal policiesamong Member States would also be required. Chapter 5discusses in detail the desirable properties of the budgetaryregime of EMU.

The necessary degree of convergence in terms of externalcurrent account imbalances is also difficult to assess. Sincethe current account is identically equal to the differencebetween overall savings and investment, it is clear that abalanced current account is not always a desirable result. Asdiscussed in Chapter 6, economies with a low capital/labourratio and therefore with a relatively large potential for invest-ment would actually gain from being able to partially financeinvestment through foreign savings, thus running currentaccount deficits. This implies that current account imbal-ances should not be a cause for concern if they reflectpatterns of this type and are financed automatically throughprivate capital flows, which is in principle the case in a wellfunctioning EMU with a single currency. However, in thetransitional stages current account imbalances will still re-main highly visible and might at times signal the need forpolicy adjustments. It is difficult to decide, a priori, whethercurrent accounts disappear as an issue already when capitalmarkets are completely integrated and exchange rate fixitybecomes credible, or only when there is a single currency.Chapter 6 further discusses these matters.

A well functioning EMU does not need only convergence inthe macroeconomic performance of its member countries.An efficient management of macroeconomic policy for theunion also requires that there is convergence in the policy

preferences, or at least agreement on the policy objectives,and therefore on the weighting of targets and the choice ofinstruments of economic policy. For example, if there is alarge, common, inflationary shock, such as an increase inthe price of oil, member countries should have a reasonabledegree of convergence regarding the desirable degree ofsmoothing of the consequences of the shock on outputand inflation. Otherwise, some countries might for exampledecide to suppress its inflationary consequences using pricecontrols. This would inevitably lead to distortions within thesingle market. However, the same requirement for conver-gence implies policy autonomy in the use of domestic (e.g.budgetary) instruments while facing asymmetric shocks.

The common goal of price stability is therefore not sufficientby itself to lead to full convergence and ensure a stableEMU. It would need to be underpinned by other fundamen-tal ground rules, such as the 1992 rules for market compe-tition, and rules on fiscal discipline. Inside a consistentset of such ground rules (which together may be termed'Ordnungspolitik' in German) there would then be roomfor subsidiarity and competitive processes at the firm andgovernment level. Without a common framework for econ-omic policy, national reactions to large shocks might befundamentally different and endanger the cohesion of theunion.

Stages of EMU

The three-stage approach outlined in the Delors Reportprovides a blueprint for the realization of EMU. Therefore,the precise economic content and implications of each ofthese stages has to be made explicit. Table 2.3 presents in asimplified fashion the major features of each of those stages.Since Stage III is identical to monetary union as alreadydescribed, the comments below concentrate on Stages I andII.

The major decisions to be taken in Stage I are alreadyeffective since in the economic field the internal marketprogramme is on track, the reform of the structural Fundsis in operation and the new decision governing coordinationand surveillance was finalized in early 1990. In the monetaryfield capital market liberalization is effective in most Mem-ber States and monetary coordination is being strengthenedwithin the framework of the Committee of Central BankGovernors. The remaining aspects mainly relate to country-specific adjustments, i.e. participation in the ERM, furtherconvergence towards low inflation, and budgetary adjust-ments.

16 GDP per employed person, at standard purchasing power exchangerates. Source: Eurostat.

In contrast to Stage I, which is basically defined in economicterms, Stage II is mainly institutional in nature. It is meant

39


Table 2.3Major features of the three EMU stages

Economic Monetary

Stage Io Completion of the internal marketo Strengthened competition policyo Full implementation of the reform of the structural Fundso Enhanced coordination and surveillanceo Budgetary adjustments in high debt/deficit countries

Stage II

o Evaluation and adaptation of Stage I policieso Review of national macroeconomic adjustments

Stage IIIo Definitive budgetary coordination systemo Possible strengthening of structural and regional policies

o Capital market liberalizationo Enhanred monetary and exchange rate coordinationo Realignments possible, but infrequento All EC currencies in the narrow-band ERMo Extended use of the ecu

o Establishment of EuroFedo Possible narrowing of EMS bands

o EuroFed in charge of monetary policyo Irrevocably fixed exchange rates or ecu as single currency

to be a transitional stage only. The Delors Report placedemphasis on the gradual transfer in Stage II of responsibilityfor monetary policy. After discussion, the emphasis hassince been placed more on the technical preparation of theEuroFed institution in Stage II, on the grounds that policyresponsibility must be clear-cut. For this reason it is nowwidely considered that Stage II should be quite short.

The economic logic of Stage II is that of a transition betweenan asymmetric system where monetary policy is coordinatedthrough the exchange markets and a system in which monet-ary policy is set at the Community level and exchangemarkets lose their coordination function. The single mostimportant step in this process is the irrevocable fixing ofparities at the outset of Stage III.

The economic and monetary content of Stage III has alreadybeen described. On the economic side application of theprinciple of subsidiarity results in an evolutionary set ofpolicy responsibilities, with considerable room for debateand experimentation over the best regime at any one time.This concerns especially the rules and competences of na-tional and EC budgets, which Chapter 5 discusses in moredetail.

The timing of the transition through the Stages is largely amatter for political choice, although as developed in Chapter8 pure cost/benefit considerations tend to argue in favourof a rapid move toward full EMU. The degree of conver-gence of inflation, budget deficits and other variables needed

in the move to the definitive EMU is discussed in Chapter8. However, the speed of this convergence will depend hea-vily on the timing and credibility of the political commit-ments to move to Stages II and III.

2.1.2. Alternatives to EMU

Against what benchmark situation should the effects ofEMU be assessed? This question is less obvious than it seemsat first sight since the situation of today cannot be taken asa stable point of reference for two reasons: first, decisionshave been taken whose effects are not yet visible, e.g. regard-ing the completion of the internal market and capital marketliberalization; second, commitments have been made whichare not yet translated into effective decisions, e.g. the Britishcommitment to participate in the ERM during Stage I. Thepresent situation can therefore only be viewed as a snapshotin an ongoing process. Choosing this situation as a baselinewould not be satisfactory. It would in addition introducesomewhat artificial differences among Member States de-pending on their present situation within or with respect tothe ERM. This section reviews the alternative baselinesagainst which EMU can be evaluated.

Stage I as a baseline

A natural starting point for comparisons is Stage I, or a"EMS + 1992' hypothetical baseline, which is assumed to

40


incorporate all the effects of decisions and commitmentsdescribed in the first panel of Table 2.3. This is the basichypothesis which is retained throughout this study, althoughcomparisons are in some cases also made with alternativebaselines. Given the importance of the 1992 programme inthis regard, a relatively large part of economic union effectsare therefore already present. Most of the additional gainsfrom EMU would thus arise from progress in the monetaryfield, whose effects may be evaluated against a clear anduniform benchmark.

Since Stage I is not yet observed, assumptions have to bemade regarding its functioning. Major issues regard (1) itsbasic monetary logic, (2) the practice of realignments and(3) the micro and macroeconomic effects of capital marketliberalization.

1. As to its basic monetary logic, Stage I would not be verydifferent from the EMS of the late 1980s. Like in any otherfixed-but-adjustable exchange rate system, one can assumethat it will remain asymmetric in the sense that one particularnational central bank acts as a de facto leader. Whether this'anchor country' would be Germany as in the 1980s, orwhether a competition for the leadership might arise is anopen issue (see Chapter 8). Whichever the leader, however,the logic of such a system is that overall monetary policy isset by a particular central bank, while the policy of the(N - I) other members is devoted to the peg of the exchangerate. Therefore, as long as central parities remain fixed, theonly room for manoeuvre for autonomous monetary policiesis provided by the possibility of exchange rate movementswithin the bands.

This relates to the monetary logic of a fixed-but-adjustableexchange rate system without capital controls. Admittedly,monetary policy coordination could become more symmetricin the perspective of EMU through informal cooperationamong central banks, as a preparation for Stages II and I I I .However, for the evaluation of EMU, Stage I has to beregarded as much as possible as a steady state, setting asidefeatures that would stem from future developments of EMU.

2. The most uncertain issue concerning the evolution of theEMS in Stage I concerns the practice of realignments. Twoextreme models are (a) some kind of loosely defined softEMS, and (b) a de facto monetary union where realignmentswould become exceptional, an image of which can be pro-vided by the present situation between Germany and theNetherlands.17 Capital market liberalization within the con-

text of a fixed-but-adjustable exchange rate system points inthe direction of the second model: realignments would stillbe possible, but since anticipated realignments can give riseto potentially unlimited capital movements, one cannot as-sume that this instrument could be used by governmentssystematically.

Realignments come onto the agenda either when countriesare characterized by different inflation trends or subject tocountry-specific shocks. Whether or not significant differ-ences in inflation trends could still be compensated by per-iodic realignments in Stage I is in principle an open issue.The disciplinary effect of the EMS has proven effective sincethe second half of the 1980s when realignments becamerare and compensated only half of the inflation differentials(Table 2.4). This is widely considered to be a major factorof inflation convergence within the narrow-band ERM. Thisrecord should continue with the removal of the last capitalcontrols, for the only way to discourage speculative attacksis (i) to make the realignments rare and randomly distrib-uted, (ii) to offset only partially inflation differentials, and(iii) to avoid discrete jumps in the market exchange rate, i.e.to realign inside the band. Since the first two are costlyin terms of relative prices while the third either implies arenunciation of the EMS (if bands are so wide than thesystem becomes closer to a flexible rate system) or frequentrealignments, these techniques cannot really accommodatelarge and permanent inflation differentials. This is the reasonfor expecting further inflation convergence in Stage I.18

Actually, there is already a clear tendency within the coreERM towards a de facto narrowing of the bands of fluctu-ation vis-a-vis the Deutschmark.19

An opposite strand of arguments states that capital controls,by allowing for delayed realignments and therefore realexchange rate appreciations in the more inflationarycountries, were a key component of inflation discipline in theEMS. Another argument states that exchange rate credibilitywould lead to a decrease in nominal interest rates in in-flationary countries, thereby leading to low real interest ratesand a weakening of inflation discipline.20 Those arguments,however, can only be considered of some validity in theshort run. They do not demonstrate how large differencesin trend inflation could be accommodated at low cost withina Stage I EMU without leading to speculative attacks. There-

in May 1990, Belgium announced its intention to commit itself toexchange rate stability vis-a-vis the Deutschmark. The French Govern-ment has also repeatedly stressed its intention to rule out devaluations.

The requirements for a realignment rule in an EMS without capitalcontrols are discussed by Drifiill (1988) and Obstfeld (1988). See alsoAnnex E.See Annex E.See Giavazzi and Giovannini (1989), Giavazzi and Spaventa (1990).

41

sduquenoy


Table 2.4Realignments in the EMS, 1979-89

(vis-a-vis the Deutschmark)

B/L Narrowband

ERM

1979-83

Number of realignmentsAverage size (%)Average cumulated price differentialDegree of offsetting (%)

1984-87Number of realignmentsAverage size (%)Average cumulated price differentialDegree of offsetting (%)

54,91,65

296,6

21,55,3

28,5

74,43,1

139,8

22,56,8

36,5

47,16,7

105,4

24,58,3

54,5

45,2

12,043,5

34,76,1

77,2

56,39,8

64,4

34,79,3

50,7

221,0

203,2

225,15,0

101,2

93,56,6

52,7

275,35,9

89,7

123,87,3

52,1

Note: All data refer to bilateral Deulschmark exchange rales. Cumulated price differentials (CPI, monthly data) are measured between two realignments. Offsetting of inflation differentials ismeasured by the ratio of central rate variation to the price differentia! by the time of the realignment.Source: Commission services.

fore it can be considered that regarding inflation discipline,the implications of Stage I would not be very different fromthose of a full EMU.

However realignments might still be possible in the case ofcountry-specific shocks, provided those shocks are unexpec-ted. A country hit, for example, by a fall in the demand forits exports (a demand shock) or a surge in nominal wages(a supply shock) could use the exchange rate instrument inorder to modify its real exchange rate or to accommodatethe inflationary shock. As discussed in Chapter 6, this iswhere Stage I would retain nominal exchange rate flexibilitythat would be lost in the full EMU.

3. Capital market liberalization is already in effect for mostcountries, however its full impact is not yet observed. It willnot only affect potential short-term capital movements, butalso the integration of financial markets. One can expecttherefore, the supply of financial products to increase, pre-sumably providing larger hedging opportunities, and also,some reduction in transaction and hedging costs becauseof economies of scale, technological change, and greatercompetition.21 Although the extent of these microeconomiceffects cannot be assessed with precision, they have to be

taken into account in the cost-benefit evaluation of EMU(Chapter 3).22

Summing up, the following hypotheses are made regardingStage I:

(i) The working of the exchange rate system will remainasymmetric, with an anchor country setting the overallmonetary policy while the (N-l) others are coordinatedthrough the exchange markets.

(ii) Exchange rate discipline in Stage I will see furtherconvergence towards low inflation, practically eliminat-ing trend differences in inflation rates which do not arisefrom equilibrating mechanisms. However, realignmentswould still be an available instrument in the case ofunexpected country-specific shocks.

(iii) The removal of capital controls will yield savings intransaction and hedging costs as well as a closer inte-gration of capital markets, but fall short of what agenuine EMU would bring.

21 See the special issue of European Economy (1988b), and Gros (1990).

Other effects of the removal of capital controls concern the allocationof savings, i.e. the micro and macroeconomic effects of capital marketsintegration which are discussed in Chapters 3 and 6. Here also theremoval of capital controls is expected to yield some benefits, althoughthis is incomplete as long as exchange rate risk remains a barrier tocapital mobility.

42


The above characterization of Stage I is based on the as-sumption that it is stable, at least as a transitional stage.However, this stability has been questioned by a number ofauthors, either directly or with reference to the 'evolutionaryapproach to EMU', proposed in 1989 by the British Govern-ment (HM Treasury, 1989), which bears close resemblanceto a protracted Stage I. The issue of stability in the transitionis taken up in Chapter 8, where it is argued that Stage I canbe considered stable as long as economic policies in theMember States are consistent with the constraints of thesystem, but that 'systemic instability' could arise from arefusal of the asymmetric character of the system or from acompetition for leadership. Indeed, the stability of a Stage Ilasting for a few years and considered to be a transitiontowards full EMU, and that of the same system without theprospect of moving on to the definitive EMU are two differ-ent issues: as long as governments and agents expect a newregime to be established in the near future, their behaviouris conditioned accordingly. Historically, the coexistence offixed exchange rates and completely free capital movementsnever lasted for very long.23 It cannot be taken for grantedthat a renunciation of EMU would not lead, sooner or later,some countries to opt for more exchange rate flexibility or,alternatively, to reintroduce some capital controls. There-fore, the costs and benefits of EMU have to be discussedin the wider context of the possible alternative choices ofexchange rate regimes.

GRAPH 2.3: A graphical representation of alternativeexchange rate regimes

Financial market autarky

Exchange ratestability

Monetarypolicy autonomy

Monelaryunion

Exchangerate Ho at

Financial market integration

Alternative exchange rate regimes

The well-known inconsistency between financial integration,fixed exchange rates and monetary policy autonomy pro-vides an appropriate starting point for the discussion of thealternatives to EMU. All three objectives are legitimate, thefirst two because of the economic benefits they yield, andthe third because policy autonomy is, ceteris paribus, desir-able on subsidiarity grounds. A tri-dimensional trade-off istherefore at work, which can be represented by Graph 2.3.24

The three objectives are represented along the three axes ofthe figure. Any point within the triangle can therefore becharacterized by the degree to which each of these objectivesis fulfilled. The three sides of the triangle are each character-ized by complete fulfilment of one particular condition. Itsthree corners correspond to situations where two conditionsare simultaneously fulfilled, the third being excluded:

(i) financial market autarky (complete fixity of exchangerate and policy autonomy, but total control of capital

See Carli (1989) and Giovannini (1989).Adapted from Aglietta (1988).

movements; current account imbalances are settledthrough official bilateral capital transactions);

(ii) exchange rate float (complete financial integration andautonomy, with exchange rate determined by the mar-ket without any unsterilized intervention);

(iii) monetary union (complete financial integration and ex-change rate fixity, with zero policy autonomy).

The evolution of the exchange rate regimes within the Com-munity can be illustrated on the figure. The compromise ofthe 1960s was characterized by low financial integrationdue to widespread capital controls and high exchange ratestability. The breakdown of the Bretton Woods system ledto greater autonomy at the expense of exchange rate stabilityand while the 1970s witnessed to some extent greater finan-cial integration, some countries reverted to capital controlsto prevent exchange rate crisis. The creation of the EMS in1979 has been a reaction to this experience as well as astatement of preference for exchange rate stability, and the1980s have been characterized by both greater exchange ratefixity and greater financial integration (at the expense ofautonomy). Stage I basically represents a continuation ofthis trend. This is also the case for full EMU.

43

sduquenoy


Neither financial autarky nor a clean float represent realalternatives to EMU. It is hard to imagine why and howEuropean governments would choose to renounce the ben-efits of a Community-wide financial market, or to abandontheir aim of exchange rate stability after the successes of theEMS. Both moves would represent a dramatic reversal ofthe historical trend. However, the two cases are useful bench-marks for a discussion of the costs and benefits of EMU.

The 'hard ecu' proposal

Another proposal which has been put forward in the summerof 1990 as an alternative to the Delors Committee approachis the 'hard ecu' of the UK Chancellor of the Exchequer.

In this case the ecu would be developed as a parallel currencymanaged by a new institution, which could be called theEuropean Monetary Fund (EMF). As a first step, the Chan-cellor proposes setting up a new institution (the EMF) toact as a currency board and provide ecu bank notes ondemand for Community currencies. To ensure no new moneycreation, EMF ecu notes would have to be fully backed byits own holdings of component currencies. The ecu wouldremain a basket currency, with its interest and exchangerates being determined as now.

A further step would involve a switch from a basket ecu toa 'hard ecu', managed by the EMF, which would fullyfeature in the EMS on equal terms with other currencies,staying within ERM margins and never being devalued atrealignments in relation to any Community currency. TheEMF would issue hard ecu deposits in the Community onlyin return for national currencies. Any excess of nationalcurrencies held by the EMF would have to be repurchasedby the central banks concerned for hard currencies. Thehard ecu is intended specifically to strengthen the pressuresfor anti-inflationary convergence in the Community by put-ting it at the centre of the EMS.

The hard ecu plan represents a considerable developmentof the previous position expressed by the Treasury (Anevolutionary approach to EMU, HM Treasury, 1989): itacknowledges the need for a new Community monetaryinstitution and allows for the possibility of a single currencyat the end of the process.

An assessment of the benefits and costs of the hard ecuproposal must take into account its uncertain evolution. Toits proponent, this uncertainty is one of its main attractions,since it stems from the freedom of choice over whether touse the hard ecu, which is left to the markets. This meansthat, in relation to the existing situation, no economic agenthas to undergo extra costs by using the hard ecu, since

national currencies remain as alternatives. On the otherhand, this uncertainty also means unpredictable currencysubstitution possibilities. This could have costs in terms ofmacroeconomic stability, with monetary control made moredifficult.

As regards the hard ecu's likely economic impact, this wouldof course depend on whether it was actually taken up andused in macroeconomically significant amounts: clearly, ifthe amounts used were small, the effects would be veryslight.

If, however, the hard ecu were to become widely used, itsimpact may be considered in terms of the effects discussedin this study. To the extent that the hard ecu were used forinternational commerce and achieved the economies of scaleobserved with major currencies, there could be savings intransaction costs for some users. However, these costs wouldnot be eliminated as with a single currency. Similarly someexchange rate uncertainty would remain. While the mechan-ism contains an incentive in favour of stability-orientedmacroeconomic policies, it is not clear that this would addto the disciplines already experienced by currencies thatparticipate in the narrow-band of the exchange rate mechan-ism of the EMS (or the incentive to be faced by the otherCommunity countries when they put their currencies intothe narrow band). It is likely to be in the least convergentcountries that the use of the hard ecu would grow fastest. Ifin some countries the hard ecu acquired a significant marketshare in relation to the national currency, there would arisethe question of what the denominator would be for domesticwage and price contracts. Presumably this transitional stagewould also mean considerable uncertainty for economicagents, until the time came for the switch to a single currency.Externally, the hard ecu could take on a wider internationalrole, but not to the extent that would a single Communitycurrency.

2.2. The economic impact of EMU

The aim of the present section is to outline the costs andbenefits of EMU as well as to discuss the methods andtechniques of their evaluation: Subsection 2.2.1 reviews thetheoretical aspects, and Subsection 2.2.2 the nature of costsand benefits, together with the more technical questions ofeconomic evaluation and measurement.

The basic message of this investigation is that both becauseof the present state of the theory of monetary unificationand because of the diversity of the effects involved, an overallquantitative evaluation of the gains from EMU would beboth out of reach and inappropriate. The present study has

44

sduquenoy


accordingly the more modest aim to propose a frameworkfor the analysis of the economics of EMU and to providesome partial quantifications. However, qualitative con-clusions can be drawn from a comprehensive approach tothe costs and benefits of EMU.

2.2.1. Building blocks for the analysis of EMU

In contrast to the trade field, where standard economictheory has for long offered straightforward arguments infavour of trade liberalization and economic integration moregenerally, the analysis of EMU cannot be based on a simple,unified, ready-to-use theory of the benefits of monetaryintegration. The two issues of economic integration andmonetary integration are indeed treated very differently ineconomic theory. The case for economic integration is basedon a unified and secure microeconomic approach since what-ever the latest developments in the theory of internationaltrade, the basic rationale for the internal market programmeis still rooted in the approaches of Adam Smith and DavidRicardo.25 Monetary integration, however, immediatelyraises a more complex set of theoretical and empirical is-sues.26

This may seem paradoxical since, as pointed out by variousauthors,27 the use of a single money seems to be as essentialto the unification of the US internal market as the absenceof any direct or indirect trade barriers. Though the diversityof industrial structures among US states or regions is asgreat as among the Member States of the Community, theidea of appreciating the Texas dollar when the price of oilrises, or depreciating the Michigan dollar when Detroit isoutpriced by Japanese car exports, sounds as pointless asintroducing custom duties inside the US market. The sameapplies to other federal States, which without exception havechosen to have a single currency. History also shows thatmonetary unification has most frequently gone hand in handwith trade integration.28 One could therefore expect econ-

omic theory to provide the rationale for these empiricalobservations.

There is indeed a theory whose aim is to aid the analysis ofmonetary integration: the theory of optimum currency areasdeveloped in the 1960s by Robert Mundell, Ronald McKin-non and others. The basic insight of this approach is thatthe adoption of a single currency involves a trade-off be-tween, on the one hand, the benefits arising from monetaryintegration and, on the other hand, the costs incurred whenthe exchange rate is lost as an adjustment instrument. Thisinsight remains valid and is taken as a basis for the presentstudy. However, a precise and comprehensive identificationof those costs and benefits is not provided by the theoryof optimum currency areas, since (i) some of the benefitsassociated with monetary unification are assumed withoutfurther investigation, while others are simply omitted, and(ii) the framework for the analysis of costs is rather limitedand outdated. The most well-known conclusion of this ap-proach, namely that labour mobility across an area is theessential criterion for deciding whether monetary union isdesirable, appears to be largely overstated in early formu-lations of the theory, at least as regards the EC. Moreover,economic theory has evolved substantially in several relevantbranches since the early 1960s (Box 2.3), but the revision ofthe theory of optimal currency areas has not kept up withthis. Thus, the analysis of EMU does not need to be limitedto this rather narrow approach.

Recent developments in several branches of economics pro-vide elements of a comprehensive approach to the costs andbenefits of EMU.

1. Although almost no recent reserch has been specificallydevoted to the microeconomics of monetary union — exceptthe contribution by R. Baldwin, some recent developmentsin the microeconomics of imperfect markets (i.e. in presenceof externalities, or adjustment and information costs) leadto a better if still incomplete understanding of the channelsthrough which the adoption of a single currency wouldprovide output and welfare gains.29

However, as pointed out by Ingram (1973), most estimates of the benefitsof the Common Market, made in advance of the economic realities,were close to zero.Paul Krugman (1990) states: The economics of international money,by contrast [to those of trade integration], are not at all well understood:they hinge crucially not only on sophisticated and ambiguous issues likecredibility and coordination, but on even deeper issues like transactioncosts and bounded rationality'.For example Krugman (1990), Baldwin (1990).A classic example is that of the German monetary unification in thenineteenth century, which followed a few years after the trade andeconomic unification under the Zollverein. See Holtfrerich (1989).

2. A large body of literature has been devoted to the ques-tion of the choice of an optimal exchange rate policy for asmall economy in a stochastic environment, and thus in anenvironment of random disturbances. 30 The issue is whichmonetary and exchange rate policy performs best for reduc-

29 See Chapter 3 and Baldwin {1990).30 See, for example, Marston (1985), Aizenman and Frenkel (1985). For a

recent survey, see Argy (1990).

45


Box 2.3: EMU and the theory of optimum currency areas

The theory of optimum currency areas was developed in the1960s in order to determine what should be the appropriatedomain within which exchange rates are fixed.31

The question relates to the debate at that time between theproponents of exchange rate fixity and those of floating ex-change rates. The basic argument for floating was twofold: firstthat exchange rate changes were a less costly way of correctingcurrent account imbalances, and more generally of adjustingrelative prices between two countries when prices are sticky, andsecond that it would allow countries to pursue independentmacroeconomic policies (e.g. to choose different inflationrates).32

The optimum currency areas approach uses an implicit stabiliza-tion framework and introduces the issues of asymmetries andalternative adjustment channels when countries face exogenousshocks. Mundell (1961) points out that exchange rate flexibilityis of no use between the US and Canada if, due to a shock tothe industrial structure, both countries are hurt by a shift in therelative economic conditions of their eastern and western parts.What is needed instead is a change in the relative prices ofeastern and western products. Therefore, the right criterion fordesigning a currency area should be the degree of factor (capital,i.e. physical investment, and labour) mobility within the region,since a high degree of factor mobility would provide the adjust-ment channel which is lost in a fixed exchange rate regime. Itwas precisely because of a low labour mobility within Europethat Meade (1957) argued that exchange rates should remainflexible within that zone. It became conventional wisdom to saythat Europe was no! an optimum currency area.

Although the overall approach remains a useful starting point,this conclusion only holds within a specific and rather limitedframework whose adequacy for today's analysis is questionable.First, the microeconomic benefits of a monetary union aresimply assumed without further exploration: in MundelFswords, 'money is a convenience and this restricts the optimumnumber of currencies'. Second, although labour mobility acrossEurope remains low — but it is also limited within the UnitedStates — the mobility of both physical and especially financialcapital is now much higher than in the early 1960s. As pointedout by Ingram (1959, 1973), the degree of international financialintegration has to be introduced as a specific criterion and thealternative adjustment channel of cross-country financing has

therefore to be taken into account. Third, Mundell's implicitframework is a world of rigid wages and prices. Although wage/price stickiness is still a fact of life, their rigidity could be lessthan in Mundell's framework, and the possibility of marketadjustments has to be considered. Fourth, there is the implicitassumption that no inefficiencies arise from flexible exchangerates, such as because of instability in the exchange markets andnon-cooperative or suboptimal policies. Obviously this can nolonger be considered an acceptable hypothesis in the light ofrecent experience.

Fifth, the whole approach ignores policy credibility issues whichhave been stressed by recent macroeconomic theory. Six, thetheory of optimum currency treats the whole area as a smallcountry within a given world and thus it ignores the externaleffects of monetary integration.

The approaches of McKinnon (1963) and Kenen (1969) to thissame topic are different. McKinnon points out that if exchangerate changes are used to offset the effects of domestic demandshocks on the current account, price instability is bound toincrease in line with the degree of openness (or the share oftradables in production) in a floating rates regime. This isparticularly relevant within the EC and is a major reason behindthe long-standing choice of a peg system by smaller open econ-omies like the Netherlands and Belgium. Moreover, Krugman(1990), building on this argument, argues that the costs ofmonetary union decrease and the benefits increase with theintensity of trade within the currency area. Kenen focuses onthe degree of product diversification and argues that countriescharacterized by a low degree of diversification should retainexchange rate flexibility in order to offset product-specificshocks, while a high degree of product diversification, by averag-ing product-specific shocks, could compensate for low labourmobility. In practice EC countries typically have highly diversi-fied industrial structures. The general issue of the impact ofdifferent industrial structures is discussed in Chapter 6.

Summing up, the optimum currency area approach providesuseful insights but cannot be considered a comprehensive frame-work in which the costs and benefits of EMU can be analysed.Empirical applications of this approach are scarce and hardlyconclusive.

31 Major contributions include Mundell (1961), Me Kinnon (1963). and Kcnen (1969).For more recent surveys, see Ishiyama (1975), and Wihlborg and Willetl (1990).

32 See, for example. Friedman (1953).

46

sduquenoy


ing aggregate macroeconomic variability, i.e. insulates betterthe economy from the effects of shocks. This line of researchhas established (i) that the choice between exchange rateregimes not only depends on the pattern of shocks (domesticor foreign, real or monetary) affecting the economy, but alsoon the degree of wage-price indexation; and that with full andinstantaneous indexation, there is no difference whatsoeverbetween fixed and floating rates,33 and (ii) that neither fixednor floating exchange rates are generally optimal, but ratherthat some country-dependent mix of the two regimes ispreferable.34

of exchange rate instability and on possible remedies.38 Thisresearch is of special relevance in the European case sincethe creation of EMU would bring more symmetry in theinternational monetary system and could therefore be abuilding block for wider changes in this system.

5. Last but not least, there is already a substantial andgrowing body of research on the economics of EMU whichdraws on all the above theories, and of which the paperspresented in "The economics of EMU' (European Economy(1990)) offer an up-to-date sample.

3. Applications of game theory in the field of macro-economic policy have led to emphasize issues in the choiceof an exchange rate regime which had been for long stressedby policymakers but were largely ignored by the literatureof the 1960s. This approach leads to questioning the standardpolicy autonomy argument in favour of flexible exchangerates from two different angles. First, this autonomy mightnot be optimal from a political economy point of view, e.g.if domestic monetary authorities suffer from low credi-bility. 35 In such instances, policy discipline through thepegging of the exchange rate to a hard currency can be ashort cut to a long and painful reputation-building. This lineof research appears to be especially relevant in the Europeancase, since for several countries, membership of the ERMhas been perceived as a way to borrow the credibility of theBundesbank.36 Second, the risks of suboptimal outcomeswhen monetary policies separately aim at contradictory re-sults have been stressed by the literature on policy coordi-nation. 37

4. Another domain of research concerns the relative per-formance of alternative monetary arrangements at the worldlevel. Two decades of disappointing experience with floatingexchange rates have stimulated new research on the causes

These developments in economic research do not yet providean integrated approach to monetary union. One might doubtthey ever will, since very different fields of economic theoryare involved in the analysis of EMU: micro-economic ef-ficiency has little to do with stability in a stochastic environ-ment or with the rules-versus-discretion debate in macro-economic policy. Recent contributions which, in the spiritof the optimum currency area approach, attempt to isolatea trade-off in the choice for or against EMU (seigniorageversus transaction costs, or insulation from shocks versustime-inconsistency 39} only highlight partial aspects of thechoice problem raised by EMU. In any case, much moreresearch has to be done before these different branches canbe linked together within a unified model of the welfareeffects of international money.

However, empirical analysis can obtain results from thesevarious fields as building blocks for a comprehensive ap-proach to EMU. As detailed below, it turns out that thisapproach leads to basing the case for EMU on a muchbroader framework of analysis than that provided by thetheory of optimal currency areas. While this report has notheoretical pretensions, the nature of the effects discussed inthe analytical chapters can be taken as an indication of thescope that a full theory of monetary unions would have toencompass.

See Annex D for an analysis of the links between the speed of indexationand the cost of losing the exchange rate as an instrument.Another, related approach stresses the relative information content ofexchange rate regimes in rational expectation models. The emphasis ison the degree to which prices provide the right information to agentsregarding the nature of the shocks affecting the economy under alterna-tive exchange rate regimes. No strong result in favour of one regime oranother emerges from this literature. See Wihlborg and Willett (1990)for a survey.This might arise from public choice considerations, but also from thetime-inconsistency argument. See Kydland and Prescott (1977), Barroand Gordon (1983).See, for example, Giavazzi and Pagano {1988), and several contributionsto Giavazzi, Micossi and Miller (1989).For a recent survey, see Currie, Holtham and Hughes Hallett (1989).

Before turning in more detail to the nature of costs andbenefits, three important methodological points have to bemade:(i) A number of questions arise whose answers rely on

different paradigms. The use of a single currency yieldsefficiency gains which are much of the same natureas those arising from the elimination of physical andtechnical barriers to trade. The natural framework for

38 See. for example, Witliamson (1985), McKinnon (1988), Krugman(1990).

39 On the first aspect, see Canzioneri and Rogers (1989). On the second,see, for example, Van der Ploeg (1990).

47


the analysis of these effects is the classical microeco-nomic paradigm of competitive markets and flexibleprices. The costs of losing the exchange rate instrument,however, only arises in a world of imperfect integrationand sticky prices, i.e. within a Keynesian paradigm,since the whole macroeconomic issue of choice of anexchange rate regime is irrelevant if markets are per-fectly integrated and prices fully flexible. Finally, theanalysis of the credibility of EMU and of its effects forinflation discipline has to be based on a third paradigm,where considerations of political economy play an im-portant role, together with theories of expectations andstrategic behaviour. In all, therefore, one has to accepta somewhat eclectic approach.

(ii) Due to the present state of the theory, an attempt tomake an overall quantitative evaluation of the costs andbenefits of EMU would hardly be meaningful. Some ofthe effects that can be expected from EMU cannot beyet quantified, and others would only be evaluated witha considerable margin of uncertainty. For example,few results are available regarding the welfare costs ofinflation, despite the fact that most economists andpolicymakers agree on the importance of these effects.Moreover, the lack of a unified theory would raiseproblems of inconsistency in the aggregation of effectsof a different nature. This does not preclude partialquantifications, and indeed several such assessments ofspecific effects are given throughout this report.

(iii) Reliance on the most recent literature has the advantageof bringing in the most recent insights of current re-search, but also has its cost: results can be model-dependent, and are therefore less secure and widelyaccepted than with established theories. Empirical as-sessments also remain fragile and frequently disap-pointing.

2.2.2. Costs and benefits

According to the method sketched out above, the presentcosts-benefit analysis of EMU is based on the combinationof different approaches. Major categories of effects havebeen presented in Chapter 1. The aim of this section is topresent in more detail the categorization of effects and themethodology of the present study.

Direct and indirect effects of EMU

As presented in Graph 1.2 of Chapter 1, the final economicimpacts of EMU will arise both directly from changes in theeconomic system of the Community, and indirectly throughpolicy and behavioural changes induced by the systemic

changes. Although this is not a water-tight distinction sinceany change immediately impacts on the system as a whole,it is important to sort out indirect effects from direct onessince the latter are unconditional while the former are con-ditional on the policy stance of EuroFed and national fiscalauthorities, and on the modifications in the behaviour ofprivate agents that can be expected from both systemic andpolicy changes. For example, savings in transaction costsare an unconditional consequence of the adoption of a singlecurrency which directly affects economic efficiency, whileprice stability is conditional on the policy of EuroFed andon wage and price discipline.

It could be argued that those effects which are conditionalon the quality of economic policies should not be assessedas arising from EMU per se: whatever the institutionaldesign of the system, it cannot be taken for granted that itwill lead to more appropriate policies. However, as discussedabove, policy discipline effects cannot be ignored as they arean integral part of the motives behind the choice of EMU.Indeed, it is well known that any exchange rate regime hasdisciplinary effects on economic policies, and this has forlong been considered one of the prominent aspects of thechoice of an exchange rate regime.

What can be expected is that an adequate design of theinstitutional system, notably in the statutes of EuroFed, orin the framework for budgetary policies, will yield benefitseither because of a greater effectiveness of Community andnational policies, or through a removal of the inefficienciesassociated with low policy credibility, which have a strongimpact on expectations and risk premiums incorporated ininterest rates and other financial variables. As discussed inChapter 4, an independent EuroFed with a clear mandateto achieve price stability would inherit the reputation of themost stability-oriented central banks of the Community. Asto the budgetary regime, governments should retain fiscalflexibility for stabilization and adjustment purposes, but aspublic debt monetization would be ruled out, the avoidanceof unsustainable fiscal positions would become an absoluterequirement. There would also be a strengthening of coordi-nation and surveillance procedures.

Throughout this study, it has been assumed that EuroFedwould fulfil its mandate to aim at price stability. Regardingbudgetary policies, it is generally assumed that MemberStates would not depart from disciplined policies. As theimportance of these and of other explicit policy assumptionsis highlighted when needed in the text, the reader should beable to distinguish between direct and indirect effects ofEMU.

The same kind of issues arise for behavioural changes. It iswell known that the behaviour of private agents cannot be

48

sduquenoy


supposed to remain invariant with respect to systemic orpolicy regime changes. Wage bargaining and price-settingbehaviours, for example, are not independent from the policyof the central bank. However, there is no sure way to identifyprecisely the consequences of EMU on agents' behaviour,and excessive reliance on behavioural changes could also bemisleading.

In most of this study, behaviour has only been supposed tobe affected through the impact of the exchange rate andmonetary policy regime on expectations. This means forexample that wage-setting behaviour is supposed to bechanged in so far as price expectations would be modifiedby the exchange rate and policy regime. ̂ This assumptionis warranted since EMU will not be a mere unilateral com-mitment to exchange rate stability, but rather an institutionalchange whose impact should be clear for all agents (especiallywith the adoption of a single currency). It has also beenassumed that the policy of EuroFed would be as credible tothe public as the policy of the more stability-oriented centralbanks of the Community. The possibility of deeper changesin behaviour is discussed (for example in Chapter 6 asregards wage and price flexibility), but the assessment ofEMU does not rely upon this type of assumption.

The distinction between direct and indirect effects, whichrefers to the channels of action, should not be mistaken forthat between static and dynamic effects, which refers to thetime dimension. Static effects are supposed to arise at theoutset of EMU, or possibly with some lag. These are forinstance savings on transaction costs or the benefits of stableprices. Dynamic effects are those which only build up pro-gressively over time, because capital accumulation or otherstock adjustments are involved. They translate into an in-crease in the rate of economic growth for a substantialperiod. This is the case for the effects of a lower riskiness ofinvestment, but it also arises when hysteresis is at work, as forexample in the emergence of the ecu as a major internationalcurrency.41

The exact definition of a behavioural change depends on the modelwhich is used to represent behaviours. For example, changes in thereaction of wages to an inflationary shock as a consequence of monetaryunion would be assessed as an exogenous change in behaviour in amode! with adaptive expectations, though it would only be an endogen-ous reaction to the regime change in a model with forward-lookingexpectations. See Annex E for further discussion.There is a slight difference in terminology on this point between thisstudy and the Report on the economics of 1992 in European Economy(I988a).

Costs and benefits in outline

Regarding the permanent effects of EMU, three major cat-egories of final impacts have been distinguished in Chapter1, which relate to microeconomic efficiency, macroeconomicstability, and regional equity. In addition, EMU will give riseto important transitional effects, mainly of a macroeconomicnature. Costs and benefits of EMU, grouped in these fourmajor categories, are summarized in Table 2.5. For the sakeof clarity, external effects resulting from the consequencesof EMU for the Community's relations with the rest of theworld are presented separately from purely internal effects,although they have the same type of final impact.

The major costs and benefits of EMU are discussed in thechapters of Part B below together with some associatedpolicy issues. As this analysis does not follow the abovecategorization, but rather isolates groups of policy issues(the 5 domains and 16 mechanisms detailed in Chapter 1),Table 2.5 gives a summary presentation of the main effectswith reference to the chapters in which they are discussed.

(a) Microeconomic efficiency effects, which are presented inChapter 3, are basically of the same nature as those arisingfrom the completion of the internal market.42 They proceedfrom two microeconomic phenomena in the monetary field(the disappearance of exchange rate uncertainty and trans-action costs) and one in the economic field (measures tostrengthen the internal market, e.g. competition policy). Al-though, as detailed in Chapter 3 and in Baldwin (1990), themicroeconomics of monetary union raise difficult technicalissues, the basic logic is straightforward: EMU yields(i) direct benefits arising from savings on transaction andhedging costs; (ii) indirect benefits due to the impact ontrade, cross-border investment and capital flows of the disap-pearance of exchange rate barriers; and (iii) further dynamicbenefits due to the impact on domestic investment decisionsof an overall reduction in uncertainty (conditional on thestability gains brought by EuroFed policy).

Efficiency gains are microeconomic in nature but wouldobviously translate into macroeconomic effects. They canultimately be expected to raise permanently the level ofoutput and real income in the Community because of ahigher overall factor productivity and a higher capital stock,in the same way as the completion of the internal marketis expected to yield output and income gains. During thetransition to this higher steady-state, growth would be in-creased. According to 'new growth models', there is eventhe possibility of a sustained rise in the GDP growth rate ofthe Community. This remains, however, a conjecture.43

See European Economy (1988a).See Chapter 3 and Baldwin and Lyons (1990).

49

sduquenoy


Table 2.5Major categories of effects of EMU

Nature of the final impact Main channels of action

Indirect

Microeconomic efficiency Steady state level of output andincome in the Community

Savings in transaction and hedg-ing costs [3]Disappearance of exchange rateuncertainty and instability [3, 6]

Macroeconomic stability

Regional equity

External effects

Transitional effects

Steady state rate of inflationVariability of output and in-flation, with associated conse-quences for welfare

Distributional effects amongMember States and regions

Both micro and macro effects forthe Community

Changes in the macroeconomictransition path towards EMU

Loss of an adjustment instrumentfor asymmetric shocks [6]Disappearance of non-cooperat-ive exchange rate policies [6]Greater availability of externalfinancing [6]

Spatial distribution of directmicro and macro gains [9, 10]

Additional savings on trans-action costs, revenues from inter-national seigniorage and re-duction in the need for exchangereserves [7]

Disinflation costs [8]Costs of fiscal adjustment [5, 8]

Strengthening of the internalmarket [3]Integration of goods and capitalmarkets [3]Effects of lower uncertainty onthe riskiness of investment [3]Induced dynamic effects [3]Seigniorage revenue losses forsome Member States [5]Effects on taxation and provisionof public goods [5]

Price stability as consequence ofEuroFed policy [4]Sound, coordinated fiscal poli-cies [5]

Differentiated impacts of EMUdepending on initial macro-economic conditions and eco-nomic policy strategies [9, 10]

Better coordination inter-nationally [7]Changes in the internationalmonetary system [7]

Drop in real interest rates [5]Effects on the dollar/ecu ex-change rate[7]

Note: numbers in brackets refer to the chapters of this study.

As for any efficiency gains, there will also be temporarynegative effects and adjustments in the short run, for exam-ple as the foreign exchange activities of banks decrease.

To a large extent, gains of this category are not conditionalon macroeconomic policies. Some would nevertheless disap-pear if inadequate policies were to give rise to macro-economic instability. In addition, for EMU to yield efficiencygains in the medium run, real exchange rate levels withinthe Community have to be consistent with economic funda-mentals. Although an initial misalignment could be eventu-ally corrected through relative price movements withoutchanging the nominal exchange rate, it could in the meantime

have severe welfare-reducing effects. It is therefore importantthat initial levels are in line with fundamentals. This assump-tion is maintained throughout this study.

In addition to its effects on the private sector, EMU willalso affect taxation and the provision of public goods, andmore generally public sector efficiency. This topic is dis-cussed in Chapter 5.

(b) Macroeconomic stability effects, which are discussed inChapters 4 and 6, are of a different nature. These effectsconcern the variability of macroeconomic variables, bothnominal and real. Variability arises as a consequence of

50

sduquenoy


policy and non-policy shocks of a domestic (e.g. non-perma-nent changes in the behaviour of private agents) or external(e.g. a change in a partner country's economic policy) origin.This is an important topic for two reasons. First, variabilityin macroeconomic variables like the price level, inflation, **GDP and employment is welfare-reducing per se for risk-adverse agents, even if their average level remains unaffected(see Box 2.4). Second, aggregate variability also permanentlyinfluences the level of output, at least if it goes with greateruncertainty, because it reduces the information content ofprices and leads to an increase in risk premiums.45

Regarding stability, EMU yields both permanent costs andbenefits. The net impact in terms of macroeconomic stabilityin EMU is therefore an empirical issue. The main costsarise from the loss of a policy instrument for stabilization(Chapter 6). Once exchange rates are irrevocably fixed, na-tional monetary policies can no longer be used to cushion theeffects of asymmetric disturbances, either through marketadjustments (as in a floating rate regime) or through dis-cretionary exchange rate adjustments (as in the EMS). Inthis respect the loss of the monetary and exchange rate policyat the national level has to be considered a cost as long aswages and prices do not fully and instantaneously respondto a disequilibrium in the labour or goods markets. This isthe basic optimum currency area argument, which obviouslyonly applies to relativities, since the common monetarypolicy can be used to cushion the impact of common shocks.

In addition to the benefits already discussed resulting fromthe extension of price stability to all Member States(Chapter 4), significant gains would arise from the removalof inefficiencies associated with exchange rate flexibility(Chapter 6). Exchange rate instability, i.e. that part of itsvariability which does not result from fundamental changesin policies or in the economy more generally is harmful,because it impacts on macroeconomic variables like inflationand output, therefore increasing their variability.46 The re-

Variability in prices, i.e. inflation, has to be distinguished from variabilityin inflation, i.e. in the change in the rate of increase of prices. Theoreticallyprices can be rising while the rate of inflation is stable. However, inpractice inflation and its variability are correlated (see Chapter 4).See Chapter 3 and Baldwin (1990).Exchange rate instability has to be distinguished from variability and

uncertainty. Variability refers to the observed changes in exchange rates.Uncertainty refers to the unanticipated component of these changes.Instability refers to that part of variability which cannot be explained bychanges in policies or other economic fundamentals.

moval of this market failure would be stabilizing.47 Anothereffect concerns failures of governments and authorities toact in a cooperative fashion. Non-cooperative monetary andexchange rate policies are welfare-reducing, for exampleif each government separately aims to export the adverseconsequences of a common inflationary shock through ex-change rate appreciation. This risk would disappear inEMU. In addition, the alternative adjustment mechanism ofexternal financing exists, and its availability is increased byexchange rate fixity.

Admittedly, the EMS already removes at least part of ex-change rate instability, and reduces the risk of non-cooperat-ive policies since realignments are not unilaterally decidedbut negotiated. The same would hold in Stage I. This is aclear benefit, but this also goes with a cost because ofasymmetry in monetary policy. As long as one nationalcentral bank sets the pace for monetary policy in the Com-munity as a whole, policy reactions to asymmetric shocksare unlikely to be optimal since the legitimate objective ofthe national central bank is to seek the stability of domesticprices in its national economy. For example, if the Frencheconomy were hit by an inflationary shock there would beno case for the Bundesbank to tighten its policy.48 In EMU,the policy of EuroFed would be more symmetric and wouldonly react to country-specific developments which affectprice stability in the Community as a whole.49

(c) Regional equity effects concern the distribution of costsand benefits among Member States and regions. Differencesin industrial structure, size and level of development, orforeign trade patterns can significantly affect the balance ofcosts and benefits across Member States. For example, theanalysis of transaction costs shows that these are higher forsmaller countries, which rarely use their own currency intrade invoicing and whose financial markets lack depth.Initial macroeconomic conditions are also an importantfactor to take into account, as differences in inflation ratesor fiscal conditions can alter the extent of policy changesbrought about by EMU, at least transitionally. These issuesare taken up Chapters 9 and 10. The specific role of theCommunity budget is discussed in Chapter 6.

(d) External effects which are the topic of Chapter 7 orig-inate in the large size of the Community in the world econ-omy. Due to its size, the effects of EMU would not be

Obviously, this is only true in so far as the variability of other variables(e.g. the dollar exchange rate) does not increase as a consequence ofintra-Community exchange-rate fixity. This issue is discussed in Chapter3.Except for offsetting the effects of spill-overs of French inflation onGermany.See Chapter 6, especially the results from stochastic simulation.

51

sduquenoy


limited to the Community but would also have an impacton the international monetary and economic system. First,the ecu would become one of the major world currenciesalongside the dollar and the yen, and could acquire a vehicu-lar role in trade and finance for EC residents as well as non-residents. This could bring additional savings on transactioncosts, yield some revenues from international seigniorage,and impact on macroeconomic stability. Second, the Com-munity would be able to engage more effectively in inter-national policy coordination and to speak with a single voicein international monetary affairs, More generally, policycoordination at the G7 level would be transformed as thenumber of large players would be basically reduced to three.This would also impact on macroeconomic stability. Finally,both moves could ultimately induce wide-ranging changesin the international monetary system.

(e) Transitional effects are discussed in the analytical chap-ters and summarized in Chapter 8. These effects differ frompermanent ones which arise in comparing a steady-stateEMU to an alternative steady-state, generally assumed to be'1992 + EMS' (but alternative baselines are also discussed).Only these effects represent the durable benefits and costsof EMU. However, since the starting point may not be asteady state, and indeed is surely not such for some countries,the costs and benefits of a relatively rapid move towardsEMU have to be considered compared to those which wouldarise from a lengthy convergence within the framework ofStage I. In particular, as discussed in Chapter 5, a numberof Member States still suffer from low exchange rate andmonetary policy credibility, and therefore from relativelyhigh real ex-post interest rates. Although most if not all ofthese phenomena would disappear in the due course in asteady-state Stage I, experience shows this is a lengthy pro-cess. A more prompt move towards EMU could significantlyaffect the balance of macroeconomic costs and benefits inthe transition towards low inflation and budgetary positions(Chapter 8). Transitional issues also arise in the externalfield, as an increase in the demand for ecu assets could leadto an appreciation of the Community currency (Chapter 7).

Methodology of evaluations

Strictly speaking, the effects of EMU should only be deemedpositive or negative with respect to an explicit welfare func-tion. Aggregation of these effects would also imply the choiceof a common yardstick, which as discussed in Box 2.4 couldonly rest on some rather arbitrary assumptions. In practice,this has not been necessary, since no overall quantitativeevaluation has been attempted. However, as several partialevaluations are given in this study, issues of comparabilityand relative importance arise. Table 2.6 attempts to summar-ize the major elements which contribute to this evaluation.

The categories of effects distinguished above relate to differ-ent fields of economic theory, and to some extent todifferent paradigms, as summarized in the first two col-umns of Table 2.6. This has consequences for the methodsand instruments of evaluation that are available and onthe possibility of quantifying the different effects.

(a) For microeconomic efficiency effects, the natural focusis on their permanent impact on the levels of output andwelfare. The associated framework of analysis is thereforethe classical full employment/flexible prices model, sincethese assumptions are standard for the long run.51 How-ever in the textbook versions of the standard microecon-omic model, where transaction costs are zero and adjust-ments are costless, and where all agents have perfectinformation and no aversion towards risk, exchange rateregimes are neutral. Gains from a move towards monetaryunion arise when these oversimplifying assumptions arerelaxed, which frequently implies using recent insightsfrom research in progress.

Evaluation techniques are diverse. Gains from the elimin-ation of transaction costs can be directly assessed usingsurveys and data on bank balance sheets, provided assump-tions are made regarding the respective effects of capitalmarket liberalization and the adoption of a single currency.The assessment of indirect effects is more partial and tenta-tive since, firstly, only some aspects are quantified, andsecondly evaluations with present models are bound to pro-vide orders of magnitude, rather than precise figures. Dy-namic gains are evaluated using either calibrated standardgrowth models, or the more recent endogenous growth mo-dels.

The time horizon corresponding to these effects is short forthe most direct ones, like the elimination of transaction andhedging costs. Indirect and dynamic effects can normally beexpected to accrue over many years if not decades, but theexperience with the internal market tends to show thatfavourable expectations can lead firms to increase investmenteven before the policy measures are effectively implemented.

(b) The basic framework of analysis for macroeconomicstability effects is that of neo-Keynesian models, which arecharacterized by sticky wages and prices in the short run.Only in this kind of model can the loss of the exchange rateinstrument be assessed. However, this framework has to beextended to encompass also policy discipline effects.

'Full employment' does not mean that there is indeed full employmentin the baseline, but only that the long run refers to the horizon whereshort-term unemployment effects of efficiency gains are eliminated.

52


Box 2.4: Measurement issues in the cost/benefits analysis

EMU will have three concurrent types of impact:(i) on the level of output in the steady-state and the path

towards this new equilibrium;(ii) on its variability in a stochastic environment (i.e. character-

ized by the occurrence of random economic shocks);(iii) inflation and other macroeconomic variables are also affec-

ted in both the foregoing ways.

(i) Effects which impact on the level of output (mostly efficiencyeffects) can be measured either in terms of output or in termsof welfare. The output measure can be related directly to usualmacroeconomic statistics (GDP), but it is 'gross', since it doesnot take into account the cost of forgoing present consumptionin order to accumulate capital. The welfare measure, which isderived from microeconomic utility considerations, is theoreti-cally more appropriate. However it is less easy to grasp andrests on assumptions concerning among other things the rate oftime preference. Both measures are close to each other in thecase of direct, static efficiency gains, but can differ widely assoon as intertemporal issues are involved.

(ii) Variability in macroeconomic variables raises in additionthe issue of risk aversion. Only for risk-neutral individualsand firms would temporary deviations from equilibrium not bewelfare-reducing, provided they are averaged over time. The

welfare cost of variability per se is however not of the samenature as the welfare cost of, for example, a permanent incomeloss. Aggregation of level and variability effects could only reston a measure of risk aversion.

(iii) Taking into account other macroeconomic variables impliesrelying on a weighting of these variables. Due to the non-availability of a social welfare function derived from the utilityfunctions of individuals, economists usually use macroeconomicwelfare functions that are supposed to reflect policy preferences(and the welfare of the general public, provided policymakersexpress the preferences of this public). Here also, evaluationshave to rest on some rather arbitrary quantitative assumptionsregarding the weighting of deviations of output and inflationfrom their optimal level. ̂

An overall measurement of the effects of EMU would have beenvulnerable to changes in technical assumptions regarding theparameters discussed above. For the partial quantificationswhich are given in this study, reliance is placed mostly on usualstatistics, i.e. output rather than welfare, and generally to discussthe results of macroeconomic model simulations in terms ofchanges in familiar variables rather than on the basis of a macrowelfare function.

50 A specific problem is that the weight of inflation in usual macro welfare functions(which attempts to reflect the preferences of policymakers} cannot be based on micro-economic grounds. See Blanchard and Fisher (1989), Chapter I I .

Table 2.6Assumptions and methods of quantitative evaluations or EMU

Microeconomicefficiency

Macroeconomicstability

Regional equity

Underlying theories

Microeconomics of imper-fect marketsGrowth theory ('old1 and'new')

Macroeconomic theoryGame-theoretic approachto economic policyPolitical economy

Both micro and macroapproaches

Main assumptions

Flexible prices and full em-ployment (long run)Information and adjust-ment costs, risk aversion

Sticky prices and under-employment (short run)Imperfect credibility

As for micro and macro

Methods and instruments

Direct evaluations andsurveys (transaction costs)Computations with cali-brated growth models

Econometric techniquesDeterministic simulationswith macro modelsStochastic simulations

No specific instrument

Quantitative assessment

Substantial for direct ef-fectsPartial and tentative forindirect and dynamic ef-fectsPartial evaluation for ana-lytical purposesAttempt at a syntheticquantification throughstochastic simulationsNo quantitative assess-ment for individual Mem-ber States

External

Transitional effects

grationInternational monetaryeconomicsTheory of internationalpolicy coordinationAs for macro stability

Relaxation of the 'smallcountry' assumption

As for macro stability



Only for specific effects

Only for specific effects

53

sduquenoy


The degree of wage-price rigidity, and the associated costsof losing the exchange rate instrument, can be evaluatedusing econometric techniques or the estimates incorporatedin empirical macroeconomic models (Box 2.5). Howeverstandard methods can provide only some analytical elementsrather than comprehensive assessment, since (i) deterministicsimulations only evaluate the effects of single, specificshocks, and (ii) the impact of a change in the monetaryregime on the behaviour of agents is not taken into account.Both limitations can be overcome through stochastic simula-tions with a model incorporating forward-looking features.The results of these simulations represent the average behav-iour of an economy facing randomly distributed shocks,under different exchange rate regimes. They provide there-fore a synthetic assessment. But this also is not fully compre-hensive as some other gains resulting from the availabilityof alternative adjustment mechanisms or from the elimin-ation of non-cooperative exchange rate behaviour are nottaken into account.

Generally speaking, this type of macroeconomic stabilityeffects will arise as soon as exchange rates are irrevocablyfixed. The time horizon corresponding to this category ofeffects can be assessed to be about five years for specificshocks since as discussed in Chapter 6, the real effects ofnominal exchange rates changes do not last longer.

Policy discipline effects also concern macroeconomic stab-ility, but instead of focusing on the availability and effective-ness of policy instruments, the approach here focuses on theirpossible misuses. As already mentioned, the correspondinganalytical framework stresses both the nature of economicpolicies that monetary authorities and governments can beexpected to follow within a given framework of incentivesand their interaction with private agents who form expec-tations regarding future policies.

Quantification of such effects is difficult since (i) the effec-tiveness of the EMU policy framework can only be a matterof judgment, (ii) there is only a weak causal link between agiven framework of incentives and the policies which areeffectively followed, and (iii) credibility varies over time andcannot generally be directly observed. Experience with theEMS provides a good reference since the system has evolvedfrom merely an exchange rate stabilizing device into a veryeffective disciplinary device without major institutionalchanges. Only partial computations are therefore given(mostly as regards transitional aspects), which aim at indicat-ing orders of magnitude; they do not represent a comprehen-sive assessment of the policy discipline effects of EMU.Model simulations are also provided for illustrative pur-poses.

It is difficult to establish which part of policy disciplineeffects can be expected to be reaped at the outset of StageIII, and which part will only arise after the reputation ofEuroFed is established and fiscal coordination and surveil-lance procedures have proven effective.(c) The evaluation of regional equity effects relies on thesame approaches as micro and macroeconomic evaluations,supplemented by insights from the theory of economic inte-gration. Conclusions regarding the regional impact of EMUare drawn from this analysis, but no specific attempt at acountry-by-country quantitative evaluation has been made.The time horizon for these effects is the same as for microand macroeconomic effects.

(d) The analysis of external effects rests on the standardframework for the study of international monetary regimes.Data are provided to give an assessment of possible changesin the international use of currencies, but no quantitativeevaluations of the associated gains are given except for verydirect and specific effects like the reduction on transactioncosts or the revenues from international seigniorage. Indeed,a large part of the gains one can expect in this field dependson the behaviour of third countries, since EMU will onlycreate opportunities for change in international monetarycooperation.

A large part of the external effects of EMU can be expectedto arise only in the medium to long run, because of hysteresisin the international role of currencies and of the delaysrequired for systemic changes in international monetaryrelations.(e) Transitional effects are basically of a macroeconomicnature. The same approaches have been used as for evaluat-ing macroeconomic stability effects, especially as regardsissues of credibility and policy discipline.

Costs and benefits of alternative exchange rate regimes

It has been argued in Section 2.1.1 that for comprehensive-ness, benefits and costs of EMU should not only be assessedwith respect to the '1992 + EMS' baseline but also incomparison to alternative exchange rate regimes. Since thisis obviously a very wide issue, only the main arguments willbe given here. For this purpose, Table 2.7 gives in a schematicfashion the costs and benefits of the three 'pure1 regimes ofGraph 2.3: financial markets autarky, free float and EMU.

The costs of financial autarky are immediately apparent:microeconomic efficiency is bound to be low due to a com-plete segmentation of capital markets; no significant externalinfluence can be expected for the Community as priority isgiven to autonomy over coordination and collective influ

54


Box 2.5: Macroeconomic models for EMU simulations

Two different multi-country macroeconomic models have beenused for evaluating some of the effects of EMU.

The Quest model of the Commission's services is a medium-sized quarterly model in the neo-Keynesian tradition. As astructural model, it involves a complete representation of thenational economies and of their trade linkages. The 1989 versionwhich has been used for this study incorporates individual struc-tural models of four Member States (France, Germany, Italy,United Kingdom) and the US.52 Exchange rates in Quest areexogenous, i.e. the exchange rate is considered as a policy instru-ment. This approximates to the situation within the EMS, as acountry facing an adverse shock can settle upon a realignmentin order to offset the impact of the shock. Simulations thereforeprovide some assessment of the cost of losing this instrument.

Quest, however, cannot be used for an evaluation of the macro-economic stability properties of alternative exchange rate re-gimes (including float), for such an evaluation requires, first,that exchange rates are endogenous, and second that the modeltakes into account that a change of regime impacts on thebehaviour of agents. This second requirement originates in the'Lucas critique' of policy evaluation with econometric models,according to which changes in policy regime alter the behav-ioural parameters of the model.53 Its theoretical relevance inthe case of a regime change like EMU is indisputable, especiallyfor the adoption of a single currency, since the nature of thischange, its permanent character and the associated constraintswould be highly visible for and credible to all agents. The extentof these changes in behaviour is however an empirical issue.

The Multimod model developed at the IMF responds to a largeextent to these two problems.54 The forward-looking Financialblock of this small annual model of the G7 economies involvesan endogenous determination of exchange rates through a finan-cial arbitrage condition (under the assumptions of perfect capitalmobility, complete asset substitutability, and perfect foresight),whereas the real sector incorporates in a simplified fashion bothbackward- and forward-looking features. When simulating theeffects of shocks under different regimes, it is assumed thatagents have full knowledge of the structure of the economy and

52 For an overall presentation of Quest, from which the version used in this study howeverdiffers in some respects, see Bekx. Bucher, Italianer and Mors (1989). Models for otherEC Member States and Japan are under development. In the present Quest model,feedbacks from the rest of the world are however incorporated in a reduced form.

" See Lucas (1976).54 Commission services arc grateful to the IMF for providing the latest version of

Multimod as well as technical support, but the simulations have been carried oulunder their sole responsibility. Neither the results nor possible errors should beattributed in any way to the IMF.

therefore of the consequences of the shock.55 However, wage-price stickiness (which can be due to multi-year contracts) andliquidity constraints imply short-term departures from equilib-rium in the Keynesian tradition.

The specific features of Multimod make it a suitable instrumentfor an examination of macroeconomic stability in EMU, eitherthrough deterministic simulations which attempt at evaluatingthe effects of the same shock under alternative regimes, orthrough stochastic simulations whose aim is to provide an over-all quantitative assessment of these regimes in an environmentof random disturbances. However Multimod has its weak pointstoo. First, it is a highly aggregated model which sometimesrests on simplifying assumptions, especially when compared tostructural models like Quest. Second, the estimation strategyfollowed by the IMF team has been to incorporate only signifi-cant parameter differences across countries, which is both astrength (because of the uncertainty surrounding these asymmet-ries) and a weakness (because the precision of individual equa-tions is reduced). Third, as a forward-looking model Multimodtends to exhibit much less nominal rigidity than standard macromodels, and more generally a stronger tendency to revert to thebaseline equilibrium. Since the estimation of forward-lookingmodels raises a number of technical difficulties, there is a certaindegree of uncertainty in this field too. Fourth and conversely,although regime changes impact on behaviour through expec-tations, it is still possible that some genuine behavioural changesregarding, for example, investment or wage behaviours arisingfrom EMU, are ignored.

Due to the different characteristics of the two models, it hasbeen chosen to specialize each in a certain type of simulations.The main purpose of the Quest simulations is to evaluate thecost of losing the exchange rate as an instrument in a standardtarget-instrument policy framework, whereas Multimod simula-tions are devoted to the overall assessment of the consequencesof EMU for macroeconomic stability. However, results fromboth models can also be considered as representing the uncer-tainty surrounding the true model of the economy and thereforethe range of possible effects of EMU.

Model simulations are presented in Chapters 5 and 6. Detailedresults, methodology and technical elements are given inAnnexes D and E.

Technically, expectations are model-consistent, which means that In deterministicsimulations future values of variables and the impact on the behaviour of agents areknown (obviously, this is not the case for unanticipated shocks). For a generalpresentation of Multimod, see Masson, Symansky and Meredith (1990), The versionused in this study differs from the one presented in this paper, due to minor changesintroduced in the version used at the Commission.

55


Table 2.7A schematic presentation of costs and benefits of alternative exchange rate regimes

Financialautarky

Freefloat

Microeconomic efficiency low medium high

Macroeconomic stability:(a) In the presence of shocks

Asymmetric shocksSymmetric shocksExchange rate instability

(b) Resulting from policy disciplineMonetary credibilityFiscal discipline

lowmedium

high

mediumcountry-dependent

highlowlow

country-dependentmedium

lowhighhigh

highmedium

External influence low low high

ence-building. The average macroeconomic performance canalso be expected to be poor in a regime characterized by theneed for each country to achieve external balance: asymmet-ric shocks but also symmetric ones, if policy reactions differ,would give rise to balance-of-payment crises as in the 1970s.Only for shocks that arise from exchange rate instabilitycan this regime achieve a fair performance. Finally, policydiscipline is not high in such a regime since external pressuresonly arise when the country experiences a current accountdeficit. As past experience consistently shows, this is at mosta second best since current account deficits exhibit only aweak and at least delayed correlation with either monetaryor fiscal mismanagement.

The arguments for free float are well known: each countryhas maximum policy autonomy and, provided wage-priceindexation is not complete in the short run, exchange-ratechanges can cushion asymmetric shocks. In spite of thesearguments, however, the performance of flexible exchangerates is generally considered disappointing for severalreasons. First, exchange rate variability is an obstacle tomicroeconomic efficiency: as shown by the recent US experi-ence, wide exchange rate swings degrade the quality of pricesignals and lead firms to adopt 'wait and see' attitudes;56

moreover, exchange rate misalignments imply significantwelfare losses, especially when adjustment costs are import-ant. Secondly, stabilizing properties of floating exchangerates are only apparent while facing country-specific, i.e.asymmetric real shocks; symmetric shocks, especially supplyshocks, give rise to beggar-thy-neighbour exchange rate poli-56 See Baldwin (1988), Dixit (1989), and for a general discussion Krugman

(1989 b).

cies as each country tries to export inflation or unemploy-ment; moreover, monetary shocks to the exchange rate itself,which arise from failures in the international financial mar-kets, are a source of instability. Thirdly, policy autonomyin a floating rates regime implies that no strong effectson policy discipline can be expected. Finally, individualparticipation of Member States in a floating rates regimecannot be expected to increase the influence of the Com-munity as a whole.

Over the whole range of effects considered in Table 2.7, theonly important disadvantage of EMU concerns macro-economic stability in the presence of asymmetric shocks. Thisis indeed a well-known argument and has unambiguously tobe considered a cost, but one that should be weighted againstthe clear advantages EMU yields in other fields.

Although for most Member States neither a pure floatingregime nor financial autarky are real alternatives to EMU,the qualitative framework of arguments presented points tothe basic features of its costs and benefits when comparedto hybrid regimes that could represent possible alternatives.As already discussed, an important methodological choiceof the present study is to assess the costs and benefits ofEMU as compared to a '1992 + EMS' baseline, in spite ofthe rather hypothetical character of some of its features andof the diversity of present situations among Member Statesregarding their participation in the ERM. In so far as therisk of systemic instability in Stage I could lead the systemto a reversion to some kind of pre-Stage I regime (i.e. someloosely defined mix of capital controls and crawling peg asin the early EMS), it is apparent that the net benefit of EMUcould only be greater.

56

sduquenoy


References

Aglietta, M. (1988), 'Regimes monetaires, monnaiesupranational, monnaie commune'. Communication a laConference Internationale de Barcelone sur la theorie de laregulation, June.

Aglietta, M., Coudert, V. and Delessy, H. (1990), 'Politiquesbudgetaires et ajustements macro-economiques dans la per-spective de 1'integration monetaire europeenne', in Lecturescritiques du rapport Delors, De Pecunia Vol. II, No 2-3,Brussels, September.

Aizenman, J. and Frenkel, J.A.(1985), 'Optimal wage index-ation, foreign exchange intervention, and monetary policy',American Economic Review, Vol. 75, No 3, June.

Argy, V. (1989), 'Choice of exchange rate regime for asmaller economy — A survey of some key issues', in Choos-ing an exchange rate regime: the challenge for smaller indus-trial countries, edited by V. Argy and P. De Grauwe, CEPS/IMF.

Artus, P. (1990), 'Epargne nationale, investissement et in-tegration Internationale', Document de travail I990-17/T,Caisse des Depots et Consignations.

Backus and Drifill, J., (1985), "Inflation and reputation',American Economic Review, June.

Balassa, B. (1964), The purchasing power parity doctrine:a reappraisal', Journal of Political Economy, December.

Baldwin, R. (1988), 'Hysteresis in import prices: the beach-head effect', American Economic Review, Vol. 78, No 4,September.

Baldwin, R. (1990), 'On the microeconomics of EMU', inEuropean Economy (1990).

Baldwin, R. and Krugman, P. (1987), The persistence ofthe US trade deficit', Brookings Papers on Economic ActivityNo 1.

Baldwin, R. and Lyons, R. (1990), 'External economicsand European integration: the potential for self-fulfillingexpectations', in European Economy (1990).

Barro, R.J. and Gordon, D. (1983), 'A critical theory ofmonetary policy in a natural rate model', Journal of PoliticalEconomy, 91, pp. 589-610, August.

Bekx, P., Bucher, A, Italianer, A. and Mors, M. (1989),The Quest model (version 1988)', EC Economic PapersNo 75, March.

Blanchard, O. J. and Fisher, S. (1989), Lectures on macro-economics, MIT Press, Cambridge.

Canzoneri, M. and Rogers, C. A. (1989), 'Is the EuropeanCommunity an optimal currency area? Optimal taxationversus the cost of multiple currencies', American EconomicReview, November.

Carli, G. (1989), The evolution towards economic and monet-ary union: a response to the HM Treasury Paper, Ministeriodel Tesoro, December.

Casella, A. (1990), 'Participation in a monetary union',NBER Working Paper No 3220, January.

Currie, A., Holtham, G. and Hughes Hallett, A. (1989), Thetheory and practice of international policy coordination:does coordination pay?' in Macroeconomic policies in aninterdependent world, edited by R. Bryant, D. Currie, R.Fortes, CEPR/The Brookings Institution/IMF.

Dixit, A. (1989), 'Entry and exit decisions under uncertainty',Journal of Political Economy, Vol. 97, pp. 620ff.

Driffill, J. (1988), The stability and sustainability of theEuropean Monetary System with perfect capital markets',in The European Monetary System, edited by F. Giavazzi,S. Micossi and M. Miller, Banca dTtalia/CEPR.

European Economy (1988a), The economics of 1992', No 35,March.

European Economy (1988b), 'Creation of a European finan-cial area', Special issue No 36, May.

European Economy (1990), The economics of EMU', specialissue.

Flood, R., Bhandari, J. and Home, J. (1989), 'Evolution ofexchange rate regimes', IMF Staff Papers, Vol. 36, No 4,December.

Friedman, M. (1953), The case for flexible exchange rates',in Essays in Positive Economics, The University of ChicagoPress.

Genberg, H. (1989), 'In the shadow of the mark: exchangerate and monetary policy in Austria and Switzerland*, in

57


Choosing an exchange rate regime: the challenge for smallerindustrial countries, edited by V. Argy and P. De Grauwe,CEPS/IMF.

Giavazzi, F. and Giovannini, A. (1989), The European Mon-etary System, MIT Press, Cambridge.

Giavazzi, F., Micossi, S. and Miller, M, eds (1988), TheEuropean Monetary System, Cambridge University Press.

Giavazzi, F. and Spaventa, L. (1990), The new EMS', CEPRDiscussion Paper No 369, January

Giavazzi, F. and Pagano, M. (1988), The advantage of tyingone's hands', European Economic Review, June.

Giovannini A. (1989), 'How do fixed exchange rate regimeswork? Evidence from the gold standard Bretton Woodsand the EMS', Blueprints for exchange rate management,January.

Gros, D. (1990), The EMS without capital controls', EcuNewsletter, No 30, pp. 22-26, October.

HM Treasury (1989), An evolutionary approach to economicand monetary union, London, November.

Hochreiter, E. and T6rnqvist,A. (1990), 'Austria's monetaryand exchange rate policy — Some comparative remarks withrespect to Sweden', in Lectures critiques du rapport Delors,De Pecunia Vol. II, No 2-3, Brussels, September.

Holtfrerich, C. L. (1989), The monetary unification processin nineteenth-century Germany: relevance and lessons forEurope today', in A European Central Bank?, edited by M.De Cecco and A. Giovannini, IPMG/CEPR, CambridgeUniversity Press.

Ingram, J. C. (1959), 'State and regional payments mechan-isms', Quarterly Journal of Economics No 73, November,pp. 619-632.

Ingram, J. C. (1973), The case for European monetaryintegration', Princeton essays in inter national finance. No 98,Princeton University, April.

Ishiyama, Y.(1975), The theory of optimum currency areas:a survey', IMF Staff Papers 22.

Kenen, P. B. (1969), The theory of optimum currency areas:an eclectic view', in Monetary problems of the internationaleconomy, edited by R. A. Mundell and A. K. Swoboda,University of Chicago Press.

Krugman, P. (1989a), Exchange rate instability, MIT Press,Cambridge, Massachussetts.

Krugman, P. (1989b), 'Policy problems of a monetaryunion', paper prepared for the CEPR/Bank of Greece Con-ference on the EMS in the 1990s.

Kydland, F. E. and Prescott, E.C. (1977), 'Rules rather thandiscretion; the inconsistency of optimal plans'. Journal ofPolitical Economy, Vol. 85, No 3.

Lucas, R. E. (1976), 'Econometric policy evaluation: a cri-tique', in Brunner, K. and Meltzer, A. H. (eds), The Phillipscurve and labor markets, Carnegie-Rochester ConferenceSeries on Public Policy 1, pp. 19-46.

Marston, R. C. (1984), 'Exchange rate unions as an alterna-tive to flexible rates: the effects of real and monetary disturb-ances', in J. O. Bilson and R. C. Marston (eds), Exchangerate theory and practice, NBER/University of Chicago Press.

Marston, R. C. (1985), 'Stabilization policies in open econ-omies', in Handbook of international economics, Vol. II,edited by R. Jones and P. Kenen, Elsevier Science Publishers.

Masson, P., Symansky, S., and Meredith, G. (1990), 'Multi-mod Mark II: A revised and extended model', IMF Oc-casional Paper No 71, July.

McKinnon, R. I. (1963), 'Optimum currency areas', Amer-ican Economic Review, Vol. 53, pp 717-725, September.

McKinnon, R. I. (1988), 'Monetary and exchange rate poli-cies for international financial stability: a proposal', Journalof Economic Perspectives, Vol. 2, No I, winter.

Meade, J. E. (1957), The balance-of-payments problems ofa European free-trade area', Economic Journal, Vol. 67, pp.379-396, September.

Mundell, R. A. (1961), 'A theory of optimum currencyareas', American Economic Review, September.

Obstfeld, M., (1988), 'Competitiveness, realignment, andspeculation: the role of financial markets', in The EuropeanMonetary System, edited by F. Giavazzi, S. Micossi and M.Miller, Banca d'ltalia/CEPR.

Padoa-Schioppa, T. (1987), Efficiency, stability, equity, Ox-ford University Press, Oxford.


Poloz, S. (1990) 'Real exchange rate adjustment between Van Rompuy, P., Abrahams, F. and Heremans, D. (1990),regions in a common currency area1, mimeo, Bank of Can- 'Economic federalism and the EMU', in European Economyada, February. (1990).

Wihlborg, C. and Willett, T. (1990), 'Optimum currencyareas revisited', mimeo, April.

Van der Ploeg, F. (1990), 'Macroeconomic policy coordi-nation during the various phases of economic and monetary Williamson (1985), The exchange rate system, Institute forintegration in Europe1, in European Economy (1990). International Economics, Washington.

59

PartB

The main benefits and costs

sduquenoy

Chapter 3. Efficiency gains

Chapter 3

Efficiency gainsThis chapter discusses and estimates the gains in terms ofeconomic efficiency from EMU, in particular those arising inthe private sector (i.e. to business and households). The publicsector will also reap efficiency gains, but these are discussedin Chapters 5 and 7 below.

Since the efficiency gains from monetary union are easier toestimate than those of economic union the chapter discussesfirst, in Sections 1 and 2, the two main sources for directefficiency gains from monetary union which are the eliminationof exchange rate related transaction costs and the suppressionof exchange rate uncertainty, The gains from building on1992 in moving towards economic union are then discussed inSection 3. However, an economic and monetary union willalso have a number of further indirect and dynamic effectswhich cannot be measured with the same precision as themore direct and static effect of the suppression of exchangerate variability and transaction costs. These indirect and dy-namic effects are therefore discussed separately in Section 4of this chapter. Section 5 concludes by discussing how improvedbusiness expectations because of EMU could reignite growthand thus reduce unemployment.

The main findings of this chapter are:

(a) A single currency eliminates the present cost associatedwith converting one EC currency into another. The result-ing savings can be estimated at more than ECU 15 billionper annum, or about 0,4% of Community GDP. Thelarger part of these gains are 'financial', consisting of thedisappearance of the exchange margin and commissionfees paid to banks. The other gains take the form ofreductions in costs and inefficiencies inside firms,

(b) Transaction cost savings differ strongly from country tocountry. The gains for the larger Member States whosecurrency is extensively used as a means of internationalpayments and belongs to the EKM may be of the order ofbetween 0,1% and 0,2% of national GDP. In contrast,the small open and the less developed economies of theCommunity may stand to gain around 1% of their GDP.

(c) Transaction costs are more harmful to small and medium-sized enterprises engaged in intra-EC trade than to largemultinational companies. Whereas total transaction costsincurred by firms can be estimated to amount on averageto some 15 % of their profits on turnover in other ECcountries, they can easily be twice as great in the case of

small firms, in particular when they are located in non-ERM countries.

(d) When accompanied by measures that remove the technicalbarriers which currently complicate international banksettlements, a single currency will also enable significantcuts to be made in other banking costs in connection withcross-border payments.

(e) EMU would eliminate nominal exchange rate variabilityamong Community currencies. However, some variabilityin national price levels might remain. Comparisons withother monetary unions indicate that the level of real ex-change rate (the nominal exchange rate adjusted for move-ments in the prices) variability existing at present insidethe original narrow band ERM members is not far fromwhat one could expect in EMU. However, aside from thisgroup of countries, EMU should lead to a sharp reductionin real exchange rate variability.

(f) The gains from the suppression of exchange rate varia-bility in terms of increased trade and capital movementsare difficult to measure because firms can in many casesinsure against this risk using sophisticated foreign ex-change market operations. However, business surveys pro-vide strong evidence that despite this possibility, which isin itself costly, foreign exchange risk is still considered amajor obstacle to trade. The suppression of exchange ratevariability will be more important for small firms andcountries with less-developed financial markets that do nothave access to sophisticated hedging techniques.

(g) The gains from economic union consist mainly in a betterformulation and implementation of Community policies inareas where there are Community-wide external effects oreconomies of scale and Community action is thereforejustified under the principle of subsidiarity. These gainscan be shown to be potentially important in a number ofpolicy areas.

(h) A potentially very important gain arises if EMU reducesthe overall uncertainty for investors associated with the

v existence of national currencies and independent monetarypolicies. A reduction in overall uncertainty could lower therisk premium firms have to pay on equity and would greatlyincrease investment. Preliminary estimates show that evena reduction in the risk premium of only 0,5 percentagepoints could raise income in the Community significantly,possibly up to 5-10% in the long run.

(i) Recent research suggests that by improving the expec-tations of business EMU could lead the Community to anew growth path along which unemployment could be re-duced decisively.

63

Part B — The main benefits and costs

(j) In fact, opinion surveys of European industrialists indicatethat business leaders do expect significant gains from asingle currency. As shown in the introduction to this study,the addition of a single currency to the single marketincreases the share of industrialists expecting a very posi-tive impact on the business climate from 10 to 45%.

3.1. Exit exchange rate transaction costs

With the introduction of a single currency all exchangerate related conversion costs disappear on intra-Communitytransactions.

These costs can be split into two parts. First, there are thedirect transaction costs households and firms pay to thefinancial sector in the form of foreign exchange commissionsand the difference between buying and selling rates. Second,there are the costs borne inside companies, arising for in-stance from the need to allocate personnel and equipmentto foreign exchange management. These latter costs mightbe called 'in house1 costs to distinguish them from the formerthat are more visible since they arise in transactions with thefinancial sector and might therefore be called external or'financial* costs.A recent strand of research in economics (see Akerlof andYelen (1989)) suggests that even small transaction or infor-mation costs can have significant economic effects. Thisresearch would imply that the economic losses from ex-change rate transaction costs are much larger than the directcosts themselves that are estimated in this section. An illus-tration of the large indirect effects of even small transactionand information costs is provided by de Jonquieres (1990)where it is shown that even for goods to which there are notrade barriers in the Community prices can diverge by asmuch as 100% (with a range going from 40 to 170%)from one country to another. These price differences implyconsiderable welfare losses which should be reduced andmay be even eliminated under a single currency because witha single currency consumers will immediately be able tocompare prices. These additional potential gains are, how-ever, not taken into account in this study since they cannotbe quantified with any precision.

3.1.1. External or financial costs

The financial costs Community firms and individuals incurowing to the absence of a single currency can be measuredin several ways. The most direct way would be to ascertainthe income financial institutions obtain from their customersfor foreign exchange services, or, what amounts to the samething under competitive conditions, to measure the foreign

exchange related resource costs of banks. A more indirectapproach is to determine the foreign exchange costs bankscharge their customers and combine this information withdata on the total volume of foreign exchange transactionsbetween EC currencies on behalf of non-bank enterprisesand individuals. These data can be derived either from surveyresults on the turnover on EC foreign exchange markets orfrom Member States' current and capital account relatedreceipts and payments in EC currency other than the dom-estic money.

In the present study both the direct and indirect approacheswere followed. They lead to very similar estimates of overallfinancial costs. A detailed presentation of the basic data,working hypotheses and calculations underlying these esti-mates is given in Annex A of this report. The presentationhere is limited to a discussion of the chief findings.

Foreign exchange related revenues for banks

This approach relies on confidential data collected by theCommission services on the foreign exchange related rev-enues of banks in the Community. Since only informationon total foreign exchange revenues is available it was as-sumed that one half is attributable to foreign exchangebetween EC currencies. This can be justified by the factthat the value of Member States' current account relatedpayments and receipts denominated in foreign EC currenciesclearly exceeds the value in non-EC currencies. The bankrevenue data suggest that the total external transactionscosts that could be saved by a single currency are about0,25% of the GDP of the Community. This results from theestimate that a little less than 5% of bank revenues comefrom foreign exchange activities between EC currencies andthat the banking sector accounts for about 6% of the GDPof the Community.

Firms* and households' financial costs of foreign exchange

Approaching the question of quantifying exchange trans-action costs from the viewpoint of the buyers of foreignexchange, it is necessary to know the prices banks chargefor their intermediation services as well as the total value oftransactions in foreign EC currency these prices should beapplied to.

Bank foreign exchange charges vary considerably, dependingon the currency of exchange, the nature of the foreignexchange 'product' (spot, forward, swaps, options...), thesize of the transaction and, as in any other market, theimportance of the bank customer.

64

sduquenoy

sduquenoy


Box 3.1: Bid-ask spreads and exchange rate variability

While it is apparent that the introduction of a single currencywill eliminate all exchange rate related conversion costs it is notclear a priori to what extent these costs will already decline asexchange rates become increasingly fixed.

The available evidence on bid-ask spreads, which are a majorcomponent of overall exchange rate conversion costs, indicatesthat lower exchange rate variability might not have a strongimpact on conversion costs. (See for example Boyd, Gielens andGros (1990) and Black (1989).) Whether the irrevocable fixingof exchange rates would decisively lower the costs of exchangingcurrencies is difficult to decide since there are only very fewhistorical examples that could be used as a guide. The extent towhich this might happen depends on the degree to which oper-ators consider different currencies just different units of account.For example, the exchange rate between the Belgian and theLuxembourg franc has been fixed at one to one for over 50years and the probability that it might change in the future is

considered very small. Belgian and Luxembourg francs cantherefore be exchanged without any conversion costs. However,it is unlikely that the same would occur for the other Communitycurrencies for which it would be more difficult to achieve thesame degree of credibility in the irrevocable nature of the ex-change rate and for which the actual conversion rates would notbe round numbers. It is therefore likely that financial institutionscould consider Community currencies as distinct and wouldcharge fees for converting one into another even if exchangerates are declared to be irrevocably fixed by the authorities.

It has been suggested by Dornbusch (1990) that conversion costsfor the corporate sector could be eliminated even in a system withseparate national currencies through a system of'par clearing', inwhich financial institutions would be forced to exchange bal-ances in Community currencies at par, i.e. without a bid-askspread and without charging foreign exchange fees. Providednational payments and clearing systems are made compatible,such a scheme would therefore create a close substitute for asingle currency, at least in terms of transactions costs savings.

(a) Transactions by households

The highest transaction costs arise when exchanging cash.These costs were illustrated vividly by the admittedly theor-etical example worked out by Bureau Europeen des Unionsde Consommateurs1 (BEUC) in 1988. A traveller is assumedto start out with 40 000 Belgian francs in Brussels embarkingon a clockwise tour of all Community capitals (except Lux-embourg and Dublin). At the end of the journey the accumu-lated loss is about 47% if he exchanges his cash at each legof the roundtrip into local banknotes. Table 3.1. shows howmuch he would lose at each of his 10 consecutive conver-sions. The largest losses (14 and 21%) occur when buyingor selling in countries with strong currencies the banknotesof weak currencies, like the drachma or the escudo. It willbe argued in Annex A that a weighted average of the costof banknote conversion in the Community is likely to am-ount to about 2,5%. Using the latter percentage, total bank-note transaction costs that will be eliminated by a single ECcurrency can be estimated to lie between ECU 1,3 and 2billion, depending on the hypotheses retained to make upfor missing data on the volume of banknote conversions insome Member States.

See BEUC (1988b).

The exchange margin for traveller's cheques is usuallysmaller than for cash but there is a 1 % commission charge.Eurocheques, which in many Member States are free ofcharge (upon the payment of a fixed fee) when used dom-estically, cost normally between 2 and 3 % when written ina foreign currency. In the case of international credit cardsforeign currency costs vary between 1,5 and 2,5%. Witharound 40 million international eurocheques in the EC foran average value of ECU 125, a reasonable estimate ofthe cost savings a single currency would allow for presenteurocheque users is between ECU 100 and 150 million perannum. The gains with respect to EC currency denominatedtraveller's cheques, of which the annual sales in the Com-munity represent a value of around ECU 5 billion, can alsobe put at around 150 million. The total foreign EC currencypayments volume by means of credit cards can be roughlyestimated at ECU 10 billion per year. The associated econ-omies a single currency would allow are therefore likely tolie between ECU 150-200 million.

(b) Transactions by the corporate sector

Given the relatively high minimum fee, bank transfers tendto be a relatively costly international payments instrumentfor small amounts. They are the standard means of inter-national settlements between enterprises, with bank chargesbeing a function of the amount and the currency.

65

sduquenoy


Table 3.1

Currency transaction losses in a (hypothetical) round-trip through 10 countries

Exchanged inon 1 March 1988

B (begin)UKFEPIGRDDKNLB (end)

Total

Exchange rateapplied in local

currency

UKL1FF 9,8065PTA 19,47ESC 1,18LIT 7,75DR 10,575DM 0,98DKR 378,44HFL 27,75BFR 18,14

= BFR 64,95= UKL1- FF1= PTA 1= ESC1- LIT 100= DR 100= DM 100= DKR 100= HFL 1

Amounts afterexchange

transaction

BFR 40 000UKL 615,86FF 6039,43PTA 117587,49ESC 138753,49LIT 1 075 339,52DR 113717,15DM 1 114,43DKR 4 21 7,45HFL 1 170,34BFR 21 300

In ecu1

925,18891,30863,55843.69820,35706,43686,97539,42534,42504,71492,66

Loss in %

-3,662

-3,H2

- 2,302

-2,77- 13.892

-2,75-21,462

-0,952

- 5,562

-2,39-46,75

Official exchange rate published in the Official Journal of the European Communities, I March 1988.Additional bank charges can occur.

Source: BEUC(198Bb).

Bankers' replies to a questionnaire submitted by the Com-mission services suggest that when the amount involved isequivalent to ECU 10 000, foreign EC currency bought onthe spot market costs around 0,5 %, with reported extremesranging from 0,1 to 2,5 %. Foreign currency conversion ofan amount equivalent to ECU 100 000 was reported to costabout 0,3 %; nevertheless foreign exchange charges veryoften still exceed 1 % for payments in reputedly weak cur-rencies that are hardly used in international transactions,like the drachma or escudo. Very large amounts, equivalentto ECU 5 million or more, involve costs of the order of 0,05and 0,1 %, which is the size of the spread that can beobserved in the interbank market for foreign exchange.

Bank charges declining with the amount to be converted, anestimate of average exchange transaction costs for firmsrequires also information on the size distribution of foreigncurrency payments and receipts. Such data exist for a num-ber of Member States and show that in- and outflows witha value equivalent to ECU 100000 or more claim about55 % of the total value of current account transactions inforeign EC currency. For capital account transactions bythe non-bank sector this figure climbs to more than 90 %.

As explained in Annex A, combining these figures with theaforementioned information on the banking sector's pricesfor foreign exchange services allows one to advance therough estimate that converting the domestic currency intoanother EC currency costs non-bank firms on average about0,3-0,35 % as far as current account transactions are con-

cerned; for capital account transactions this percentage canbe thought to diminish to about 0,1-0,15%, leading to anoverall average of 0,15-0,2 %.

(c) Grossing up financial costs

The BIS survey in April 1989 on foreign exchange marketssuggests that the total net turnover on EC foreign exchangemarkets arising from the foreign currency needs of the non-bank sector equals some USD 13000 billion per annum.Depending on assumptions (see Annex A) 34 to 43 % of thisturnover can be thought to be directly or indirectly betweenEC currencies, representing an amount between ECU 4 100and 5 200 billion. Applying the just mentioned overall costaverage of 0,15-0,2 % to the latter values, this would meanthat in 1989 exchange transaction costs were situated be-tween ECU 6,2 and 10,4 billion.

To these Figures must be added the costs associated withconverting banknotes, travellers' cheques and eurochequesas the latter were not taken into account in the BIS survey.Consequently, 'financial' transaction costs borne by the ECeconomy due to the absence of a single currency can beestimated to range in 1990 from ECU 8 to 13 billion, orfrom 0,17 to 0,27% of EC GDP, which is very similar tothe aggregate estimate derived from the banking revenuedata,2

The economics literature has so far not come up yet with well-foundedtransaction cost estimates. Cukierman (1990) provides a back-of-lhe-envelope assessment, putting financial transaction costs as high as 1 %of Community GDP.

66


An alternative method of arriving at the transaction volumein EC currency other than the domestic money is to deter-mine, on the basis of balance of payments statistics, eachindividual Member State's gross current and capital accountflows in foreign currency and to isolate the EC component.Unfortunately, the necessary data are not available for allcountries. However, the available evidence permits to shedlight on the fact that in relative terms the transaction costsavings will be distributed unevenly over the Member States.

These potential savings grow larger;(i) the lesser the use of the national currency as a means of

international payment;(ii) the more intense the trade in goods, services and assets

with other Member States;(iii) the lower the technical and price efficiency of domestic

foreign exchange services;(iv) the greater the variability of the national currency's

exchange rate as it necessitates more systematic hedging(see Section 3.2.4.) and causes bankers' margins towiden.

Small open economies with 'small' currencies like Belgium-Luxembourg, Denmark, Ireland and, to a lesser extent,the Netherlands, or countries with as yet unsophisticatedfinancial markets like Greece, Portugal and Spain will benefitrelatively more from the elimination of transaction coststhan Germany and France whose currency belongs to theERM and is a well-accepted means of international settle-ments. The available material indicates that whereas theexchange transaction cost savings for the latter MemberStates is likely to oscillate between 0,1 and 0,2 % of GDP,for the small open and less developed economies the gaincould be as high as 0,9 %.

Reductions in cost and time of cross-border payments

Apart from eliminating exchange transaction costs, a singlecurrency could also make an important contribution tocutting the present expenses and delays associated with cross-border bank payments.

In comparison to the situation in the USA, where a coast-to-coast cheque costs a fixed money transfer fee of 20 to 50US cents and takes two working days, these costs and delaysare substantial in the Community. A recent study by BEUC3

found that a bank transfer from one Member State toanother of ECU 100 in the beneficiary's money cost onaverage more than 12% — of which less than 25% was

caused directly by currency conversionfive working days.

and took generally

BEUC(1988a).

The creation of a single currency and a single system ofcentral banks will permit a major simplification of banks'treasury management, accounting and reporting to monet-ary authorities. When flanked by internal market measuresto remove the technical barriers that still complicate theprocessing of international bank transfers, these simplifi-cations could make cross-border payments as fast and cheapas domestic ones today.

With an estimated number of 220 million cross-border banktransfers in the Community per year and the difference infixed processing fee between a domestic and an internationalsettlement (net of exchange transaction costs) around ECU6 the potential supplementary gain could be set at ECU 1.3billion.

3.1.2. In-house costs

The existence of different currencies leads also to costs thatare internal to the non-financial corporate sector. Thesecosts arise for a variety of reasons. First, multiple currenciesrender the treasury and accounting functions more compli-cated so that firms need to devote more personnel to thesetasks. They also raise the managerial complexity in transna-tional firms in that they complicate central management'stask of control and evaluation. Second, multiple currenciesfragment cash management and thereby lead to companycash being poorly remunerated or, conversely, to interestcosts on debit positions. Third, they lengthen the delaybetween debiting and crediting bank accounts. Fourth, firmsmay incur opportunity costs in their attempt to avoid, ratherthan manage, exposure to foreign exchange risk.

These costs are difficult to measure with any precision, sincethe sources of costs are distributed over a wide number ofdifferent departments inside each corporation. They could,in principle, be quantified for each corporation through acareful audit which could determine the resources that couldbe saved through the introduction of a single currency. Sinceit would clearly be impossible to do this for a significantnumber of corporations it was necessary to rely on a sampleof case studies.

As explained in Annex A, these case studies suggest a lowerbound estimate for these in-house costs equals 0,2 % offirms' turnover from business in other EC countries. Smalland medium-sized enterprises tend to suffer relatively largercosts because of the overhead nature of these expenses.Given that value added represents by and large 55 % of

67

sduquenoy


corporate turnover and that intra-EC exports of goods andservices equal around 16% of EC value added or GDP,these in-house costs amount to 0,1 % of EC GDP ((0,2 %x 0,16)/0,55).

3.1.3. Summary evaluation of transactioncost savings

This section has estimated the savings in transaction coststhat will arise from the introduction of a single currency.Two approaches to estimate the costs of the conversionservices performed by banks yielded similar results, namelythat the total cost of these services is equivalent to about0,25 % of the GDP of the Community. To these costs it isnecessary to add the in-house costs, of the order of 0,1 %of GDP, which the corporate sector faces. Moreover, a singlecurrency is also a necessary condition for a reduction in costand time of international bank transfers, which could yieldanother ECU 1.3 billion.

As shown in Table 3.2, the total quantifiable savings interms of transaction costs are therefore around 0,3 to 0,4 %of the GDP of the Community or about ECU 13 to 19billion per annum.

Table 3.2

Cost savings on intra-EC settlements by single EC currency(in billion ECU, 1990>

Estimated range

1. Financial transaction costsBank transfersBanknotes, eurocheques, traveller's cheques.

6,4 10,6

credit cards

2. In-house costs

Total

3. Reduction of cross-border payments cost

Total

1,88,2

3,6

1,3

13,1

2,513,1

4,8

1,3

19,2

Noie: Exchange transaction costs associated with several sources of in-house costs are notincluded in this table.

The latter figures relate only to the direct 'mechanical' ex-penses occasioned by the need to convert currencies andcover against exchange risk. They do not take into accountthe implicit tax on cross-border business, hence the fragmen-tation of markets which the existence of a multitude of ECcurrencies gives rise to due to the heightened complexity anduncertainty it induces.

The direct costs to households and cross-border shoppersare large in proportional terms, but their total cannot beestimated very accurately. The costs with respect to bank-notes and eurocheques can be set at around ECU 2 billion,Under the plausible assumption that these expenses areincurred by households, the total 'financial' and in-housecost firms undergo with regard to their current accounttransactions with other Member States can be estimated atECU 7 to 8 billion. As this represents about 1 % of the totalvalue of intra-EC exports of goods and services and profitsamount on average to 5-6 % of exports, transaction costsborne by firms are equal to more than 15 % of their profitson exports to other Member States.

Against these efficiency gains one would have to set the costof introducing the single currency. However, since this is ofa once-and-for-all nature it should be small relative to theefficiency gains from a single currency which can be reapedfor the indefinite future.

Besides, the Community's banking sector will need to gothrough an adjustment phase as its resources that havebecome redundant following the introduction of a singlecurrency are redeployed. These resources could be directedtowards the expanding financial intermediation activities inecu-denominated assets. As argued in Chapter 7, both theworld supply of and demand for such assets are likely toincrease significantly upon the creation of a single currency.

3.2. Exit exchange rate uncertainty

Monetary union obviously eliminates exchange rate move-ments and hence uncertainty about intra-EC exchange rateswhich should stimulate trade and investment. In order toestimate the gains which the suppression of exchange ratevariability brings, this section presents first some data aboutintra-EC exchange rate variability to indicate the reductionin exchange rate variability member countries can expect.However, the economic welfare gains that should result fromthis are difficult to measure because economic theory has notcome to definite results concerning the relationship betweenexchange rate uncertainty and trade (or investment). Thetheoretical arguments are therefore examined before pro-ceeding with a discussion of the empirical research on theimpact of exchange rate uncertainty on trade and capitalflows.

Before going into the measurement of exchange rate uncer-tainty it is necessary to clarify four general points:(i) At the theoretical level it is clear that only unexpectedchanges in exchange rates constitute exchange rate uncer-

68

sduquenoy

sduquenoy

Chapter3. Efficiency gains

tainty. However, since a large body of empirical literature4

has demonstrated that most short run (i.e. over a quarter ora year) changes in exchange rate are unexpected, the varia-bility in actual exchange rates is a good proxy for exchangerate uncertainty. In the remainder of this section the terms'exchange rate uncertainty' and 'exchange rate variability'will therefore be used interchangeably and will be taken torefer to short run changes in exchange rates.

(ii) It is apparent that a monetary union eliminates onlyvariability in nominal exchange rates. Changes in real ex-change rates (i.e. the nominal rate corrected by some priceindex) are still possible, and at times even desirable, if econ-omic conditions and therefore prices develop differently inthe regions that are part of the union. The theoretical andempirical literature on exchange rate variability indicates,however, that both nominal and real exchange rate varia-bility may be important. For floating exchange rates thisdistinction has usually not been important since short run(up to a year) changes in nominal rates are usually equivalentto changes in the real exchange rate, given that nationalprice levels move much more slowly than exchange rates.

For floating exchange rates it is therefore approximatelytrue that a reduction in nominal exchange rate variability isequivalent to a reduction in real exchange rate variability.However, this does not apply to the tightly managed ex-change rates that belong to the ERM. Indeed there are somedata that suggest that the low degree of real exchange ratevariability achieved in the ERM corresponds to the levelone could expect to find even inside a monetary unionconsisting of rather diverse regions as would be the case forthe Community.5 This section will therefore concentrate onthe effects of (short run) nominal exchange rate variabilityalthough this is only partially equivalent to real exchangerate variability.

Even in EMU long run movements in real exchange ratescan occur through movements in wages and prices (seeChapter 2). However, these long run movements are gener-ally not unexpected and should therefore not have the sameimpact on trade and investment as the largely unexpectedshort run changes. Economic benefits can therefore be ex-pected only from the suppression of the short run exchangerate variability defined above.6

See Mussa (1986) for a survey.See Poloz (1990).This section does not consider separately the issue of medium-termmisalignments in exchange rates since they are generally taken not toplay an important role among Community currencies, especially thosethat belong to the ERM.

Table 3.3 below illustrates these points with data on thevariability of the DM vis-a-vis four other currencies: the twoextremes are formed by the Dutch guilder and the USdollar.7 For the US dollar nominal and real exchange ratevariability are very high, and about equal in size for allsubperiods. For the Dutch guilder nominal and real varia-bility are very low, but the variability of the real rate is muchhigher especially during the last subperiod. Indeed for 1987-89 the variability of the real DM/HFL rate is of the sameorder of magnitude as the variability of the real DM/LITrate, although the variability of the nominal DM/LIT rateis about six times as high as that of the DM/HFL rate.Finally, the DM/DR exchange rate is, for the period 1979-89, over twice as variable as the DM/LIT rate in nominaland real terms, attaining almost the level of the DM/USDrate.

Table 3.3

Bilateral exchange rates

Variability as standard deviation of monthly percentage changes

1974-78 1979-89 1979-83 1984-86 1987-89

DM/HFL nominalDM/HFL real

DM/LIT nominalDM/LIT real

DM/DR nominalDM/DR real

DM/USD nominalDM/USD real

0,640,85

2,402,32

1,82,15

2,312,41

0,320,52

0,850,91

2,152,33

2,932,97

0,460,59

0,961,04

2,332,60

2,642,68

0,130,38

0,770,78

2,562,53

3,113,10

0,110,52

0,690,72

0,751,41

2,932,97

(iii) Any judgement about the impact of an elimination ofexchange rate variability on trade, investment and finallyeconomic welfare must be based on the assumption thatthe existing exchange rate variability is not warranted byvariability in underlying economic conditions. If mostchanges in exchange rates could be regarded as an efficientadjustment to changes in productivity, investment oppor-tunities or other so-called fundamental factors, the sup-pression of this adjustment mechanism (by irrevocably fixingexchange rates) might actually lower welfare.3

7 All the tables in this chapter refer to monthly averages of exchange ratesobtained from Commission sources.

8 Mussa (1986) also offers a cautious assessment of the welfare conse-quences of exchange rate variability.

69

sduquenoy

sduquenoy

sduquenoy


However, there is considerable evidence that, in a regime offree float, most exchange rate variability is not related tovariability in fundamental factors such as productivity orinvestment opportunities. It has been impossible so far torelate systematically the evolution of the so-called funda-mentals to exchange rate movements (see for example Meeseand Rogoff (1983) and Gros (1989)).

Among the currencies participating in the EMS no suchovershooting seems to occur and the remaining fluctuationsinside the bands just offset short run differences in the stanceof monetary policy. However, even though there might beno overshooting among ERM currencies, the remainingintra-ERM exchange rate variability is caused mainly bydifferences in national monetary policy stances and hence isnot related to the fundamental real factors that determineinvestment and trade, such as productivity and comparativeadvantage. One could therefore argue that even for EMScurrencies the remaining exchange rate variability is notwarranted by policy-independent fundamental factors thatoperate even in EMU. This does not imply that fixingexchange rates involves no economic costs, indeed Chapter6 below discusses the costs that arise when the adjustmentin labour and goods markets is slower than in the foreignexchange market. However, this section concentrates on thegains that might arise from the suppression of exchange ratevariability.(iv) Any positive effect of the suppression of exchange ratevariability among Community currencies on intra-EC tradeand investment can be considered a benefit of EMU only ifthere are no side-effects on the variability of the system ofCommunity currencies vis-a-vis the other major currencies.If the elimination of intra-EC exchange rate changes leadsto more variability among the three remaining currencyareas (dollar, yen and ecu) there might be an offsettingnegative effect and vice versa if EMU leads to a morestable international monetary system. Since the external andinternal trade of the Community are about equal in size thispotential offsetting or reinforcing effect could alter substan-tially the size of the benefits that can be expected from morestable exchange rates within the Community.

The impact of EMU on the global monetary system isdiscussed in more detail in Chapter 7 of this study, where itis argued that it is difficult to predict a priori what effectthe creation of EMU will have on the stability of the globalmonetary system. This section therefore assumes that theformation of EMU will at least not increase uncertainty inthe global monetary system.9

3.2,1. Exchange rate variability in theCommunity

As discussed in the introduction to this section the variabilityin short run changes in nominal exchange rates is the appro-priate measure of exchange rate variability for the purposeof measuring the benefits of irrevocably Fixing exchangerates. The exact meaning of 'variability' and of 'short runchanges' are always subject to discussion on technicalgrounds. However, the following tables use the most widelyaccepted measure of variability, namely the standard devi-ation of percentage changes in monthly nominal exchangerates. The large literature on the effects of the EMS onexchange rate variability shows that the choice of the exactmeasure of exchange rate variability is largely arbitrary anddoes not influence the results since most measures lead tothe same conclusion.10

Table 3.4 shows the overall variability of the Communitycurrencies (plus the US dollar, the yen and the Swiss franc)using nominal exchange rates. Table 3.5 displays the varia-bility of the currencies considered against Community cur-rencies only. For Community currencies this measures intra-Community variability. Finally, Table 3.6 displays the varia-bility against ERM currencies only. For ERM members"this measures intra-ERM variability. These tables show onlyaverages for the pre-EMS and selected post-EMS periods.The study by Weber (1990) also analyses exchange ratevariability in some detail.

It is apparent from these tables that global exchange ratevariability exceeds intra-Community variability, which, inturn, exceeds intra-ERM variability. EMU will, however,eliminate only intra-Community exchange rate variability.The effect of EMU can therefore be read from Table 3.5which indicates that the main benefit from irrevocably fixingexchange rates is to reduce the average variability (as meas-ured by the standard deviation) of Community currenciesagainst each other from at present (i.e. 1987-89) 0,8 % (permonth) to zero.

As can be seen from Table 3.4 this average hides largedifferences among member countries. For the UK the re-duction would be almost two times as high (1,6 %) and forBelgium it would only be 0,5 % . For the original membersof the ERM the average in 1987/89 (and therefore thereduction through irrevocably fixing exchange rates) wasonly 0,7 %; whereas the other member countries start from

The stochastic simulations presented in Annex E suggest that the varia-bility among the remaining major three currencies would not changesubstantially.

10 See Ungerer el al. (1986) and Ariis and Taylor (1988) for more references.11 Since Spain did not participate in the ERM until 20 June 1989 and the

United Kingdom until 8 October 1990, these two countries are notconsidered ERM members in the calculations in the present chapter.

70


Table 3.4Bilateral nominal exchange rates against 20 industrialized countries

Variability as weighted sum of standard deviation of monthlypercentage changes

1974-78 1979-89 1979-83 1984-86 1987-89

B/LDKDGREFIRLINLPUK

USAJapan

1,5 1,4 1,6 1,11,6 1,5 1,7 1,31,8 1,5 1,6 1,41,9 2,3 2,5 2,72,9 1,9 2,1 1,42,1 1,5 1,7 1,41,7 1,6 1,6 1,72,3 1,5 1,6 1,51,4 1,2 1,3 1,02,9 1,8 2,3 1,32,1 2,4 2,4 2,5

2,1 2,5 2,3 2,62,4 2,7 2,9 2,7

Switzerland 2,4 1,9 2,1 1,8

AVIAV2AV3AV4

A'o/f: AVIAV2AV3AV4

1,8 1,5 1,6 1,31,9 1,6 1,7 1,52,4 2,2 2,4 2,22,1 2,0 2,1 2,0

= Weighted average of ERM currencies, ecu weights.= Weighted average of EC currencies, ecu weights.= Weighted average of EC non-ERM currencies, ecu weights.= Unweighted average of non-ERM currencies.

0,91,21,21,01,51,11,21,20,91,01,92,42,41,6

1,11,21,71,6

EMU

0,050,080,090,050,070,070,060,070,050,050,10

2,42,41,5

0,070,080,090,13

a much higher level of variability, they can expect a reductionin (intra-Community) variability of about 1,9 % .

Table 3.6 can be used to indicate the benefits participationin Stage I might bring to those countries that have not untilrecently participated in the narrow margins of the ERM. Acomparison with Table 3.5 suggests that for a country likethe UK the absolute reduction in exchange rate variabilitythrough Stage I is equal to about 1,2 % (i.e. from the valueof 1,6 in Table 3.5 to 0,4 in Table 3.6), i.e. three times largerthan the additional reduction that is obtained by going fromStage I to EMU.

While it is clear that EMU reduces nominal exchange ratevariability to zero it is not as clear what level of real exchangerate variability will remain. The evidence from Canada dis-cussed in Chapter 2 above suggests that the real exchangerate variability that exists at present between narrow bandERM members might actually constitute the level to beexpected under EMU. This implies that Member States thatat present do not participate in the narrow band ERM willexperience some reduction in real exchange rate variabilityas well.

Assuming that the variability of the other major currenciesagainst the ecu remains unchanged the last column ofTable 3.4 shows the average variability in nominal exchangerates member countries can expect once intra-Communityexchange rates have been irrevocably fixed. This table sug-gests that the countries that will gain most from EMU underthis aspect are countries like Spain, Portugal and the UKwhich trade relatively more with the rest of the world andwhere exchange rates are at present more variable.

3.2.2. Exchange rate variability and trade

The preceding subsection has measured the reduction inexchange rate variability from EMU. But what economicbenefits will result from this reduction in exchange ratevariability? This subsection discusses the impact which morestable exchange rates might have on trade starting with thetheoretical arguments.

Table 3.5Bilateral nominal exchange rates against EUR 12 currencies


1974-78 1979-89

B/LDKDGREFIRLINLPUK

USAJapan

1,3

1,51,71,82,92,01,52,21,32,82,0

2,22,3

Switzerland 2,2

AVIAV2AV3AV4

Note: AVI -AV2 -AV3 =AVd =

1,7U92,32,2

Weighted average

1,01,11,12,21,71,11,31,10,91,72,2

2,82,41,5

1,11,32,02,0

ofERMcurrenciWeighted average of EC currencies.Weighted average ofECnon-ERM

1979-83

1,31,31,32,42,01,31,31,21,02,22,4

2,62,81,7

1,31,42,32,1

1984-86

0,70,80,92,61,11,01,41,10,71,12,2

3,02,31,30,91,11,91,9

1987-89

0,50,70,70,71,20,70,80,80,50,81,6

2,81,81,10,70,81,91,5

EMU

00000000000

000

1,4

es, ecu weights., ecu weights.currencies. ecu weights

Unweighted average of non-ERM currencies.

71

sduquenoy

sduquenoy


Table 3.6Bilateral nominal exchange rates against ERM currencies


1974-78 1979-89 1979-83 1984-86 19B7-89 EMU

B/LDKDGREFIRL1NLPUK

USAJapanSwitzerland

AVIAV2AV3AV4

1,11,21,51,82,91,82,02,21,12,72,0

2,22,32,1

1,61,72,22,1

0,80,70,72,21,60,80,90,80,61,62,2

2,92,31,30,71,02,01,9

1,10,90,92,42,01,00,81,00,72,12,4

2,62,71,5

0,91,22,32,1

0,50,50,52,50,90,71,30,80,40,92,2

3,02,11,10,60,81,81,8

0,30,40,40,61,20,40,40,60,30,61,7

2,81,91,0

0,40,61,41,5

00000000000

000

1,4

Nole: AVI = Weighted average of ERM currencies, ecu weights,AV2 = Weighted average of EC currencies, ecu weights.AV3 = Weighted average of EC non-ERM currencies, ecu weights.AV4 = Unweighted average of non-ERM currencies.

ERM currencies as selected here do not include the Spanish peseta or the pound sterling.

Theoretical considerations

The main theoretical argument as to why exchange ratevariability should adversely affect trade is that risk-adverseagents will reduce their activity in an area, such as trade orinvestment for export, if the risk, i.e. the variability of thereturn they can obtain from this activity, increases. Thetheoretical literature on the effects of exchange rate varia-bility usually refers implicitly to floating exchange rates anddoes not always distinguish clearly between real and nominalexchange rate variability because, as mentioned above, forfree floating exchange rates nominal and real exchange ratevariability are equivalent. This subsection discusses thereforehow both nominal and real exchange rate variability canincrease the riskiness of international trade.

its accounting and most of its costs in the domestic currency.The exporter can eliminate this risk for himself by agreeingonly to contracts in his own domestic currency. This mightbe the reason why most trade among the major industrializedcountries is invoiced in the exporters' currency. However,this practice only shifts the risk from the exporter to theimporter who will then face the uncertainty that the price inhis domestic currency can change between the time he placesthe order and when he receives the merchandise (or service).The only way the risk can be eliminated for both partnersis through the use of forward or future markets.

The forward and future markets for foreign exchange haveindeed expanded considerably since the advent of floatingexchange rates and it is now possible to hedge exchange raterisk among the major currencies. For the simplest oper-ations, for example a simple forward contract, the directcosts involved are of the same order of magnitude as thespreads on spot transactions. However, in many cases asimple forward contract is not sufficient to cover all exchangerate risk. For example, a firm that submits an internationalbid, say for a large investment project, does not know inadvance whether it will obtain the contract. In order to beable to submit a bid in foreign currency it will have tobuy a foreign exchange option, which involves much highertransaction costs.

A further limitation of the usefulness of forward and futuremarkets for foreign exchange is that they are not completein terms of maturity and currency. Forward cover on ma-turities of up to one year are readily available in mostmajor currencies, but no developed markets exist for longermaturities. For long-term delivery contracts, for examplefor aeroplanes for which delivery lags of several years arecommon, this is a potentially important limitation. It ispossible to construct forward cover indirectly by issuing debtin foreign currency,12 but this technique involves more costsand can be used only by large firms which have access toforeign capital markets or the market for foreign currencyswaps. For firms located in small countries it is generallymore difficult to obtain forward cover since the range offorward contracts available in the less important currenciesis usually much more limited.13

The most direct channel for nominal exchange rate varia-bility to affect international trade arises because most inter-national trade contracts involve a time lag between the timethe contract is made, and when the exporter obtains hispayment. All exporting therefore involves an exchange raterisk from the point of view of the exporting firm which has

An exporter expecting a foreign currency payment in the future can becertain about the domestic currency equivalent if he sells the correspond-ing amount of foreign currency forward. He can obtain the same resultby contracting a loan in foreign currency (that matures at the time ofthe expected payment) and converting the foreign currency he obtainsagainst the loan on the spot market into domestic currency.At present currency options exist against the US dollar.

72

sduquenoy

sduquenoy


Exchange rate variability is, therefore, a more importantfactor in countries with less developed financial marketsbecause of the higher cost and limited availability of forwardcover. However, the 1992 internal market programme, whichwill open all national financial markets to competitionshould mitigate considerably this particular problem. Oncethe internal market programme has led to an integratedEuropean financial market all enterprises in the Communitywill have access to the common financial market whichshould be as efficient and complete in terms of coverage asany of the existing national financial markets.

The time lag between contract and payment represents themost direct way in which exchange rate variability, throughunexpected changes in the nominal rate, can make traderiskier. However, even if exporters and/or importers per-fectly hedge against this risk exchange rate variability willstill increase the risk of foreign trade because it introducesan additional source of uncertainty regarding the real ex-change rate that will prevail in the future. Movements of thereal exchange rate, i.e. the nominal rate adjusted for domesticcosts and the costs of the competition, will determine futureprofits, which thus can become highly uncertain.l4 This riskcan, of course, not be hedged with simple forward contracts.However, it is possible to hedge against this risk, also calledeconomic exposure, by a suitable choice of the currency inwhich to denominate the assets and liabilities of the firm.Sophisticated techniques to hedge against this economicexposure are, however, followed mostly by large firms, forwhich exchange rate changes have a more complicated im-pact on profits since both costs and revenues are often inseveral different currencies.

The risk of unpredictable large swings in real exchange rateswould, however, not exist in EMU (nor in the EMS) sincein EMU real exchange rates would move only through theslow and relatively predictable movements of overall pricelevels. EMU (and the EMS) provide therefore also a hedgeagainst the risk of economic exposure.

Financial markets offer therefore a variety of ways to obtaininsurance against exchange rate risk in international trade,but this insurance is not costless and not always available.The price of this insurance diminishes and its availabilityincreases with the degree of sophistication of financial mar-kets and the size of the firm.

For a small competitive firm costs and the market price are given andchanges in the nominal exchange rate will directly determine the realprice it can obtain for ils output. For the often more realistic case of afirm which can influence the market price the nominal exchange ratewill not directly determine the price, but will shift the entire demandschedule expressed in domestic currency.

Empirical evidence on exchange rate variability and trade

There exists a large body of empirical studies on the effectof exchange rate variability on trade. However, this literaturehas not been conclusive. A number of studies have foundno convincing effect and a major survey undertaken by theIMF (see IMF (1984)) concluded that it had not been ableto find a systematic link between short term exchange ratevolatility and the volume of international trade.

Some individual studies (De Grauwe (1987) and Peree andSteinherr (1989)) have found a significant effect of exchangerate variability on international trade. However, these stud-ies generally refer to the major floating currencies. It istherefore difficult to decide whether a similar effect operatesalso at the much lower level of exchange rate variabilityamong EMS currencies. The only two studies that concen-trate specifically on European currencies and intra-EC tradeappear to be Bini-Smaghi (1987) and Sapir and Sekkat(1989). However, even these studies find only very smalleffects. Bini-Smaghi finds a significant, but very small, effectof exchange rate variability on bilateral trade for Franceand Italy and no effect for Germany; Sapir and Sekkat findno significant effect at all.

Since the empirical research has not found any robust re-lationship between exchange rate variability and trade it isnot possible to estimate the increase in intra-EC trade thatmight derive from the irrevocable fixing of exchange rates.However, this does not imply that no such link exists. Thetheoretical discussion suggests that exchange rate variabilityshould have no impact, if exchange risk is either hedged orirrelevant to the firm because it is diversified. Neither ofthese two conditions is satisfied in reality as only a fractionof total trade is hedged through forward operations andthe available data on international portfolio diversificationsuggests that exchange rate risk is not diversified by share-holders.

One way to reconcile the lack of empirical results with theseconsiderations might be that the effect might be just toosmall to be detected in the available samples. This is sug-gested by Gagnon (1989) who simulates a theoretical modelin which exchange rate variability should have an impact ontrade. He finds that the impact is numerically very small. Inhis model the switch from the Bretton Woods to the floatingexchange rates regime in the 1970s, which was followed bya very large increase in exchange rate variability, would havereduced global international trade only by about 1 %.

The same model also predicts that, despite the small quanti-tative impact of exchange rate variability on trade, the switchto floating exchange rates might have had a considerable

73


impact on the welfare of traders.15 This result would be ableto reconcile the scientific economic literature with regularopinion surveys of business leaders, which repeatedly indi-cate that exchange rate uncertainty has adverse effects ontrade and investment. For example surveys conducted bythe Confederation of British Industry found that over halfof all the companies questioned, and all firms with less than1 000 employees, considered that exchange rate stabilitywas important to their operations.16 Similar results wereobtained in a European-wide survey conducted on behalf ofthe Association for the Monetary Union of Europe whichfound that among the advantages from a single currency'reduction in monetary fluctuations' ranked first amongeight other reasons.

3.2.3. Exchange rate variability and capitalmovements

Exchange rate variability affects not only international tradein goods and services, but also international capital move-ments. Since the determinants of short- and long-term capitalmovements differ considerably they will be discussed separ-ately.

For long-term capital movements, and direct foreign invest-ment, which is usually based on a long-term horizon, theshort-term variability of nominal and real exchange ratesdiscussed here should not be important as long as movementsup and down average out over time.17 So-called misalign-ments that persist over several years would, of course, bemore important, but such misalignments are not typicallyobserved among Community currencies.

The evidence of a direct impact of exchange rate variabilityon foreign direct investment is somewhat stronger than the

A large effect of exchange rate variability on welfare is compatible witha small effect on trade if trading is very profitable. An increase inexchange rate variability might reduce the welfare of traders consider-ably (in technical terms it might reduce the consumer and producersurplus) but if there is no good alternative use for the resources used inforeign trade the effect on trade will be small.See CBI (1989).If fluctuations in the exchange rate do not average out over time toallow exchange rates to return to some long-run equilibrium level, itcan be shown that in the presence of sunk costs even a very low levelof exchange rate variability can induce firms not to react to sizeablechanges in the exchange rate because there is still a small probabilityleft that the investment might be lost, or that the conditions might beeven better in the future. This idea implies that eliminating even a lowlevel of residual exchange rate variability might considerably increasethe elasticity with which trade reacts to the exchange rate. See Dixil(1989) for an analytical framework for this idea.

one relating to trade, as reported in Molle and Morsink(1990) which analyses exchange rate variability and intra-EC direct foreign investment. The estimates reported thereimply that EMU might increase intra-EC direct foreigninvestment considerably by reducing an important 'friction'factor.

Short-term capital movements may be more strongly affec-ted by exchange rate variability. However, it is difficult tosay what the impact of the elimination of exchange ratevariability will be on short-term capital movements. On theone hand many capital movements may actually be causedby exchange rate variability, for example if they serve tohedge against exchange rate risk in trade. Moreover, ifexchange rate variability reduces the correlation in the re-turns from investments in otherwise similar assets, to theextent that this also reduces the correlation with the marketportfolio this makes international investment more attract-ive. On the other hand, exchange rate changes also increasethe absolute risk of investing in foreign currency denomi-nated securities and might therefore deter individuals fromengaging in full international diversification. In practice,however, the extent of international portfolio diversificationis much lower than one would expect given the gain in areduction of risk that could be achieved through it by mostinvestors. But it is difficult to say whether this fact is a resultof the high transactions costs that arise often in internationalinvestment (i.e. a lack of economic union) or of exchangerate variability.

Given these conflicting theoretical arguments it is thereforedifficult to decide whether a reduction in exchange ratevariability would increase short-term capital movements andwhether any induced increase in asset trade should be re-garded as welfare enhancing.1S

3.2.4. Efficiency gains from the elimination ofexchange rate uncertainty

Given the difficulties in estimating empirically the impact ofexchange rate variability on trade and investment it mightbe more convenient to characterize the direct19 economicbenefits that may be obtained from the elimination of ex-change rate uncertainty by saying that the irrevocable fixingof exchange rates provides free unlimited hedging for alltrade and capital movements. Since the cost of hedgingvaries with the size and type of transaction (trade versus

See also Persson and Svenson (1987) which also comes to the result thatthe relationship could go either way.As mentioned above the indirect and dynamic effects of EMU arediscussed separately in Section 3.4 below.

74


investment, contract already concluded versus tender offer,etc.) the implied benefit from this 'free' hedging is difficultto evaluate. The savings for firms from the elimination ofthe need for hedging in intra-EC trade have therefore to beestimated by combining an estimate of the total, directplus indirect, costs of hedging standard international tradetransactions with information about the frequency of thevarious transactions considered. Moreover, it appears thatonly a small part of intra-EC trade is hedged in reality. Thisestimate would therefore provide only a lower limit on thereal savings from irrevocably fixing exchange rates to theextent that smaller firms do not hedge because it is tooexpensive or difficult for them.20

The irrevocable fixing of exchange rates might bring anadditional benefit by leading to the complete equalizationof interest rates. The experience in the EMS has shown thateven if exchange rates are de facto fixed for some timeinterest rates do not converge completely as long as thepossibility of exchange rate changes remains. The Dutchguilder/German mark rate provides a good example for thisphenomenon given that since 1983 the Dutch guilder has ex-post not depreciated against the DM and the exchange ratehas never moved outside a corridor of about + /- 0,5%from the average. Despite all this, three months' Euro-interest rates on Dutch guilder deposits have on averagebeen about 50 basis points higher than comparable DMdeposits. The reason for this difference is usually taken tobe the uncertainty that was created in financial markets,when in the 1983 realignment the Dutch guilder did notfollow the DM. Only an irrevocable commitment of lockingexchange rates could be expected to eliminate these residualinterest rate differentials.

These unwarranted differences in interest rates which con-tinue to exist in the present EMS can be considered equiva-lent to differences in the cost of capital which lead to welfarelosses in the traditional sense (see Chapter 2). In PriceWaterhouse (1988) it was estimated that the elimination ofthe residual differences in interest rates would lead to welfaregains of about 0,05 % of Community GDP. While the wel-fare gains that come from the elimination of interest ratedifferentials appear to be modest, the gains that can bereaped from a reduction in the overall level of interest rates,and therefore the cost of capital, can be very important asdiscussed more in detail in Section 3.4 below.

3.3. Building on 1992

The main pillar of economic union is the completed internalmarket within which persons, goods, services and capitalmove freely. The potential economic impact of the com-pletion of the internal market was estimated in the so-calledCecchini Report. Using a microeconomic approach the Cec-chini Report found welfare gains in the range of 2,5 to 6,5 %of Community GDP. Macroeconomic simulations indicatedthat these microeconomic welfare gains should translate intoa medium-term increase in GDP of 4,5 %, a decrease in theprice level of 6 % and an increase of employment of 1,5 %,i.e. almost 2 000 000 jobs. Following the Cecchini Report,the economic implications of the internal market have beenthe subject of considerable additional research efforts. Thesecomplement the original global assessment but they alsobegin to look into the regional and sectoral impact of thecompletion of the internal market.21 On a global level,theadditional research effort largely confirms the findings ofthe Cecchini Report, and goes further in analysing dynamiceffects not encompassed in the above figures (see Section 3.4below).

In some specific fields the Commission has taken or willtake initiatives going clearly beyond the 1985 White Paper.This is particularly the case for energy and transport. Recentstudies have found that the benefits from going beyond themeasures contained in the White Paper can be substantial.In the field of energy it is estimated (Commission of the EC(1989)) that the long-term gains from a free internal energymarket in electricity are ECU 11 billion as compared with theECU 6 billion estimated in the Cecchini Report. Similarly, inthe field of transport, the benefits of a greater degree ofcompetition among airlines are now estimated (McGowanand Seabright (1989)) to be about twice those found in theCecchini Report. Moreover, the cost of fragmentation ofEuropean airspace due to different air traffic control systemshas been estimated by the Association of European Airlinesand an independent group of economists at ECU 4-5 billionper annum.

The term "economic union' as used in this study goes beyondthe internal market. It also includes the assignment of certaineconomic policy functions to the Community level whereverthis is necessary to achieve all the economic gains frommarket integration, i.e. wherever there are Community-wideexternal effects. The areas where this is the case are compe-tition policy, commercial policy, R&D, human resource de-velopment, European-wide infrastructure and the environ-

20 If firms do not buy cover because exchange rate variability is irrelevantto them this estimate would not constitute a lower limit. 21 For a review of some of the additional literature see Italianer (1990).

75


ment.22 In all these areas Community policies can be justi-fied on the principle of subsidiarity, as explained in Chapter2. The increasing degree of market integration has alsounderlined the need for Community programmes of transna-tional collaboration in areas such as higher education (e.g.Erasmus and Comett), vocational training and permanenteducation, in order to stimulate the development of humanresource skills.

A Community involvement in these policy areas based onthe principle of subsidiarity yields net economic benefitsfor the Community as a whole since the Community willintervene only in those cases where the efficiency gainsdue to the Community involvement outweigh the cost ofadministration. While these benefits can be appreciated inqualitative terms, a reliable quantification is often very diffi-cult, if not impossible. First, there is sometimes no clearblueprint for the greater Community involvement, for exam-ple the political decisions on European-wide infrastructurenetworks have begun only recently. Thus the degree andintensity of future Community involvement is uncertain. Buteven where the Community has been involved in a policyarea for some time, such as R&D, no global assessment ofthe economic cost and benefits has been made so far, evenif, as part of the Monitor programme, specific researchprogrammes such as Brite (concerning new industrial ma-terials) have been evaluated and found to yield significantbenefits.

In the field of commercial policy, the Community seeks toensure that the removal of internal barriers is paralleled byprogress towards a more liberal multilateral trading system.The potential economic benefits of such progress are con-siderable and have been repeatedly demonstrated by econ-omic research. 23

In the field of environmental policy a greater Communityinvolvement can be expected to yield considerable net ben-efits. First, some global problems, such as the greenhouseeffect, require a world-wide solution, the value of whichcould be crucial even if it is not only economic in natureand cannot yet be quantified with any precision. If the

Community makes a constructive contribution towards this,the probability of securing a satisfactory global outcome islikely to be considerably greater than the chances in theevent of disparate individual contributions from the MemberStates. Secondly, a Community recognition that economicefficiency becomes more and more interrelated with ecologi-cal efficiency could provide an anchor for environmentalconcerns in the Member States, including those where suchconcerns have hitherto had a rather low priority. Thirdly,in the context of the internal market, national environmentalpolicy could be misused as an instrument to create newsegmentations of markets in Europe; this would be costly ineconomic terms and could be largely avoided through anactive Community role in this field.

In the field of competition policy the need for a Communitypolicy is apparent since only the Community is able tomonitor the internal market and make sure that competitionis not distorted by dominant market positions or State aids.The importance of surveillance over State aids can be seenby considering that, on average over the period 1986-88,aids to manufacturing in the Community amounted to about4 % of value added. With EMU the need to restrict Statesubsidies is much increased since it will no longer be possibleto compensate for differences in the overall level of aidthrough the exchange rate.

Major economic benefits can therefore be expected from animproved competition policy, in particular with respect to

'the surveillance of State aids.

Table 3.7 gives a breakdown of the average annual aidvolume by principal objectives.

In the context of the completion of the internal market,the Commission is determined to make additional forcefulefforts to restrict State subsidies. This should bring economicwelfare gains by reducing a drain on public budgets andremoving distortions of competition which are a source ofeconomic inefficiency. For the Community as a whole thesepotential benefits have not been assessed quantitatively butcountry-specific and sectoral results suggest that they arelikely to be substantial.

A case can also be made Tor Community functions in the area ofmacroeconomic policies (other than monetary policy), especially fiscalpolicy, but this is discussed in Chapter 5.For example, the negative overall effect of protection on economicgrowth has been estimated by Donges (1986) who found that protectionmay have slowed down the rate of economic growth by approximately2 % a year in the world economy as a whole, by 1,3 % for the group ofindustrial countries and by 2,3 % for the developing country group.It has also been shown that the global gains from eliminating quotasand tariffs in 14 key textiles and clothing categories exceed USD 15billion annually.

For example, a recent comprehensive study for Germany(Weiss et al. (1988)) estimated that the elimination of Stateaids would result in a gain of 0,9 % of GDP. The study usesa general equilibrium model and assumes that the budgetsavings from the dismantling of subsidies are used to reducedirect taxes. The gains estimated in this study may be con-sidered a lower bound since the model does not allow foreconomies of scale nor for a reallocation of factors of pro-duction between Community Member States.

76


Table 3.7State aids in the European Community, by main objective

Annual average 1986-88 in MECU

Horizontal objectives — TotalInnovation/R&DSMEsTrade/exportEnvironment/energy savingGeneral investmentOther objectives

Sectoral objectives — TotalSteelShipbuildingOther sectors

Regional aids — TotalLeast developed regionsOther regions

Manufacturing and services excluding transportGrand total

12581333028383239

6301 5081 036

91021 3651 5636 174

1203752526785

33720

Source: Commission of the European Communities (1990),

3.4. Indirect and dynamic gains from economicand monetary union

The previous sections of this chapter discussed the directgains in terms of hedging and transaction costs from monet-ary union and additional gains from economic union. Thesedirect gains can best be understood as increases in potentialincome with given endowments of factors of production,such as capital and labour. This section shows that thesedirect productivity gains should increase the capital stockover time and therefore lead to additional indirect gainswhich are called 'dynamic' because they arise over time asthe capital stock responds to efficiency gains.24 Moreover,this chapter also indicates how the reduction in uncertaintythrough EMU can have dynamic effects that are similar tothe ones that derive from the direct efficiency gains.

Since the internal market programme and EMU form twoparts of an interlinked system whose effects reinforce eachother mutually, it will, however, not be possible to dis-

tinguish sharply between the effects of the internal marketprogramme and the additional effects of EMU.

This section starts with a discussion of the economic effectsof market integration and then turns to an estimate of theoverall efficiency gains from EMU.

3.4.1. Integration and dynamic gains

Although most economists agree that integration of marketsbrings large benefits it is difficult to estimate them quantitat-ively within the context of the usual models of economicgrowth. In part this may be due to the large gap betweenthe formal theoretical models explaining the gains fromeconomic integration and the benefits as they are perceivedby economic agents. Most of the difTiculties arise from thefact that the standard theories of economic growth thatincorporate constant returns to scale and perfect competitiondo not leave any room for integration to affect growth inthe long run. This is explained briefly in Box 3.2. Theevaluation of dynamic gains made here is, however, basedon the standard and widely used models which do notincorporate economies of scale that lead to continuinggrowth. A class of newer models that does allow for continu-ing (endogenous) growth is briefly discussed at the end ofthis subsection.

Recent developments in the theory of international trade 25

have started to bridge the gap between theory and the waymarkets are perceived to work by recognizing that marketsare not always characterized by constant returns to scaleand perfect competition. Monopolistic competition andeconomies of scale are known to prevail in many markets,and theoretical and empirical models taking these propertiesinto account have started to develop. Nevertheless, thisapproach does not lead to essentially different conclusionsin terms of long-run growth and can therefore not be usedto explain dynamic gains. 26

'Dynamic gains] are defined here as the gains from integration takenover a period long enough for ihe capital stock to have adjusted to thenew equilibrium. In mosi economic analysis the capital stock en-compasses human as well as physical capital. This implies efficiencygains lead not only to increased investment in plant and equipment, butalso education and training.

25 See Helpman and Krugman (1985) for further references.26 As explained in Romer (1989) economies of scale lead to continuing

(endogenous) growth only if they affect the accumulation of capital.This is not the case in the models referred to in this paragraph.

77


Box 3.2: The neoclassical growth model

The standard 'neoclassical' theory of growth is based on a modelin which firms produce one (possibly composite) product withcapital and labour under constant returns to scale and perfectcompetition. Given an exogenous labour supply the steady stateor long-run equilibrium is reached when the marginal pro-ductivity of capital is equal to the discount rate of consumers.This can be illustrated using a diagram that relates the capitalstock (per capita) on the horizontal axis to output (per capita)on the vertical axis.

The curve F(k) shows how much output can be produced giventhe per capita capital stock k. The long-run level of output isdetermined at the point of tangency between the straight lineDD whose slope is given by the discount rate and the productionfunction F(k). This point of tangency determines the steady statecapital stock denoted by kss. It is apparent that this frameworkexplains only the level of steady state income (per capita).Continuing growth is possible in this framework only if pro-ductivity grows. (Growth in the labour force leads only togrowth in total output, but does not affect income per capita.)But the factors that cause growth in productivity are taken tobe exogenous to the model. In this sense this framework cannotexplain continuing growth.

The specific curve used in this graph was obtained by settingthe elasticity of output with respect to capital equal to one half•(F(k) = B.k0'5). This corresponds to a multiplier of two as inBaldwin (1990) and as assumed widely in this chapter. Forillustrative purposes it was assumed that the discount rate isequal to 10 %. The units of output can be freely chosen by the

choice of the scale factor B and have no meaning; in this graph Bwas arbitrarily equal to 10. This yields a steady state level ofoutput of 5 'units'.

The long run equilibrium in die neoclassical growth model

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

Capital (k)

The Cecchini Report, while drawing on results incorporatingeconomies of scale and monopolistic competition, did notgive an estimate of the dynamic gains to be expected fromthe completion of the internal market, although these gainswere presaged.27 Part of these dynamic gains were ascribedto the effects on business strategies of the completion of theinternal market. The buoyant investment climate in 1988and 1989 and the currently much increased activity in termsof mergers and acquisitions 28 seem to indicate that firmsare responding to the opportunities they^ expect from theinternal market even before this programme is completed.This is an example of the adaptation of business strategiesto 'news' about future economic integration.The improved business climate could of course also be dueto other factors, but a major business survey conducted inthe spring of 1989 on behalf of the European Commissionhas shown that the internal market positively influencesinvestment intentions for the period up to and after 1992.

27 See Emerson et al. (1988).28 See European Economy No 40, May 1989.

This effect of the 1992 programme on investment intentionsand growth expectations can be explained in terms of theneoclassical growth framework as a medium-run responseof the capital stock to the static efficiency gains. This impliesthat there exist theoretical arguments for the existence ofdynamic gains which are briefly explained in Box 3.3 (for arigorous exposition see Baldwin (1989)). It is shown in thisbox that if the efficiency gains from the internal market areequivalent to shifting the production function upwards, themarginal productivity of capital will increase at any givencapital stock. Starting from an initial equilibrium positionwhere the opportunity cost of capital was equal to its mar-ginal productivity this implies that firms will increase theircapital stock until the marginal efficiency of capital hasreturned to the opportunity cost of capital (which did notchange). This increase in the capital stock, until the newequilibrium level has been reached, leads, of course, to anincrease in output.

This effect does not change the long-run or steady-stategrowth path, but there will be an increase in output over themedium run. It might therefore be called, as in Baldwin

78


Table 3.8The influence of the internal market on the expectations of firms

Effect on

SalesDomesticOther ECRest of the world

InvestmentDomesticOther ECRest of the world

EmploymentDomesticOther ECRest of the world

Up 10 1992

Increase1

266

337

2828195

139

133

Neutral'

127

1013

158

1820

178

2021

From 1993 to 1996

Increase

270

3710

2623197

137

154

Neutral

20171720

22172527

25182730

Productivity 38 10 38 17

Source: European Economy No 42, November 1989.1 Increase: Difference between the percentage of firms expecting an increase and thos

ex peeling a decrease.! Neutral: Don't know/no answer.

(1990) the 'medium-run growth bonus'. On the basis ofproduction function estimates for a number of Europeancountries, Baldwin (1989) calculates that the medium-runbonus may range between 24 % and 136 % of the initialstatic gains (see Table 3.9). Since he also provides reasonsto believe that his calculations could prove to be underesti-mates, he concludes that the point estimate of the medium-run growth bonus should be close to the upper bound andthat the static efficiency gains should therefore be doubled.

Table 3.9The medium-run growth bonus as a percentage of static efficiencygains

B D F NL UK Average

Low estimateHigh estimate

38136

36129

3080

35124

2493

32,6112,4

Souce: Baldwin (1989),

It has to be emphasized, however, that the welfare impli-cations of the medium-run growth bonus are quite differentfrom those of the direct efficiency gains. The direct efficiencygains are equivalent to a gain in welfare since they justincrease the output that can be produced with given factors

of production. The medium-run growth bonus, in contrast,works through an increase in the capital stock that has tobe 'earned1 by forgoing consumption. The welfare gains (inthe sense explained in Chapter 2) from the medium-rungrowth bonus are therefore much smaller than the ones fromthe direct efficiency gains even if the medium-run growthbonus is as large as 100%. Baldwin (1989) shows that thewelfare gain (in present value terms) of the direct efficiencygain is almost 20 times higher than the welfare gain (againin present value terms) of the indirect dynamic effects.

The increase in income through the medium-run growthbonus comes through additional investment and thereforetakes time to materialize until the capital stock has reachedits new equilibrium level. The build-up of the capital stockand associated production capacity may well take severalyears. An example may illustrate this effect: with a lineardepreciation scheme, Baldwin expects half of the growthbonus to take 10 years to be realized. If the static efficiencygains of the completion of the internal market in the run-up to 1992 and in the years thereafter take about the sametime span, one may thus expect, all other things being equal,an increase in the Community growth rate of two-thirds ofa percentage point during a period of 10 years and a quarterof a percentage point thereafter.29 These would be averageincreases per annum. If firms anticipate correctly the gainsto be expected in the future they might take their decisionsin the earlier phase of the period, thereby frontloading theeffects.

As argued in Section 3.1 of this chapter exchange ratestability can lead to additional gains because it reduces theriskiness of investment to the extent that firms cannot hedgeor diversify exchange rate risk. Box 3.4 explains how areduction in the riskiness of investment can be translated intoan increase in output in the standard neoclassical framework.

It has long been recognized that the standard growth theorywith constant returns to scale was not satisfactory since itcould not explain continuing economic growth, except byinvoking exogenous technological progress. Models incorpo-rating increasing returns to scale have therefore been de-veloped recently that explain endogenous economicgrowth.30 These models are, however, not yet widely ac-cepted and must therefore be considered somewhat morespeculative.

Calculated, for the first 10 years, by taking one-tenth of 6,75%, beingthe sum of 4,5 % static efficiency gains plus half of 4,5 % dynamic gains.The other half of the dynamic gains, 2,25 %, divided by 10 representsthe average growth over the remaining period.See in particular Romer (1986); Romer (1989) provides a survey of thisapproach.

79


Box 3.3: Medium-term growth effects of efficiency gainsi

The neoclassical framework explained in Box 3.2 can be usedto explain the 'medium-term growth bonus' mentioned in thetext. If EMU increases the overall level of productivity theproduction function F(k) shifts upwards, say to AF(k), whereA is a number that exceeds one. At an unchanged capital stockthe increase in production is exactly equal to the increase inproductivity, i.e. the difference between F{kss) and AF(kss).However, the point of tangency with the discount rate line willnow shift. The new long-run equilibrium has to be at the highercapital stock, k'ss so that output increases by more than theamount of the gain in productivity.It is apparent that the assumption that the production functionis increased proportionally is crucial for this result. If the pro-duction function shifts up by a constant amount the point oftangency would remain unchanged at kss and no additionalinvestment would occur.Algebraically this can be seen most easily by considering achange in the productivity factor A. If the production functionis AF(k) any increase in A increases also the marginal pro-ductivity of capital, which is AF'(k). However, if the productionfunction is F(k) + A, the value of A does not affect the marginalproductivity of capital which would be given by F'(k). In thetext it is assumed that the former is the case, which seemsreasonable since a doubling of production of income and tradeinside the Community should also double the transactions costsof keeping different currencies.

This graph uses the same values as the preceding one. In orderto obtain a convenient graphical representation the productionfunction was assumed to shift upwards by 25 % (A goes from

0 to 12,5). Given the multiplier of two, this illustrative examplemplies that output increases by 50 %, i.e. from 5 to 7,5 'units'.

1514131211109

I 8

* I5A3210

.^^<AF(k)

: /^/!^^w

sSs^"^

1 . . ! Y , " .0 0,1 0,2 3,3 0,4 0,5 0,6 0,7 0.8 0,9 I

Capita] (10

Two of these so-called 'new growth models' are analysed byBaldwin (1989) who comes to the conclusions that the staticefficiency gains from the internal market could permanentlyincrease the Community growth rate. The first of thesemodels relies on a specific value for the elasticity of outputwith respect to capital. In a second model, any increasein the profitability of innovations permanently increasesgrowth. Calibrating these models, Baldwin finds that thecompletion of the internal market could increase the Com-munity's growth rate permanently by 0,3 to 0,9 percentagepoints. These results obviously rest on very specific assump-tions, nevertheless they are important because they providean analytical hinge for the relationship between integrationand long-term growth conditions.

It has to be emphasized, however, that the standard frame-work with constant returns to scale and the new growthmodels are not compatible in the sense that the mechanismsthat determine growth in the long run are radically different.It is therefore not possib e to add permanent growth effects(from the new models) to the level effects (from the standardframework).

3.4.2. Quantifiable dynamic gains fromeconomic and monetary union

The preceding analysis can be applied in general to thecase of economic and monetary integration. Whenever staticefficiency gains occur which raise the marginal productivityof capital, it may be expected that, in the medium term, thecapital stock and therefore output will increase to a level thatexceeds the initial shift in output. Similarly, any reductionin the riskiness of investment would also increase capitalaccumulation, and, over time, output. In order to estimatethese effects it is therefore necessary to first obtain estimatesof the static efficiency gains and the reduction in the riskpremium which form the basis to which the medium growthmultiplier can be applied. 31 This is done separately for thesetwo elements in the remainder of this subsection.

If one accepts the more speculative new growth models these directefficiency gains could also be translated into a permanent increase inlong-term growth.

80

sduquenoy


Box 3.4: Growth effects of a reduction in the risk premiumThe medium-term growth effect of a reduction in the risk ad-justed discount rate can be illustrated using the standard neo-classical growth framework already explained in Box 3.2. In thisframework the steady state level of the stock of capital (percapita) and hence of the steady state level of income (per capita)is determined by the condition that the marginal productivityof capital equals the discount rate. Although the standard neo-classical growth model does not take into account uncertaintythe discount rate which in that framework represents the purerate of intertemporal time preference of consumers could bereinterpreted as the risk adjusted required rate of return firmsface. A reduction of the risk premium would then be equivalentto a reduction in the slope of the discount rate line.

This implies that a reduction in the risk premium leads to anincrease in the capital stock from kss to k'ss and hence an increasein (per capita) income equal to the difference between F(kss)and Ffk'^). Mathematically this can also be seen by using thesteady state condition that the marginal product of capital F'(k)equals the discount rate. If the discount rate declines F(k) hasto decline as well and this can happen only if k increases.

Using once again the same production function it is assumed inthis illustrative example that the discount rate goes from 10 %to 8%, which is equivalent to a reduction of 20%. Giventhat the multiplier for this type of change is equal to one thisillustrative example implies an increase in output of 20%, i.e.from 5 to 7 'units'.

!k0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 I

Capital (k)

(i) Dynamic effects of static efficiency gains

For the static efficiency gains there are two sources sinceEMU comprises the internal market programme and theadditional benefits from monetary union: (a) The CecchiniReport estimated the efficiency gains from the internal mar-ket programme to lie between 2,5 and 6,5 % of the GDP ofthe Community; (b) Sections 3.1 and 3.2 above estimatedthe direct efficiency gains from the elimination of transactionand hedging costs to be about 0,4 % of GDP.

Applying the range of parameters for the medium-termgrowth bonus discussed in the previous section this impliesthat the total direct static plus dynamic gains from EMUshould be between 3,6 and 16,3 % of GDP with the centralestimate equal to 9,8 % of GDP. 32

The lower bound is obtained by multiplying the lower bound of thetotal static efficiency gains (2,9 %) with the lower bound of the (mediumterm growth) multiplier (1,24); the upper bound is obtained by multiply-ing the upper bound of the total static efficiency gains (6,9%) by theupper bound of the multiplier (2,4). The central estimate is based ontotal static efficiency gains of 4,9 % and a multiplier equal to 2.

By assuming the static gains to be spread out over a periodof 10 years, and half of the dynamic gains to be realized inthe first 10 years,33 the impact on the medium-term growthrate of GDP could be in the order of 0,7 percentage pointsper annum in the first 10 years and 0,25 percentage pointsthereafter.

A point estimate for the long-term effect on growth of theefficiency gains would be in the order of an additional 0,7 %per annum, with a range varying between 0,4 % and 1,0 %per annum, but this estimate is surrounded with even moreuncertainty.

As mentioned above the medium-term growth effect and thelong-term growth effect cannot be added because they arebased on two different underlying growth models. Theyshould therefore be identified with two different scenarios,one where the increase in growth eventually dies out, 34

See Baldwin (1989).The increase in growth dies out in these models. But afterwards growthcan continue at the same pace as before since the exogenous factorsthat are the underlying cause should continue to operate as before.

81

sduquenoy


and another where it has a permanent effect. Despite theirdifferent implications for the long run the two models comeapproximately to the same conclusion for an initial periodof 10 years for which they imply that growth increases byabout 0,7 % points.

(ii) Dynamic gains from a reduction in risk

However, the potentially most important source of gainsfrom EMU comes from the reduction in overall uncertaintyEMU might provide. EMU should not only eliminate ex-change rate fluctuations, which should have the direct effectsthat are discussed in Section 3.2 above, but it should alsoreduce the overall uncertainty that affects investment in theCommunity. The overall uncertainty affecting investmentmight be reduced not only through the elimination of ex-change rate movements, but also through a reduction in theuncertainty about monetary policy when there are no longerdistinct national central banks, and possibly through a morestable fiscal policy.

Baldwin (1990) analyses a number of reasons why uncer-tainty, especially exchange rate uncertainty should affectinvestment. Only the most important of these, risk aversion,is briefly discussed below. This is not to deny the otherchannels through which uncertainty may affect investment,but this particular one is the most widely known and itillustrates the general principle that the market imperfectionsthat exist in the real world may make uncertainty muchmore important than theory would suggest.

It is generally accepted that if firms are risk averse theywill reduce investment if profits become more uncertain.However, the theory of corporate finance also indicates thatin a perfectly functioning financial market firms should notbe averse to exchange rate risk because they could hedgeagainst exchange rate exposure through a variety of financialinstruments. Moreover, even if a firm cannot hedge againstexchange rate variability its investors could diversify thisrisk by holding shares of different firms, perhaps located indifferent countries.3S

The extent to which exchange rate risk will affect the riskadjusted discount rate which a firm applies to a particularinvestment project depends therefore on the degree to whichfinancial markets allow managers and investors to hedgeagainst exchange rate risk. This issue is discussed in moredetail in the Baldwin study already referred to where it isfound that in general financial markets do not seem toprovide substantial hedging against exchange rate risk, since

investors are usually not diversified internationally and asignificant proportion of trade is not hedged. This impliesthat risk, especially exchange rate risk should affect invest-ment.

There also exists, however, a broader theoretical argumentthat exchange rate variability might not affect trade andinvestment adversely after all. This argument starts from theidea that trade always represents an option; and it is a well-known principle that the value of an option increases withits variability. This basic argument in less technical termssimply means that changes in the exchange rate representnot only a risk, but also an opportunity to make profits. Afirm involved in trade can export, it does not have to, andit will do it only if it is profitable. If the exchange ratebecomes more variable the probability of very favourableexchange rates and therefore high profits increases. Theprobability of very unfavourable exchange rates also in-creases, of course, but this does not have to lead to offsettinglosses since the firm can always stop exporting.36 A highervariability of the exchange rate leads therefore to a higherprobability of making high profits, which is not matched byan equivalent probability of high losses, only somewhatlower profits. This implies that higher exchange rate varia-bility offers on average the opportunity to make higherprofits. 37

The argument that exchange rate variability might increasethe expected value of profits does not deny that it certainlymakes profits more uncertain. But it does imply that theimpact of exchange rate variability on investment is difficultto determine on theoretical grounds alone. Investment isusually taken to increase with the difference between theexpected rate of return and the risk adjusted discount rate.Since exchange rate variability affects both the expected rateof return and the risk adjusted discount rate it follows thatfrom a theoretical point of view it is not possible to decidea priori what effect a reduction of exchange rate variability

For small investors this service can be provided cheaply by investmentfunds, the so-called Ucits, which can since October 1989 be sold freelythroughout the Community.

The classic reference is Oi (1961); see also Pindyck (1982) and Gros(1987).This argument is not invalidated by the existence of adjustment costs.It is sometimes argued that exchange rate uncertainty should have anegative impact on trade because it is costly for firms to adjust theirproduction plans to changing market conditions. However, adjustmentcosts do not affect the nature of the argument that a firm will react toa change in the exchange rate only if it is profitable, taking into accountall factors, i.e. also adjustment costs. Exchange rate variability cantherefore have a positive effect on trade even in the presence of adjust-ment costs (see Gros (1987)). Adjustment costs may, however, reducethe speed with which the firm reacts to realized changes in the exchangerate, especially if they are asymmetric, i.e. if there are sunk costs. Asdiscussed above, in the presence of sunk costs even a very small degreeof exchange rate uncertainty could delay investment considerably. Butthis does not imply that there has to be necessarily less investment, itonly implies that investment might react slowly io changes in theexchange rate.

82


(and other risk) will have on investment if financial marketscan be used to hedge and diversify this risk.

Moreover, the empirical economics literature has not foundany strong relationship between investment and exchangerate variability (or other sources of risk). One of the difficult-ies in doing this is that only a fraction of total investment isdirectly affected by the exchange rate. Moreover, it is ingeneral difficult to isolate the factors that determine invest-ment.

In contrast to this lack of decisive theoretical and empiricalresults business surveys indicate consistently that businessleaders judge exchange rate variability as one of the factorsthat most inhibit investment. The results of business surveysare compatible with the available data on portfolio diversifi-cation which indicates that investors even in countries withsophisticated financial markets are generally not well diversi-fied internationally. This can be taken to imply that in realitymost exchange rate risk is not diversified. It is this non-diversified risk that should affect investment. However, ithas not been possible to detect empirically a direct linkbetween exchange rate variability and investment.

In spite of the weak empirical evidence concerning the impactof exchange rate variability on investment it is clear from atheoretical point of view that all exchange rate (and other)risk that is neither hedged nor diversified increases the riski-ness of investment.38 Empirical studies might not be able tomeasure this effect because it is in general difficult to accountfor the observed variability in investment, maybe becauseinvestment decisions have to be taken on the basis of expec-tations of the future course of a number of variables andthese expectations are difficult to measure in empirical work.EMU should, however, as argued above, reduce the uncer-tainty about some important variables, for example theexchange rate, the price level and national fiscal policies.This reduction in overall uncertainty, which goes beyondthe mere suppression of exchange rate variability, shouldincrease investment by reducing its perceived riskiness.

The 'riskiness' of investment in real capital is usually meas-ured by the so-called equity premium, i.e. the difference inthe rate of return on (risky) equity and riskless securities,such as government paper. This risk premium has beenfound to be about 8 to 10% in industrialized countries. Itis therefore an important overall component of the cost ofcapital for firms. If EMU reduces the riskiness of investmentin the Community the risk premium should fall; to estimateby how much is, however, very difficult since little is knownabout the determinants of the risk premium.

Baldwin (1990) shows that a reduction in the risk premiumyields the same medium growth bonus in terms of a higheroutput in the long run as the static efficiency gains. The onlydifference is that in this case there is no direct effect onoutput so that the medium-term growth bonus is the onlysource of gains. The gains from a reduction in uncertaintyare therefore equal to the proportional reduction in therisk adjusted rate of return times the estimated multipliersdisplayed in Table 3.9, which range from 0,24 to 1,4, withthe central estimate equal to one.

The main difficulty of this approach lies in determining thepotential reduction in the risk premium because little isknown about its determinants. Baldwin (1990) argues thata reduction of the risk premium of up to one percentagepoint cannot be ruled out, and that a drop of 0,5 percentagepoints constitutes a reasonable central estimate. The avail-able data suggest to him that the risk adjusted rate of returnlies between 5 and 10 %; using the lower value this impliesthat a 0,5 percentage point reduction in the risk premiumrepresents a drop of 10 % in proportional terms.39 It followsthat even a reduction in the risk premium of only 0,5 %could raise the GDP of the Community by 5-10 % in the longrun if the central estimate for the medium-term multiplier isapplied to a risk adjusted rate of return varying between 5and 10 %. 4° EMU could therefore lead to important gainsin output if it reduces the overall uncertainty affecting invest-ment in the Community.

3.5. Business expectations and growth

Up to now the discussion of this chapter has been basedon relatively well-established microeconomic principles. Thepurpose of this brief section is to discuss some recent research(see Baldwin and Lyons (1990)) of a more macroeconomicnature which suggests that if EMU affects the businessclimate favourably it could lead to a self-reinforcing cycleof stronger growth which would reduce unemployment con-siderably.

This research is based on the idea that due to economies ofscale investment in manufacturing is subject to forces thatdrive it to either very high or very low values. According tothis framework, a high rate of investment increases overallproductivity over time and makes additional investment evenmore profitable; this can lead to a cycle of investment that

Baldwin (1990) discusses a number of additional different 'market fail-ures' that provide channels through which exchange rate variability candistort investment decisions.

The same proportional drop could result from a reduction in the riskpremium of 1 percentage point starting from an initial value of 10 %.These are gains in terms of increased production. As mentioned abovethe gain in terms of welfare that can be reaped from the medium-termgrowth effect is much smaller than the increase in output.

83


increases until the entire labour force is fully employed. (Theconverse case of ever-falling investment leads, of course, tohigh unemployment.) This framework also suggests thatexpectations can be decisive in determining whether theeconomy experiences high investment growth or the oppositesince it is well known that expectations are an importantdeterminant of investment.

If EMU raises the expectations of business leaders thatfuture investment will become more profitable it could in-itiate a cycle of increasing investment'and lower unemploy-ment whatever the precise magnitude of the microeconomicgains discussed at some length in this chapter. Given thepersistent high unemployment in the Community, which stillstood at 9 % in 1989 it is clear that this effect could lead toeconomic gains that are of the same order of magnitude asthe efficiency gains discussed so far. However, since thisresearch is still tentative and of a qualitative nature it isnot possible to give a precise estimate of the reduction inunemployment and the increase in investment which EMUcould bring about through this channel.

Nevertheless, the interest of these newer analyticalapproaches is enhanced by the results of a business survey

of European industrialists undertaken for the present study.A sample of industrialists was asked how the business climatewould be affected (i) with completion of the single marketby 1992; (ii) by supposing in addition that exchange rateswere fixed; (iii) by introducing a single currency to add tothe single market. The results, portrayed in Graph 1.1, showthat somewhat over 80 % of industrialists view the singlemarket as having a positive impact on the business climate,but only 10 % judge it to be very positive. Adding fixity ofexchange rates changes the result only a little, perhaps be-cause the industrialists may not perceive this to be verydifferent to the existing EMS. On the other hand, the singlecurrency is rated as a much more important development,and in this case there is a large number (45 %) judging thelikely impact on the business climate to be very positive.

Taken together with the experience of the already moredynamic business environment in the run-up to 1992, it seemsjustifiable on grounds both of theory and of expectations ofindustrialists to consider a significant growth bonus fromEMU, while unproven, to lie well within the bounds ofplausibility.

84


References

Akerlof, G., Yelen, J. (1989), 'Rational models of irrationalbehaviour', American Economic Review, May, pp. 137-142.

Artis, M. (1989), 'The call of a common currency', in Europewithout currency parities, Brittan, S., and Artis, M., SocialMarket Foundation, London, 1989.

Artis, M., Taylor U. (1988), Exchange rate, interest rates,capital controls and the European Monetary System: assess-ing the track record, Chapter 7: 'The European MonetarySystem', Giavazzi, F., Micossi, S., and Miller, M., etc., USA,1988.

Baldwin, R. (1989), The growth effects of 1992', EconomicPolicy, October 1989, pp. 248-281.

Baldwin, R. (1990), 'On the microeconomics of the Europeanmonetary union', in European Economy — 'The economicsof EMU', special issue, 1990.

Baldwin, R., Lyons, R. (1990), 'External economies andEuropean integration: the potential for self-fulfilling expec-tations', in European Economy — "The economics of EMU',special issue, 1990.

BEUC - Bureau europeen des unions de consommateurs(1988a), Transferts d'argent a 1'interieur de la CEE', April.

BEUC - Bureau europeen des unions de consommateurs(1988b), 'Holiday money', July.

Bini-Smaghi (1987), 'Exchange rate variability and tradeflows', mimeo, University of Chicago and Banca d'ltalia,1987.

Black, S. (1989), 'Transaction costs and vehicle currencies',working paper, International Monetary Fund, November.

Boyd, C, Gielens, G., Gros, D. (1990), 'Bid/ask spreads inthe foreign exchange markets', mimeo, Brussels, February.

Commission of the European Communities (1989), Thebenefits of integration in the European electricity system',DG XVII, December.

Commission of the European Communities (1990), 'Secondsurvey on State aids in the European Community', Brussels.

Confederation of British Industry (1989), European monetaryunion: a business perspective, London, November.

Cukierman, A. (1990), 'Fixed parities versus a commonlymanaged currency and the case against "Stage II'", Ministryof Finance, Paris, 21 June.

De Grauwe, R. (1987), 'International trade and economicgrowth in the European Monetary System', European Econ-omic Review, No 31, pp. 389-398.

De Jonquieres, G. (1990), 'Counting the costs of dual pricingin the run-up to 1992', Financial Times.

Dixit, A. (1989), 'Entry and exit decisions under uncertainty',Journal of Political Economy, 1989, Vol. 97, pp. 620-630.

Donges, J. (1986), 'Whither international trade policies?Worries about continuing protectionism', Institut furWeltwirtschaft, Kiel.

Dornbusch, R. (1990), 'Problems of European monetaryintegration', Massachusetts Institute of Technology.

Emerson, M., et al. (1988), The economics of 1992', Euro-pean Economy, No 35.

European Economy (1990) — The economics of EMU',special issue.

Gagnon, J. (1989), 'Exchange rate variability and the levelof international trade', Board of Governors of the FederalReserve System, International Finance, Discussion PaperNo 369, December 1989.

Gros, D. (1987), 'Exchange rate variability and foreign tradein the presence of adjustment costs', Working PaperNo 8704, Departement des sciences economiques, UniversiteCatholique de Louvain, Louvain-la-Neuve.

Gros, D. (1989), 'On the volatility of exchange rates — Atest of monetary and portfolio models of exchange ratedetermination', Weltwirtschaftliches Archiv, June 1989, pp.273 - 295.

Helpman, E., JCrugman, P. (1985), Market structure andforeign trade, MIT Press, Cambridge, May.

International Monetary Fund (1984), 'Exchange rate varia-bility and world trade', Occasional Paper No 28, IMF.

Italianer, A. (1990), '1992 hype or hope: a review', EconomicPapers, No 77, Commission of the European Communities,February.

McGowan, F., Seabright, P. (1989), 'Deregulating Europeanairlines', Economic Policy, No 9, Cambridge, October.

85


Meese, R., Rogoff, K. (1983), 'Empirical exchange ratemodels of the seventies: Do they fit out of sample', Journalof International Economics, 1983, pp. 3-24.

Molle, W., Morsink, R. (1990), 'Direct investments andmonetary integration', European Economy — The econ-omics of EMU', special issue.

Mussa, M. (1986), 'Nominal exchange rate regimes and thebehaviour of real exchange rate: evidence and implications',Carnegie Rochester Series on Public Policy (1986), pp. 117-214.

Oi, W. Y. (1961), 'The desirability of price instability underperfect competition'. Econometrica, Vol. 29, No 1, January,pp. 58-64.

Peree, E., Steinherr, A. (1989), 'Exchange rate uncertaintyand foreign trade', European Economic Review, No 33, pp.1241-1264.

Persson, M., Svenson, L. (1987), 'Exchange rate variabilityand asset trade', presented at the Conference on exchangerate variability, Toronto, September 1987.

Pindyck, R. S. (1982), 'Adjustment costs, uncertainty andthe behavior of the firm', American Economic Review, No72, June 1982, pp. 415-427.

Poloz, S. (1990), 'Real exchange rate adjustments betweenregions in a common currency area', mimeo, Bank of Can-ada, February.

Price Waterhouse (1988), The cost of 'non-Europe' in financialservices, Commission of the European Communities.

Romer, P. (1986), 'Increasing returns and long-run growth',Journal of Political Economy, No 94, October, pp. 1002-1037.

Romer, P. (1989), 'Increasing returns and new developmentsin the theory of growth', NBER Working Paper No 3098,September.

Sapir, A., Sekkat, K. (1989), 'Exchange rate variability andinternational trade: the effects of the European MonetarySystem', mimeo, Universite libre du Bruxelles.

Ungerer H., Evans, O., Mayer, T., Young, P. (1986), 'TheEuropean Monetary System: recent developments', Oc-casional Paper No 48, International Monetary Fund, Wash-ington DC, December,

Weiss, F.D., et al. (1988), Trade policy in West Germany',Kieler Studien, No 217, Tubingen.

86

sduquenoy

Chapter 4. Benefits of stable prices

Chapter 4

Benefits of stable prices// is generally agreed that the EuroFed which determinesmonetary policy in EMU should aim at price stability. Thefirst section of this chapter argues that this choice is based onsound economic criteria because price stability brings econ-omic benefits. The second section then briefly discusses thecosts of reaching price stability, i.e. the cost of disinflation.Finally, the third section then discusses what kind of monetaryregime is most likely to yield the benefits of stable prices.

The main findings of this chapter are:(i) Inflation involves substantial costs that are, however,

difficult to measure. The nature of these costs differsbetween anticipated and unanticipated inflation.

(ii) Standardmicroeconomic theory suggests that anticipatedinflation of 10 % leads to direct welfare losses that are ofthe same order of magnitude, about 0,3 % ofGOP, as thedirect transaction costs savings through EMU.

(Hi) The post-war macroeconomic experience of the indus-trialized world suggests that, on average, high inflationcountries have a higher unemployment rate and a lowerper capita income.

(iv) High inflation is usually associated with highly variableinflation rates and therefore also with unanticipated in-flation. Since unanticipated inflation can affect outputtemporarily this explains why countries with higher in-flation have also on average more unstable growth rates.

(v) The costs of disinflation are minimized if there is acredible commitment to stable prices and backward-looking wage indexation is abolished or reformed into aforward-looking scheme.

(vi) A stable and credible monetary regime requires an inde-pendent central bank with the statutory mandate toguarantee price stability. Otherwise the public mightexpect that the authorities might be tempted to usesurprise inflation to temporarily increase output or tem-porarily reduce the real interest rate on public debt.These anticipations can lead to higher inflation and makeit difficult to reduce inflation. This idea is supported byempirical experience which shows that there is a stronglink between central bank independence and inflation inthe long run.

4.1. The cost of inflation

The economic benefits from stable prices, or rather theeconomic costs of inflation, have been discussed extensively

in the literature. This discussion has not come to any definiteresults concerning the overall analytical basis on which tomeasure the costs of inflation as summarized in a well knownstudy on the costs of inflation:'

'It is well known that the costs of inflation depend on thesources of the inflation, on whether and when the inflationwas anticipated, and on the institutional structure of theeconomy. There is, therefore, no short answer to the questionof the costs of inflation. Further, since the inflation rate isnot an exogenous variable to the economy, there is somelogical difficulty in discussing the costs of inflation per serather than the costs and benefits of alternative policychoices.'

Despite these difficulties in establishing an overall analyticalbasis on which to measure the costs of inflation there is wideagreement that inflation does cause considerable costs. Thischapter discusses the main sources of economic costs ofinflation and illustrates their likely magnitude with the helpof some simple, but forceful, statistical indicators. It doesnot, however, try to present an overall evaluation of the costsof inflation because these costs come from many differentsources and are therefore not comparable, as suggested inthe above quotation. For the purpose of the discussion theterm 'inflation' is taken to mean sustained increases in thegeneral price level. A once-and-for-all jump in the price levelshould therefore not be regarded as inflation.

The quote above also suggests strongly that the economiceffects of inflation are different if it is anticipated than if itcomes as a surprise. The discussion in this chapter willtherefore distinguish between anticipated and unanticipatedinflation. The effects of anticipated inflation are discussedfirst, because they can be more precisely determined.

4.1.1. The effects of anticipated inflation

In discussing the effects of anticipated inflation it is con-venient to distinguish between the effects that are based ongenerally accepted microeconomic theory and the macro-economic effects that are not as well established.

At the microeconomic level the main effects are;(i) inflation reduces the demand for money,(ii) it forces economic agents to obtain additional balances

from the central bank in order to keep the real valueof their money holdings constant, and

(iii) it forces producers to change price lists continually.

Fischer(198l), p. 5.

87


The first effect is an immediate implication of generallyaccepted microeconomic theory and the resulting economicwelfare losses can in principle be estimated. The secondeffect is also called the inflation tax; it constitutes the onlypotential argument for a positive rate of inflation. The directcosts of changing prices, called 'menu costs', are also aclear consequence of inflation, but their importance is moredebatable.

At the macroeconomic level it is less clear on theoreticalgrounds whether anticipated inflation should have any ef-fects, and it is therefore also much more difficult to estimateempirically the macroeconomic costs of inflation.

The three abovementioned effects are now discussed in turn.Additional effects that come from non-indexed tax systemsor non-indexed contracts, which lead to changes in incomedistribution, are discussed under the third heading since theyare all related to the cost of changing contracts.

Suboptimal money holdings

The earlier contributions concerning the costs of inflationfocused on the reduction in money holdings it causes. Themain argument in this line of thought is that the social costof producing money can be taken to be zero because thecost of printing additional bank notes is negligible. It followsthat an economic optimum is attained only if the privatecost of holding money is also equal to zero. However, if thealternative to holding money is to buy storable consumptiongoods, the private opportunity cost of holding money is notzero, but equal to the rate of inflation. If the alternative toholding money is holding bonds, or other 'near money'assets that yield interest, the private opportunity cost ofholding money is equal to the rate of interest. Dependingon what the alternative to holding money is taken to be, theoptimal inflation rate would therefore be either zero ornegative.2

The welfare losses from a rate of inflation that is differentfrom the optimal rate can then be calculated, according tothe standard 'triangle' methodology, as the area under themoney demand curve. The general idea behind this approachis discussed in Chapter 2 of this volume. In the presentapplication the source of the welfare loss lies in the lowerbalances households and firms hold if inflation goes up.With lower balances available to finance their regular flowof expenditures, households and firms have to incur more

often the cost of going to a bank to obtain additional money.These more frequent transactions with banks imply a lossof time and expense for households and firms and they alsoimply higher costs for banks which have to execute moretransactions. These transaction costs determine the slope ofthe demand curve for money and the 'triangle' area belowthe curve therefore represents the additional transactioncosts caused by inflation. The only input required to measurethese costs are the two variables that determine the slopeand position of the money demand curve, i.e. estimatesof the elasticity of money demand and of the velocity ofcirculation.3 These two variables can, however, not be esti-mated without first determining the proper definition ofmoney in this theoretical context. This is done in Box 4.1,which shows that this theoretical framework leads to theconclusion that a 10% rate of inflation would cause awelfare loss of 0,1 to 0,3 % of GDP depending on whetherthe appropriate definition of money is taken to be the monet-ary base (cash plus required reserves of commercial banks)or Ml (cash plus sight deposits). This effect alone is of thesame order of magnitude as the transaction costs that canbe saved by a common currency and indicates the importanceof stable prices.

Inflation tax

The concept of economic welfare loss from inflation used sofar is based on the assumption that inflation is the onlydistortion in the economy. It assumes in particular that thegovernment can finance its expenditure through lump sumtaxes. This is a crucial condition because a lower inflationrate implies also a lower 4 revenue from the inflation tax andin reality most taxes cause distortions since they are notlump sum. The literature on the optimal inflation tax there-fore argues that inflation should just be viewed as any othertax that causes distortions and should therefore be used tosome degree alongside other taxes. However, the existenceof lump sum taxes represents only a sufficient condition thatmakes it possible to ignore public finance considerations.Even admitting that in reality no lump sum taxes exist doesnot necessarily lead to the result that the optimal inflationrate is positive. The large amount of literature on this issue 5

has not come to any definite results; the theoretical issueconcerning the optimal inflation rate in the absence of lumpsum taxes therefore remains open.

The latter result is based on the requirement that the nominal interest bezero; see Friedman (1969).

See, for example, Lucas (1981), p. 43.This is true for low to medium rates of inflation. Most money demandresearch estimates that for inflation rates in excess of 100% p.a. furtherincreases in inflation lead to lower seigniorage revenue because the taxbase, i.e. money demand, becomes very sensitive to inflation.See Spaventa (1989) for a recent survey.

88


Box 4.1: The triangle welfare cost of inflation

It is usually assumed that money refers to a means of paymentand that the main alternative to money is interest bearing assets.This implies that the proper definition of money in this contextshould be the sum of all non-interest bearing assets, or all assetson which interest payments are restricted to below-market rates.Cash and required reserves on deposits with commercial banks,i.e. the monetary base, would therefore certainly have to beincluded in this definition of money. Bank accounts would haveto be included only if interest payments on them are restricted.Since the degree to which interest is paid on bank accountsvaries at present considerably from country to country it isdifficult to measure exactly this definition of money for theentire Community. However, by using the monetary base it isstill possible to obtain a lower limit for the welfare loss.

Following Lucas (1981) the welfare loss as a percentage of GDPcan be based on a standard money demand function and isapproximately equal to 0,5*(b/v)p2 where b is the semi-elasticityof money demand and v is velocity at the optimal inflation rate.If money is taken to be the monetary base the average velocity

of circulation in the Community would be about 10. Recentestimates of money demand for the EC (see Bekx and Tullio(1989) and Kremers and Lane (1990)) give an interest rateelasticity of about 2. A conservative estimate of the welfare lossfrom a 10 % rate of inflation in the Community would thereforebe 0,5*(2/10)*0,01, or about 0,1 % of Community GDP. Anupper limit for the welfare loss can be obtained by using Ml,i.e. cash plus sight deposits, as the definition of money. Sincethe average velocity of circulation of Ml in the Community isabout one third of that of the monetary base the upper limit forthe welfare loss would be about 0,3 % of Community GDP.

These two estimates rely on the present average velocity inthe Community, which hides large differences among membercountries. It can be expected, however, that with EMU thesedifferences will to a large extent disappear. National reserverequirements will be abolished and it is not clear whether theEuroFed will impose a uniform Community wide reserve coef-ficient. This alone would imply that the average velocity of themonetary base should increase. A further reason to expectvelocity to increase in EMU is that modern payment instrumentswill be adopted also in countries which at present have aninefficient banking system. This suggests that the lower estimatemight be closer to the true value under EMU.

It has been suggested,6 however, that as a practical matterthe inflation tax is important in a number of Communitycountries so that convergence to a low inflation rate in EMUwould at least imply a certain loss of revenue for thesecountries. The revenue loss some member countries wouldexperience if their inflation rates go to th*e level experiencedby the more stable countries is calculated in Chapter 5. It isfound there that the loss would be minor for most countries,except possibly Portugal and Greece, where it might beabout I % of GDP. It should be apparent, however, thatthis loss of revenue is not a welfare loss. The revenue losscan be made up through lower expenditures and highertaxes. Only to the extent that this is impossible would therebe any welfare loss. Chapter 5 on fiscal policy issues discussesthis in more detail.

Menu costs

Even a completely anticipated inflation forces all economicagents to revise their prices frequently. Just to change pricelists, i.e. 'menus', should not be very costly in itself. Thesedirect costs of inflation, called menu costs, are thereforeusually taken to be negligible. However, some recent theor-etical contributions suggest that even if the direct costs arevery small they can still have sizeable indirect effects on

production and prices.7 The reason for this result is thatsmall direct costs can be magnified if they interact withother important distortions that exist somewhere else in theeconomy. For example, if the marginal productivity of lab-our is far above the wage rate (maybe because of incometaxes), even a small fall in employment due to the menucosts can have a large cost for the economy. However, atthe present stage of knowledge it appears that it is notpossible to quantify these indirect effects. 8

The expression 'menu costs' refers to the cost of changingprice lists. This cost might be considered negligible in general,not only for the private sector, but also for the publicsector since the cost of changing for example the income taxschedule should also be trivial. In reality, it seems, however,that these costs are substantial in the area of taxation. Inmost countries, income tax rates are progressive and notindexed; moreover, in the computation of the income ofcapital all nominal interest earnings are considered incomealthough a part of them constitutes a reimbursement of theprincipal in real terms. This implies that any positive in-flation rate increases the effective rate of taxation of labour

See notably Dornbusch (1988).

7 See Akerlof and Yeelen (1989).8 Benabou (1989) contains an attempt to quantify the welfare costs of

anticipated inflation based on a framework in which discrete price adjust-ments lead to speculation in storable goods.

89


and capital over time. It is difficult to say a priori whetherthis increase in tax revenue constitutes a cost or a benefit.To the extent that it increases a distortion it should beclassified as a cost. However, to the extent that the govern-ment can use the revenue productively it should be classifiedas a benefit.

Macroeconomic effects

The macroeconomic effects of anticipated inflation havebeen thoroughly discussed in the literature. Macroeconomictheory of the 1960s held that there was a stable relationshipbetween inflation and unemployment (the so-called Phillipscurve) so that the authorities could attain and maintain anydesired level of unemployment by choosing the appropriateinflation rate. This view of the world is no longer acceptedon theoretical and empirical grounds. On theoretical groundsthe main objection is that it assumes that economic agentsnever learn about inflation. On empirical grounds the 1970s,which had on average higher inflation and higher unemploy-ment, showed that there was no stable trade-off betweeninflation and unemployment.

This observation has been confirmed by a large body ofempirical literature. Graph 4.1 visualizes the same resultby displaying a scatter diagram of average inflation andunemployment rates for OECD countries over the 15-yearperiod 1970-85. It is apparent that higher inflation is notassociated with lower unemployment in the long run. Onthe contrary, the regression between the two variables yieldseven a statistically significant positive coefficient which im-plies that, at least over the 15-year period considered here,countries with higher inflation had also, on average, higherunemployment. The point estimate of the inflation coef-ficient suggests that a one percentage point higher inflationrate was associated with a 0,3 percentage points higherunemployment rate.9

This does not imply that the Phillips curve necessarily has apositive slope, but it does show that in the long run higherinflation is certainly not associated with lower unemploy-ment.

Further evidence that inflation does not have positive effectsis provided by Graph 4.2 which depicts the relationshipbetween inflation and per capita GDP. It is apparent thatthere is a strong negative relationship between income

GRAPH 4.1: Unemployment and inflation in OECDcountries, !9*'0-85

1970-85 - OECD countries (excluding Yugoslavia and Iceland)

8 12 16 20 24

Ave-ageCPl inflation (% p.a.)

Source OECD. daia available f'om 1970Regression resuli Unemployment rale = 1,6 + 0.33 inflation

(-1.82)T -statistic in parentheses R squjred - 0,41, 2) D.o.F

28 32

It is apparent from Graph 4,1 thai the outlier in the upper right-handcorner strongly influences the regression result. The equation was there-fore re-estimated without this outlier (Turkey). The coefficient was stillpositive, but the t-statistic fell to 1,63.

and inflation. This is what one would expect if inflationreduces the overall efficiency of the economy. Although therelationship is very strong, it cannot be regarded as proofthat inflation reduces the efficiency of the economy sostrongly that differences in inflation can account for theobserved differences in per capita income. Nevertheless, thisstage suggests that countries with higher inflation on averagedo not perform better than countries with low inflation.

Other economic theories of the 1960s held that inflation wasnecessary to achieve growth in actual output. This might betrue even if there is no relationship between the unemploy-ment rate (which represents only the gap between potentialand actual output) and inflation, as suggested above. How-ever, there is now wide agreement even at the theoreticallevel that inflation does not increase growth. This is alsosuggested by Graph 4.3 which displays average growth rates(of real per capita GDP) and inflation rates (of the consumerprice index (CPI)) for all OECD countries for the period1955-85. Inspection of this scatter diagram suggests thatthere is no systematic relationship between the two variables.A regression confirms this impression since it shows thatthere is no statistically significant relationship betweengrowth and inflation.

90

sduquenoy


GRAPH 4.2: Real GDP per capita and inflation in OECDcountries, 1975-85

1,3,1.2 i-t.lI

0.90,8

0.7

0.60.5

0.40.3

0.20.1

0

AH OECD countries. 1975-85

20 40

Average inflation 1975-85

GRAPH 4.3: Inflation and growth of real GDP per capita

6,5

6

5,5

5

4,5

4

3,5

3

2,5

2

1.5

All OECD countries. 1955-85

7 ° 9 11 13 15 17 19

Average CPI inflation (% p.a.)

21 23

ftaunr IPS.Note except for the Netherlands. Ani-ina. Italy and Yugoslavia, the period of obser-vation is 1955-K5Regression result. Growth - 2.K * (1,025 inflation

10.67)T-tdtistit m parentheses R squared := 0.02, 25 D.o F

Theoretical considerations therefore suggest that in the longrun, i.e. when expectations have had time to adjust andcorrectly anticipate inflation, inflation does not yield anymacroeconomic benefits in terms of unemployment orgrowth. On the contrary, some evidence suggests that higherinflation is associated with higher unemployment and lowlevels of real per capita income and that there is no significantcorrelation with growth.

4.1.2. The effects of unanticipated inflation

Up to this point the discussion has focused on the effects ofa steady rate of inflation that is entirely predictable. How-ever, in reality inflation is never constant and therefore neverentirely predictable. Moreover, economic theory suggeststhat surprise inflation has much stronger adverse economiceffects than anticipated inflation. This is true in particularfor the macroeconomic effects and the effect on governmentrevenues. The potential welfare losses from a highly variableinflation rate are therefore much larger than the losses thatcould result from a high, but stable, inflation rate. Moreover,unanticipated inflation also tends to create variability inrelative prices which should have similar effects to variability

in exchange rates. A detailed discussion of the various chan-nels that link inflation, the variability of inflation and thevariability in relative prices is set out by Cukierman (1983).

This subsection therefore first presents some evidence aboutthe link between inflation and its variability. This evidencesuggests that in reality high inflation is usually linked tohighly variable — and hence also highly unpredictable —inflation. Having established this link it then discusses themain channels through which unanticipated inflation causeseconomic costs.

The available evidence suggests that high but stable inflationrates are very rare because there is a strong link between thelevel of inflation and its variability. One reason for thismight lie in the stop-and-go policies that are discussed brieflybelow. This link has been well documented in the economicsliterature 10 and is visually apparent in Graph 4.4 whichshows the average and the standard deviation of inflation forall OECD countries between 1955 and 1985. The existence ofa strong link between the average and variability, asexpressed by the standard deviation, of inflation is also

10 See, for example, Cukierman (1981).

91


GRAPH 4.4: Inflation and its variability in OECD countries

Average inflation rate 1955-85

Soum IPSRegression result: Standard deviation of inflation := 1.N + 1,29* average inflation

rate (18.2)T-siatistic in parentheses, 21 D.o.F

confirmed more formally by the regression result reportedbelow Graph 4.4. The estimated coefficient suggests that aone percentage point increase in the average inflation leadson average to a 1,3 percentage point increase in the standarddeviation of inflation. " These results suggest that if inflationis high on average it also becomes largely unpredictable.

The remainder of this subsection discusses the main channelsthrough which surprise inflation causes welfare losses.

Surprise inflation tax

Unanticipated inflation has much stronger effects on govern-ment finances than anticipated inflation which worksthrough the inflation tax. The base for the inflation tax isthe monetary base, i.e. cash and required reserves which areoften remunerated at below-market rates. However, the basefor the surprise inflation tax is the entire stock of governmentdebt that carries a fixed nominal interest rate because the

It is clear that on a theoretical level variability and predictability aretwo separate concepts. However, as shown for example in Cukierman(1981). an increase in the variability of inflation is usually equivalent (toa reduction in its predictability.

real value of that debt is reduced by a surprise inflation.Since the total debt in almost all countries of the Communityexceeds the monetary base by a large multiple, a 1 % surpriseinflation yields a much higher government revenue than a1 % inflation that is anticipated (see Chapter 5 for a dis-cussion of the public finance effects of inflation).

However, the large potential revenue effect of surprise in-flation cannot be considered beneficial. To begin with, itcannot, of course, be reaped in the long run since financialmarkets anticipate the incentive for the government to use itand interest rates will then contain an inflation risk premium.This implies that the mere possibility that the governmentmay use the surprise inflation tax leads to higher interestrates even if the government does not intend to create anysurprise inflation. The higher interest rates that result fromthis situation might be considered the cost of weak credi-bility. This cost should be most important for countries thathave a high public debt and a poor inflation record.

Menu costs

In principle the menu costs from unanticipated inflationshould be similar to the ones from anticipated inflation. Thisimplies that they should correspond to the costs of printingnew price lists (i.e. 'menus') or new income tax schedules.However, there might still be substantial indirect effects,especially through a non-indexed tax system which can leadto considerable distortions in the price system. However,these indirect consequences cannot be measured quantitat-ively.

Macroeconomic effects

Modern macroeconomic theory holds that unanticipatedinflation can temporarily increase output and employmentabove the equilibrium level (i.e. a short-run Phillips curvedoes exist). However, since the reverse holds when inflationis lower than anticipated it follows immediately that a highlyvariable inflation rate keeps output continuously away fromthe equilibrium level and thus leads to losses.

The discussion of the effects of surprise inflation on publicfinance and output suggests that the authorities may perceivean incentive to use surprise inflation to reduce the real valueof their public debt and to increase output and employment.Modern economic theory assumes, however, that the publicanticipates this so that the government cannot obtain anyadvantage from an inflationary policy. However, even if theauthorities realize that the public anticipates their actionsthey will still create inflation because otherwise inflationwould be much lower than expected, and temporary unem-

92

sduquenoy


ployment might result. To reach a situation in which pricesare stable and expected to remain so therefore requires thatthe public can trust that the authorities will not give in tothe temptation to use surprise inflation.12 In other words,this raises again the issue of 'credibility', which is taken upin Subsection 4.3 which discusses the requirements for acredible anti-inflationary monetary regime.

A further reason to believe that inflation involves macro-economic costs is that in most countries with high inflationthe authorities periodically try to suppress it with tightmonetary and fiscal policies. However, as these policiesare often not maintained, but reversed when they causeunemployment, a cycle of 'stop-and-go' policies can develop.Since this cycle can never be predicted with accuracy itimplies that on average high inflation rates may be associatedwith periods of disinflation and (re)accelerating inflation.This involves considerable macroeconomic costs becausedemand and output are destabilized.

This idea is borne out to some extent in the evidence pre-sented in Graph 4.5 which relates average inflation to thevariability of output growth.

See Barro and Gordon (1983) for an analytical description of how aninflationary equilibrium can arise even if only unanticipated inflationaffects output.

GRAPH 4.5: Inflation and variability of output growth forOECD countries, 1955-85

5

4,5

4

3,5

3 -

1rft l t

3 5 7 9 11 13 15 17 19 21 23

Average CP1 inflation (% p.a.)

Sour ft IPSNote the observed period is 1956-85 for Ihe Netherlands and 1%I>85 for Austria andYugoslavia (earlier years unavailable).Regression result Standard deviation of growth 2.1 •+• O.OH inflation

13,52)T-slaHsue in parentheses R-squared - 0.35. 23Do.F

This graph suggests that higher inflation indeed tends to beassociated with a higher variability of real growth. Giventhe high correlation between average inflation and its varia-bility it is difficult to decide whether this relationship comesfrom the level of inflation or its variability. Several of theeffects discussed in this subsection, as well as the previousone on anticipated inflation, could be the cause for thisresult.

Relative price variability

Higher inflation and therefore also higher variability ofinflation has been shown to lead to higher variability inrelative prices as well.13 This higher variability of relativeprices, which is not justified by changes in demand andsupply conditions, causes welfare losses in much the sameway as excessive variability of exchange rates. Estimates forGermany suggest that a 1 % increase in the variability ofrelative prices lowers potential output by about 0,3 %.14

The arguments of this subsection indicate that inflation isalso costly because high inflation rates are associated withhighly variable inflation and a higher variability of outputgrowth. Moreover, an unpredictable inflation rate is undesir-able because it makes real interest payments on governmentdebt unpredictable.

4.2. The cost of disinflation

While it is widely accepted that inflation is costly it isoften argued that inflation should not be reduced becausedisinflation also involves costs. l5 Modern macroeconomictheory argues that these costs arise for two reasons: (i) wagerigidity; and (ii) lack of credibility in the adjustment process.Box 4.2 visualizes the adjustment path during disinflationthat might be caused by either of these two reasons.

See Cukierman (1983).See Neumann and von Hagen (1989).This consideration does not really apply to EMU, because convergenceto price stability should already be substantially complete in Stage I.However, as discussed further in Chapter 8 of this volume, the cost ofdisinflation might be an important consideration for some countriesduring Stage I.

93

sduquenoy

sduquenoy


Box 4.2: Disinflation and the short-run Phillips curve \Ii

The adjustment path from a high to a low inflation rate can bevisualized as in the adjacent diagram. The two negatively slopedlines denote two short-run Phillips curves that may be thoughtof as being the result of either a different initial nominal wagerate (effect (i) or different inflationary expectations of the public(effect (ii)).

Starting from the initial high inflation rate p (associated withthe natural rate of unemployment because it is assumed to havepersisted for some time) any reduction in inflation leads to anincrease in unemployment because it involves a movement alongthe short-run Philips curve PP. However, as time passes, the rateof growth of nominal wages (and/or inflationary expectations)adjusts downward, implying that the short run Phillips curveshifts downwards, say to PP'. In the long run the economyeventually will settle down again at the natural rate of unemploy-ment, but with the lower inflation rate.

Since the position of the short-run Phillips curve depends onthe inflation rate wage earners expect to prevail in the futurethis picture can also be used to illustrate the effect of incomepolicies which can be thought of as lowering inflationary expec-tations 'by decree'. A successful income policy would reduceincreases in wages and other income to the rate that is compatiblewith the lower inflation, thus allowing the economy to movedirectly to a lower inflation rate without transitional unemploy-ment. In terms of the picture this means that income policy

would shift the short-run Phillips curve downwards at the sameime as actual inflation falls, thus leading to a vertical drop innflation from p to p'.

n flat i on

P

P's ^\PP

^\PF

Unemployment_________ ____________ ______ ._ ___

(i) It is widely accepted in the economics literature thatrigidities in the wage-setting process can create a short runtrade-off between inflation and output. This implies thatdisinflation can be associated with temporary losses of out-put. One example of wage rigidity is backward-looking wageindexation schemes which lead to a rise in real wages ifinflation diminishes. The rise in real wages reduces demandfor labour and hence causes unemployment (and losses ofoutput). Lack of credibility in the adjustment process canalso lead to output losses because if the decline in inflationis not anticipated nominal interest rates remain high. Withactual inflation low this implies that realized real interestrates are high. This in turn increases the debt service burdenon the public debt and might depress investment.

Although the precise reasons for wage rigidity vary fromcountry to country, experience shows that most episodesof disinflation coincide with considerable, but temporary,unemployment. However, the precise value of the 'sacrificeratio', i.e. the output loss per percentage point reduction ininflation, is not a constant. It changes over time and it isdifferent across countries. For this reason alone it would be

extremely difficult to estimate quantitatively the output costof disinflation.16

(ii) An even more important reason why it is impossible toestimate the cost of disinflation is that economic theorysuggests that it depends decisively on the policy regime,especially the exchange rate regime, under which it occurs.In the absence of an exchange rate commitment the mereannouncement of a disinflation programme might not becredible and therefore lead to large losses if it is actuallyimplemented. In contrast, a credible commitment to a zoneof monetary stability, such as the EMS, should reduce outputlosses, because it should reduce nominal interest rates andshould also lead to a faster adjustment in wage setting.17 It

16 The seigniorage loss from disinflation is discussed in detail in Chapter 5.17 Under the assumption that institutional rigidities, such as backward-

looking wage indexation, are remedied during the disinflation period.

94

sduquenoy


is often argued that an exchange rate target in the EMS issuperior to other forms of domestic commitments (such asa commitment to a low rate of monetary expansion) becausein the EMS no country can change its exchange rate withoutthe consent of its partners and because the cost of deviatingfrom price stability comes immediately in the form of a realexchange rate appreciation that reduces output and demand.

It has been argued that this effect has allowed some EMSmember countries to minimize the losses from disinflationduring the 1980s. However, the statistical evidence on thispoint is weak, basically because it is impossible to determineexactly the degree to which the EMS commitment was 'cred-ible' given that realignments did always provide some leewayfor more inflation.18

The two arguments discussed so far suggest that disinflationshould involve a cost in terms of unemployment unless thecommitment of the authorities is completely credible andthere is no wage rigidity. It has been suggested, however, thatdisinflation through a credible exchange rate commitmentmight not involve any unemployment, but on the contraryan increase in demand and economic activity.19 This line ofargument does not imply that disinflation imposes no costs,but it does suggest that with a credible exchange rate commit-ment the disinflation process might follow a quite differentpath than the one sketched out above.

The possibility that a credibly fixed exchange rate may leadto a different disinflation process depends on the degree ofcapital mobility and therefore the adjustment of nominalinterest rates. With a high degree of capital mobility, acredibly fixed exchange rate leads to an equalization ofinterest rates, but if inflation has a certain inertia, realinterest rates (the nominal rate minus inflation) will not beequalized.20 A country that starts out with some inflationand fixes its exchange rate to a low-inflation (and thereforelow-interest-rate) currency would therefore immediatelyhave a low nominal domestic interest rate, and, until in-flation has also been reduced, also a very low real interestrate.

In the beginning of the disinflation process real interest ratescould therefore be low. This should stimulate investmentdemand. According to this model of disinflation there mighttherefore initially be an expansion of output. The cost of

disinflation in this case arises through the current accountdeficit that is needed to finance the increase in investment.^

The problems arising from external current account deficitsthat may occur from the fixing of exchange rates beforeinflation rates have converged should only be of a tran-sitional nature. After the initial adjustment period the expec-tations that underlie the process of wage and price formationshould adjust and inflation rates should converge. However,the countries with initially high inflation might have accumu-lated in the mean time a considerable stock of external debtbecause they will have had a current account deficit in themean time. In EMU an external debt should net createany particular problems by itself since the intra-Communitybalance of payments constraint will disappear. But the debtservice will, of course, reduce the standard of living ofthe population, unless the debt has been used to financeproductive investment.

4.3. Requirements of a stable and crediblemonetary regime

The benefits from price stability outlined in the precedingsections can be assured only if the institution that is respon-sible for the aggregate monetary policy of EMU follows, andis seen to follow, the appropriate anti-inflationary policy.Moreover, as was also argued, a credible strict monetarypolicy would also reduce the costs of disinflation. However,this condition cannot be taken for granted. The extent towhich the EuroFed will (be able to) conduct a credible policygeared towards assuring price stability for the Communitydepends on a number of factors, the most important ofwhich are its constitution and the budget policy followed byMember States. Other important features are the aversionagainst inflation of the public and the behaviour of therepresentatives of employees and employers. However, asthese are, partly at least, dependent on the behaviour of thecentral bank this section concentrates on the role of theconstitution of the central bank in EMU, called EuroFed,in assuring a credible anti-inflationary policy. The role ofbudget policy is analysed separately in Chapter 5 below.

18 See the evidence in Giavazzi and Giovannini (1989). de Grauwe andVansanten (1989) and Begg (1990).

19 See Giavazzi and Spaventa (1990).20 Idem.

It has even been suggested that the increase in demand through higherinvestment would tend to put pressure on prices and that capital inflowswould make it impossible for the national monetary authorities torestrict liquidity. In this case the fixed exchange rate commitment mightinitially slow down the disinflation process. An imperfect degree ofcapital market integration could also lead to a process of'overfinancing1

of current account deficits. The still imperfect degree of capital marketintegration of some Community countries gives the authorities thepossibility to keep national interest rates higJi to restrict liquidity. Butgiven that capital markets are highly, even if not perfectly integrated,such a policy leads to large capital inflows which would exceed thecurrent account deficit.

95


It is widely recognized that two constitutional features of acentral bank are of great importance in determining to whatextent it pursues a monetary policy geared towards pricestability: its statutory duty to assure price stability and itspolitical independence. These two elements are now dis-cussed in turn.

4.3.1. A statutory mandate for price stability

A statutory mandate to aim at price stability will give Euro-Fed a clear direction for its policy. It is therefore, withoutany doubt, a necessary element of the constitution of theEuroFed. The issues that remain open in this context arethe definition of the price index which EuroFed shouldstabilize and whether the mandate should also contain otherprovisions.

The concept of price stability

The discussion of the costs of unanticipated inflation in theprevious section implies that price stability should not betaken to mean only an average inflation rate close to zero,but also that prices should remain predictable, i.e. inflationshould not vary around the average value of zero. The taskof EuroFed should therefore be not only to keep inflationat or close to zero in the medium term, but also to makesure that it does not fluctuate in the short term.

In order to make the concept of price stability operationalit is necessary to specify what price index is to be stabilized.22

This choice is rarely made explicit even in countries whereprice stability is the main mandate for the central bank,probably because at the national level most prices moveclosely together. However, since prices can diverge muchmore at the European level than inside any member countryit might be preferable to give some indication to EuroFedwhat price index it should look at. From a theoretical pointof view the appropriate target index is the one that is mostclosely related to the source of the cost of inflation. Forexample, if one considers the main cost of inflation to besuboptimal money holdings the appropriate index would bethe consumer price index (CPI). But if the main cost ofinflation is taken to derive from the variability in relativeprices which leads to lower investment and production theappropriate price index to stabilize might be the producerprice index. It is therefore difficult to decide on purely

theoretical grounds what price index should be stabilized byEuroFed.

The main factor in this choice might therefore be availabilityand comparability across countries. The CPI has the advan-tage that it is published monthly everywhere, in contrast towholesale and producer prices which are not always availablewith this frequency. A further advantage of the CPI is thatit is widely understood and used in wage negotiations.

Price stability as the overriding mandate

A statutory duty for price stability is, of course, effectiveonly to the extent that the central bank does not have otherpolicy goals that can make this goal impossible to achieve.The provisions that are most likely to make price stabilityimpossible to achieve concern the exchange rate and themanagement of government debt.

It is apparent that if the central bank has to support theexchange rate23 it may not always be able to combat theimported inflation that results if the exchange rate is peggedto an inflationary currency. However, since the decisionabout the exchange rate regime (i.e. whether the exchangerate of the ecu should be fixed, be kept inside a target zone,or be floating vis-a-vis the other major currencies) could beconsidered of wider political importance, it should thereforebe taken by the appropriate democratically legitimated Com-munity institution. To be able to pursue the goal of pricestability EuroFed would, however, have to be independentin its management of the exchange rate within the limitsimplied by the exchange rate regime that has been decidedat the political level.

Similarly, if the central bank has to finance governmentdeficits, or if it can be required to buy certain amountsof government debt, it may be forced to follow a moreexpansionary policy than would be compatible with stableprices.

A provision that might be compatible with the goal of pricestability is the requirement to support the general economicpolicy of the government.24 If given a subordinate rank,i.e. if price stability remains the overriding goal, such arequirement would effectively be suspended if it conflictswith the need to pursue restrictive policies. Moreover, sinceit does not provide a direct channel through which it couldaffect monetary policy it would seem to pose less of a danger

It is assumed here that EuroFed should not be concerned with thevariability of prices inside the union because these relative prices willjust represent the effects of regional shifts in demand, supply or othernon-monetary factors. EuroFed should therefore not be concerned withthe lowest or highest regional inflation rate, but only with the average.

23 At present the government decides on the exchange rate in all membercountries.

24 This is the case at present in the Federal Republic of Germany.

96


to a consistent anti-inflationary policy than the other twoprovisions discussed so far. In the Community such a re-quirement would, however, be difficult to interpret since thepolicy stance of different Member States can be expectedto diverge considerably at times even in EMU and theCommunity itself would not be able to establish an overallfiscal policy stance because its budget would remain smallrelative to national budgets.

For the case of the Community this implies that EuroFedwould have the means to pursue price stability only if monet-ary financing of public deficits (at the national or the Com-munity level) is excluded and if it does not have to stabilizethe exchange rate of the Community currencies (or the singlecurrency at the end of Stage III) against the rest of the world(i.e. effectively the US dollar).

The relationship between central bank independence andprice stability has been investigated empirically. The degreeof political independence of any given central bank is, ofcourse, difficult to measure in practice. However, it is poss-ible to find some objective indicators of independence, suchas the formal institutional relationship between the centralbank and the government, the extent of formal contactsbetween the two and the existence of rules forcing the centralbank to automatically accommodate fiscal policy. Theseindicators have been used in empirical research to establisha measure of independence, which was then compared withthe price stability record.27 This research shows that thereis a strong link between political independence and perform-ance in terms of low inflation. In particular the two mostindependent central banks (in Germany and Switzerland)had also the two lowest inflation rates.

4.3.2. Central bank independence and pricestability25

Even if a central bank has a formal statutory duty of pricestability and the means to pursue this goal, it may not alwaysbe able to do so if it is not politically independent. Thereare strong theoretical reasons to believe that independenceis a necessary condition for price stability.

The main reason is that, as shown above, unanticipatedinflation has the potential to stimulate, even if only tempor-arily, economic activity (especially when facing short elec-toral timetables) and reduce the real value of public debt.Even well-intentioned policy-makers would therefore alwaysface the issue of how to convince the public that they willnever succumb to the temptation to create surprise inflation.Since in democratic societies elected officials are in generalfree to determine economic policy at their discretion it is verydifficult for political bodies to acquire enough credibility toconvince the public that inflation will always stay low. Anindependent central bank, however, does not face this temp-tation to create surprise inflation because, if its statutoryduty is to safeguard price stability, it has no interest intemporarily increasing economic activity or lowering thevalue of public debt through surprise inflation. Central bankindependence can therefore solve, at least to some extent,the credibility problem.26

Much of the following draws on Neumann (1990).The theoretical and empirical contributions concerning the issue ofcredibility always take as their point of departure countries where thereis a political cycle with well-defined dates at which national electionstake place. This will not be the case in the Community and one couldtherefore argue that the issue of credibility and the political businesscycle is less important for the Community than for countries that alsorepresent a political union.

As explained in Box 4.3 the elements of the statutes of acentral bank that determine the degree to which it is politi-cally independent are the following:

(i) Independence of instructions from government bodies(in the EC this means vis-a-vis the Community and thenational level);

(ii) Personal independence of Board and Council members;

(iii) Legal rank of its statutes.

(i) The first element is indispensable. If the central bank hadto accept instructions from political institutions it would notbe able to fulfil its mandate to maintain price stabilitysince these political institutions will often have differentobjectives.

(ii) The second element ensures that individual members ofthe policy-making bodies of the central bank cannot facepersonal conflicts of interest, because, for example, theirreappointment might be subject to political pressures. Per-sonal independence could be achieved if the term to whichmembers of the Board and the Council are appointed issufficiently long and if the appointment process cannot besubject to political pressures. It has also been argued thatthe degree of personal independence would be strengthenedif a reappointment is not possible.

(iii) The last element determines the conditions under whichthe statutes of the central bank can be changed; the moredifficult this is, the more secure the central bank — andhence the public — could be that its independence is perma-nent. In the case of the Community the structure of theconstitution of EuroFed would be laid down in an amend

27 See Alesina (1989).

97


Box 4.3: Inflation and central bank independence Alesina (1989) classifies central banks on a scale going from theleast independent, associated with a value of 0,5, to the most

GRAPH 4.6 : Central bank independence and inflation

]5 —— —————————————————————— __14

13_ !•>*o \&

<K

g 11

B 10o1 9

.5 g&fe '< 6

5

43

° o. -.

\>\D

\, D

\v^V °

EKa N.a ^v

a \\o^x.

D N.\v

(two countries) o, , , , , , , ,

0,5 1,5 2,5 3,5

Index of central bank independence

Source based on Alesina (1989).Regression result: Independence = 4,6 - 0.29 inflation

( - 7.50)T-statistic in parentheses R squared = 0.79; 1 5 D.o.F

independent, with a value of 4. This classification, based in turnon Bade and Parkin (1985) and Masciandaro and Tabellini(1988), takes into account several institutional characteristics ofdifferent central banks, such as the the formal institutionalrelationship between the central bank and the Executive (forinstance who appoints the head of the central bank, and howoften, the presence of government officials on the executiveboards of the central bank, and so on); the extent of informalcontacts between the Executive and the central bank, and theexistence of rules forcing the central bank to automaticallyaccommodate fiscal policy.28

18 The ranking in this graph is based on the situation in the early 1980s (correspondingroughly to the end point of the period given for average inflation). It therefore doesnot reflect the current degree of independence of these central banks. For example, inthe case of Italy the graph does not take into account the so-called divorce (betweenthe Treasury and the Banca d'ltalia) that in 1985 relieved the Italian centra! bankfrom the requirement to automatically accommodate fiscal policy, which is one of theparameters mentioned above.

V

ment to the Treaty of Rome which can be changed only ifall Member States agree. In the case of the Community thestatutes of the central bank would therefore certainly havea special legal rank.

This section has argued that an effective statutory duty forprice stability and political independence appear to be necess-ary conditions for a consistent and credible anti-inflationarypolicy. However, these two elements would not be sufficientto guarantee that the central bank responsible for the commonmonetary policy in EMU, i.e. EuroFed, will alwayspursue stable prices. Its task may become difficult if29

Or if budget policy puts pressure on financial markets which is possibleeven in the absence of direct monetary financing of the deficit asdiscussed in Chapter 5 below.

tensions on the labour market result in excessive increasesin nominal wages because this would leave EuroFed onlythe choice of accommodating the inflationary pressure orpursuing a restrictive policy with adverse consequences foremployment. Such a conflict is less likely to arise to theextent that the value of stable prices is generally recognized.The environment in which EuroFed operates is therefore animportant element in determining to what extent EuroFedwill be able to attain the goal of price stability.

It has therefore been suggested that regular contacts betweenEuroFed and democratically elected institutions could in-crease public support for the policy of EuroFed and makeit easier for it to achieve its target of price stability. Somedegree of democratic accountability, for example in theform of regular reports from EuroFed to the EuropeanParliament, might therefore actually facilitate the task ofEuroFed.

98

sduquenoy


References

Akerlof, G. and Yeelen, J. (1989), 'Rational models of ir-rational behaviour', American Economic Review, May, pp.137-142.

Alesina, A. (1989), 'Politics and business cycles in industrialdemocracies', Economic Policy 8, pp. 57-89.

Bade, R. and Parkin, M. (1985), 'Central bank laws andmonetary policy', unpublished.

Barro, P. and Gordon, D. (1983), 'A critical theory ofmonetary policy in a national rate model', Journal of Politi-cal Economy, 91, August, pp. 589-610.

Begg, D. (1990), 'Alternative exchange rate regimes: Therole of the exchange rate and the implications for wage-priceadjustment', in European Economy (1990).

Bekx, P. and Tullio, G. (1989), 'A note on the EuropeanMonetary System and the determination of the DM-dollarexchange rate', Cakiers economigues de Bruxelles, No 123,pp. 329-343.

Benabou, R. (1989), 'Optimal price dynamics and specu-lation with a storable good', Econometrica, Vol. 57, No 1(January 1989), pp. 41-80.

Cukierman, A. (1981), 'Interest rates during the cycle, inven-tories and monetary policy — A theoretical analysis', Carn-egie- Rochester Conference Series on Public Policy 15, pp. 87-144.

Cukierman, A. (1983), 'Relative price variability and in-flation: A survey and further results', Carnegie-RochesterConference Series on Public Policy 19, pp. 103-158.

Cukierman, A. (1990), 'Central bank behaviour, credibility,accommodation and stabilization', draft, May 1990.

de Grauwe, P. and Vansanten, K. (1989), 'Deterministicchaos in the foreign exchange market', University of Leuven.

Dornbusch, R. (1988), The EMS, the dollar and the yen', inGiavazzi, F., Micossi, S. and Miller M. (eds), The EuropeanMonetary System, Banca d'ltalia, CEPR.

European Economy (1990)special issue.

'The economics of EMU',

Fischer, S. (1981), Towards an understanding of the costsof inflation: II', Carnegie-Rochester Conference Series onPublic Policy 15, 5-42, p. 5.

Friedman, M. (1969), 'Nobel lecture: Inflation and unem-ployment', Journal of Political Economy, 1977, Vol. 85,No 3.

Giavazzi, F. and Giovannini A. (1989), Limiting exchangerate flexibility, MIT Press, Cambridge, Mass.

Giavazzi, F. and Spaventa, L. (1989), The new EMS',CEPR, paper No 363, January 1989.

Giavazzi, F. and Spaventa, L. (1990), The new EMS', forth-coming in The European Monetary System in the 1990s, Paulde Grauwe and Lucas Papademos (eds), Longman 1990.

Kremers, J. J. M. and Lane, T. (1990), "Economic andmonetary integration and the aggregate demand for moneyin the EMS', IMF Working Paper 90/23, March.

Lucas, R. E. (1981), Studies in business cycle theory.MVTPress, Cambridge.

Masciandaro, D. and Tabellini, G. (1988), 'Fiscal deficitsand monetary institutions: a comparative analysis', in ChengH. (ed.), Challenges to monetary policy in the Pacific-Basincountries, Kluwer Publishers.

Neumann, M. (1990), 'Central bank independence as a pre-requisite of price stability', in European Economy (1990).

Neumann, M. and von Hagen, J. (1989), 'Conditional rela-tive price variance and its determinants: Open economyevidence from Germany', forthcoming International Econ-omic Review (February 1991).

Spaventa, L. (1989), 'Seigniorage: old and new policy issues',European Economic Review 33, pp. 557-563.

99


Chapter 5

Implications for pubic financeEMU will have strong implications for economic policy attarge, including policies for product and factor markets. Thesepolicies will be regulatory and financial in character. Aspectsof these policies, for example in the structural and incomedomains, are discussed elsewhere in this study (Chapters 3and 6). The aim of this chapter is to discuss two major issuesraised by EMU in the public finance domain and to assess theassociated costs and benefits:

(i) What are the logical implications and requirements aris-ing from EMU for budgetary policy? (Section 5.1.)

(ii) How will EMU affect the income and expenditures ofgovernments, and their ability to take autonomous taxingand spending decisions? (Section 5.2.)

Although these two questions are closely linked in practice,they are analytically distinct since the first mainly refers todemand policy and debtjdeficit issues, whereas the secondprimarily regards the level and structure of taxes and publicspending, which are more in the domain of supply policy.Table 5.1 summarizes these two types of effects while empha-sizing the distinction between those of economic union andmonetary union.

The main conclusions of this chapter are the following:(i) The challenge that has to be faced in the design of the

fiscal regime of EMU is to provide an adequate mixof autonomy, discipline and coordination, for those arethe logical requirements of a well-functioning economicand monetary union.

(ii) The need for fiscal autonomy and flexibility arises fromthe loss of the monetary and exchange rate instrumentfor individual countries. Indeed, EMU will place newdemands on fiscal policy ar the national level for short-term stabilization and medium-term adjustment pur-poses in the case of country-specific disturbances.

(Hi) The mere creation of a monetary union requires, how-ever, long-term consistency between the common mon-etary policy and the fiscal policies of the MemberStates. Unsustainable budgetary positions in a MemberState, ultimately leading to either default or debt mon-etization, would be a major threat to the overall monet-ary stability. High and growing public debt ratios wouldlead to pressures on EuroFed to soften its policy stanceand more generally on the Community as a whole toprovide financial relief. Fiscal discipline is therefore a

vital component of EMU. Since the present fiscal pos-ition of some Member States cannot be considered assustainable, this is a serious matter of concern.

(iv) An important issue is whether EMU could weaken theincentives towards fiscal discipline. Effects in oppositedirections can be expected. On the one hand, partici-pation in EMU is indeed disciplinary since it impliesthe acceptance of monetary discipline and therefore therenunciation of debt monetization. Financial integrationshould also lead to a better market assessment of na-tional fiscal positions, although the effectiveness of mar-ket discipline cannot be taken for granted. On theother hand, markets cannot be expected to behave asif solidarity across Community Member States werecompletely ruled out, since concerns for solidarity areintegral to the philosophy of the Community. Other,frequently discussed arguments regarding interest ratecosts or current account constraints do not lead to clear-cut conclusions either. Uncoordinated policies mightalso lead to inappropriate deficits in the medium term,but only to the extent that coordination and surveillanceprocedures are not effective. On balance, there is nocompelling evidence that EMU would have strong ad-verse effects on fiscal discipline, but there is a case foraddressing the risk of failures of market discipline.

(v) Although macroeconomic interdependence betweenMember States will be affected by EMU, there isno evidence that short-term spill-over effects of fiscalpolicies would be so strong that fiscal policies shouldbe tightly coordinated on a day-to-day basis. However,coordination would be required if the common exchangerate and current account needed correction, and moregenerally in ihe context of policy coordination at theG7 level. In the medium term, surveillance will have tocorrect possible tendencies for budget deficits to becometoo large as their interest rate cost is spread throughoutthe Union.

(vi) As inflation rates converge to a low level, four southernMember States where the 'inflation tax' is presentlyabove average due to higher inflation, wider use of cash,and higher bank reserves, will suffer seigniorage revenuelosses. However, taking present inflation rates as areference, this revenue loss of monetary union amountsto about I % of GDP in two Member States, and isbelow 0,5 % of GDP in the two others. Moreover, sinceseigniorage revenues can be replaced by explicit taxes,the welfare cost of this revenue loss can be consideredto be much lower.

(vii) On the spending side, a timely move towards EMUwould substantially reduce the ex-post cost of publicborrowing during the transition to price convergencesince present interest rates carry inflation expectations

100

sduquenoy

Chapter 5. Implications for public finance

and risk premiums. Some countries would also experi-ence a more permanent decline in the cost of publicborrowing. However, this gain should not be regardedas general.

(viii) As the Community progresses towards EMU, the gen-eral issues of efficient taxation and provision of publicgoods arise. Even with a high degree of integration.

there is no case for an overall harmonization of the taxsystems and Member States would remain free to choosetheir spending and taxing levels. However, care shouldbe taken with respect to spill-over effects in the cross-frontier incidence of taxing and spending, and to prob-lems of migration of tax bases. In the absence of cooper-ation these can lead to undertaxation and an insufficientprovision of public goods. Therefore, minimal standardsand common rules should be set when necessary.

Table 5.1Summary presentation of the fiscal policy and public finance effects of EMU

Monetary union(I)

Economic union(II)

Aggregate effectin the medium term

(I + II)

1. Fiscal policy(a) Need for autonomy(b) Need for discipline(c) Incentives to discipline1

(d) Need for coordination

2. Government income and expenditure(a) Seigniorage revenues(b) Debt service(c) Tax revenues2

(d) Provision of public goods2

Government balance (a + c —b-d) 3

increasesincreases

increases

decreasedecreasesneutralneutral

decreasesneutral

decrease+ /-

decreasedecreases

increasesincreases

increases

decreasedecreasesdecreasedecreases

Without specific Community budgetary rules.Without coordination.Increase means increased budget surplus.

5.1. Budgetary policy in a monetary union

The optimal design of the fiscal system for the Communityis only partially a matter of economic efficiency. Analysiscan help to identify those public goods and services whichshould be supplied at the Community level in order tomaximize the overall welfare, and those which, according tothe principle of subsidiarity, should be provided by nationalor regional governments. This approach, known as thetheory of fiscal federalism, can provide insights regardingthe desirable allocation of different functions among differ-ent levels of government.' However, the design of a fiscal

' See Van Rompuy el al. (1990) and, for an early attempt to define theadequate fiscal regime for the Community, the MacDougall Report(1977). See also Section 5.2.3 below.

model is more fundamentally a matter of political choice,and it remains for the governments and citizens of Europeto decide whether or not to transfer new fiscal functions atthe Community level. Indeed, the diversity of the budgetarysystems in existing political federations confirms the viabilityof alternative models.

The purpose of the present analysis is therefore not todiscuss what should be in abstracto the fiscal regime of theCommunity. Throughout this section, it is assumed withoutdiscussion that the Community budget remains too limitedto have any significant macroeconomic role for the Com-munity as a whole, and therefore that any fiscal policymeasure belongs to the autonomous or coordinated actionof the Member States. Taking as given this institutionalframework, the main focus is rather on the logical impli-

101


cations and requirements for national fiscal policies of thenext developments of the Community: the completion ofthe internal market and monetary union.An appropriate starting point is to analyse the demandsEMU will place on national fiscal policy. This is the purposeof Section 5.1.1, from which the conclusion emerges thatthere is indeed a case for national fiscal autonomy in amonetary union. Then the key point is whether and towhat extent EMU makes the fiscal position and policy of aMember State also a matter of concern for the Communityas a whole. In dealing with this question, a distinction is bemade between those aspects which relate to fiscal discipline(Section 5.1.2) and those which relate to fiscal policy coordi-nation (Section 5.1.3). The root of this distinction, which isfully spelled out in the text, is that discipline mainly refersto the risks that unsustainable debt/deficit paths of nationalbudgets would present for the monetary stability of theunion, i.e. to the externalities associated with a violation ofthe government's intertemporal constraint, whereas coordi-nation refers to the appropriateness of the fiscal policy stancein the Member States and the Community as a whole, i.e.to the standard demand and interest rate externalities.2 Inother words, fiscal deficits are envisaged as problems in thefirst case, while they are treated as instruments in the second.To supply the adequate mix of autonomy, discipline andcoordination is the challenge the fiscal regime of the Com-munity has to meet. It is not an easy task since these objec-tives are not spontaneously assured. Nor are they antinomic,either. Autonomy and discipline refer to different time hor-izons. Indeed, it is inherent to a monetary union that itrelaxes the constraints to fiscal policy in the short run (be-cause of an easier and wider access to external financing),but simultaneously makes the long-term budget constraintmore strict (because the possibility of monetizing the publicdebt is eliminated). In other words, EMU replaces short-term constraints by an intertemporal constraint.3 However,even if the budgetary policy of a Member State conforms tothe discipline conditions, changes can be necessary in orderto achieve the right policy stance for the Community as awhole. This is what coordination and surveillance of fiscalpolicies are about.

Basically, fiscal policy will be considered here as demandpolicy, and one whose impact is primarily measured by theevolution of the budget deficit. Autonomy, discipline andcoordination mainly concern those deficits. The implicationsof EMU for taxation and spending are discussed inSection 5.2.3.

5.1.1. The need for autonomy

One of the most obvious consequences of monetary unionis that monetary policy is lost as an instrument of nationalmacroeconomic policy. This is the primary basis for thepresumption that monetary union leads national fiscal policyto assume a larger function. However, this argument needsto be further elaborated in order to specify what fiscal policyshould aim at and to assess the degree of fiscal flexibilitywhich is required. A good starting point is to examinehow the relative macroeconomic positions of economiesparticipating in EMU (relativities) could be affected, know-ing that monetary policy would only be directed towardsthe aggregate management of the Community economy. Thisis an important issue because the need for a modification ofrelativities would arise from any kind of asymmetry, what-ever its precise origin, namely different initial positions,shocks, structures, behaviours or preferences.4

Obviously, not all asymmetrical reactions or differences ineconomic performance would need to be corrected, and evenless through fiscal demand policy only. Chapter 6 deals atlength with the cost of losing the exchange rate as an adjust-ment instrument and discusses the effectiveness and assign-ment of alternative adjustment mechanisms, including fiscalpolicy. Without entering this discussion, it is worth mention-ing here that wage-price flexibility remains the basic adjust-ment channel as a substitute for the nominal exchange rate.Neither economic theory nor the practice of economic policyrecommend replacing the loss of such a nominal instrumentby fiscal policy. However, in so far as nominal rigiditieshamper market adjustments, fiscal policy measures can al-leviate temporary country-specific disequilibria. This is in-deed the traditional role of fiscal policy as a tool for stabiliza-tion. Budgetary adjustments can also be a necessary medium-term component of the path towards a new equilibrium inthe case of permanent shocks, although this adjustment roleis not specific to EMU. Both roles can best be illustratedwith the help of an example. Corresponding model simula-tions are given in Box 5.1.

Budgetary policy as a tool for stabilization

Suppose that a national economy which is initially in internaland external equilibrium is hit by a demand shock, e.g. atemporary increase in households' propensity to consume (afall in the saving ratio) lasting for a few periods. Due toimperfect goods market integration, most of the increase indemand falls on home goods, with the associated conse-

This dislinction, which was not made in the Delors Report (1989), hasbeen stressed inter alia by Bredenkamp and Deppler (1989) and Padoa-Schioppa(l990).See Artus( 1988). The nature of asymmetries in EMU is discussed in Chapter 6.

102


quences for inflation. In a floating rate regime, monetarypolicy would respond to these developments by a rise ininterest rates which would reduce domestic demand andappreciate the exchange rate. Both effects, would contributeto reducing the inflationary impact of the shock. In so faras the shock is temporary, this would in addition reduce theneed for further wage-price adjustments when the savingratio returns to the baseline level.

In EMU, EuroFed monetary policy would only very par-tially respond to the shock affecting a specific country sinceits primary task is to aim at stability in the union as awhole. But since goods markets would still not be perfectlyintegrated, equilibrium in the goods markets would stillrequire a real exchange rate appreciation.5 Most of theburden of stabilization would fall on spontaneous price andwage adjustments. The main channel would in that case bethe real demand impact of the loss of price competitivenessinduced by higher domestic prices, which would offset theeffect of the rise in household demand. However, with imper-fectly integrated markets, this adjustment could well taketime. This is why fiscal policy could be used in order tooffset directly the impact of the increase in domestic demand.Assuming a timely fiscal tightening, prices would thereforerise less and the subsequent adjustment when saving resumeswould be less costly in terms of lost output and unemploy-ment. This illustrates the stabilization role of autonomousbudgetary policy in EMU.

This role has to be qualified. Although both exchange ratefixity and free capital movements would increase the effec-tiveness of fiscal policy in EMU,6 most of the reasons whichhave led governments to opt for medium-term orientedbudgetary policies instead of short-term fine-tuning remainvalid: the lack of budgetary flexibility will still be a majorobstacle, and, as argued below, concern about excessivedeficits should only grow with monetary union. Moreover,fiscal policy should not be used to delay market adjustments(e.g. real wage adjustments) when those adjustments arerequired. Therefore, a reason for not using fiscal stabilizationpolicy systematically in EMU is that it could well amountto trading-off short-term stabilization for a more sluggishadjustment in the longer term.7

EMU would therefore not require a reversion to fiscal fine-tuning management. Indeed, no such tendency has beenobserved in the ERM which to a large extent raises the samekind of issue. Only in the case of well-identified and severecountry-specific shocks would the case for fiscal demandmanagement be compelling (an example of this in practiceis described in Annex B, in the case of hydrocarbon priceshocks and the Netherlands). In addition, the need for na-tional stabilization in the short-term would be reduced bythe development of 'shock-absorber' assistance mechanismsat the Community level as discussed in Chapter 6.

5 On the role of imperfect goods market integration, see Krugman (1989)and Aglietta, Coudert and Delessy (1990).

In a standard Mundell-Fleming framework, the conditions for fiscalpolicy to be the most effective are (i) fixed exchange rate, and (ii) freecapita! movements, since in that case the interest rate cost of fiscal policyis minimum.See Chapter 6 and Begg (1990).

Box 5.1: Macroeconomic effects of a fall in the saving rate underalternative policy regimes

The Multimod model developed by the IMF (see Box 2.5) hasbeen used in order to evaluate the effects of a shock to thehousehold saving rate under different exchange rate regimes.Multimod, a small-scale multinational model with model-con-sistent expectations, is adequately tailored for this task. It in-cludes separate models for the four largest Member States (Ger-many, France, Italy and the United Kingdom).

Three exchange rate regimes are considered:8 a pure float, theEMS in Stage I (with the UK participating in the ERM) whichis modelled according to the German anchor hypothesis, andEMU. Monetary policy is set according to the same rule for allcountries in the float regime, for Germany in the EMS regime,and for EuroFed in the EMU regime; this rule gives priority tothe maintenance of price stability. Only its scope changes, i.e.national monetary policies respond to deviations in nationalvariables whilst EuroFed considers the performance of the Com-munity as a whole. For non-German ERM members, a high8 Details on Multimod, the exchange rate regimes and simulation results are given

Annex E.

priority is given to Deutschmark exchange rate stability; how-ever, bands allow for some flexibility in the short term.

Dollar exchange rates are in each case determined by a forward-looking uncovered interest rate parity condition under the as-sumptions of perfect capital mobility, asset substitutability andperfect foresight. In other terms, the evolution of exchange ratesequalizes the expected return on short-term assets denominatedin different currencies. With such a determination of the ex-change rate and such a monetary rule, a rise in domestic demandboth deteriorates the current account and appreciates the ex-change rate because of the rise in domestic interest rates. Thisresult might be contrary to European conventional wisdom.However, it holds once full capital mobility and asset substituta-bility are assumed.9 The model is used to simulate the macro-economic consequences of a permanent fall in the British savingrate by two percentage points under three alternative exchangerate regimes:

(i) Under a float regime, a fall in private saving provokes animmediate 2 % exchange rate appreciation since monetarypolicy is expected to be tight in the years ahead. Domestic

9 The US experience of the early 1980s shows that ihis is not only theoretical.

103

sduquenoy


'GRAPH 5.1 : Effects of a fall in L'K household saving

under three exchange rate regimesRoal GDP

,'jr-K^1,1 -

1 -0,9-0,80,70,6

a 0,5* 0,4

0,30,20,1 -

0 .-0,1 --0,2

/N Vl\ \^\ \ "^\ . •* EMU

// \ \ \// \ \ \// \ EMS\ \

// \ \\,. Float\ \/ * ~/ 'V Xîv

- - - - - - - - - ^NT ; ^^r— 1^ - -e- - ^"\J

__._ . ___ . __ . _ . . T T J0 1 2 3 4 5 6 7 8

YearsAbsorption deflator2,4, - - ^ * ;

2.2 •> ^. - ^ +. ^2 4 / '

1,8 < EMU ̂ '1,6 '•, ' '"EMS1,4 H / ,

1,2 4 .,

* - ' i /'0,8 J y0,6 H // 4

•0*2 ; ^/-''0 JK "*""' ' - •** 1

-0,2 1 ~ **" Roat-0,4 i ' ,-fr"-0,6 1 r" . , , i

0 1 2 3 4 5 6 7 8Years

Current account balance°1x -

-0,1

o

\ ,• v - "V\ \F,oa,

J-0,3 \\S| \\ N5 -0,4 4 u'o ' '\ + EMS^ -0,5 J \ i /^ r,

\\ /-0.6 j ^ ^, EMU w

-o, ~S r ,-0,8-1 . . . . . .

0 1 2 3 4 5 6 7 8Years

Sourer Multimod simulations by the Ccm minion services

demand therefore crowds out external demand, the expan-sion is short-lived and inflation remains low, as shown inGraph 5.1.

(ii) In the EMS, monetary policy can no longer be directedtowards demand management although bands allow for alimited interest rate rise. 10 The exchange rate appreciatesonly by 0,4 % in the first year. Domestic demand is higher,and translates more into output growth. However, the risein inflation induces a real appreciation, which deterioratesthe trade balance.

(iii) The behaviour of the economy under EMU does not differmuch from the experience under the EMS. Interest-rateand nominal exchange-rate effects of the fall in saving aresmaller, with the consequence that growth and inflation areslightly higher (Graph 5.1).

In order to illustrate the potential role of fiscal policy in EMU,Table 5.2 gives the effects of the shock to the saving rate underthe three regimes and its effect under the EMU regime assumingan (unanticipated) reduction in government expenditures in year1 calibrated in order to minimize the deviation of inflation fromthe baseline over eight years. It is apparent that fiscal policy, byoffsetting most of the effects of the decline in the saving rate,could help to reduce the adverse effects of such shocks, Obvi-ously, this could also be the case in the EMS. Alternatively, thecountry could settle for a realignment.

10 The shock is not large enough to trigger a realignment, which is supposed 10 happenonly when the inflation differential with respeci to Germany reaches 8 %.

104

sduquenoy


Table 5.2Medium-term effects of the saving decline

(Present value of deviation from baseline over eight years)

GDPPrivate consumptionInflation (absolute deviation)Current accountReal effective exchange rate1

Float

1,8921,07

1,47-1,81

1,9

EMS

3,6619,95

1,87-2,84

1,4

EMU

4,2122,59

2,05-3,28

1,7

EMU with fiscal policy

-0,1323,70

0,450,030,2

1 Present value of average appreciation.Source : MuHimod simulations by the Commission services.

Budgetary policy as a tool for adjustmentSuppose now that instead of being temporary, the fall in thesaving rate is permanent." The medium-term impact is thesame whatever the exchange rate regime. In all regimes, thefall in the saving rate leads to a permanent real exchangerate appreciation (in order to achieve internal equilibrium)and to an external deficit. In EMU, real appreciation takesplace through a rise in domestic prices, whereas in a floatingregime the main channel is a rise in the nominal exchangerate. The difference lies in the possible reactions to theexternal deficit.

Ultimately, external balance must be achieved in all exchangerate regimes. Among possible adjustment mechanisms are(i) a build-up of the external debt, which erodes wealthand disposable income, and leads agents to reduce theirconsumption, (ii) inflation, which also reduces wealth bywiping out the real value of monetary balances, and (iii) apermanent reduction in the budget deficit, which raises theaggregate national saving rate.12 The first mechanism isindependent from the exchange rate regime; however, it onlyacts in the very long run; relying on it implies one is readyto accept the build-up of net foreign debt or assets in so faras it arises from private sector behaviour. The second restson the possibility of adjusting the nominal exchange rate tooffset the effects of higher inflation. It is not available inEMU where inflation differentials would entirely translateinto real exchange rate changes. Thus, except for the Com-

munity as a whole in relation to the rest of the world, anadjustment mechanism is lost in EMU. This is the basicrationale for expecting more reliance on the third mechan-ism, namely fiscal policy.

The above example is only a specific case for what could becalled the adjustment role of fiscal policy in EMU. Thegeneral point is that if there is a need to change the realexchange rate within EMU because of the build-up of unde-sirable external imbalances, the only available macro-economic instrument is fiscal policy.13 Whether the need forusing fiscal policy as a tool of adjustment will increase ordecrease in EMU depends on the balance between twoopposite effects: on one hand, as developed in Chapter 6,current account surpluses and deficits should be less of aproblem within the Community once capital moves freelyand exchange rate risk is eliminated; on the other hand, inso far as changes will be required in the real exchangerate or the external account, fiscal policy will be the mainavailable adjustment mechanism. However, the instrumentwhich is lost, i.e. adjustment through inflation, is neither avery powerful nor a desirable one, since it is at odds withthe objective of price stability. The loss of inflation andassociated real wealth effects as an adjustment mechanismcannot be considered of real significance in the Communitycontext.i4 On balance, it can therefore be assumed that theneed for fiscal policy adjustments will decrease.

This permanent fall in the saving rate could arise, for example, fromchanges in the rate of time preference.Provided Ricardian equivalence does not hold. For a discussion of someof these mechanisms and the consequences for fiscal policy in EMU,see Begg (1990), and Masson and Melitz (1990).

Masson and Melitz (1990) exhibit another case for fiscal differentiation:assuming the common EMU exchange rate has to appreciate in themedium run in order to contribute to the correction of US currentaccount imbalances, they show that only fiscal policy could be used toachieve the desired sharing of this adjustment between Member Statespresently in surplus (Germany) and in deficit.Begg (1990), which bases the role of fiscal policy as an adjustmentmechanism on the loss of the real balance effect, reaches therefore adifferent conclusion.

105

sduquenoy


A related issue is whether fiscal convergence is desirable overthe cycle, i.e. whether cyclically-adjusted deficits should beat the same level (presumably zero) across the Community.In such a context, and assuming the fiscal position is sustain-able, the deficit can best be viewed as an intergenerationaltransfer mechanism since it must be repaid by the followinggenerations in the form of higher taxes. The issue thereforeis whether such transfers are warranted and whether thereare good reasons for socially desirable deficits to differ fromone Member State to another.

The need for sustained surpluses or deficits mainly arisesfrom either long-term social security equilibrium, publicinvestment financing or intertemporal market failures. Thefirst case can be expected to imply wide variations acrosscountries since both the features of the pension regimes andthe demographic perspectives differ within the EC. In someMember States which mostly rely on pay-as-you-go systemsand face the perspective of rapid ageing, it could be desirablefor the general government to run substantial surpluses inthe years ahead, allowing these to accumulate in a trust fundin order to be able to meet the commitments to futurepensioners, while no such need exists in other MemberStates.15 The second, which generally arises from the searchfor an optimal financing of public investment yielding socialbenefits in the long run, can be significant in catching-upMember States, and especially relevant in the case of theinfrastructure effort required by the modernization of EastGermany. The last argument concerns, for example, theadequate size of the capital stock. It can be desirable thatthe government runs a deficit if the capital stock is toolow and the current generation unwilling to reduce its ownconsumption for the benefit of future generations.16 There-fore, there is no need for fiscal surpluses or deficits toconverge in the medium term at any particular level withinthe Community under any of the three arguments presented.

Summing up, the stabilization role of fiscal policy at thenational level is bound to remain important in EMU as longas asymmetries persist, at least in the period before the fulleffects of economic integration are felt, but only in the caseof severe country-specific disturbances. As to the role offiscal policy as an adjustment instrument, it mainly arises

As an illustration, the immediate rise in the budget surplus needed tomaintain the real value of retirement benefits in a 30-year perspectivemay exceed 10 % of GDP in some Member States. Actually, the existenceof implicit liabilities arising from contractual commitments to futurepensioners raises also sustainability issues. For an analysis of the budgetpositions along these lines, see Hagemann and Nicoletti (1989) andOECD(1990).This argument, first raised by Diamond, is discussed by Wyplosz (1990)together with other rationales for differences in the level of publicdeficits.

from the need to achieve external equilibrium. It would onlyremain important in so far as governments still have externaltargets in the medium run. But legitimate domestic targetscould also imply that fiscal deficits remain different in oneMember State from another in the medium run. Fiscalautonomy is therefore warranted. However, this does notmean that all fiscal deficits are acceptable. This issue is takenup in the next two sections.

5.1.2. The need for discipline

Since the publication of the Delors Report, budgetary disci-pline has been a matter of discussion in both academic andpolicy circles. Among the most debated issues are the extentof need for discipline, and whether it should be achievedthrough market discipline or through rules governing defi-cits. In order to clarify this question, three points are success-ively addressed below:(i) what fiscal discipline is about and why it is a major

concern in a monetary union;(ii) why fiscal discipline is empirically a serious issue in the

Community; and(iii) whether the incentives to discipline will be stronger or

weaker in EMU.

The whys and wherefores of discipline

Discipline is an intuitive notion, but not an easy conceptto define and quantify. Its most straightforward and onlyindisputable definition relates to the fact the governmenthas to ensure that it does not become insolvent. This defi-nition, which is both referred to as the intertemporal budgetconstraint and as the condition of sustainability of the debt/deficit path, is admittedly narrow (see Box 5.2). Whether ornot a budget policy which does not put the sustainability ofthe debt/deficit path at stake can be deemed undisciplinedis also a matter of discussion. However, a first step is toassess the consequences of a lack of discipline in the narrowsense.

Budgetary sustainability is always a concern for monetarypolicy because monetary and budgetary policy are interde-pendent in the long run: protracted deficits leading to unsus-tainable budget positions end up either in debt monetizationor government default. In the first case, monetary authoritiesgive up their autonomy in order to rescue the government;in the second, they stick to their own objective but force thegovernment to repudiate part of its debt. l7 These are surely

This has been shown by Sargent and Wallace (1981).

106


extreme cases. But they are still a legitimate matter of con-cern in the context of EMU since the consequences of sucha situation would be very serious: Euro Fed would have todeal with the risk of a financial crisis and, theoretically atleast, central bank independence could be endangered.

More precisely, lack of discipline in one or more MemberStates could jeopardize the policy stance of an independentEuroFed and affect the other Member States through threechannels.

(a) The first and most obvious would be a pressure onEuroFed to soften its overall monetary stance. It is wellknown that in highly indebted countries, the room for ma-noeuvre of monetary authorities tends to be restricted be-cause monetary tightening aggravates the budgetary prob-lem: both the reduction in tax income brought by a tempor-ary slow-down and the increase in the interest burden onthe public debt worsen the deficit.I8 Even if monetary policyis formally insulated from Treasury pressures, the very factthat a monetary tightening could turn a difficult budgetarysituation into a genuine financial crisis acts as a de factoconstraint to the central bank.19 This risk could be aggra-vated in the case of a short maturity of public debt or theuse of floating-rate debt instruments since in such cases evena temporary tightening can have very severe consequenceson the debt burden and precipitate at least a liquidity crisis,as the experience of the city of New York in 1975 demon-strated. Moreover, markets may expect the central bank tobe tempted to resort to inflation to alleviate the budgetaryproblem. This immediately affects its credibility and there-fore interest rates.20 These considerations could well berelevant for the Community if due to a lack of discipline,the budgets of some Member States remain on the edge ofsustainability.

(b) Even if EuroFed sticks to its predetermined stance with-out consideration for public borrowing costs or governmentdefault risks, violation of the budget constraint in one Mem-ber State could affect the others. This is because this countrywould either have to declare default or to withdraw from

18 With a 100% debt to GDP ratio, any significant monetary tighteningwould cost an indebted government several times more than the coun-try's contribution to the Community budget.

19 It can be recalled in this context that in 1982, the US Federal Reserveis widely reported to have softened its monetary stance in order tocontribute to a solution of the emerging Latin-American debt crisis.Although it can be argued that domestic considerations were pointingin the same direction, this can be taken as an example of de factopressure towards monetary softening without any restriction of formalmonetary independence.

20 For a discussion along these lines, see Van der Ploeg (1990), andBovenberg, Kremers and Masson (1990).

the EMU in order to be free to resort to debt monetization.To the extent that its debt would still be denominated in thehome currency (instead of ecus), this second option couldalleviate its public finance crisis. However, such a withdrawalcould have an impact on the other members of EMU sinceit would signal that adhesion to the union was not irrevo-cable after all, and therefore reduce the market assessmentof the substitutability of assets issued by agents of differentcountries.

(c) The third channel is of a distributive character.21 Finan-cial difficulties in one Member State would raise the issueof financial solidarity across the Community. At the extreme,this would take the form of pressures to bail out an insolventgovernment. But milder forms of solidarity can exist, e.g.through the purchase by EuroFed of a disproportionateshare of public bonds from a specific country (which wouldbe equivalent to a Community loan) or in the form of explicittransfers. In each case, the ultimate result would be felt inthe overall market for government paper instead of beingbottled up in the country of origin.

In all three instances, it is apparent that externalities are atwork, which make unsustainable deficits and debts a matterof concern for the Community as a whole.

Fiscal discipline defined as the avoidance of an unsustainablebuild-up of public debt is therefore a vital condition for thesuccess of EMU. Has this definition to be extended, i.e. isthere a legitimate concern that sustainable deficits could bedeemed excessive because they would have damaging effectson the Community? Although there are obvious instanceswhere a sustainable budget deficit in a Member State couldunambiguously be considered inappropriate from a Com-munity point of view, it is also apparent that as the problemsarising from such a deficit would be of a different naturefrom those arising from an unsustainable fiscal situation,they would call for a different type of solution.

Such a situation could arise for example if due to a lack ofdomestic (private and public) savings, a Member State runsa large current account deficit. Without accommodation byEuroFed, this could provoke a rise in the common interestrate, an external deficit for the Community as a whole, andtherefore affect the common dollar exchange rate. Thereforethe interest of the Community as a whole would call for areduction of this deficit even if it would not be appropriatefrom the domestic point of view of that Member State.However, symmetric instances could also happen where fis-cal policy in a Member State could be considered too tight

21 The distinction between global and distributive issues in this context hasbeen made by Padoa-Schioppa (1990).

107


Box 5.2: The arithmetic of budget constraints

The standard expression of the intertemporal budget constraintis a straightforward application of the accounting equationdescribing the dynamics of debt and deficits:22

(l)dB/dt = -S + i B

where B is the public debt, S is the primary surplus (i.e. thegovernment balance less interest payments), i is the nominalinterest rate, and d is the usual derivation operator. Seignioragerevenues are neglected for the sake of simplicity (this issue istaken up in Section 5.2.1 below).(1) can be rewritten in terms of ratio to GDP. Simple calculationslead to:(2) db/dt = -s + (r-y)b

where b and s are respectively the ratios of debt and primarysurplus to GDP, r is the real interest rate and y is the growthrate of real GDP. This condition states that when the debt ratiomultiplied by the difference between the real interest rate and

the growth rate exceeds the primary surplus, the debt to GDPratio grows.

Integrating (2) yields the expression of the intertemporal budgetconstraint :

C\\ h =s f s e~(r~y)1 rlt\_j) u ^ j o s c ( - "u i

i.e. the initial debt must at most equal the present value of futureprimary surpluses, the discount factor being the excess of thereal interest rate over the growth rate. This condition only holdsif the real interest rate exceeds the growth rate. If not, growthwipes out any debt which can be issued.

Equation (3) shows that the sustainability criterion is clearly ofa forward-looking nature. Its empirical implementation has torely either on forecasts or on assumptions regarding themaximum future primary surplus.23 Without resorting to suchnorms, equation (2) can be used to assess the gap between thepresent primary surplus and that which would stabilize thedebt ratio under normal long-term conditions. This is done inTable 5.3. Graphs 5.2 and 5.3 are based on the same approach.

22 This derivation follows Blanchard (1990). Sec also Wyplosz (1990). 23 The first approach is followed by the OECD (1990), the second by Blanchard (1985)and Wyplosz (1990).

from a Community point of view, because it would lead tonational and Community current account surpluses, and toundesirable exchange rate effects. Whether or not a sustain-able deficit is inappropriate should therefore primarily re-main a matter of judgment. When there is a problem of thiskind, regular policy coordination and surveillance pro-cedures should address the problem and pave the way for apolicy correction.

Assessing fiscal discipline in the Community

Were fiscal positions in all Member States basically sound,then the above considerations could be dismissed as excess-ively theoretical. However, this is not the case, since in spiteof favourable economic conditions several countries are stillcharacterized by high and rapidly growing public debt ratios.Thus, an assessment of these fiscal positions is required.

The empirical assessment of sustainability is not straightfor-ward because it is inherently a forward-looking condition:at any point in time, whether or not a budgetary policymeets the substainability criterion depends on the the futurecourse of taxes, spending, and macroeconomic variables likethe growth rate and the real interest rate. Technically, thiscondition states that if the real interest rate exceeds thegrowth rate, the present value of future primary surplusesmust exceed the debt level, the discount factor being thedifference between the real interest rate and the growth rate(Box 5.2).

A sustainability criterion of frequent use in policy discussionsis the evolution of the public debt. The fiscal position isconsidered sustainable if the debt to GDP ratio is stable ordecreasing. This kind of criterion approximates the formalbudget constraint condition and is useful, with qualificationshowever.24 Table 5.3 and Graphs 5.2 and 5.3 are based ona related approach, but instead of using the current valuesof the growth rate and the real interest rate, the sustainabilitycondition is computed with normal values of these twofactors which presumably reflect more adequately long-termgrowth conditions.25 Table 5.3 gives for all CommunityMember States the gap between the primary budget surplusand the one that would stabilize the debt ratio in the mediumterm.

In order to synthesize this material, Graph 5.2 presents ac-tual debt levels together with the gap between the primarysurplus and that which would be consistent with a stabiliza-tion of the debt ratio assuming a 5 % real interest rate.

Major limitations of this indicator are the following: (i) it does not takethe level of debt into account, although stabilization of the debt ratiois more an issue when ihis ratio reaches 100 % than when it amountsto 20 %; (ii) it depends on short-term conditions regarding growth rateand real interest rates rather than on long-term conditions; and (iii) itignores future implicit or contractual liabilities arising from social secur-ity commitments.A specific reason for that choice is that real ex-post interest rates areprobably above their long-term levels in ERM Member States whosedisinflation and exchange rate commitment are not yet fully credible.See 5.2.2 below.

108


Table 5.3Debt sustainability conditions in the Community

BelgiumDenmarkGermanyGreeceSpainFranceIrelandItalyLuxembourgNetherlandsPortugalUnited KingdomEC

Debtratio1

(1989)

128,463,543,086,243,835,5

104,998,99,0

78,473,144,358,4

Currentsurplus1

(1989)

-6,3-0,7

0,2-17,6

-2,1-1,3-3,1

-10,23,3

-5,1-5,0

1,6-2,8

Primarysurplus1

(1989)

2,44,71,6

-10,1-0,9

0,87,7

-2,31,21,1

-3,24,31,2

Growthrate

(assumed)

3333,53,533,53333,533,2

Required primarysurplus

r = 4%

1,30,60,40,40,20,30,510,10,80,40,40,5

r = 5%

2,61,30,91,30,70,71,620,21,61,10,91,1

Susiainabilitygap

r = 4*/o

-1,1-4,1-1,210,5

1,1-0,4-7,2

3,3-1,1-0,3

3,6-3,9-0,7

r = 5%

0,2-3,4-0,7

11,41,6

-0,1-6,1

4,3-1

0,54,3

-3,4-0,1

Commission services. Data refer to general government. Debt ratios are gross public debt as a percentage of GDP al market prices. Primary surpluses (government surpluses net of initrespayments) for 1989 include slock-How adjustments in order to be made consistent with the evolution of public debt.

GRAPH 5.2: Debt ratios and sustainability gaps in theCommunity, 1989

121110987654321 *0 +

-I J-2 .-3 .-4 H-5 1-6 1

(as a percentage of GDP)

°GR

NL °B

"UK DK

20 40 60 80

Debt raiio

° I R L

100 120 140

Source Commission servicesNote the sustainability gap is defined as in Table 5 3 Countries above the horizontalaxis are characterized by growing debt to GDP ratios under normal long-termconditions.

Critical situations, i.e. large and growing debt ratios, arerepresented by points in the north-east region of the graph.

The above material shows that the higher the debt ratio, thehigher the primary balance compatible with a stability ofthis ratio. Therefore the limit to the debt ratio effectivelybecomes the ability of the government to devote a largeamount of taxes to the primary surplus and the service ofpublic debt, i.e. to tax heavily the present generation in orderto pay for previous expenditures. Although no confidentassessment of the maximum debt/GDP ratio can be made,it is fair to assume that debt ratios above 100% of GDPshould imperatively be stabilized since at that level, taxesdevoted to debt stabilization already amount to about 9 %of GDP.26 Moreover, closer integration within EMU putsfurther limits to the possibility of increasing future taxes inorder to service high debt ratios. A Member State whichwas compelled to raise taxes without supplying a proportion-ate level of public goods, because tax revenues were devotedto the sole purpose of servicing the debt, could face the riskof large-scale migration of mobile factors of production.Large intergenerational transfers are indeed only possible tothe extent that factors are of a limited mobility. Taking this

With 3 % growth, 2 % inflation and 5 % real interest rate, this corre-sponds to 2 % primary surplus plus 7 % debt service. Wyplosz (1990)considers the maximum debt ratio is about 150 to 250 % of GDP.

109


element into account leads to a lower upper bound forsustainable debt ratios.27

Table 5.3 and Graph 5.2 highlight the wide dispersion offiscal positions within the Community. Although most Mem-ber States can be considered to be in a sustainable position,the debt ratio is both high and increasing in Greece, Italy,and Portugal, and the present fiscal position of these Mem-ber States can be considered as violating the discipline whichwould be required in EMU. Fiscal action is needed also inthe case of very high debt ratios, even if these ratios arestable. This concerns Belgium, Ireland and to a lesser extentthe Netherlands.

In order to illustrate possible stabilization paths, Graph 5.3depicts for two heavily indebted Member States the path ofthe actual effective primary deficit and debt burden, togetherwith the equilibrium condition which gives the primary defi-cit required in order to stabilize the debt ratio at any particu-lar level. This highlights the contrast between the Belgiansituation, where debt sustainability has been approximatelyachieved in spite of a very unfavourable initial situation andthat of Italy where the situation has not improved much inrecent years. The graph also shows how the Italian debtratio began to increase in the 1980s, after remaining stablefor a long time because of a very low real effective cost ofborrowing.

Are the incentives to discipline weaker in EMU ?

Fiscal discipline ought to be primarily a matter of concernfor national governments and parliaments, not the Com-munity. However, it has been demonstrated that a lack offiscal discipline by a Member State would affect the unionas a whole. This is a classic case of externality, which a prioricalls for Community rules and procedures aiming at theenforcement of discipline. Yet the nature of the rules andprocedures that could be most effective is a difficult question,since it immediately raises issues of incentives, and of marketversus government failures. Before turning to these issues indetail, it can be useful to describe briefly why and how theproblem arises and what is the relation between the presenttopic and the revenue issues which are the topic of the nextsection.

A useful benchmark case is the one where all necessaryconditions for effective market discipline hold. These con-ditions are as follows. First, there would be a precise distri-bution of roles between different types and levels of fullyindependent government institutions (EuroFed, EC budget

GRAPH S.3: Evolution of public debt and primary deficit inselected Member States

5T-432-11

(as a percentage of GDP)Belgium, 1971-89

-5-6-

:J1-9 <

-101-11 \- I2 i -

0

5 ,

>12 H1 i0 i-

-1 .

=51-6-7-8 *-9

-10 j- I I 1-12 i

0

20

20 40 60 80Debl ratio

100 120 140

Sourir Commission servicesfor each of the Lbovt tigjrc'., the evolution of ihe debt tc GDP ratio is reportedalong the horizontal sus and th*: cf the primary defici! along the vertical axis. Theupward-sloping OA schedule .Jcsmbes the equilibrium condition for a J % discountfactor (c.g 5°-c real interest rale and J% growth)

authorities, national governments) without any financialsolidarity between them. Second, the design of the systemensures the independence and credibility of EuroFed. Statu-tory clauses guarantee that it is fully sheltered from govern-ment pressures.28 It would also be committed to follow

This point is made by Bovenberg, Kremers and Masson (1990). See alsoSection 5.2.3 below.

Examples of provisions of the first type, i.e. regarding the prudentialtreatment of government paper in bank portfolios are given by Bishop(1990). Regarding the provisions aiming at isolating EuroFed fromgovernment pressures, specific proposals are given in Neuman (1990)and Padoa-Schioppa (1990). All three authors suggest EuroFed couldbe forbidden to hold public sector debt.

110


certain rules in order to avoid time inconsistency.29 Third,prudential rules have to be modified in order to make sure(i) that markets have full information on the state of publicfinance, (ii) that the maturity of government debt is not tooshort, and (Hi) that regulations neither force private agentsto hold government securities nor lead them to consider thesecurities as free of risk.

Assuming these conditions hold, market discipline wouldoperate through increases in the marginal cost of borrowingaccording to the assessment of default risk by market partici-pants. If the sustainability of the debt-deficit path of a givengovernment were put into question, markets would react bychanging the credit rating of the undisciplined governmentand by increasing the marginal cost of borrowing it faces,presumably triggering a timely fiscal adjustment programme.Neither quantitative norms regarding the amount of borrow-ing by governments nor procedures regarding the monitoringof their behaviour would therefore be needed. Moreover,any change in the cost of public borrowing consistent withan exact assessment of default risks would be welcomed,whether a rise or fall.

The above image helps to clarify the discussion but relies onsimplifying assumptions. First, whether financial solidarityamong Member States can be effectively ruled out is disput-able. As the default of a single Member State would affectthe Community as a whole, there would be a rationale forproviding support to an insolvent government. Only anabsolute 'no bail-out' clause, prohibiting financial supportnot only from EuroFed, but also from the Community as awhole, could persuade the markets that no solidarity meas-ures can be expected. Moreover, the Community has alreadychosen to set up a regional policy^nd to organize a systemof cross-country transfers on solidarity grounds. This doesnot imply automatic solidarity, and a 'no guaranteed bail-out' clause could feature in the rules of EMU, but it never-theless means that markets cannot be expected to behave asif financial solidarity were completely ruled out in such anenvironment. Second, the record of markets in assessing thedegree of government default risk is disputed. Althoughsome evidence points in the direction of a true assessmentof default risks, the experience of the Latin American debtcrisis is the most recent case that exemplifies the weaknessesof pure market discipline. Third, there is no guarantee thatgovernments would effectively react to a worsening of their

credit rating as private borrowers would. As high publicdebts frequently result from political polarization or distri-butional conflicts over the allocation of the fiscal burdenrather than from intertemporal maximization, one can doubtthat a rise in the cost of borrowing would necessarily triggerthe required fiscal adjustment.30 Furthermore, the size of thepublic debt of large countries with respect to the CommunityGDP could limit the effectiveness of market discipline.

Once these elements are taken into account, it is no longerclear whether constraints which are suboptimal in a first-best situation because they tend to limit fiscal flexibilityremain so, or whether the removal of such constraints perse would have perverse incentive effects.

On the basis of the above discussion, incentive effects ofEMU can be classified under three different headings:(i) those which arise from the rules governing the behav-

iour of EuroFed and other Community institutions;(ii) those which arise from the effectiveness of market disci-

pline in the context created by capital market liberaliza-tion;

(iii) those which arise from the macroeconomic role andstance of fiscal policies in the EMU context since in animperfect context it could happen that a loosening ofthe macroeconomic incentives to sound fiscal policieswould ultimately raise sustainability problems.

(i) The first issue is whether the behaviour of EuroFedor other Community institutions will enhance or weakendiscipline. Concerning EuroFed, as argued in Chapter 3,both its independence and its commitment to price stabilitywill contribute to a high standard of monetary discipline.One should therefore expect governments to be aware thatby adhering to EMU they would ipso facto accept a strictintertemporal budget constraint, and renounce future debtmonetization. Moreover, as demonstrated by Wyplosz(1990), even assuming that without EMU monetary authori-ties would already rule out debt monetization, the budgetconstraint in EMU should become tighter.31 For someMember States where the government still has access toprivileged central bank financing, this could significantly

In monetary policy, time-inconsistency refers to the dilemma of policieswhich aim at price stability, but face an incentive to attempt a surpriseinflation, for instance in order to reduce the real value of public debt.In so far as the public is aware of this incentive, it expects monetarypolicy to depart from price stability. The result is that inflation is toohigh, but without any gain regarding the public debt. See Kydland andPrescott (1977), Barro and Gordon (1983).

See Alesina and Tabellini (1990).This is because in the case of government default the reduction in thewealth of private agents does not have as much impact on goods marketequilibrium as in a non-EMU. The argument runs as follows: in a non-EMU, the fall in private wealth due to partial government defaultreduces the demand for goods, and therefore the real interest rate fallsto restore goods market equilibrium, therefore reducing the debt serviceburden. In the EMU, interest rate equalization leads to a much smallerfall in the real interest rate. See Wyplosz (1990) for the demonstration.

Ill

sduquenoy


change the framework of incentives faced by policy-makers.This would come with costs, namely that of fiscal adjustmentin the medium term and the reduction in seigniorage rev-enues which would result from lower inflation., but as shownin Section 5.2 below these costs should not be overestimated.Concerning the other Community institutions, a strengthen-ing of Community integration inevitably leads to an expec-tation of more solidarity between its members, and couldtherefore weaken discipline. A specific issue is whether Com-munity assistance in the case of severe country-specificshocks could have such discipline weakening effects. Thisissue is taken up in Chapter 6 where it is argued that anappropriate design of such a transfer scheme, includingconditionality provisions, could reduce the risk of moralhazard.

(ii) The second issue relates to the effectiveness of marketdiscipline, i.e. the ability of markets to evaluate correctlydefault risk premiums and to trigger the appropriate re-sponse in the case of government borrowing. Empiricalevidence confirms that markets do differentiate among sov-ereign borrowers, but that spreads are usually small oncetax factors are taken into account. Market differentiationacross provinces or states exists in Canada, where borrowingcosts differ by about 50 basis points, for example, and inthe US, where the spread is about 40 basis points.32 In theUS case, there is only weak evidence that higher debt burdensincrease the cost of borrowing, and no evidence at all that atsome point high-debt states get rationed out of the market.33

Evidence from Eurobonds markets does not show any strongrelation either between the yield to maturity of public bondsof the same currency denomination and characteristics, andthe budgetary situation of the borrower.34

Although both the developing countries' debt crisis of the1980s and that of New York City in 1975 exemplify the risksof imprudent behaviour of sovereign borrowers, there is noagreement on the causes of this poor record. Some authorsanalyse it as illustrating the inherently limited capacity ofmarkets in assessing system risks (i.e. risks which involve notonly standard random shocks, but also uncertainty regardingevents for which there is no relevant prior experience, e.g.because they depend heavily on outside macroeconomic

factors).35 Others point to violations of the conditions whichare required for market discipline to work properly (e.g.regarding the access to full information or the exclusion ofany possibility of bail-out) or to the inadequate reactions ofthe borrowers as the major explanations of that record.36

In the Community context, capital market integrationshould contribute to a better evaluation of risks; thereforemarkets can be expected to exert disciplinary effects. How-ever, even assuming an appropriate degree of risk assessmentby the markets, it cannot be taken for granted that marketdiscipline would be sufficient, due to expectations of Com-munity assistance and/or inadequate response of govern-ments to market signals.

(iii) Although a tendency for fiscal policies to be too laxwould not necessarily imply a failure to achieve sustain-ability, it has to be discussed whether EMU would give riseto such a tendency by relaxing the constraints which pre-sently press in favour of fiscal tightness. This could in prin-ciple happen even if the relaxation of these constraints couldfrom a different angle be considered as a benefit from EMU.

A first issue is whether the removal of constraints whichpresently act as incentives to fiscal restraint would weakendiscipline. Major constraints are at present high real ex-postinterest rates in non-German ERM Member States, and the'external constraint' which acts as a barrier to fiscal expan-sion when the country experiences a current account deficitor when a currency devaluation is at risk. Both will to alarge extent disappear or be transferred to the Communitylevel with the creation of EMU. Therefore, discipline couldbe weakened.

However, the relevance of this argument can be disputed,since none of these constraints directly addresses fiscal disci-pline. High ex-post real interest rates primarily reflect slug-gish adaptation of exchange rate expectations rather thandefault risks: for example, ex-post real interest rates havebeen almost at the same level in France and Italy over the1985-89 period in spite of very different budget situations.37

Figures are taken from Bishop, Damrau and Miller (1989).See Eichengreen (1990).Alesina, Prati and Tabellini (1989) report that risk premiums are appar-ent on Italian Government bonds." They find evidence in interest ratedifferentials between Treasury bills and private certificates of deposit,and between public and private medium-term bonds. However, asmentioned by the authors themselves, differences in the degree of liquid-ity could also be an explanation. Fiscal bias could also intervene.

More precisely, this failure can be related to the inherent uncertainty ofan evaluation of the creditworthiness of public agents who can alwaysrely on taxation to service their debt.The first line of reasoning is illustrated by Aglietta, Brender and Coudert(1990), the second by Bishop, Damrau and Miller (1989).Bredenkamp and Deppler (1989) remark in addition that the reductionin interest rates brought by EMU would only arise because of theacceptance of monetary discipline by Member States. Hence, one shouldnot consider as anti-disciplinary what in reality arises from a strongerdiscipline.

112

sduquenoy


The Italian case also exemplifies the fact that a public debtbuild-up is perfectly compatible with external equilibriumprovided the private saving rate is high. More generally,Graph 5.4 shows that there is no apparent correlation be-tween the budget balance and the current account balancewithin the EC. To the extent that these constraints act asdisciplinary devices, it is only as a kind of surrogate disciplinewhose optimality and efficiency are both highly question-able38. Their replacement by an explicit commitment tofiscal discipline would clearly be beneficial.

GRAPH 5.4: Budget and current account balances, 1989

V3cM

"aAeDo8TJ

C

1u

*-,.

(as a percentage of GDP)

16 , ,15 j fr141312 '1110^ I8

I5 *43 N L i D i

\ " °" F0 j +

-' ' „ P DK- T P F C-^ -„ r c. \-_ 3 G R o , , UK

-4 c-S + , T T T , , T , . . + - 4

-22 -18 -14 -10 -6 -2 2

Budget balance

Comn.ôn.erv^

A second point is whether coordination failures in EMUwould lead governments to pursue excessively loose fiscalpolicies. Standard coordination models show that in a non-cooperative equilibrium, fiscal policies are too loose in thepresence of cross-border welfare-reducing effects because inthat case each country would only care about its own welfarewithout taking into account adverse spill-over effects onother EMU members. This issue is taken up in Section 5.1.3

below, where it is argued that no compelling evidence indi-cates that EMU would give rise to a bias in favour ofexcessively expansionary fiscal policies. There is however arisk that without coordination, budget deficits would not becorrected in time and remain too large in the medium runbecause their interest rate cost would be spread across theCommunity. This problem should be addressed throughcoordination and surveillance procedures.

Summing up, there is no compelling evidence that EMUwould have strong adverse effects on fiscal discipline, butthere is no reason either to rely exclusively on markets toenforce discipline, since the conditions for this discipline tobe effective cannot be assumed to hold fully. Thus, there isa case for addressing the risk of failures of market disciplinethrough Community rules and procedures.

5.1.3. The need for coordination

Policy coordination i*, after autonomy and discipline, thethird component of the budgetary regime of EMU. Thepurpose of this section is to analyse the rationale forstrengthening fiscal policy coordination and surveillance.

Much of the policy debate about coordination has so farfocused on global issues at the G7 level, and much less onintra-EC coordination.39 This could be interpreted as a signof success since the EMS has been set up partly in order totrigger monetary and exchange rate coordination. However,there is another explanation: intra-EC and especially intra-ERM policy discussions in the 1980s have been dominatedby the search for a convergence of inflation rates at a lowlevel. To a large extent, fiscal policy discussions have beendominated, and still are, by the need for an adjustment inhigh-deficit and high-debt countries. Due to this priority forconvergence, genuine coordination issues have so far beenleft in the shade. However, it remains an important topicthat could come to the forefront in the years ahead.

Coordination has in reality two different meanings. In thepolicy debate, it often refers to all kinds of discussionsamong governments and central banks. Coordination maythen be implicit and not formalized. Recent developmentsof economic theory have emphasized coordination in thenarrow sense of a formal and negotiated coordination inpolicy-making whose need arises from the game-theoreticaspects of interdependence, and specifically from the risksof collective welfare losses if governments and/or monetary

If one assumes Ricardian equivalence, i.e. that public dissaving is fullycompensated by private saving, the disciplinary effect of the currentaccount constraint disappears entirely. Interestingly, a recent OECDstudy by Nicoletti (1988) rejects Ricardian equivalence for mostcountries, but with the exception of Belgium and Italy.

Exceptions are, for example, Oudiz (1985), Coudert (1987), Amalric andSterdyniak (1989), Cohen and Wyplosz (1989, 1990).

113

sduquenoy


authorities independently aim to achieve contradictory re-sults.40 In what follows, coordination will be used in thisnarrow sense.

Coordination of economic policies is called for in all in-stances where spill-over effects of domestic policies on part-ner countries are not taken into account by the policy-maker.When such externalities arise, independent policy-making isbound to be sub-optimal, but this sub-optimality can gener-ally be corrected by coordinated policy changes, with theresult that the welfare of all participants improves. A pre-requisite to the analysis of coordination is therefore to reviewhow EMU will affect cross-country budgetary policy spill-overs. This provides the basis for discussing whether EMUcalls for a strengthening of coordination procedures. A finalpoint comments on possible costs or benefits of EMU inthis respect.

Are spill-over effects of budgetary policystronger in EMU?

EMU will affect cross-country spill-overs through three dif-ferent channels:

(i) economic union will foster goods and capital marketintegration, thereby affecting the intensity of cross-country spill-overs;

(ii) the adoption of a common monetary policy handledby EuroFed will remove the possibility of beggar-thy-neighbour monetary and exchange rate policies, but atthe same time increase spill-over effects of policies thataffect the interest rate since all EC countries will sharethe same interest rate;

(iii) the common exchange rate vis-a-vis the rest of the worldwill be turned into a major spill-over channel, sincebudgetary policy in any country will affect this EC-widevariable. In the absence of coordination, the currentaccount of the Community becomes for Member Statesa kind of 'public good' whose determination is beyondthe reach of any single policy body.41

A classic example in the monetary field drawn from the experience ofthe 1970s is the attempt to export the inflationary consequences of asupply shock (e.g. oil) by appreciating the currency: when all countriessimultaneously (but in vain) try io achieve this result, monetary policyis too tight, such that policy coordination, alternatively, would achievea Pa re to-superior outcome which provides welfare gains to all participat-ing countries.Both of the last two effects arise from the same cause, namely monetaryunion. However, analytically they are of a different nature because theformer arises in a two-country world whereas the latter depends on therelations with third countries. See Cohen and Wyplosz (1989).

Thus on the one hand, EMU will remove the risk of non-cooperative monetary policies, but on the other hand itcould increase externalities of fiscal policies and thereforecall for stronger fiscal coordination. For assessing fiscalpolicy spill-overs, different approaches may be followed. Afirst approach relates to the conventional cross-country ef-fects of fiscal policies in the short to medium term, i.e. tothose effects which arise while adjustments to goods andlabour markets are still under way (this can be related towhat have been called the 'stabilization' aspects of fiscalpolicy in Section 5.1.1 above). A second approach focuseson the specific, but important, type of externalities arisingfrom the sharing of a common current account and a com-mon exchange rate. A third approach relates to longer-termchannels of externalities which can arise from long-lastingchanges in fiscal position after adjustments in relative priceshave been completed (i.e. to the 'adjustment' role of fiscalpolicy).

(i) In the short to medium term, i.e. in a period up to fiveyears, fiscal policy primarily impacts on aggregate demand,and the main concern of policy-makers is with respect togrowth and inflation. Spill-overs arise, first from incomeand price transmission through trade flows, and second,from interest rate effects on demand. Interest rate changesalso influence trade volumes and prices via the exchangerate. The standard framework of analysis for these effects isthe Mundell-Fleming model in which economies character-ized by sticky prices and unemployment are linked by goodsand capital markets. Although some of the simplifying as-sumptions of the original Mundell-Fleming model have tobe relaxed, it remains a very useful framework from whichclear conclusions can be drawn.42 Major results are thefollowing:(a) Goods market integration increases spill-overs through

both demand and relative price effects.(b) In a fixed exchange rate regime, capital market liberaliza-

tion, which increases capital mobility and asset substitu-tability, increases the effectiveness of fiscal policy athome because access to world capital markets reduces(or, for a small country, eliminates altogether) crowding-cut effects. It also reduces the 'locomotive' character offiscal policy, i.e. its impact abroad, possibly leading tonegative spill-overs of expansionary policies.

(c) With free capital markets and therefore high capitalmobility, the domestic impact of fiscal policies on outputis minimal with fully flexible exchange rates, and maximalwith fixed exchange rates.

Major modifications of the Mundell-Fleming model relate to pricedetermination and the exchange rate. For a recent and comprehensivepresentation, see Frenkel and Razin (1987). For a discussion of theeffects of EMU in a Mundell-Fleming framework, see Van der Ploeg(1989, 1990), from which this presentation draws.

114


The move towards EMU therefore amplifies the domesticeffectiveness of national fiscal policy for stabilization pur-poses. The assessment of spill-overs is less straightforwardbecause they depend both on the relative magnitude ofincome and interest rate effects, and on the asymmetries ofthe exchange rate regime. In a pure float, fiscal policy tendsto be expansionary abroad due to both demand and competi-tiveness effects (Table 5.4). Indeed, under perfect capitalmobility a fiscal expansion appreciates the exchange rate,and hence increases the competitiveness of the partnercountries. Within an asymmetric EMS, fiscal expansion inthe anchor country might be beggar-thy-neighbour becauseinterest rate and exchange rate effects could dominate in-come effects. This is because Germany determines interestrates and dollar exchange rates for the whole EMS, whilstexports to Germany do not exceed 3 % of GDP in France,Italy or the UK; hence, the 'monetary' weight of Germanyexceeds its 'economic' weight within the system. Althoughthe sign of the aggregate spill-over effect of a fiscal expansionin Germany is an empirical issue, there is a presumptionthat it could be negative. In contrast, a fiscal expansion ina non-anchor country has a 'locomotive' effect because ithas no direct interest rate effect.

In a symmetric regime like the EMU, under the assumptionof common, stability-oriented monetary policy, the sign ofthese spill-over effects is theoretically ambiguous: fiscal pol-

Table 5.4Short-term spill-over effects of a bond-financed fiscal expansion underalternative exchange rate regimes

Country originating the policy moveExchange rate regime

EMS anchorcountry

Othercountries

FloatEMSEMU

+ /- to -

Note: Signs in this table refer to those of spill-over effects on the GDP of other Member StalesIn all regimes, full capital mobility is assumed and European currencies are supposed to floatwith respect to non-European currencies. Monetary policy is assumed to be directed towardsprice stability.

icy in one Member State can either increase or decreaseoutput abroad since the 'monetary' weight of each MemberState, which depends on its share in the aggregate indicators(prices, money demand, etc.) upon which EuroFed bases itspolicy, is close to its 'economic' weight. Thus, the sign ofthe spill-over effect depends on the relative magnitude of theincome effects with respect to interest rate and exchange rateeffects. However, the direction of change when one movesfrom the EMS towards EMU is clear since fiscal policyin the anchor country tends to become less 'beggar-thy-neighbour' while it tends to become less 'locomotive' in theother countries.

Model simulations with Multimod and Quest broadly con-firm the above qualitative conclusions (Box 5.3). Concen-trating on the effects of the move from the EMS towardsEMU, simulation results show that in EMU the impact ofa bond-financed expansion in the EMS anchor country(Germany) becomes more expansionary at home and lesscontractionary abroad since monetary policy reacts to theEC's economic performance rather to that of Germany. Theopposite is true for France, for which the major changeconcerns the interest rate/exchange rate nexus. Generallyspeaking, short-term spill-over effects of fiscal policies ap-pear to be small in EMU due to the offsetting effects ofimport growth and interest rate/exchange rate increases.

These results tend to show that under normal circumstances,national budgetary autonomy in EMU should not leadgovernments to opt for suboptimal policy stances becauseof a failure to take spill-over effects properly into account.In other words, there does not seem to be an a priori casefor a major strengthening of coordination in day-to-daypolicy-making. However, these results depend both on theassumption of no monetary accommodation and on thecharacteristics of the model. Furthermore, they cruciallyhinge on the assumption that under a float, a fiscal expansionappreciates the exchange rate.

Although this normally holds once capital mobility is effec-tive, exchange rate determination is admittedly not preciseenough to rule out the possibility of opposite effects. In thatcase, positive spill-over effects of fiscal policy would be muchstronger.

115


Box 5.3: Shorter-run spill-over effects of fiscal policies

Model simulations can help to assess the sensitivity of spill-overeffects to different exchange rate regimes. However, only someof the linkages are sensitive to the exchange rate regime. Demand

impacts on goods markets due to a bond-financed fiscal expan-sion are quite independent of the exchange rate regime. On theother hand interest rate and exchange rate linkages are, incontrast, a function of the exchange rate regime. Four regimesare considered in Table 5.5 below.

Table 5.5Spill-over effects of a rise in government expenditures by 2 % of GDP, Multimod simulation

Country originating the policy move:Germany

EMS AEMU Float EMS AEMU

GDP (first year effect)

Germany 1,1France 0,06Italy 0,12United Kingdom -0,02

Short-term interest rate (first year effect)

GermanyFranceItalyUnited Kingdom

0,20,180,260,26

Dollar exchange rate (first year effect)

Germany - 2,2France - 0,48Italy -0,38United Kingdom - 0,44

Absorption deflator (fifth year effect)

GermanyFranceItalyUnited Kingdom

1,30,40,520,4

1,06-0,24-0,24-0,26

0,260,40,420,42

-2,26-1,9-1,88-1,84

1,24-0,34-0,44-0,22

1,040,34

-0,36-0,32

0,260,260,260,26

-2,28-2,28-2,28-2,28

1,2-0,62-0,76-0,36

1,600

-0,02

0,180,180,180,18

0,9•0,9•0,90,9

2,00,160,10,1

0,121,180,10,04

0,120,280,180,14

0,44-3,260,28

-0,28

0,262,060,340,28

0,141,620,100,06

0,12•0,120,180,14

-0,421,300,46

-0,42

0,343,480,240,24

0,181,740,10,06

0,10,10,10,1

-0,30,3

-0,3-0,3

0,263,40,180,16

-0,161,60,08

-0,1

0,140,140,140,14

1,221,221,221,22

03,66

-0,34-0,14

Source: Multimod simulations by the Commission services. Standard monetary policy, A decrease in the dollar exchange rate means appreciation. AEMU: asymmelric EMU.

Since the EMS is asymmetric, it is necessary to consider boththe effects of a fiscal expansion in Germany and another ERMmember, here France (the results would be similar with anotherlarge country). The asymmetry between these two countries isapparent for both the EMS and the asymmetric EMU, since afiscal expansion in France has much smaller effects on theinterest rate and the exchange rate than a fiscal expansion inGermany.

In a pure float regime, most of the effects of the expansion arebottled up in the originating country, which experiences a risein interest rates and an appreciation of the currency. Spill-overeffects are mainly positive, but small.

In the EMS, a fiscal expansion in Germany is 'beggar-thy-neighbour' in character and provokes disinflation abroad be-cause interest rates and exchange rates increase for all members

116


of the ERM {although the fluctuation bands of the EMS allowfor some offsetting of this effect); in contrast, fiscal policy inFrance is 'locomotive' in character (and inflationary for ERMmembers) because trade linkage effects dominate.43

The hypothetical 'asymmetric EMU' is intermediate betweenEMS and EMU. Since monetary policy is entirely set by Ger-many (without the flexibility that bands allow for), the fiscalpolicy multiplier is higher than in any other regime in France;44

in contrast, the beggar-thy-neighbour effects of the Germanfiscal expansion are amplified. Price effects are also amplifiedwith respect to the EMS case.

In EMU, the effects of a fiscal policy are qualitatively the samefor Germany and France, since in both cases the reaction ofmonetary authorities is supposed to be identical. For the Ger-man expansion, trade and interest rate linkages turn out tooffset each other almost completely. In the French case, beggar-thy-neighbour effects dominate slightly.45 Inflation spill-oversare small.

43 The small decline in the French interest rate is due lo the reaction of the Frenchauthorities to [he appreciation of the currency within CMS bands.

44 It is, however, still low in comparison with the results of standard macro models. Thisis due, first, to the remaining interest rate and exchange rate effects (because Franceis not a smaller economy within (he EC), and second lo the forward-looking natureof the model, which reduces the effectiveness of fiscal policy.

4S Different sizes, degrees of openness, and wage-price behaviour in Germany and Franceaccount for a large part of the difference between the simulation results for the twocountries. Another factor is a technical hypothesis made for these simulations, namelythat Ihe weighting of country variables for Ihe compulation of EC aggregates whichenter the reaction function is equal for France and Germany. A more realistic assump-tion would be to weight countries according to their GDP. This would reduce thecontrast between France and Germany in the EMU regime.

The spill-over effects broadly conform to the qualitative resultswhich can be derived from a standard Mundell- Fleming model.However, the results could be model-specific since the sign ofthe spill-over effects depends on the relative intensity of tradeand financial linkages. Therefore, a Quest linked simulation hasalso been carried out to test for the size of spill-over effects of aGerman fiscal expansion under an EMU regime, setting theexogenous value of interest rates and exchange rates at levelsconsistent with the monetary policy reactions function and theopen interest rate parity condition (Table 5.6). Under the samekind of assumptions regarding monetary policy and exchangerates, these results broadly confirm the small size of spill-overeffects of fiscal policy in EMU.

Table 5.6Spill-over effects of a rise in German government expenditures by2% of GDP under an EMU regime, Quest simulation

GDP Interest rate Dollar Absorpiionexchange deflator1

rate

Germany 2,1 1,5 -2,7 4,5France 0,14 1,5 -2,7 -1,5

. Italy 0,18 1,5 -2,7 -0,9UnitedKingdom 0,34 1,5 -2,7 -1,71 First-year effect for GDP, interest rates and exchange rate. Fifth-year effect for absorp-

tion deflator. Interest rate and exchange rate effects are exogenous.Source: Quest simulation by the Commission services.

(ii) The sharing of a common exchange rate vis-a-vis thirdcurrencies and of a common external balance will be directconsequences of EMU. As no individual Member State willbe able to target those variables, the fact that these objectiveswill be shared will appear to policy-makers as the mostvisible sign of the increased interdependence across the Com-munity.

To what extent will this channel of externalities be importantin practice? The current account and the exchange rate ofthe Community will probably remain two among manypolicy objectives, as long as their level and evolution do notdepart from normal limits. However, since the variability ofboth variables is high, circumstances will presumably arisewhere their evolution will be regarded as of paramountimportance by European policy-makers. For instance, if the

external deficit of the Community reaches certain limits, itwill become the key policy target. Since monetary policy hasan ambiguous action on the external account (because ofthe opposite effects on foreign trade of a slower domesticdemand and of exchange rate appreciation), correction ofthis deficit will presumably need a fiscal policy move. How-ever the benefits of a fiscal correction by a national govern-ment would accrue to the Community as a whole (through areduction in the external deficit), whereas the home economywould bear almost all the costs. This is a textbook exampleof strong externality.

(iii) In the long run, the main focus shifts from growthjunemployment and inflation spill-overs to externalitiesthrough real interest rates and terms of trade effects.46

The following draws on Wyplosz (1990).

117

sduquenoy


Assuming that in the long run price flexibility ensures marketclearing, and that all economies are at full employment, theissue is whether and how a protracted deficit in a MemberState affects its neighbours through higher real interest ratesand a deterioration of the terms of trade.47 This can berelated to the ongoing discussion regarding what has beencalled the risk of 'undue appropriation of EMU savings'(Lamfalussy, 1989).

Since EMU removes the possibility of adjusting the nominalexchange rate, the rise in real interest rates can no longer belimited to the country running a budget deficit and is felt inthe union as a whole. As the deficit country experiences a realexchange rate appreciation, this also affects its neighbours.These terms of trade effects are not specific to EMU, butEMU makes two differences: first, some of the change in

1 the terms of trade would be common to all EMU members,because it would arise from changes in the value of the ecuwith respect to third currencies; second, intra-EC terms oftrade changes would have to be achieved through changes inthe price system instead of the exchange rate, with associatedadjustment and menu costs.

Budgetary policy coordination in EMU

The above analysis provides the basis for a discussion of theneed for budgetary coordination in EMU. What matters is(i) whether in EMU coordination is likely to be a majorrequirement in policy-making and (ii) whether uncoordi-nated policies would tend to be too lax or too tight.

Turning first to coordination of short and medium-termpolicies, the evidence presented above does not call for atight coordination of day-to-day national demand policies,since within the EC cross-country spill-over effects ongrowth and inflation are not likely to be much greater thanat present. As long as no special circumstance calls for jointaction, existing coordination and surveillance procedures,which provide channels for information exchange and con-sultation, would be adequate. However, this conclusion im-mediately has to be qualified because of the externalitiesthrough the common exchange rate and current account. Asexemplified by the experience of the industrialized countriesin the 1980s, conflicts over shared targets like the exchangerate and the current account are in practice much strongerincentives to coordination than an inadequate overall policy

stance.48 Although simulations (see Box 5.3) confirm thatspill-over effects of fiscal policies through the exchange rateare already present in the EMS, non-anchor Member Statestheoretically retain the possibility to settle for a realignment.This possibility would no longer exist in EMU.

When needed, coordination with respect to the commonexchange rate and the common external account would haveto be tight since failure to take joint action would severelyimpact on these important variables. In the absence of coor-dination, no national government would be willing to deflatefor the benefit of the Community as a whole. Even if anumber of Member States were tempted to take action, otherMember States could be willing to free-ride, i.e. to reap partof the benefits of this action without taking part in it. Thus,in such a situation a prima facie case for coordination and,surveillance would arise. A related motive for greater fiscalcoordination within EMU would stem from the need todefine a policy for the Community in the larger context ofthe world economy.49 Even if intra-EC spill-overs did notcall at all for fiscal coordination, as argued in Chapter 7,there would still be a need for coordination in the search fora good policy mix because the Community budget wouldremain too small to have a significant macroeconomic im-pact, Without fiscal coordination, only EuroFed would beable to engage in policy coordination at the world level and,in the absence of a fiscal policy actor, could be subject topressures to adopt an inadequate stance. In this respect, thepreservation of EuroFed's independence could well call forfiscal coordination.

As the need for coordinated policies would presumably notbe a day-to-day feature of fiscal policy in the Community,but could become the priority in some circumstances, specialcoordination procedures could be designed in order toguarantee an efficient handling of these external aspects ofinterdependence. These procedures would have to ensurethat commonly agreed decisions would be enforced and toaddress the risk of free-rider behaviour.

These needs for coordination would not conflict with theneed for autonomy since what would be called for is achange in the budgetary policy of the Community as a whole.

The discussion here focuses on 'protracted' deficits and not on perma-nent deficits since as discussed by Wyplosz (1990), in an intertemporalsetting the budget constraint has to be fulfilled and a budget deficit hasultimately to be financed through higher taxes. Therefore, a 'permanent'deficit is in reality a rise in the size of the government. These EMUeffects are discussed in Section 5.2.3 below.

In other words, although the economic literature emphasizes the gainsfrom absolute coordination, real coordination tends to be relative, andto focus on the settlement of possible disputes over these shared targets(Currie, Holtham and Hughes Hallett, 1989). To some extent, theexperience with the ERM confirms the prevalence of relative coordi-nation since discussions over the right level of the exchange rates weremuch more frequent than discussions about the overall policy stance ofthe Community.See Lamfalussy (1989), Padoa-Schioppa (1990), Cohen and Wyplosz(1989).

118

sduquenoy


This would be perfectly compatible with differentiated na-tional positions. In other words, coordination would haveto ensure that the overall policy mix of the Communityis adequate, whereas differentiated national policies wouldaddress relativities, i.e. national macroeconomic perform-ances with respect to the rest of the Community.

A related issue is whether coordination failures would induceexcessively lax short-term fiscal policies. According to stan-dard coordination models, this would be the case if fiscalexpansions were systematically beggar-thy-neighbour incharacter in EMU, since governments would tend to over-look the adverse consequences of their actions on partnercountries. As discussed above, the evidence derived fromempirical models does not unambiguously point in thatdirection. Although fiscal policies in non-anchor countrieswould become less expansionary in their external impactthan in the present EMS, this would mainly be a correctionfor an existing bias since the interest rate and exchangerate crowding-out effects would roughly compensate forthe income effect. Indeed, an opposite bias could arise ifindividual Member States were seeking employment andoutput gains through real depreciation.50

As to coordination in the medium to long term, more clear-cut conclusions can be drawn from analysis. Assuming fullemployment, spill-over effects of a protracted deficit wouldbe felt by partner countries through higher real interest ratesand a worsening of the terms of trade. Since these adverseconsequences of domestic policy would be overlooked with-out coordination, a tendency for budget deficits to be toolarge could appear.51 In this instance, coordination andsurveillance would be needed in order to avoid too lax apolicy.

Summing up, there is no compelling evidence that EMUwould call for close coordination in day-to-day fiscal policy.However, coordination would be required in two instances:(i) in order to ensure an appropriate policy mix of the

Community in the context of the world economy, par-ticularly with respect to the common exchange rate andcurrent account;

(ii) in order to correct possible medium-term tendencies forbudget deficits to be too large.

In both cases, mutual surveillance procedures should ex-plicitly address the risk of free-rider behaviour.

Does EMU yield coordination gains?

International macroeconomic policy coordination is gener-ally considered as yielding welfare gains to the participatingcountries.52 Therefore, the question arises whether sucheffects should be considered in the cost-benefit analysis ofEMU.

Since EMU would remove an instrument (the nominal ex-change rate) whose cooperative setting would be welfare-improving, there is no doubt that it would entail a loss incomparison to full coordination in the management of theexchange rates.53 This is just another way to present thecost of losing the exchange rate instrument which is discussedin Chapter 6. However, this is not fully relevant, because thereference situation to which EMU is to be compared is notone where coordination can be assumed to be perfect. Withrespect to a floating rates regime, EMU would removethe inefficiencies associated with non-cooperative monetarypolicies. It could therefore be considered a form of "surrogatecoordination' but at the cost of nominal exchange raterigidity.54 With respect to a '1992 + EMS' baseline, EMUwould not yield the same kind of gains. Indeed, the EMS isalready a form of surrogate coordination which practicallyrules out non-cooperative exchange rate management, theassociated cost being the suboptimality of an asymmetricsystem when non-anchor countries are hit by shocks (plus,at least in Stage I, part of the cost resulting from nominalexchange rate rigidity). Thus, EMU would substitute therigidity cost for the asymmetry cost.

The relevant comparison therefore has to be made betweentwo second bests whose relative performance would have tobe assessed (i) with respect to a given distribution of shocks,and (ii) taking into account the degree of policy coordinationthat could be expected in each case. A quantitative compari-son is clearly out of reach.

See Cohen and Wyplosz (1990), Wyplosz (1990),This assumes that each government determines the size of its budgetdeficit according to domestic considerations, i.e. by maximizing a welfarefunction which takes into account the impact of higher interest rates athome but not abroad. Whether this could be called'undue appropriationof Community saving' is not so clear. As pointed out by Bredenkampand Deppler (1989), this appropriation cannot be qualified as undue ifthe return on government investment equals or exceeds the market realinterest rate.

See, for example, Oudiz and Sachs (1984). Gains from coordination arehowever disputed. For a recent survey, see Currie, Holtham and HughesHallet(1989).This is demonstrated by Giavazzi and Giovannini (1987) and discussedby Van der Ploeg (1989, 1990).See Hughes Hallet, Holtham and Huston (1989).

119


5.2. Impact on income and expenditure ofgovernments

This section focuses on the effects of EMU on the resourcesand expenditure of governments.

Three types of effects have to be considered. First, as in-flation rates converge, governments lose the possibility ofderiving resources from a higher inflation rate, eitherthrough seigniorage revenues or, in the short run, becauseof the non-indexation of tax brackets and rates. This issueof seigniorage losses has been much discussed in the recentacademic literature. However, as shown in Section 5.2.1, itsempirical relevance appears to be relatively minor in theCommunity context.

A second effect concerns government expenditure. As thenominal interest rate is reduced in the most inflationaryMember States, the public debt service and therefore alsothe recorded deficit shrink. This nominal effect also goeswith a real, but mostly transitional one since real interestrates are presently significantly higher in the most inflation-prone countries. This is the topic of Section 5.2.2.

These two effects are direct consequences of monetary union.However the broader issue of tax and spending convergencein the Community has also to be raised in relation to EMU.To what degree, and through which mechanisms, should thisconvergence be achieved? This theme is briefly addressed inSection 5.2.3.

5.2.1. Seigniorage revenue losses

Seigniorage is the ability of the government to finance itsexpenditure by issuing money. Governments extract seig-niorage revenues (also called the inflation tax) through theissuance of non- or low-interest bearing debt which is heldby the public in the form of currency or by the commercialbanks in the form of reserves at the central bank.55 For thisreason, even fully anticipated steady-state inflation is notneutral with respect to public finance.

The possibility of Financing government expenditure by issu-ing money was for long a seigniorial privilege, later accruing

to the State. Such a financing of large budget deficits has inthe past been at the origin of major inflationary crises. Butseigniorage can also accrue to governments without large-scale debt monetization, and constitute a steady-state chan-nel of government revenue. It is therefore a legitimate matterof concern in the design of a monetary union. There are, inaddition, other reasons for inflation to affect public finance.As a rule, tax systems are not neutral with respect to inflationsince many taxes are based on nominal instead of realincomes. This could be called another form of seigniorage(or inflation tax) and will be briefly discussed at the end ofthe present section.

Seigniorage is a major source of government revenues onlyfor economies suffering high inflation. However, it has alsobeen significant in recent years for some Member States dueto above average inflation, the wide use of cash for day-to-day transactions and high bank reserve ratios (Graph 5.5).EMU will not lead seigniorage to disappear, but as arguedbelow both the likely convergence of compulsory bank re-serve ratios on a low level and the low inflation performanceof the Community will reduce this resource to a modestlevel.56 For those countries which still significantly rely onthe inflation tax to finance their budget, EMU will entailpermanent ex-ante government revenue losses, which willhave to be compensated by a rise in taxes or a cut inexpenditure. Therefore, two questions arise:

(i) What is the gross public finance cost of reducing in-flation, taking into account that independently of monet-ary union, changes in the banking legislation broughtby the single market will induce convergence in the levelsof required reserves?

(ii) Is there a net economic welfare cost of permanentlyreplacing the inflation tax by explicit taxes?57

Obviously, the issue of seigniorage revenues only arises incomparison to a situation where Member States are free tochoose independently their inflation rates. As argued inChapter 2, this would already not be the case in Stage I.Therefore, this section implicitly refers to another baseline,

Bank reserves yield zero interest rate in most EC countries. However,reserves are remunerated, but below market rates, in countries character-ized by high reserve requirements. Hence, seigniorage revenues are notproportional to the size of the monetary base. These revenues do notonly appear as central bank profits, but also as lower interest paymentsin the government's accounts since in some countries the Treasuryborrows from the Central Bank on subsidized terms. For an examinationalong these lines of the case of Italy, see Mohlo (1989).

We do not discuss how seigniorage revenues will be redistributed amongMember States. This issue is related to the distribution of the capital ofEuroFed.For high-debt countries, the reduction in seigniorage revenues can im-pact on the dynamic of the public debt as shown in Section 5.1. Whetherthis could be an argument for giving priority to fiscal consolidation overdisinflation is discussed in Section 5.2.2 below.

120


which is either a floating regime or a soft exchange ratearrangement (like a crawling peg) which allows for perma-nent inflation differentials.

GRAPH 5.5: Seigniorage revenues in the Community, 1988

(as a percenlage of GDP)

4 , - —

3,5 H

3 •

». 2,5 1

1,5 >i

I 4

0,5 i

0

v/.

UK DK NL F B IRL D I P OR

and Greece and, to a lesser extent, Spain and Italy). Closerexamination shows that as the four countries are character-ized by both a wider use of cash (except for Italy) and higherreserve requirements than in the rest of the Community, thisdifference is rooted in both technological and regulatoryfactors. For these countries, an evaluation of the publicfinance cost of EMU is therefore warranted.

However, this cost cannot be assessed readily on the basisof present seigniorage revenues, since, following the fullimplementation of the Second Banking Directive by 1993,bank competition in the internal market will already imposeconvergence in reserve requirement ratios. Although thisdirective does not address monetary regulation per se, higherreserve requirements act like a tax on home banks whichwould be a disadvantage in international competition. Mar-ket pressures can therefore be expected to lead to conver-gence in the reserve ratios. In the same period, the use ofcash is also expected to decline due to the modernization ofthe payment system. In order to evaluate the specific effectof monetary union, one has to make hypotheses regardingthe convergence of reserve ratios and ratios of currency toGDP brought by technological change and the completionof the internal market.58

Aod Seigniorage revenues have been computed using the opportunity cost definitionof seigniorage, i e s i*c » (i-ir)*r where i is the market short-term interest rate, i,is the mteresi rate on (compulsory) bank rc^rves. and c and r are respectively theratios of currency and bank reserves to GDP It is implicitly assumed that the sameinterest rate applies on free reserves. See Gros (19891 for details.

Gross public finance cost of seigniorage losses

Seigniorage can be measured either as a flow of cash or asan implicit revenue. The first measure, which is the mostcommonly used in the literature, is given by the ratio of thechange in monetary base outstanding to nominal GDP.However, it can exhibit wide variations over time and is notappropriate if bank reserves, which are part of the monetarybase, are remunerated. The second measure is based on thedifference between the effective interest rate on the monetarybase (which is zero for cash, but can be positive for requiredbank reserves) and the market interest rate. It is much morestable over time, and is therefore used in what follows.However, it should be kept in mind that the correspondingrevenues result from an accounting imputation.

Graph 5.5 shows that seigniorage revenues, which are closeto 0,5 % of GDP or even below in most Member States, arenevertheless significant in the southern countries (Portugal

Table 5.7 gives, for the four Member States whose seignior-age revenues were above 1 % of GDP in 1988, an assessmentof the gross public finance cost of EMU. This evaluationis based on the comparison of two scenarios. Scenario 1incorporates in '1993' the full effects of the internal marketprogramme: convergence in the reserve ratios at a 2 % level,elimination of interest payments on reserves, further re-ductions in the use of cash due to technological changes;but while real interest rates are assumed to converge, noconvergence in inflation rates is assumed. Scenario 2 or*EMU' assumes in addition convergence in inflation rates ata 2 % per annum level.59

Interestingly, Spain has already in 1990 legislated a progressive reductionin the required reserve ratio from 17 % to 5 %.The 2% reserve ratio hypothesis in scenario 1 ('1993') corresponds tothe average level of the other EUR 8 countries. Giovannini (1990)assumes 1,5% reserve ratios after the liberalization of the bankingindustry. Currency to GDP ratios have been projected using the 1979-88 time trends. Real interest rates are supposed to converge at the 5 %level. Inflation is supposed to remain at the level forecast for 1991. Inboth scenarios, seigniorage revenues are distributed among MemberStates according to their monetary base.

121

sduquenoy


Table 5.7

Gross seigniorage revenue effects of monetary union(Seigniorage revenues 01 a percentage ofGDP)

GreecePortugalSpainItaly

1982-84(1)

2,464,391,932,23

1985-87(2)

2,342,851,031,21

19880)

2,752,231,361,13

•1993'(4)

1,841,621,200,72

•EMU'(5)

0,710,710,860,51

Single marketeffect <6) = <4)-<3)

0,910,610,160,41

EMU effect(7) = (5)-(4)

1,130,910,340,21

Source: Commission services.

As shown in Table 5.7, except for Greece seigniorage rev-enues are already well below their levels of the early 1980s.Even assuming that without EMU governments would notchoose to reduce inflation, the gross public finance cost ofseigniorage revenue losses implied by monetary union onlyexceeds 1 % of GDP in Greece, because inflation in thiscountry would still be at a high 15% level in the '1993'scenario. However, this is clearly an upper bound since evenwithout participation in EMU, Greece would be very likelyto disinflate. The same is true to a lesser extent for Portugalwhose inflation is still above 10 % in the first scenario. ForSpain and Italy, however, the EMU effect is below 0,5 % ofGDP.

Welfare effects of seigniorage losses

To the extent that a reduction of seigniorage revenues elimin-ates a perverse incentive towards inflation and fiscal laxity,it should be seen as a source of welfare gains whatever itsgross public finance cost. However, the question has to beraised whether there are good economic reasons to preferseigniorage finance over tax finance. The optimal taxationtheory has pointed out that since taxes are distortionary,a welfare-maximizing government would not choose zeroinflation, but rather combine explicit taxes and the inflationtax in order to minimize distortions.60 Hence, countriescharacterized by distortionary taxes, less efficient tax collec-tion, or a larger underground economy should accept ahigher inflation rate. In this respect, it has been argued that

participation in a low-inflation EMU could be suboptimalfor southern and/or catching-up Member States because itwould lead them either to raise taxes or to lower publicspending excessively.61

Although this argument has its logic, its pertinence is ques-tionable since it relies on the assumption that economicpolicy can choose — and does choose — a 'socially optimalinflation rate1 and, moreover, can stick permanently to thisinflation rate without incurring credibility losses. There isno evidence, however, that past or present inflation rateswould be optimal.62 Moreover, the possibility of achievingthat optimal mix of taxation and inflation is theoreticallydisputed: since the governments of the four Member Statesunder consideration are all heavily indebted — Spain beingin a better situation than the other three — they face astrong incentive to reduce the ex-post real cost of servicingthe public debt through one-shot surprise monetization, i.e.inflation. This motive interferes with the choice of an optimalsteady-state inflation rate and, to the extent that it is awareof this temptation, leads the public to expect inflation to behigh. If this happens, the country reaches an inefficientequilibrium: inflation is permanently above the optimal leveland the country therefore endures welfare losses but, since

In technical terms, the marginal social cost of raising revenue throughdirect taxation and the marginal cost of raising revenue through seignior-age should be equalized. See Phelps (1973) and, for a recent presentation,Mankiw{1987).

This issue has first been raised by Dornbusch (1988), Subsequent analy-ses include, for example, Drazen (1989), Giavazzi (1989), Grilli (1989)andGros(1989a, 1989b).Poterba and Rotemberg (1990) find no evidence of correlation betweeninflation rates and tax rates in a sample of OECD countries. Cukierman,Edwards and Tabellini (1989), working on a sample of developed aswell as developing countries, find no correlation between inflation andstructural economic variables and argue that inefficiencies in the taxsystem leading to resort to seigniorage should not be regarded asexogenous, but rather as linked to features of the political system, e.g.polarization and instability.

122


the price rise has been anticipated, without any gain regard-ing the debt-service burden of the government. In theseconditions, participation in a low-inflation EMU is welfare-improving even if it forces the choice of an inflation ratebelow the optimal level.63

These objections notwithstanding, simple computationshave been be made in order to assess the maximum welfarecost of the seigniorage loss. Computations with a numericalexample (Appendix 1 to this chapter) show that the welfarecost of EMU is in any case much smaller than the grosspublic finance cost for two reasons: first, the optimal in-flation rate is not independent from the reserve ratio, butdecreases with it; second, the welfare cost of higher taxeshas to be weighed against the welfare gain of lower inflation.Therefore, calculations with this simple model indicate thateven assuming for illustration that the 'optimal' inflationrate is presently 10 % (which is a quite pessimistic assump-tion), and that seigniorage revenues amount to 1,5% ofGDP, the welfare cost of seigniorage losses due to monetaryunion would not exceed that of a ceieris paribus rise in taxesby a third of a percentage point.M

Taken together, the above developments show that the wel-fare cost of EMU for countries facing seigniorage losses canbe considered minor to negligible.

Inflation and the tax system

As mentioned above, seigniorage is not the only channel forinflation to affect real government revenues. Tax systemsare rarely neutral with respect to inflation. Among the mostimportant factors are delays in the collection of taxes (e.g.for VAT or income tax); lags in the indexation (even non-indexation) of the income tax brackets and/or the socialsecurity benefits; non-discrimination between nominal andreal interest income (also nominal and real capital gains);non-discrimination between nominal and real capital con-

The problem arises because the optimal public finance policy of choosingthe right mix of taxation and inflation is time-inconsistent. The argumentsummarized in the text is fully spelled out in Gros (1989a) and Van derPloeg (1990), where it is shown that the higher the public debt, thehigher the incentive to engineer surprise inflation and the more a countrygains from participation in EMU. Although tax distortions reduce thisbenefit, back-of-the-envelope calculations show that under the assump-tions, of the model (i.e. welfare maximization by the authorities andrational expectations by the public), the present level of the debt toGDP ratio is above its critical value in the four Member States.In a recent paper, Canzioneri and Rogers (1989) use a different approach:they consider the trade-off between seigniorage and the reduction intransaction costs involved in the choice of forming a monetary union.Even assuming that the black economy is about 20% of GDP in Italy,they conclude that even very small conversion costs outweigh the welfarelosses arising from seigniorage revenue losses.

sumption allowances and interest charges in the computationof taxable profits of the firms.

As apparent in the above list, inflation can either increaseor decrease government revenues. However, there is a funda-mental difference between the effects of seigniorage andthose of the non-neutrality of inflation with respect to thetax system. Although the latter can be significant in theshort term, there is no welfare motive for taxes not todiscriminate between nominal and real income. Actually,taxation of nominal interest income, or deductions of nomi-nal interest charges and capital allowances are highly distor-tionary because they lead individuals and firms to subopti-mal investment choices. A standard consequence of non-indexation in the tax system is therefore that a rise in therate of inflation increases distortions and therefore reduceswelfare. For this reason, even if a low-inflation EMU wouldlead to an increase in tax rates in order to compensate forrevenue losses, there is no a priori evidence that this wouldbe welfare-reducing.

5.2.2. Interest rates and public debt service

Interest payments on the public debt amount approximatelyto 5 % of GDP in the Community as a whole. In the mostindebted countries like Belgium, Greece, Ireland, and Italy,interest payments are close to 10 % of GDP or even above,and represent more than 20 % of total government expendi-ture. Hence, any effect of EMU on the interest rate on publicdebt would be a major issue for public finance.

Graph 5.6 depicts the relation between long-term nominalinterest rates and inflation in the Community. Obviously,most of the variance in interest rates is accounted for byinflation differentials. Member States experiencing higherinflation also experience higher interest rates and thereforehigher interest burden on the public debt. However, this isbasically a nominal phenomenon: increased interest pay-ments on the public debt only compensate for the realdepreciation of the principal. In so far as participation inEMU would reduce both inflation and the nominal interestrate, leaving the real interest rate unchanged, this would cutthe budget deficit, but simultaneously reduce real de-preciation on existing public debt.

Some real consequences would nevertheless arise because ofthe tax treatment of nominal interest income and capitallosses. In addition, the public perception of the deficit issuewould surely be affected since the policy debate frequentlyfocuses on the observed deficit, without consideration fordebt depreciation effects.

123


GRAPH 5.6: Long-term nominal interest rates and inflation.

u28g

_c'g500B,3

average 1985-89

1918

161514131211109876543210 — ———

•

.

.---.

-

-

PmGR

In °E.pi noDK

FdUK"Bn

o

i i i i i i i i . . i , i i i 10 2 4 6 8 10 12 14 16

Inflation (GDP deflator)

GRAPH 5.7: Ex-post real long-term interest rates andinflation, average 1985-89

IRL; DK

NL

UK

Pn°GRI

2 4 6 8 10 12 14 16

Inflation (GDP deflator)

However EMU will also affect real interest charges on thepublic debt, at least in a transitory way, because for inflation-prone Member States it will enhance the credibility of thecommitment to exchange rate and price stability. As shownin Graph 5.7, in the second half of the 1980s ex-post reallong-term interest rates were far from being equalized acrossthe Community. They were high by historical standards andabove the German level for ERM countries, quite indepen-dently of the inflation differential. For Greece and Portugal,ex-post real rates were lower because of the maintenance ofcapital controls.

EMU and real interest rate differentials

A decomposition of the factors behind interest rate differen-tials helps to clarify the potential effects of EMU on realinterest rates.

Cross-country differences in real ex-post (or realized) interestrate on assets of the same category and specific risk can beanalytically decomposed into four components:

(i) country premiums resulting from capital controls andother limitations to capital mobility;

(ii) exchange risk premiums resulting from exchange ratevariability;

(iii) real exchange rate depreciation;

(iv) expectational errors regarding the evolution of the ex-change rate.

Each of these factors is considered separately below. Thedecomposition is carried out analytically in Appendix 2 tothis chapter.

(i) Capital market liberalization has the effect of reducingthe country premium component of interest rate differentialsto zero. Hence, after-tax domestic nominal interest ratesonly differ from one country to another by the amount ofthe forward discount, i.e. the difference between current andforward exchange rates. A practical consequence is that forany agent within the Community, it becomes equivalent toborrow in currency x or to borrow in currency y and tohedge against exchange rate variations (assuming capitalmarkets provide hedging for the adequate maturity).65

This is known as the covered interest rate parity condition. Transactionand hedging costs are neglected. See Chapter 6, Section 6.5 for a quanti-fication.

124


(ij) A second factor is the existence of currency risk pre-miums, i.e. of spreads between the forward discount on theexchange markets and the depreciation expected by marketparticipants. This term cannot be observed, but is analyti-cally important to isolate: even if investors do not expectexchange rates to be changed in the foreseeable future, assetsdenominated in a foreign currency remain more risky thanhome assets of the same category. Hence, exchange rate riskacts as a barrier to capital mobility even without any capitalcontrol. Moreover, as long as exchange rates can be changedat low cost, investors are aware of the possibility of depreci-ating the weak-reputation currencies. Therefore, currencyrisk premiums can persist for previously weak currencieseven if exchange rates are kept fixed for a long period oftime.

Since EMU would eliminate exchange rate risk altogether,it would also eliminate exchange risk premiums. This wouldbe for weak-reputation countries a gain of a permanentnature, but presumably a relatively small one. Althoughthese premiums cannot be observed directly, they cannot beexpected to account permanently for a large fraction ofinterest rate differentials.

(iii) A third factor of divergence is that purchasing powerparity does not hold. As long as exchange rate realignmentsdo not compensate fully for inflation differentials, realinterest rates are bound to differ. This factor could lead toa permanent effect only if EMU were in the long run toaffect real exchange rate changes within the Community. Asargued in Chapter 6, this should not be the case. Regardingthe transition, as documented in Chapter 2, realignmentswithin the ERM have since 1985 systematically been of asmaller magnitude than cumulated inflation differentials.Therefore, this factor should actually reduce ex-post realrates in the more inflationary countries.

(iv) However a fourth factor, which probably accounts forthe largest part of differences in real rates, results fromexpectational errors regarding exchange rates. Markets ap-pear to judge the ERM commitment as not fully credibleand to expect the weak-reputation currencies to devalue. Inthat respect, exchange rate expectations appear to adjustvery sluggishly.

EMU would presumably remove all expectations of ex-change rate changes within the Community. Since it cannotbe assumed that without EMU market expectations wouldpermanently remain wrong, this cannot be considered apermanent gain with respect to the Stage I baseline. How-ever, this is an important topic for the transition because arapid move towards EMU could alleviate the costs associ-ated with a protracted period of high ex-post real rates.

Permanent effects of EMU

From the above discussion, it is apparent that permanenteffects of EMU on real interest rate differentials with respectto a stable Stage I baseline are bound to be limited. Thefirst and only undisputable permanent effect would be thedisappearance of exchange risk premiums, which have beenassessed as relatively small. Moreover, in a stable Stage Iexchange rate risk could limit capital mobility, but as ex-change rates should be stable (except in the presence ofshocks), there is no reason to expect that risk premiumswould still be biased in any significant way.

In EMU, nominal interest rates on assets of the same ma-turity and risk would be fully equalized across the Com-munity. After this equalization becomes effective, the onlyremaining issue concerns the general level of market interestrates. A straightforward assumption, which is made here, isthat it would be the baseline level in the previous anchorcountry. This assumption is warranted because as capitalmarket liberalization would already be effective in the base-line, the effect of EMU should only be to eliminate themonetary 'noise' that affects real interest rates. There is noreason to expect real interest rates to average out within theCommunity as is the case when capital controls are removedbetween two countries.66 Moreover, as EuroFed shouldbroadly maintain the same stance as the most stability-oriented central banks of the Community, there are noreasons from the monetary side why interest rates in Ger-many should change to some Community average. Thevalidity of this assumption is however obviously conditionalon policy assumptions: if the policy stance of EuroFedwere not judged by markets as convincing as that of theBundesbank, inflation risk premiums and therefore interestrates would rise.

Summing up, EMU should not have important permanenteffects on interest rates. This, however, only holds for ident-ical assets. A separate issue, which relates to the identity ofthe borrower, is whether specific effects can be expectedregarding the yield of government bonds. Since bond rateson the public debt may differ for two reasons, namelyeither because of different degrees of credit-worthiness of thegovernments, or because of portfolio effects, EMU effectsregarding these two aspects will be examined separately.

As discussed in Section 5.1.2, evidence indicates that defaultrisk premiums are at present low, which can be interpretedas reflecting the market judgment that public debt unsustain-

Since capital market liberalization is erga omnes, real interest rates withinthe Community would only differ from interest rates abroad for one ofthe reasons spelled out above.

125


ability would actually lead to debt monetization (and there-fore a depreciation of the currency), rather than to default.In so far as EMU would rule out monetization, this couldonly lead borrower's risk premiums to increase for high-debt countries. In other terms, an increase in the specificrisk premium on public debt would (at least partially) offsetthe disappearance of the currency premium. In EMU,governments whose debt cannot be regarded as sustainablecould for this reason face a higher cost of borrowing thanother public borrowers.

Portfolio effects do not arise from default risk, but ratherbecause public bonds and other financial assets (e.g. liquidassets and stocks) differ as regards their yield and its varia-bility. When choosing the structure of their portfolio, risk-adverse agents take both the yields and their variability intoaccount, thus determining the desired shares of each categoryof asset. Therefore, as long as financial markets are seg-mented and the public debt is mainly held domestically, ahigh debt ratio is likely to lead to an increase in public bondrates because private agents are reluctant to increase theshare of that asset in their financial portfolios.67

Following this approach, the effects of EMU would dependon both domestic and foreign portfolio diversification ef-fects. Since exchange rate risks would be eliminated, capitalmarket segmentation would disappear altogether. As bondsof the same maturity and risk issued by different EC govern-ments would become perfect substitutes, their yields wouldbe equalized. High-debt countries should therefore benefitfrom EMU as access to a wider capital market would lowerthe risk-adjusted cost of government borrowing.68 Whetheror not this could offset the increase in risk premiums isan open issue, but it should be noted that capital marketintegration, by permitting a better diversification of risk,would ceteris paribus reduce the size of the risk premiums.It can therefore be presumed that, at least, the net effectwould be a gain for countries whose debt is high but stable.However, possible regulatory bias favouring the holding ofgovernment paper should also be taken into account. If

This result holds in standard empirical portfolio models. However, asdocumented by Wyplosz (1990), investigations on cross-country interestrate differentials do not yield clear results in that respect. The sameholds for time-series analysis.This assumes, first, that in the reference situation agents only holddomestic financial assets, and second that agents across the EMU wouldhave the same preferred portfolio. Both assumptions are simplifying.However inspection of the Italian case shows that the budget deficit isprimarily financed by domestic agents, mainly households (53 % of thedebt in 1986) and secondly banks (30 %). Borrowing in foreign curren-cies is a tiny fraction (3 %) of the total debt outstanding. As a conse-quence, access to a wider market could have significant effects.

specific regulations allow the government to borrow at pref-erential rates, a cost-increasing effect could dominate.

Transitional effects of EMU

Examination of the behaviour of bond rates in the courseof disinflation shows that uncertainty regarding the commit-ment to exchange rate stability in the non-German ERMcountries has probably significantly contributed to their highreal interest rates record. The specific contribution of weakcredibility can be assessed by comparing both long-termrates and inflation in ERM countries with those of Germany,the anchor country. Starting in the early 1980s from asituation characterized both by higher inflation and highernominal rates (but frequently lower real interest rates be-cause of widespread capital controls), most countries havegone through a protracted phase of significantly higher ex-post real rates than in Germany. The typical movement ofinflation and interest rate differentials, shown in the toppanel of Graph 5.8 below, is a counter-clockwise move: adisinflation programme is characterized by a simultaneousdecrease in nominal interest rates and inflation towards theorigin of the diagram (complete convergence), but as thenominal rate decrease lags behind disinflation, the pathremains above the (ex-post) real interest rate parity locusrepresented by the diagonal.69

This is best illustrated by the case of the Netherlands, whichhas gone all the way down to the German inflation rate level(and even below), but as depicted in Graph 5.8 at the costof several years of high real rates. Interestingly, the interestrate differential between the Netherlands and Germanylasted a long time in spite of (i) a strong commitment toexchange rate stability, illustrated by the history of ERMrealignments (the guilder has been devalued with respect tothe Deutschmark only twice, in 1979 and 1983 by 2 % eachtime), and (ii) an early achievement of price convergence(actually better performance than in Germany since 1983).This high interest rate cost of disinflation is the effect ofthe risk premium and the persistence of expectations ofrealignments.70 By contrast, disinflation in Italy is still notcomplete and the ex-post real interest rate differential re-mains significant.

A precise discussion of the typical bond rates behaviour during a disin-flation in a fixed-but-adjustable exchange rate regime can be found inAndersen and Risager (1988).The better inflation performance of the Dutch economy is somewhatexaggerated by the choice of the GDP deflator as a measure of inflation.This is because the share of energy in the Dutch GDP is much higherthan in Germany. See Annex B.

126


GRAPH 5.8: Long-term interest rate and inflation differen-tials with respect to Germany

Theoretical example4 "

Disinflation path

0

-1-1 Inflation differential

The Netherlands, 1974-89

- 2 0 2Inflation differential

Italy, 1974-89

1 3 5 7 9Inflation differential

I I 13 15

Note moving average over three years The diagonal locus corresponds 10 ex-postreal inieresl rate parity. Points above the diagonal are characterized by positive realfate differentials. Note that scales differ

A timely move towards EMU could substantially reduce theex-post cost of State borrowing during the transition towardsprice convergence. Referring to Graph 5.8, for the MemberStates whose disinflation would not be completely achievedat the outset of Stage III, EMU would imply immediatenominal interest rate convergence instead of a slow down-ward move.71

This could be a sizeable benefit for high-debt countries whosedisinflation programmes are difficult to achieve because oftheir public finance cost. Indeed, this public finance cost hasfrequently been considered the major obstacle to full priceconvergence for high-debt countries: because disinflationwould reduce seigniorage revenues and initially increase realinterest rates, at least in the first year it could make thesituation of public finance worse. Thus, it has been arguedthat priority should be given to debt stabilization over priceconvergence, and even suggested that disinflation could becontradictory itself because it would make the public debtpath unsustainable and therefore ultimately force the autho-rities to monetize the public debt.72 By reducing significantlythe public finance cost of disinflation, EMU could removethis possible contradiction.

A precise evaluation of the quantitative gain would requiredetailed country-by-country assumptions. Only a very roughmeasure of the potential gains can be given here for illustra-tive purposes. The Dutch case provides a useful benchmarkcase since it is the only one for which complete convergenceof both inflation and interest rates with Germany has alreadybeen achieved, and because, as documented in Annex B, thetwo countries exhibit at least since 1983 a textbook case ofde facto monetary union; this is presumably what someother Member States could individually achieve within theframework of Stage I by unilaterally committing themselvesto a fixed Deutschmark exchange rate.

From the beginning of the EMS in 1979 until 1989, thecumulative nominal interest rate differential between theNetherlands and Germany amounted to 11 percentagepoints, and the ex-post real differential to 15 percentagepoints, if measured using the GDP deflator because of betterinflation performance on average in the Netherlands.73 Tak-

This is indeed the strategy already followed by several Member States,the most recent example being the Belgian announcement of a Deutsch-mark peg policy. However as this type of commitment remains unilat-eral, it cannot have the same credibility as EMU. Moreover, it presum-ably benefits already from market expectations of the next movestowards EMU.This is known as the Sargent-Wallace (1981) argument. For an appli-cation to the European case, see Dornbusch (1989, 1990).This differential is somewhat lower — 12,8 percentage points — if theconsumption deflator ts used because of the effect of energy prices.

127

sduquenoy


ing the nominal differential, which is the lowest of thesefigures, the (gross) public finance cost of disinflation can beapproximated by 0,11 times the average value of the debt toGDP ratio over the period (60 %), i.e. 6,6 % of GDP.74

This is not a measure of EMU savings, since disinflation isalready under way in most Member States and some ERMmembers are already close to the Dutch situation. However,these figures indicate that in the medium term potentialbudgetary savings could offset the consequences of seignior-age losses for high-inflation and high-debt countries. ForItaly, Spain, Portugal and Greece the inflation differentialwith respect to Germany was above 4 % in 1989, i.e. higherthan the Dutch-German differential at the beginning of the1980s. EMU is not an immediate prospect of course, andthese countries should first achieve a better convergence.But starting from a reduced inflation differential (about 2to 3 %), EMU could help complete disinflation at a lowercost than otherwise. As a benchmark, it can be assumed thatthis strategy would cut the interest rate cost by half of theDutch figure (i.e. 5,5 cumulated percentage points). Theassociated public finance benefit would range between 2,3 %of GDP for Spain, whose debt is relatively low, to 5,4 % forItaly. These would be once and for all gains; but in themedium term, their budgetary impact could offset the seig-niorage costs given in Table 5.7.

Admittedly, disinflation could also raise temporary difficult-ies for governments whose debt is of a long maturity becauseit would increase the ex-post real yield of existing bonds.This effect, which also holds for private borrowers, arises inany disinflation programme. It would only be significant forcountries whose government has mostly issued long-termfixed-interest bonds, and for which EMU can be expectedto bring a significantly speedier disinflation.

Macroeconomic implications

These interest rate effects would have a macroeconomicimpact on the economy as a whole, going beyond publicfinance issues. A sharp distinction has however to be madebetween the effects of EMU on ex-post (or realized) realinterest rates discussed above and its effects on ex-ante realrates which determine the investment and saving decisionsof economic agents. Genuine ex-ante real rate reductions inthe transition would boost investment and growth in themedium run. In the long run, lower risk premiums wouldstimulate investment as discussed in Chapter 3. In contrast,

lower ex-post real rates do not change investment decisions,but provoke transfers between borrowers and creditors andonly impact on the real economy to the extent that agentsreact to these wealth effects.

The potential for a drop in ex-ante real interest rates appearsto be smaller than for ex-post rates, as discussed in Appendix2. In addition to the exchange risk premium effect, theonly motive for a decrease would arise either from presentexpectations of real depreciation of the non-anchor countriesor conversely from expectations of real appreciation duringthe first period of EMU. Both are possible, but to a lesserextent than for ex-post rates. Regarding this second possi-bility, it should be mentioned that an early EMU commit-ment could have negative side-effects precisely because ofan excessive drop in real interest rates in the more inflation-prone countries. Since irrevocably fixed exchange rates implynominal interest rate parity, they also imply that real ratesare the lowest in the more inflationary countries. A too earlycommitment to exchange rate stability could therefore makedisinflation slower and more painful. Indeed, it has beenargued that this is already the case in countries like Italysince firms can borrow abroad at a lower real rate.75 Asfurther discussed in Chapter 8, the right strategy shouldfollow a middle route between the two extremes of 'coron-ation theory' and immediate commitment to exchange ratefixity.

5.2.3. Taxation and the provision of publicgoods in EMU

Throughout the foregoing discussion, it has been assumedthat Member States would retain their full autonomy intaxing and spending decisions, i.e. that neither the levelof non-interest spending nor the structures of taxes andexpenditures would be affected. The aim of this section is todiscuss briefly whether and to what extent EMU would inaddition impose convergence in taxing and spending. Sinceit is a broad and complex issue, only the main argumentsdirectly related to EMU will be given here.

The issue of tax convergence has already been the focus ofnumerous policy discussions within the Community, butmostly in the specific fields of indirect taxation (followingthe Commission's harmonization proposals of 1987) andcapital income taxation (in the context of capital marketliberalization). A more general discussion of taxation andof the parallel issue of provision of public goods and servicesin the context of EMU is warranted. As argued below,

The true public finance cost of disinflation is probably lower because ahigher (even constant) inflation rate in the Netherlands would haveraised risk premiums. This point is made by Giavazzi and Spaventa (1990).

128

sduquenoy


for taxation and spending in general EMU should not beexpected to introduce major qualitative changes with respectto the '1992 + EMS' reference situation. It will, however,have an impact in specific fields, especially capital incometaxation.

Fiscal federalism and EMU

The theory of fiscal federalism76 provides an appropriatepoint of departure as it envisages issues of optimum assign-ment in a system characterized by the existence of a largenumber of local jurisdictions which independently levy taxeson their citizens (supposedly mobile) and provide them withpublic goods.77 The purpose of this theory is to delimit therespective domains of competition and cooperation amonglower jurisdictions, and to determine which resources andfunctions should be assigned to higher or lower levels ofgovernment.

The basic model is one in which goods, capital and citizenscan and do move freely across jurisdictions, where citizenshave full information and where private as well as non-private goods and services are produced under competitiveconditions. In such a highly simplified model, individualcitizens could choose the level of public goods provision(and related taxation) which they prefer by 'voting with theirfeet', i.e. by moving from one jurisdiction to another in orderto find the package of goods and services that suits thembest. Jurisdictions would not be bound to uniformity; on thecontrary, some would be characterized by a higher levelof public goods provision, (e.g. in the domain of publiceducation), but also by higher taxes. Nevertheless, all juris-dictions would be concerned for efficiency in public servicessince a higher cost of production financed by higher taxes,but not matched by better public services, would lead citizensto migrate.

These conclusions are important in the Community contextas they emphasize the irrelevance of any overall requirementfor convergence in the fields of taxation and public services:for example, following this model Member States couldretain significant differences in the level of old-age socialsecurity contributions, provided they would be matched bydifferences in real wage levels. They could also still chooseto levy taxes through a variety of different systems, subjectto a general requirement of efficiency in the management of

the public sector arising from mobility. Although as alreadyargued labour mobility is bound to remain relatively lowwithin the Community, at least across language frontiers,large differences in public sector efficiency could be a motivefor enterprises and certain categories of labour to migrate.

However, the above model has immediately to be extendedto cover the issue of externalities. Benefits and costs derivedfrom public goods can spill over from one jurisdiction toanother: for example, better transportation infrastructuresor cultural facilities benefit residents from other jurisdic-tions, whereas pollution is not restricted to the area in whichit originates. With neither coordination nor assignment ofthose functions to a higher level of jurisdiction, the non-cooperative equilibrium would be characterized by an insuf-ficient provision of public goods, but an excessive provisionof public 'bads' like pollution.78 Externalities also arise inthe tax field. Problems of tax competition originate in themigration of tax bases, for example in the case of capitalincome taxation, corporate taxation or expenditure tax-ation.79 With neither coordination nor harmonization, suchexternalities would lead jurisdictions to choose too low taxrates on mobile tax bases. A similar problem arises also asregards interpersonal redistribution through taxation, sinceif households (at least some categories of them) are mobile,redistributive policies spill over from one jurisdiction toanother. This limits the possibility for any single jurisdictionto engage in redistributive policies which depart from thatof its neighbours, and can generally lead to reducing thescope for redistributive fiscal policies in comparison to eachjurisdiction's preferences.80 Negative revenue spill-overs canalso exist, since an increase in taxes in a jurisdiction reducesthe income being spent on private goods produced in theneighbouring region.81 Coordination and bargaining amonglower levels of jurisdiction, the fixing of minimum standardsin the field of taxation, revenue-sharing provisions, com-monly agreed norms (regarding, for example, the environ-ment) and assignment of certain resources and functions tothe higher federal levels are all, therefore, relevant techniquesin order to improve the overall welfare compared to thenon-cooperative equilibrium.

For a survey of the theory of fiscal federalism and its implications forEMU, see Van Rompuy, Abraham and Heremans (1990).Public goods are those whose consumption by an individual does notreduce the availability for other individuals. Classic examples are TVbroadcasting and clean air.

This risk is underlined by Van der Ploeg (1990).This should not be mixed with the mobility of citizens from one jurisdic-tion to another. Whereas this mobility is a condition for competitionamong jurisdictions to hold, mobile tax bases can give rise to problemsof tax avoidance- Examples can be found in the taxation of capitalincome (Giovannini, 1989) and of corporate income.See Van Rompuy el al. (1990), and Wildason (1990).See Eichengreen (1990), and Wyplosz (1990).

129

sduquenoy

sduquenoy


Implications for EMU

The above approach offers some basic insights on the issuesof taxation and public spending in the Community. Ad-ditions to the model, allowing for the imperfect mobility ofcitizens, information costs and economies of scale in theprovision of public goods do not alter the basic results.However, they strengthen the case for harmonization andan assignment of resources and functions to the upper,federal level: imperfect information increases the costs ofdecentralized bargaining among jurisdictions and of the en-forcement of agreements; economies of scale call for central-ization. Furthermore, restrictions in the mobility of somefactors (e.g. semi-skilled workers and employees) whileothers (e.g. skilled labour and capital) are mobile raisesissues of distortions and vertical equity.

In addition, the need for tax convergence can also resultfrom purely administrative requirements, frequently relatedto the enforcement of tax compliance. This arose in tworecent cases of tax harmonization within the Community inthe context of the single market. In the case of indirecttaxation, a certain degree of convergence in VAT ratesis required as border controls are eliminated, since widedifferences in rates would give rise to tax evasion. The sameissue of tax evasion arises also in the case of capital incometaxation where the case for minimum rates, or a commonwithholding tax, specifically results from differences in na-tional reporting systems. However on purely economicgrounds, neither of these taxes would in principle need tobe harmonized.82

A number of studies have already discussed tax convergenceconstraints which arise from the internal market and capitalmarket liberalization.83 A general conclusion of these studiesis that the need for harmonization or centralization is limitedto certain categories of taxes which account for a relativelysmall part of government revenues: in particular, neitherincome taxes nor social security contributions need to beharmonized, while for VAT rates, only a reduction in cross-country differences is warranted. Corporate and capital in-come taxation, however, are exceptional cases.

In the field of corporate income tax, a case can be made forharmonization of the tax base and for fixing a minimumrate as economic union draws nearer. Minimum conditionsto be fulfilled are tax neutrality with respect to foreign

investment, cross-border cooperation, and the prevention oftax avoidance. In the long term, some studies suggest thatas the internal market becomes the natural habitat for Com-munity companies, the corporate income tax could becomeone of the Community's own resources.84 An alternativecould be to retain different tax rates but to establish a fixedapportionment scheme (based, for example, on the share ofthe country in the total EC turnover of the firm) as in certainfederal States. However, it remains to be clarified how farexisting differences in marginal effective tax rates resultin inefficient distortions in corporate resource allocationbetween countries.

In the field of capital income taxation, the introduction ofa single currency would introduce a qualitative change, sincethere would no longer be any currency discrimination in thefinancial assets held in different Community countries. Thepotential for a migration of this tax base would thereforeencompass the totality of financial assets. This would furtherincrease the need for a Community solution, either throughthe adoption of common reporting rules or through theestablishment of a minimum withholding tax.

In other fields, the specific effects of EMU can be consideredincremental. The largest part of the effects of economicunion already result from the completion of the internalmarket. As to monetary union, apart from the fact that itwould strengthen integration, additional effects would resultfrom (i) inflation convergence, (ii) tighter sustainability con-straints on fiscal deficits and (Hi) the loss of the exchangerate instrument. None of these effects would impose a closerconvergence of the tax systems. The effects of inflationconvergence relate to the seigniorage issue discussed in Sec-tion 5.2.1 above and would only lead to a minor increase intaxes. Tighter constraints on fiscal deficits could lead to taxincreases in some countries, but without clear implicationsfor convergence. The loss of the nominal exchange rate couldlead governments to rely on tax instruments to influence thereal exchange rate, thereby leading to tax competition, butthis problem should be dealt with through coordination.

Empirical evidence derived from the experience of federalStates confirms that inter-state tax differentials are lowerwithin, for example, the United States than in Europe.Nevertheless, provided specific problems arising from taxcompetition are dealt with in an appropriate way, there isstill considerable room for tax autonomy.85 However, itshould be recalled that within EMU differences in the level

Regarding capital income taxation, it should be added that since capitalmarket liberalization is erga omnes, enforcement of tax compliancerequires cooperation with non-Community governments.See, for example, Artus (1988), Giovannini (1989), Gros (1989c), Isard(1989), CEPII-OFCE (1990a, 1990b), and the references therein.

84 See Padoa-Schioppa (1987), Devereux and Pearson (1989), CEPII-OFCE (1990a, I990b).

85 See, for example. Van Rompuy el al. (1990). and Eichengreen (1990).

130

sduquenoy


of taxation will be matched by differences in the real after-tax income of individuals. One can therefore expect a certaindegree of pressure towards public sector efficiency to resultfrom a higher mobility of persons and increased competitionin the product markets.

As cross-border mobility increases, the issue of externalitiesin the provision of public goods is bound to gain in import-ance. Specific fields of Community competence like energy,transport and the environment have been discussed in Chap-ter 3. In other fields like for instance higher education, risksof undersupply of public goods should as much as possiblebe prevented through cooperation among Member States.

Problems might also arise in high-debt countries, since, as alarge proportion of public resources is devoted to servicingthe public debt, public goods tend to be in short supply withrespect to the level of taxes (or taxes to be too high withrespect to the provision of public goods). This creates anincentive for agents to migrate, i.e. to repudiate the liabilitiesof the previous generations. The same kind of intertemporalproblems could arise in the future due to contractual liabili-ties embodied in social security arrangements. Generallyspeaking, mobility in EMU would limit the possibility toorganize large-scale intergenerational transfers within na-tional budgets.86 More precisely, tax convergence (resultingeither from harmonization or market pressures) could leadhigh-debt countries which devote a significant part of theirresources to the interest cost of the public debt to reducethe provision of public goods to a suboptimal level. Theseproblems are not severe in the short term, but deserve atten-tion in a long-term perspective. At present they only reallycall for a sound management of public finance, rather thanspecific Community initiatives.

Appendix 1: The welfare cost of a seigniorageloss: an illustration

The welfare cost of a reduction in seigniorage can be illus-trated by using a very simple numerical example.

Assume the welfare loss function can be represented in asimplified fashion by:

(1) L = ap2 + t2

where p is the inflation rate, t is the tax rate, and a character-izes the relative distortionary effect of taxes and inflation (alower a means taxes are more distortionary).

86 This relates to the discussion of sustainability in Section 5.1.2 above.

The problem of the government is to minimize L subject tothe budget constraint:

(2) mp + t = e

where m is the monetary base and e is the level of publicexpenditure, both measured as ratios to GDP (the deficit isassumed to be zero).

Minimization of L subject to (2) yields the optimal inflationrate:(3) p* = em/(a + m2)

If a is not too small, and m is not too large, (3) can beapproximated by:

(3) p* = em/a

Inflation depends positively on the level of public expendi-ture and on the distortionary effect of taxes. It also dependspositively on the size of the monetary base (except for verysmall values of a).

In order to assess the welfare cost of a seigniorage loss,suppose the monetary base is 15 % of GDP, a typical valuefor southern Member States, and suppose the optimal in-flation p* is 10%, which is high and means taxes are quitedistortionary. Seigniorage revenues amount therefore to1,5% of GDP. With e = 0,4, i.e. government spendingrepresents 40 % of GDP, this gives the value of coefficienta = 0,5775.

Suppose now that due to the completion of the single marketthe monetary base is reduced to 10% of GDP (e.g. 2%reserves + 8 % currency). This already reduces the optimalinflation rate to 6,8 %. According to (2), taxes have to beraised by 0,8 percentage point of GDP.

Assume now that due to EMU inflation is reduced to zero.Seigniorage disappears. Taxes have to be raised by an ad-ditional 0,7 percentage point. The associated welfare loss ofthe combined decrease in inflation and increase in taxes canbe computed by using (1). It comes out to be about thewelfare loss implied ceteris paribus by a rise in taxes by onethird of a percentage point.

The above model can be modified in order to take intoaccount the welfare cost of excess reserve holdings. Supposethe loss function is:

(!') L = ap2 + b(m - m0)2 + t2

where m0 = 10 % is a technically efficient ratio of monetarybase to GDP. Coefficients a and b can be identified assuming

131


that p = 10% and m= 15% correspond to an optimum.Assuming that m is reduced to m0 with the single market,reoptimization with respect to p yields the same result asabove.

Appendix 2: EMU and real interest rates

This appendix discusses the effects of EMU on real interestrate differences between Member States. The emphasis is onthe difference between the present situation, which has sometransitory elements in it, and the steady-state EMU. Bothex-ante real rates (R), which determine investment decisions,and ex-post (or realized) rates (r), which are important forthe interest cost on public debt, are considered.

Perfect capital mobility and asset substitutability are sup-posed to hold, whatever the exchange rate regime. It is alsoassumed that the overall monetary policy is set by an anchorcountry in the baseline situation and by EuroFed in EMU,but that both policies are identical, implying that the short-term nominal interest rate for the union will be the same asfor the anchor country in the baseline.

For assets of the same category, maturity and risk, the ex-post real interest rate differential between the home countryand the anchor country is :

(1) r - r* = (i -p)-(i* - p*)

where i, i* are nominal interest rates and p, p* are the ratesof inflation (starred variables denote the anchor country).However, ex-ante real interest rates depend on expectedinflation, not on realized inflation:

(2) R - R* = (i - i* -+ Ep*)

+ (i*1 - Ee) + (Ee - Ep

Where R, R* are ex-ante real interest rates, and Ep, Ep* areexpected rates of inflation, Ee is expected depreciation andis fd the forward discount. The first term in (2), usuallyreferred to as the country premium, is an effect of capitalcontrols. It is therefore assumed to be equal to zero, whichis equivalent to say that covered interest rate parity holds.The second term, i.e. the difference between the forwarddiscount and expected depreciation, is the exchange riskpremium; it represents the part of the interest rate differentialwhich is not explained by expected depreciation. The thirdterm, the expected real depreciation, is non-zero if ex-antePPP does not hold.87

Hence, real ex-ante interest rate differentials can be decom-posed in the following way i

(3a) R - R* = (P* - Ee) + (Ee - Ep + Ep*)

and for ex-post differentials:

(3b) r - r* = ((Ep - p) -(Ep* - p*)) + R - R*

Assuming that real ex-ante and ex-post interest rates in theanchor country are not affected by EMU, if subscript 0refers to the present situation, the effect of EMU on interestrates in the non-anchor countries is:

- Ee)0 + [( Ep - Ep*)EMU- (Ee - Ep +

(4b) rEMU - r0 = - (i* - Ee)0 + [(-p + p*)EMU- (e -p +p*)ol - (Ee -e)0

The first term in both expressions is due to the disappearanceof the exchange risk premium. This effect, which is due tothe suppression of exchange-rate variability in EMU, shouldcontribute to lower rates in peripheral countries in so far asin general the forward discount overpredicts exchange ratedepreciations.

The second term in (4a) represents the extent to whichEMU changes expectations regarding the real exchange rate.Assuming that real exchange rates can be expected to evolveon average in the same way in a permanent Stage I as inEMU, this effect would be zero. However it could be import-ant in the transition, either because of present expectationsof real depreciation of the non-anchor currencies or, as-suming to the contrary that nominal exchange rates arealready credible, because of remaining price divergence.88

In a similar way, the second term for ex-post real ratescorresponds to ex-post real exchange-rate change. This termwill on average be important only if there are significant realexchange rate adjustments in the transition. It can leadeither to interest rate decreases or increases depending uponwhether the non-anchor currencies experience real de-preciation or appreciation in the transition.

However the last term in (4b), which represents expectationalerrors, is specific to ex-post rates. The associated effect,which could be termed the 'peso problem', offers the possi-bility of significant reductions in ex-post real interest ratesin the transition if exchange-rate expectations are sluggishand/or if the commitment to a hard-currency option is notjudged credible. The experience of the last couple of yearsindicates that this effect might be important.

87 This decomposition follows Frankel (1989).

132

In this case real rates in peripheral countries could even become lowerthan in the anchor country at some stage during the transition.


References

Aglietta, M, Brender, A. and Coudert, V. (1990), Globalisa-tion fmanciere: I'aventure obligee, Economica — CEPII,Paris.

Aglietta, M., Coudert, V. and Delessy, H. (1990), 'Politiquesbudgetaires et ajustements macro-economiques dans la per-spective de 1'integration monetaire europeenne', in Lecturescritiques du rapport Delors, De Pecunia Vol. II, No 2-3,Brussels, September.

Alesina, A. and Tabellini, G. (1990). 'Voting on the budgetdeficit', American Economic Review.

Alesina, A., Prati, A. and Tabellini, G. (1989), 'Public confi-dence and debt management: a model and a case study ofItaly', CEPR Discussion Paper No 351, October.

Amalric, F. and Sterdyniak, H. (1989), 'Interdependance etcooperation: les lecqns d'une maquette', Observations etdiagnostics economiques No 26, January.

Andersen, T. and Risager, O. (1988), 'Stabilization policies,credibility, and interest rate determination in a small openeconomy', European Economic Review 32, pp. 669-679.

Artus, P. (1988), 'Integration europeenne et degres de libertede la politique economique', Document de travail No 1988-03, Service des Etudes Economiques et Financiers, Caissede Depots et Consignations, Paris, November.

Baldwin, R. (1990), 'On the microeconomics of EMU', inEuropean Economy (1990).

Barro, R. J. and Gordon, D. (1983), 'A critical theory ofmonetary policy in a natural rate model', Journal of PoliticalEconomy, 91, pp. 589-610, August.

Begg, D. (1990), 'Alternative exchange rate regimes: the roleof the exchange rate and the implications for wage-priceadjustment', in European Economy (1990).

Bishop, G. (1990), Creating an EC monetary union withbinding market rules, Solomon Brothers Inc, February.

Bishop, G., Damrau, D. and Miller, M. (1989), Marketdiscipline CAN work in the EC monetary union, SolomonBrothers Inc, November.

Blanchard, O. J. (1985), 'Debt, deficits, and finite horizons',Journal of Political Economy, April.

Blanchard, O. J. (1990), 'Suggestions for a new set of fiscalindicators', OECD/DES Working Paper No 79.

Bovenberg, L., Kremers, J. and Masson, P. (1990), 'Econ-omic and monetary union in Europe and constraints onnational budgetary policies', paper prepared for a conferenceon 'The political economy of government debt', Universityof Amsterdam, June.

Bredenkamp, H. and Deppler, M. (1989), 'Fiscal constraintsof a hard currency regime', in Choosing an exchange rateregime: the challenge for smaller industrial countries, editedby V. Argy and P. De Grauwe, CEPS/IMF.

Canzoneri, M. and Rogers, C. A. (1989), 'Is the EuropeanCommunity an optimal currency area? Optimal taxationversus the cost of multiple currencies', American EconomicReview, November.

CEPII-OFCE (1990a), 'Vers une fiscalite europeenne?', re-port to the French Senate, in Rapport d'information fait aunom de la Commission des Finances, No 211, Senat, Paris,April.

CEPII-OFCE (1990b), 'Vers une fiscalite europeenne?' Ob-servations et diagnostics economiques. No 31, April.

Chouraqui, J.-CL, Hageman, R. P. and Sartor, N. (1990),'Indicators of fiscal policy: a reassessment', OECD/DESWorking Paper No 78, April.

Cohen, D. and Wyplosz, Ch. (1989), 'The European monet-ary union: an agnostic evaluation', CEPR Discussion PaperNo 306, April.

Cohen, D. and Wyplosz, Ch. (1990), 'Price and trade effectsof exchange rate fluctuation and the design of policy coordi-nation', CEPR Discussion Paper, April.

Coudert, V. (1987), 'Asymetries et cooperation economiqueeuropeenne: le cas France - Allemagne', Economic Prospec-tive Internationale, 4° trimestre, No 32.

Cukierman, A., Edwards, S. and Tabellini, G. (1989) 'Seig-niorage and political instability', paper presented at theNational Bureau of Economic Research Conference on Pol-itical Economy, Cambridge, Mass., May.

Currie, A., Holtham, G. and Hallet, A. H. (1989), 'Thetheory and practice of international policy coordination:Does coordination pay?', in Macroeconomic policies in aninterdependent world, edited by R. Bryant, D. Currie, andR. Fortes, CEPR/The Brookings Institution/IMF.

133


Devereux, M. and Pearson, M. (1989) 'Corporate tax har-monization and economic efficiency', Institute for FiscalStudies Report Series No 35, London.

Dornbusch, R. (1988), 'The EMS, the dollar and the yen',in The European Monetary System, edited by F. Giavazzi,S. Micossi and M. Miller., Banca d'ltalia, CEPR.

Dornbusch, R. (1989), 'Ireland's disinflation', Economic Pol-icy, April.

Dornbusch, R. (1990), Problems of European monetary inte-gration, mimeo, Massachusetts Institute of Technology.

Drazen, A.(1989), 'Monetary policy, capital controls andseigniorage in an open economy', in A European CentralBank?, edited by M. De Cecco and A. Giovannini, Cam-bridge University Press.

Eichengreen, B. (1990), 'One money for Europe? Lessonsfrom the US currency union', Economic Policy No 8,pp. 173-209.

European Economy (1990) — The economics of EMU',special issue.

Frankel, J. (1989), 'Quantifying international capital mo-bility in the 1980s', NBER Working Paper No 2856, Na-tional Bureau of Economic Research, Cambridge, Mass.,February.

Frankel, J. and Razin, A. (1987), 'The Mundell-Flemingmodel: a quarter of a century later1, IMF Staff Papers,Volume 34, No 4, December.

Froot, K. and Frankel, J. (1989), 'Forward discount bias:is it an exchange risk premium?', Quarterly Journal of Econ-omics.

Giavazzi, F. (1989), The exchange rate question in Europe',EC Economic Paper No 74, January.

Giavazzi, F. and Giovannini, A. (1987) 'Exchange rate andprices in Europe', Weltwirtschaftliches Archiv 124, No 4,pp. 592-604.

Giavazzi, F. and Spaventa, L. (1990), 'The new EMS', CEPRDiscussion Paper No 369, January.

Giavazzi, F. and Pagano, M. (1989), "Confidence crises andpublic debt management', CEPR Discussion Paper No 318.

Giovannini, A. (1989), 'National tax systems versus theEuropean capital market1. Economic Policy No 9, October.

Giovannini, A. (1990), 'Money demand and monetary con-trol in an integrated European economy', in European Econ-omy (1990).

Grilli, V.(1989), 'Seigniorage in Europe', in A EuropeanCentral Bank?, edited by M. De Cecco and A. Giovannini,Cambridge University Press.

Gros, D. (1989a), 'Seigniorage and EMS discipline', Centrefor European Policy Studies Working Document No 38,Brussels.

Gros, D. (1989b), 'Seigniorage in the EC: the implicationsof the EMS and financial market integration', IMF WorkingPaper No 89/7.

Gros, D. (1989c), 'Capital-market liberalization and thetaxation of savings', Centre for European Policy Studies,Working Party Report No 2, Brussels.

Hagemann, R. and'Nicoletti, G. (1989), 'Ageing popu-lations: economic effects and implications for public fi-nance', OECD/DES Working Papers, No 61, January.

Hughes Hallet, A., Holtham, G. and Hutson, G. (1989),'Exchange rate targeting as surrogate international cooper-ation', in Blueprints for exchange-rate management, editedby M. Miller, B. Eichengreen and R. Portes, Academic Press.

hard, P. (1989), 'Corporate tax harmonization and Euro-pean monetary integration', CEPS Working DocumentNo 41, Brussels.

Krugman, P. (1989), Exchange-rate instability, MIT Press.

Kydland, F. E. and Prescott, E. C. (1977), 'Rules ratherthan discretion: the inconsistency of optimal plans', Journalof Political Economy, Vol. 85, No 3.

Lamfalussy, A. (1989), 'Macro-coordination of fiscal policiesin an economic and monetary union in Europe', in Collectionof papers. Report on economic and monetary union in theEuropean Community (Delors Report), Commission of theEC.

Mankiw, G. (1987), 'The optimal collection of seigniorage,theory and evidence', Journal of Monetary Economics, 20,pp. 327-341.

Masson, P. and Melitz, J. (1990), 'Fiscal policy independencein a European monetary union', CEPR Discussion Paper,No 414.

134

sduquenoy


Masson, P., Symansky, S., Haas, R. and Dooley, M. (1988),'Multimod: a multi-region econometric model', Staff studiesfor the world economic outlook, International MonetaryFund, Washington, July.

McDougall, Sir D. (1977), The role of public finances inEuropean integration, Commission of the EC.

Mohlo, L. (1989), 'European financial integration and rev-enue from seigniorage: the case of Italy', IMF WorkingPaper, No 89/41, Washington.

Neuman, M. J. M. (1990), 'Central bank independence as aprerequisite of price stability', in European Economy (1990).

Nicoletti, G. (1988), 'A cross-country analysis of privateconsumption, inflation and the debt neutrality hypothesis',OECD Economic Studies, No 11, autumn, pp. 43-87.

OECD (1990), OECD Economic Outlook, No 47, Paris, June.

Oudiz, G. (1985), 'European policy coordination: an evalu-ation', Recherches economiques de Louvain, Volume 51,No 3-4.

Oudiz, G. and Sachs, J. (1984), 'Macroeconomic policycoordination among the industrial economies', BrookingsPapers on Economic Activity, No 1.

Padoa-Schioppa, T. (1987), Efficiency, stability and equity,Oxford University Press, Oxford.

Padoa-Schioppa, T. (1990), 'Fiscal prerequisites of a Euro-pean monetary union', Conference on Aspects of CentralBank Policymaking, organized by the Bank of Israel andthe David Horowitz Institute, Tel Aviv, January.

Phelps, E. S. (1973), 'Inflation in the theory of public fi-nance', Swedish Journal of Economics, No 75, pp. 67-82.

Poterba, J. M. and Rotemberg, J. J. (1990), 'Inflation andtaxation with optimizing governments', Journal of Money,Credit and Banking, February.

Sargent, T. and Wallace, N. (1981), 'Some unpleasant mone-tarist arithmetic', Federal Reserve Bank of Minneapolis Quar-terly Review.

Van der Ploeg, F. (1989), 'Monetary interdependence underalternative exchange rate regimes: a European perspective',CEPR Discussion Paper, No 358.

Van der Ploeg, F. (1990), 'Macroeconomic policy coordi-nation during the various phases of economic and monetaryintegration in Europe', in European Economy (1990).

Van Rompuy et al. (1990), 'Economic federalism and theEMU', in European Economy (1990).

Wyplosz, Ch. (1990), 'Monetary union and fiscal policydiscipline', in European Economy (1990).

Wildason, David E. (1990), 'Budgetary pressures in the EEC:a fiscal federalism perspective', American Economic Review,May.

135


Chapter 6

Adjusting without the nominalexchange rateThe loss of the exchange rate as a policy instrument hasimportant implications for macroeconomic stability. In a worldin which countries are faced by unexpected shocks of eitherdomestic or foreign origin, real and nominal macroeconomicvariables wilt tend to fluctuate. A large variability of variablessuch as output or inflation is generally considered to be wel-fare-reducing. The disappearance of the nominal exchangerate, assuming it is instrumental in affecting macroeconomicvariability, could therefore have implications in terms of wel-fare.

How serious is this loss likely to be for EC Member States? Arethere alternative means of handling country-specific economicshocks? These are the questions addressed in this chapter.

Graph 6.1 gives an overview of the chapter. If shocks aresymmetric, intra-Community exchange rates are not needed.If shocks are asymmetric (Section 6.2), intra-EC exchangerates could be used (Section 6.1). This possibility is no longerpresent in EMU, and so either factor adjustment (Section6.3) or financing must take its place. The combination ofasymmetric shocks and factor adjustment is a major determi-nant of the impact of EMU on macroeconomic stability (Sec-tion 6.4). Financing may either be private, through capitalflows (Section 6.5), or public. In the latter case, both nationalborrowing (Section 6.6) or Community transfers (Section6.7) could be used. The choice between adjustment instrumentsdepends on collective welfare considerations. The first choicebetween factor adjustment or financing concerns the trade-offbetween the real wage and employment. Secondly, withinthe financing instrument, equity considerations determine thechoice between national or Community financing, subject tothe requirements of fiscal discipline.

The conclusions can be summarized as follows:

The fixing of exchange rates within the Community represents,at worst, only a very limited loss:

(i) Fixing intra-Community exchange rates in EMU stillleaves the possibility for the Community to change itsexchange rate with respect to the rest of the world.

(ii) For the original members of the exchange rate mechan-ism of the EMS, nominal exchange rates have hardlychanged at all for several years. The 'costs'associatedwith this nominal fixity have been borne or adjusted to

already, although the benefits of EMU are still to beobtained.

(in) Since wages and prices are rigid in the short run,nominal exchange rate changes may affect real ex-change rates for a while. This may dampen outputfluctuations, but may increase inflation fluctuations.Over a longer period, nominal exchange rates tendat best to accommodate inflation differentials withouthaving a lasting impact on real exchange rates.

(iv) Real exchange rate changes are still possible throughrelative price movements within EMU, as the examplesof existing federations and the experience of the EMSclearly show.

(v) Taking long-run trends, real exchange rates do not seemto contribute much in sustaining growth differentialsbetween Community countries, since there are manyother factors involved.

(vi) Economic integration will make the occurrence of coun-try-specific shocks less likely since product differen-tiation tends to dominate product specialization.

(vii) Enhanced competition in the internal market willensurethat profit margins carry part of the price adjustmentburden.

(viii) Wage discipline will also be more effective in a credibleEMU, as witnessed already in the EMS, but this willneed encouragement. To a minor extent, greater re-gional and occupational mobility may also contributeto labour market flexibility.

Additional financing will facilitate adjustment and help cushionshocks:

(ix) EMU removes the external constraint inside the Com-munity, facilitating external financing of temporaryexternal imbalances for individual countries.

(x) Budgetary policy, at central and national level, will alsohelp adjustment or cushion it, provided a certain trade-off between national budgetary autonomy and centralpublic finance is respected.

Overall, EMU will probably improve macroeconomic stab-ility :

(xi) The disappearance of exogenous asymmetric intra-Community exchange rate shocks, the absence of non-cooperative exchange rate policies and the disciplinaryeffect on wages and prices tend to offset the negativeimpact of asymmetric shocks.

(xii) Compared to a floating exchange rate regime inside theCommunity, EMU will reduce the variability of outputand notably inflation; compared to the EMS, variabilityalso decreases, since asymmetric monetary policy isreplaced by a common monetary policy which is con-

136

sduquenoy

sduquenoy

Chapter 6. Adjusting without the nominal exchange rate

GRAPH 6.1: Schematic overview of the chapter

Shocks (6.2)

^ r * ^Country-specific , ———— Common external/J 1 [ 1 Sector-specific

1

f t AAsymmetric:

Initial situationStructureBehaviourPreferences

^ J

Symmetr

Initial situzStructureBehaviourPreferences

^

• I

' Nc:

tion

J

Asymmetric shocks

Gains: employmenlCosts: real wage

Factor adjustment (6.3)

Gams: real wageCosts: employment

Gains: employmentCosts: real wage

Nationalflows (6.5)/ budget

(6.6)(borrowing)

Centralpublic

finance (6.7)(transfers)

Nole Figures between brackets represent sections of the chapter

137

sduquenoy

sduquenoy


cerned with macroeconomic stabilization of the Com-munity as a whole.

(xiii) The decrease in output and inflation variability may beexperienced by all EMU members, but is also dependenton national economic policies and the behaviour ofeconomic agents, notably for output stabilization.

6.1. Do nominal exchange rate changes have realeffects?

In discussing the real effects of exchange rates, it is usuallyassumed that there is a shock to external demand or inter-national competitiveness which brings the current accountout of equilibrium. A change in the real exchange rate maythen ensure the return to equilibrium. The usefulness ofthe nominal exchange rate instrument should therefore bejudged in the context of changes in real exchange rates. Inthe case of an adverse shock, a lasting real exchange ratedepreciation requires a decrease in the real wage.' Withnominal wages highly responsive to unemployment, thiswould happen immediately and the nominal exchange rateinstrument would not be needed. The potential usefulnessof the nominal exchange rate therefore derives from the factthat nominal wages are not fully flexible downward.2

In such a case, a devaluation has the benefit of 'front-loading' the required real exchange rate adjustment. Undernormal conditions,3 the improved real exchange rate andthe higher level of import prices will start to improve thetrade balance rather fast. After a while, however, importprices work through into consumption prices. These priceincreases will sooner or later feed through in nominal wagesand therefore in domestic output prices, undoing the initialreal depreciation brought about by the nominal depreciation.Consequently, unless the real wage decreases some time afterthe nominal exchange rate adjustment, there is no lastingreal depreciation. Thus, the slower wages are indexed toprices, the higher is the benefit derived from the nominalexchange rate. As a drawback, however, the increase indomestic inflation may increase inflationary expectations,implying an inflationary cost for the devaluation instrument.

These arguments may be illustrated empirically in severalways.

Firstly, the relationship between nominal and real effectiveexchange rates at the intra-EC level may be examined. Em-pirical evidence shows that the quarterly growth rates ofnominal exchange rates are highly correlated with those ofreal exchange rates for the same quarter. For the period1979-89, using effective real exchange rates with respect toCommunity partners with unit labour losts in manufacturingas deflator, these correlations range from 0,75 to nearly 1,0depending upon the country.4 This does not necessarilyimply that the levels of nominal and real exchange rates arealso correlated in the long run.

A formal test of such a correlation, which is also called co-integration, is presented using quarterly data for 1980-89 inAppendix 6.1. It shows that the levels of nominal effectiveintra-EC exchange rates of all Community countries arenon-stationary, i.e. they have no tendency to return to agiven mean. The same result is found for the correspondingreal exchange rates, except for Belgium/Luxembourg, theNetherlands and Portugal. Since for these three countriesthe real exchange rate returns to a mean but the nominalexchange rate does not, their levels are by definition not 'co-integrated'. For the remaining eight countries, which couldstill be co-integrated, a further test reveals that this is notthe case. Although the levels of nominal and real X-rates aretherefore not correlated in the long run for all Communitycountries, a change in the nominal X-rate may go hand inhand with a change in the real X-rate for quite some time.Using some plausible parameters for a wage-price block,Annex D gives an example showing that 50 % of the effectstill remains after two years, and that it only disappearscompletely after five years.

A second illustration of the theoretical argument concerningthe impact of a devaluation compared to a situation whereit is not available may be obtained from model simulations(see Box 6.1). In a first simulation, a permanent negativeshock to exports was simulated, keeping nominal exchangerates fixed. In a second simulation, the same shock wasimposed, but in addition a devaluation was assumed to takeplace at the same time as the export shock.5 With respectto real exchange rates, output and inflation, this exerciseillustrates the following points:

See Dornbusch (1980).This discussion does not treat explicitly the use of the exchange rate toneutralize imported inflation through nominal appreciation. However,this aspect is present implicitly through the real effects of inflation.Well-known is the so-called Marshall-Lerner condition for price elastici-ties, but this is a special case of a more general condition on elasticities,(see Gandolfo( 1987)).

This is a property generally observed among floating exchange rates(see Mussa( 1986)).In both simulations, the nominal exchange rate was kept constant atbaseline values, or baseline values after the devaluation. Governmentexpenditure was assumed to be unchanged in real terms, while nominalinterest rates were also kept at their baseline values. The devaluationwas chosen so as to bring nominal GDP back to its baseline value inyear 7 after the shock.

138


Box 6.1: Model simulation of an export shock with and withoutdevaluation

The impact of a devaluation in the face of negative export shockswas simulated with the Commission's Quest model.6 In a firstsimulation, the Quest submodel for France was simulated witha permanent negative shock to exports of 5%. In a secondsimulation, the same shock was imposed, but in addition adevaluation of 7,25 % was assumed to take place at the sametime as the export shock.

Graph 6.2.(a) displays the trajectory of deviations with respectto baseline (assumed to be the equilibrium value) for real GDPand the real effective exchange rate over a period of seven years.The initial equilibrium is at point A. The negative export shockshifts the demand curve D inward to D'. Without adjustmentof the real wage, a new equilibrium with output below potentialwould be established at a point like B. However, in order torestore the equilibrium for output equal to potential, at pointA', the real wage would have to decrease to move to the long-run supply curve AA'. With short-run price stickiness, however,the short-run supply curve will be completely inelastic (horizon-tal), and there will be underutilization of production capacityat point B before the economy moves upwards along the newdemand curve D', passing through E and further upwards tothe extent that real wages decrease sufficiently to reach the long-run supply curve.

The trajectory of the simulation without devaluation (solidcurve) follows this path, and comes close to a new equilibriumat point A' after seven years. The effect of the devaluation is toshift the short-run horizontal supply curve AB upwards to apoint such as C, which is still characterized by output belowpotential. If real wages do not change, the economy will stay atpoint C with output below equilibrium. To return to equilibriumoutput, a decrease in real wages is still needed. The conclusionis thus that the devaluation diminishes the output loss in theshort run, but that real wage adjustment is delayed as well asthe return to equilibrium output.

Moreover, as illustrated in Graph 6.2.(b), the delayed adjust-ment has its price in terms of inflation. Without devaluation,the negative demand shock shifts the equilibrium output levelto a point such as B. The real wage decrease shifts the equilib-rium back to its original level, but at a lower rate of equilibriuminflation, at point A\ The effect of the devaluation is, insteadof moving to a point like B, to use inflation to move upward ashort-run Phillips curve and have higher output, at E, in ex-change for more inflation. The result is that there will be lessoutput loss but for a longer period, and that given the upwardshift in the short-run Phillips curve to a point as high as D,relatively more inflation is incurred than without devaluation.

A further discussion of these issues, including the role of thePhillips curve, is given in Annex D.

6 SeeBekxefaMI989).

GRAPH 6.2: Adjustment with and without devaluation(percentage deviation from baseline)

. (a) Real exchange rate

10 ! A'

a

5 E/V

/ //

with devaluation |

/ ''> /// '1 /v/ VA---7fA- -". "

/ without devaluation

/ D

-10 . . . . -

long-runsupply curve

, _. J. J. 4- 1

A 1 1 -* A-4 -3 -2 -1 w ' *• J •»Real GDP

(b) Inflation2 T" - - - - - - - - -

iV with devaluation

1 V: /Vx_ 1 1 ^

-0 . v

t\

-1 •

.A'

/ \ ^/ I B/ J

""/AC< \V/\^*-vx\wiuioui devaluation

-2 ' x! xii

-3 . . „ . 1 , j. _ .+ 2 + 1,5 +1 +0,5 0 -0,5 -1 1,5 -2

Real GDP

Tim graph presents the time-paths for combinations of the real exchange tale orinflation (on (he y-axes) and real GDP (on the x-axis), both measured as percent.age deviations from baseline values. If Ihe baseline values are interpreted as theequilibrium values, these graphs show how fast the real exchange rate, inflationand real GDP adjust to (heir equilibrium values after a negative export shock tothe French model of 5 %. with and without devaluation.

139


(i) The degree of initial output loss without devaluation ishigher than with devaluation. 7 This is the advantage ofusing the exchange rate instrument: it cushions the sizeof the immediate output shock in the first few yearsthrough its direct effect on real exchange rates.

(ii) Without devaluation, the return to equilibrium outputis faster than with devaluation, where there is a delayin real-wage adjustment. This is the disadvantage ofthe devaluation: the output gap will take longer todisappear. This is due to the fact that the output gainsfrom a devaluation cause tensions on the labour marketwhich increase the real wage above its equilibrium level(see Annex D).

(iii) The 'soft landing' in the case of a devaluation has itscounterpart in the fact that the devaluation results ina higher inflation rate than in the situation withoutdevaluation.

A devaluation, therefore, has the benefit of not causing asmuch initial output loss as without devaluation, but thereturn to equilibrium takes more time and is accompaniedby higher inflation.

Overall, therefore, nominal exchange rates may have animpact on real exchange rates for, say, two to five years, butthis does not persist in the long run. The main advantagederived from the nominal exchange rate instrument residesin 'front-loading' the real exchange rate adjustment neededwhen a country is faced with an adverse shock. This reducesthe initial output loss, but does not substitute for real wageadjustment. Delayed adjustment of the latter may even pro-long the situation of underemployment of productive ca-pacity. Moreover, a devaluation may shift inflationary ex-pectations upward, thus also delaying the disinflation pro-cess needed to restore equilibrium.

This analysis is based on the role of nominal exchange ratechanges in adjusting to shocks, starting from a situation ofequilibrium. As is obvious from events in Eastern Europeor many developing countries, moving exchange rates to anappropriate level can play a useful role in a situation whereeconomic reform or structural adjustment is required start-ing from a situation of clear disequilibrium. This, however,is not so relevant for Community countries in EMU.

6.2. 'Will asymmetric shocks diminish in EMU?

In a perfectly symmetrical world with countries identical insize, structure, behaviour and preferences, which all undergo

the same shocks (unanticipated events), real exchange rateswould not have to change. Only if the economic systemdisplays asymmetries, may real exchange rate adjustmentsbe needed. In the context of EMU, the question thereforearises whether asymmetries will have a tendency to decreaseor to increase relative to the present situation. The cost oflosing the nominal exchange rate (assuming it to be a validinstrument in at least some degree) will be diminished ifasymmetries tend to disappear.

A useful way of analysing asymmetries is to look at differentkinds of shocks and see how they may turn into asymmetries.Shocks may be classified (see Box 6.2) according to whetherthey are:(i) common or country-specific;(ii) temporary or permanent.

Obviously, shocks are asymmetric if they are country-specific, such as policy shocks, resource shocks or changesin behaviour. Common shocks such as external shocks orsector-specific shocks, even when they are initially symmetricin nature, may however be asymmetric in their consequences,depending on differences among countries in initial situ-ations, economic structures (e.g. production structures),economic behaviour or preferences. Each of these cases hasto be analysed separately, therefore.

6.2.1. Country-specific shocks

Country-specific shocks are by definition asymmetric.

A first but simple approach to identifying the existence ofcountry-specific shocks is to compare the behaviour of thesame variable, e.g. real GDP, among two countries. If thevariance of such a variable, when aggregated for these twocountries, is greater than the variance of their difference, onemay conclude that the variables are behaving symmetricallyrather than asymmetrically.8 A disadvantage of this ap-proach is that it does not tell where the asymmetries comefrom. They may in fact be a combination of unknowncountry-specific and common shocks, the latter perhaps incombination with an asymmetric system.

Applying this approach to German and French data for realGDP, the GDP deflator, real wages and current accountbalances, Cohen and Wyplosz (1989) find symmetric behav-

7 Output is assumed to be measured by GDP.

140

In mathematical terms: if the correlation between x and y is positive(symmetry), the variance of x + y will be larger than the variance ofx — y, and vice versa when the correlation is negative.


Box 6.2: A taxonomy of shocks

In discussing the nature of shocks, it seems useful to search foran operational definition of what a shock really is. In order todo so, some sort of economic system has to be taken as reference.In the context of EMU, the reference economic system is takento be a member country of EMU.

Secondly, a set of variables has to be defined which are con-sidered to be determined endogenously and simultaneously in-side the system. These can be taken to be the usual macro-economic and sectoral variables such as output, inflation andemployment. Given these definitions, a shock may be definedas any unanticipated event which has a direct or indirect impacton the endogenous variables of the reference system without,however, being part of them. Given this definition, a country-specific or local shock may be defined as a shock having a directimpact on only one country. Similarly, a common shock maybe defined as a shock having a direct impact on all membercountries of the union. A second distinction arises from theduration of the shock. A temporary shock is an event thatdisappears after a period of time, while a permanent shock isan event that remains present over the time period considered(this does not preclude that it eventually disappears).

Evidently, common shocks may have diverging effects in differ-ent countries if these countries do not react to these shocks inthe same way, for example because they are in different initialsituations, have different economic structures, display differentbehaviour of economic agents or have different policy prefer-ences. Therefore, a distinction within common shocks has to bemade on the basis of whether their impact is comparable ordissimilar among countries. If the impact of a shock (mostly acommon shock) is comparable among countries, this will bereferred to as a symmetric shock, if it is a country-specific shockor a common shock with an asymmetric impact, it will be calledan asymmetric shock.

Common shocks versus country-specific shocks

In the EMU framework of reference, the two common shockshaving had the largest impact over the last two decades haveundoubtedly been the oil price hikes of 1973/74 and 1979 (typi-cally sector-specific shocks) and the increase in interest rates inthe United States which provoked the 1981-82 recession (i.e. acommon external shock for the Community).

Country-specific or local shocks may be subdivided in a numberof categories. A first source of local shocks arises from domesticpolicy instruments, to be broadly distinguished between monet-ary and budgetary policy. It might be argued that these instru-ments in fact react to changes in domestic economic perform-ance, are therefore endogenously determined and not to beconsidered as shocks. This is certainly true for some elementsof budgetary policy, but other elements, however, are of a morediscretionary nature. Unless it is assumed that fiscal fine-tuningis still possible, changes in these elements will therefore act asrandom shocks to the national economy.

A second important set of country-specific shocks relates tochanges in domestic natural, human or capital resources. Sinceresource shocks, notably concerning human resources, are evolv-ing gradually over time, they tend to be of a permanent nature.The same can be said of capital resources to the extent that itconcerns technical progress.

A third set of country-specific shocks resides in changes inbehaviour of economic agents (households, firms). Thesechanges may arise from changes in taste, business climate orany other form of news that has a random pattern in influencingeconomic behaviour. In econometric terms, these shocks arecontained in the residuals of the behavioural equations.

Finally, if the concept of a shock is defined rather broadly, afourth category of shocks is formed by inertia. This can beexplained as follows. A broad definition of a shock would defineit as all events without which the endogenous variables of theeconomic system would not change. Under this definition, pastvalues of endogenous variables of the system, or inertia, whichinfluence the present endogenous variables are also to be con-sidered as a shock. They are the consequence of the existenceof permanent shocks.

There is no unique measurement system available for shocks,be they common or country-specific. The main issue involvedhere is that shocks have different dimensions. How to comparean oil price increase to an earthquake, for instance? In order tocircumvent this problem, the impact of a shock on one of theendogenous variables could be taken as a yardstick. Since theendogenous variables of the system are all determined simul-taneously, a representation of the system which determines themis needed in order to be able to assess the final impact of a shockon any particular endogenous variable. A natural candidate forsuch a representation is an econometric model. See Annex Dfor an example.

141

sduquenoy


iour to dominate for these two countries. Using data for theoriginal members of the exchange rate mechanism (ERM)of the EMS before and after the start of the EMS, Weber(1990) has applied the same approach to all possible combi-nations between these countries. He finds that rates of in-flation, as well as domestic demand and supply shocks, tendto be dominated by symmetric behaviour.9 Real wages, onthe other hand, tend to be dominated by asymmetric shocks,and similarly for unemployment rates during the EMS per-iod. This shows precisely the weakness of this approach:asymmetric unemployment shocks may both have been thecause or the result of the real wage asymmetries. If they werethe cause, the real wage asymmetries may be seen as a wayof adjusting to these shocks. On the other hand, if real wageasymmetries caused the unemployment asymmetries, thiswould be perceived as less optimal.

ments. The last two sources of asymmetries will not bediscussed here, since they are more related to the issues oftransition and policy coordination. n

Product market integration

If product market integration is characterized by inter-indus-try specialization, this implies that a common shock to aspecific sector (e.g. a general drop in demand for a certainproduct) will asymmetrically affect the country in which theindustry concerned is located. On the other hand, if intra-industry specialization is taking place, the shock will bemore symmetric, affecting all industries in different countriesinvolved in the production of the product concerned.

A second approach is to use a complete model to analysethe origin of country-specific shocks. Fair (1988), for in-stance, has decomposed the variance of US real GNP andthe GNP deflator into components due to the residuals ofestimated equations for demand, supply, fiscal and monetaryvariables. While equation residuals represent shocks in econ-omic behaviour, shocks may also come from other sources(see Box 6.2). Annex C therefore presents an exhaustiveanalysis of sources of short-term macroeconomic fluctu-ations based on model simulations. The results, even thoughthey focus on the very short run, I0 suggest that shocks inthe four big Community countries are permanent ratherthan temporary. About 50 % of the temporary shocks isdue to equation residuals, i.e. shocks in the behaviour ofeconomic agents. In addition, government policies, such asshocks in monetary policy, tax rates or government expendi-ture, also act as a source of fluctuations, albeit to a minorextent. In EMU, the latter factor as a source of country-specific shocks may be attenuated through the impact ofmultilateral surveillance.

Within the Community, product market integration tendsto be of the intra-industry type, notably in the manufacturingsector.12 A recent study by the Commission of the EC(1990), for instance, finds that except for Portugal andGreece the share of intra-industry trade in infra-Communitytrade varied between 57% and 83% in 1987. Since intra-industry integration is characterized by the occurrence ofeconomies of scale and product differentiation, the removalof barriers obstructing the exploitation of these advantageswill increase intra-industry integration.13 Consequently, thecompletion of the internal market is likely to render theeffects of sector-specific shocks more symmetric. This re-lationship is confirmed empirically if an index for symmetryis compared to an index of trade barriers which may beexpected to disappear with the internal market (see Box 6.3and Graph 6.4). Relatively speaking, this also holds forthe lagging Community countries, whose specialization ispresently more of the inter-industry type, from which theytend to switch to intra-industry specialization.

6.2.2. When are common shocks asymmetric?

While country-specific shocks are by definition asymmetric,there are several circumstances under which a common shockmay also have an asymmetric impact. This depends onthe integration of product markets, differences in economicstructures, differences in economic behaviour, divergencesof initial situations or asymmetric preferences of govern-

9 The main exception is Germany for demand shocks.10 This is due to the fact that only the impact of shocks in the first quarter

was calculated.

For a discussion of the coordination problems in the case of differentstarting positions for the current account, see Masson and Melitz (1990).Begg (1990) also analyses the coordination problems in attaining anequilibrium from different starting positions. For an analysis of asym-metric government preferences, see Tootell (1990).See Greenaway and Milner (1986) for an overview.Jacquemin and Sapir (1988), for instance, find a negative effect ofeconomies of scale on the Community share in total imports of Ger-many, France, Italy, and the United Kingdom for 1983. They interpretthis as being the result of fragmented markets in the Community whichhave not allowed full exploitation of economies of scale.

142


Box 6.3: How asymmetric are sector-specific shocks?

If a common sector-specific shock occurs, the output of thesector concerned in a particular country would tend to movemore in conjunction with the output of the same sector in othercountries than with the output of other sectors in the samecountry. The more this phenomenon can be observed empiri-cally, the more there is a general tendency for (a) the occurrenceof common sector-specific shocks and (b) symmetric effects ofthe common shocks.

This has been tested on indices of industrial production for 31sectors for all 12 Community countries. On the basis of statisticaltests, an indicator has been developed which is able to saywhether, at Community-wide level, a sector has undergone sym-metric or asymmetric shocks. The indicator was constructed asfollows. For each of the 3 1 sectors s in each of the 1 2 Communitycountries j for which data on the volume of industrial productionare available, the annual growth rate of industrial productionysjt was regressed on a sector-specific variable x -t and a country-specific variable z -t:

The percentages of variance explained by the sector-specificvariable or the country-specific variable alone were aggregatedover countries using production weights. The difference betweenthe aggregated explained variances constructs the indicator,which therefore ranges between -100 and 100. For each re-gression, the sector- specific variable xsj, was defined as the firstprincipal component of the growth rates of industrial productionin the same sector in all other (i.e. excluding country j) Com-munity countries. Similarly, the country-specific variable zsj( wasdefined as the first principal component of the growth rates ofindustrial production in all other (i.e. exduding sector s) sectorsin the same country. The indicator is presented in Graph 6.3.

Inevitably, the indicator is surrounded with uncertainty. Never-theless, several broad trends may be distinguished.

In the first place, natural circumstances cause several sectors toundergo asymmetric shocks because they produce productswhich are mainly destined for their local market. This concernssectors such as the coke-oven industry, the printing industry, theconstruction sector and the production of timber and woodenproducts.

Secondly, there are sectors which produce relatively homo-geneous goods with few trade barriers and which thereforeare subject to symmetric shocks. This applies to petroleumextraction, the tobacco industry, the drinks industry and themetal industry as a whole.

Thirdly, for the remaining sectors the existence of trade barriersinside the Community seems to determine whether they aresubject to symmetric or asymmetric shocks. This hypothesis istested by looking at the correlation of the symmetry indicatorwith an indicator for trade barriers inside the Community.14

The scatter plot in Graph 6.4 and the corresponding regressionline show a significant negative relationship between the exist-ence of trade barriers and the symmetry of shocks. Conse-quently, it can be expected that the completion of the internalmarket will tend to reduce the occurrence of asymmetric sector-specific shocks.

14 The indicator for trade barriers is defined, per sector, as Ihe percentage increase inintra-Comrn unity trade of products produced by ibis sector as a consequence ofcompleting the internal market. This is the Stage I integration effect of the removalof trade barriers, see European Economy 'The economies of 1992', Commission of theEuropean Communities (1988, Table A.5, column (i)). Graph 6.4 incorporates all thesectors of Graph 6.3 for which these data were available. The regression line has thefollowing characteristics:1 = 36,7 - 12,6'TBR2 = 0,435

(33,7) (3,3)with I the asymmetry index and TB the trade barrier index (standard errors inbrackets).

Economic structures

Differences in economic structures may cause a nationaleconomy to react differently to a common shock than othermembers of the union. Consequently, common shocks mayresult in asymmetric effects on economic objective variablessuch as output, employment and inflation.

An often cited example of an asymmetric structure is thepresence of oil and natural gas production. Nevertheless,such differences in the availability of primary energy pro-duction may not necessarily show up in comparable differ-ences in all economic variables. Table 6.1, for instance, illus-trates the effects of differences in economic structure after a10% in oil prices using 1980 input-output tables. Takinginto account international price linkages, domestic prices inthe two oil/gas-producing countries the Netherlands and the

United Kingdom increase by 1,7-1,9%, whereas domesticprices in the other countries increase by 1,3-1,4%. Giventhe particularities of this example (an industry for whichproduction endowments are very different, and which hasgenerated very large shocks in the past), the figures mayserve as an upper bound.

The effects of economic structure are not confined to thestructure of production, but may also concern the structureof consumption, the labour market or international trade.Nevertheless, the production structure is an importantsource of asymmetries. In fact, production structures tendto become more similar in the case of intra-industry special-ization. Given the tendency already noted towards intra-industry specialization, this again suggests that the effectsof common shocks become less asymmetric as integrationproceeds.

143


Table 6.1Effects of economic structure on domestic prices after an increase of10% in oil prices, using 1980 input-output tables

Wilh price linkages1 Without price linkages

GermanyFranceItalyNetherlandsUnited Kingdom

,3,4,3,9,7

1,21,31,21,71,6

1 Price linkages among the live countries.Source: Giavazzi and Ciovannini (1987).

GRAPH 6.3: The symmetry and asymmetry of sector-specific shocks in the EC: 1979-88

Asymmetric SymmetricTobacco industryOffice machinery

Motor vehiclesPetroleum extraction

Chemical industryDrink industry

ElectricityMetal products

Production of metalsMeul ores

Food industryOther transport

Electric engineeringMineral products

Petroleum productsSolid fuels

InstrumentsMan-made fibresLeather products

Coke ovensPrinting industry

Paper industryMechanical machinery

Rubber productsNon-racial ores

Clothing industryTextile industry

ConstructionPlastic products

Timber and woodFootwear industry

-100-75-50 -25 0 25

Community average, index —100 to 100

This graph compares the symmetry/asymmetry index described in Box 6.3 among anumber of sectors for the Community average. The index ranges between - 100 and100 and is defined as the difference between the variance of the growth rate of sectoralproduction in a country explained by production growth in the same sector in othercountries (symmetry) compared to production growth in other sectors in the samecountry (asymmetry). A positive index fora sector means that the variation of growthrates of industrial production in that particular sector is, per country, more correlatedwith lhat of the same sector in other countries than with that of other sectors in thesame country. The interpretation is that sector-specific shocks in that case are sym-metric. A negative index implies (he converse, and is an indication of the predominanceof asymmetric shocks.

Behaviour of economic agents

Differences in behaviour of economic agents in the face ofsimilar shocks may be an important source of asymmetriceffects of common shocks. This is particularly true for wagebehaviour due to its central role in the determination ofinflation, real exchange rates and unemployment. Differentdegrees of real wage rigidity will imply different effectson unemployment. An analysis of differences in real wagerigidity may therefore reveal an important source of asym-metric shocks.

From a macroeconomic point of view the flexibility of reallabour costs depends on two factors:

GRAPH 6.4: The symmetry of shocks and trade barriers inthe EC

Symmetry of shocks (index), 1979-88

9 10

This graph plots the symmetry'asymmetry index of Graph 6.3 (y-axis) against ameasure of trade barriers. A negative correlation between the two means thai theexistence of trade barriers may be a factor causing asymmetric shocks. Consequently,the removal of trade barriers in the context of the completion of the internal marketmay be expected to make shocks more symmetric. The trade barrier indicator hasbeen taken from European Economy No 35, Table A.5. column (i). It corresponds tothe Stage I integration effect of the removal of trade barriers. J

144

sduquenoy


(i) The price elasticity of nominal wages. The smaller thedegree of indexation of nominal wages, the smaller willbe the inflationary effect of the wage-price spiral. This,in turn, implies lower costs in terms of unemploymentof a disinflationary policy after an inflationary shock.

(ii) The elasticity of nominal wages with respect to excesssupply or demand in the labour market. The higherthe sensitivity of nominal wage growth to the level ofunemployment, the lower will be the cost in unemploy-ment of a local wage push that initially sets the realwage level out of line.

Taken together, these two elasticities may be combined toform a measure of real wage rigidity.15 There appears to beevidence (see Appendix 6.2 and Graph 6.5) that this measureof real wage rigidity is more similar among Communitycountries than compared to countries such as the UnitedStates or Japan, but they remain divergent to a significantextent. Even although they are imperfect and model-depen-dent measures of wage behaviour, this points to the possi-bility that on the basis of these data wage behaviour mayhave been a component of asymmetric shocks, and willremain so in the future unless wage responses in the Com-munity to identical shocks become more similar, for instancethrough pan-European sectoral wage negotiations.

GRAPH 6.5 : Real wage rigidity

Real wage rigidity

2 -

1,5 -

' -

0,5 -

• •B DK D I NL UK USA

This graph plots an indicator of real wage rigidity from OECD (I989a) Let W be thegtowih rale of nominal wages, P the growih rate of (expected) prices, U the unemploy-ment rale, and let the wage equation have the following form:W - a.P + ajP j - Û-b jU ,+ .Then real wage rigidity is defined as a, 'b,, see Coe (1985).

More comprehensive evidence concerning asymmetries ineconomic behaviour may be obtained by comparing com-plete models. This confirms however the uncertainty sur-rounding conclusions on asymmetric behaviour. In practice,asymmetries among models may be of the same size as, orlarger than, asymmetries among countries. A comparison ofthe slopes of aggregate supply curves,16 which are a concisemeasure of the elasticity of supply, for four countries in fourdifferent international linkage models (see Appendix 6.3)shows that the disparity across models is almost double thatacross countries, on average. This indicates that the modelsare convergent in their assessment of the degree of disparityof supply reactions among countries but not in respect ofthe average size of supply reactions in each country.

The effect of EMU on the asymmetries of behaviour ofeconomic agents falls into two parts: (l)wage behaviourand (2) other behaviour. Concerning wage behaviour, thedisinflationary period of the 1980s has seen the demise ofmany automatic indexation schemes which were one of thecauses of real wage rigidity. A credible EMU regime maygenerate low inflation expectations and therefore containthis factor of asymmetric behaviour.l7 Asymmetric formsof other behaviour could be the subject of multilateral sur-veillance procedures and expected to be reduced in thatway. This, however, is not obtained automatically and isconditional on adequate surveillance procedures.

Let W be the growth rate of nominal wages, P the growth rate of(expected) prices, U the unemployment rate, and let the wage equationhave the following form:W = a,P + a2P_, - b,U - b 2 U_] + ...Then the short-run real wage rigidity may be defined as a^b,, and thelong-run real wage rigidity as (a, + a2)/(b, +b2), (see Coe (1985)).The slope of the aggregate supply curve is calculated by dividing thethree-year effect of a government expenditure shock on the level of theGDP deflator by the three-year effect on real GDP, both measured aspercentage deviations from baseline values.

6.2.3. Trend real exchange rates

A different possible rationale for changes in real exchangerates is related to permanent shocks, and concerns sucheffects as secular movements in productivity, changing trendsin tastes or demographic factors. An allegedly importanttype of permanent shock arises when countries want to

This issue is discussed in more detail in the next section.

145


maintain differences in output growth over a sustained per-iod of time (10-20 years). The faster growing country wouldimport more than the slowly growing country, and thereforeits bilateral real exchange rate should depreciate with a trendin order to maintain an external equilibrium between them.This case warrants particular attention in EMU given thecatching-up process of the poorer countries which could behampered in this way.

This section analyses, therefore, to which extent fixing nomi-nal exchange rates in EMU really implies costs as describedabove.

In the first place, there is the evidence presented before thatnominal exchange rates do not determine real exchange ratesin the long run. Trend changes in real exchange rates wouldtherefore have to be brought about by prices rather thannominal exchange rates. As discussed earlier, for countriesundergoing structural adjustment or economic reform, theexchange rate may have to be changed to reach a correctlevel, but this is less relevant for Community countries.

Secondly, historical evidence for the Community (seeGraph 6.6 and Appendix 6.4) suggests that there is only avery weak positive relationship between growth differentialswith respect to the Community average and a trend de-preciation of the real effective exchange rate.18 Trends inreal exchange rates have indeed appeared over the period1973-88, but the direction of these trends is not stable overthe two subperiods considered. Furthermore, the correlationbetween positive growth differentials and trend real de-preciation, as expected on the basis of the theoretical con-ditions discussed above, is rather weak. Regressions betweenthese two variables show a positive, but insignificant re-lationship between them with a slope which is so fiat thatreal depreciation cannot plausibly have been of use. Theevidence is even weaker for a country such as Japan, whichcoupled faster growth than its trade partners with a realexchange rate appreciation in the 1960s-70s. Apparently, the

reason for the weak link between trend real exchange ratesand faster growth lies in the existence of other factors influ-encing growth.

A third argument relates to the effects of stronger growthon the current account. There is an empirical regularityto be observed between growth differentials and incomeelasticities (see Appendix 6.5). This tends to move the tradebalance towards equilibrium without needing real exchangerate changes. An explanation for this phenomenon may liein supply factors. Krugman (1989), for instance, developeda model with monopolistic competition and economies ofscale where trade arises from product differentiation, inwhich a fast-growing country will be able through supplyeffects to increase the apparent foreign income elasticity forits exports and to lower the domestic-income elasticity of itsimports. As a consequence, the trade balance would not beaffected. This model fits in with the effects expected fromthe internal market completion as enhanced by EMU, andwill therefore in any case not work against the empiricalregularity observed between the ratios of income elasticitiesof exports and imports and domestic and foreign growthrates. Moreover, the external constraint as such is consider-ably weakened in EMU, as discussed below.

The conclusion from the above theoretical and empiricalevidence seems to be that there is only a weak link betweenfaster growth in the Community and trend nominal exchangerate changes or even trend real exchange rate changes. Fastergrowth seems mainly attributable to other, possibly supply-related, factors. To the extent that the external constraintinhibits faster growth, it will be considerably relieved inEMU.

6.2.4. Conclusion on shocks

18 This is illustrated by the results for the regression line in Graph 6.6:GDP = 0,363 + 0,095'REER Corrected R2 = -0,018

(0,296) (0,105)with GDP = growth rate of GDP minus growth rate of EC GDP, annualaverage 1973-88, REER = estimated linear trend growth rate of unit labourcosts in manufacturing industry relative to Community partners, annual rate1973-88. (Standard errors between brackets.) The coefficient for the realexchange rate is not significant; abstracting from this, it would indicate thatan annual difference in GDP growth rates of one percentage point wouldbe accompanied by a real exchange rate vis-a-vis Community partners whichwould have to depreciate annually by more than 10 %.

The nature of shocks in EMU is crucial, since it presents theorigin of the potential cost. In a simplified framework,for instance, Cohen and Wyplosz (1989) have shown thattemporary, asymmetric shocks have the most negative im-pact in terms of deviations from a social optimum.19 Thus,

19 A negative implication for fiscal policy is that it goes against the objectiveof tax smoothing.

146

sduquenoy

sduquenoy


GRAPH 6.6: Real depreciation and growth, 1973-88

GDP growth p.a., deviation from EC average

IRL

GR

UKD E ML B/L

DK

Annual Annualreal realappreciation depreciation

- 6 - 4 - 2 , 0 2 4

Average annual growth rates

This graph plots the average annual deviation of GDP growth from Ihe FC average(on the y-ams) against an estimated trend real depreciation of the real exchange ralevu-O'ii.i Community partners (estimated as linear trend). The average growth raleshave been calculated over the period 1973-H8 As teal exchange rate, the teal effectiveexchange rate against Communil) partners was used, with unit labour cost in manufac-turing as deflator The regression line shows lhal there is a very weak positiverelationship between trend real depreciation relative to Community partners andhigher growth relative to the Community average

the temporary nature of an adverse shock may induce onecountry to run a trade deficit (while it would have to adjustspending in the case of a permanent shock), whereas theasymmetric nature of the shock will create an incentive todo so through a real appreciation, which is not in the interestof the other country. The results from this section indicatethat asymmetric shocks in the Community, even though theyexist, are likely to diminish with the disappearance of tradebarriers through the completion of internal market. Further-more, shocks apparently are permanent rather than tempor-ary, which reduces their cost. On the supply side, and notablyon labour markets, there are divergencies among Com-munity countries which may cause asymmetric reactions tocommon shocks. This therefore points to the importance ofa common behaviour of the social partners when faced with

identical shocks. From the point of view of catching-up,finally, trend real exchange rates (and therefore certainlynominal exchange rates) have not contributed significantlyin the past, and so there is little reason to expect that theywould be needed in the future.

6.3. Factor adjustment

In the absence of exchange rate changes, an adverse asym-metric shock can be countered in two other market-clearingways, and the question is how these adjustment mechanismscan substitute for the exchange rate in EMU.

The first is through lower labour costs as a contribution tothe relative price decrease needed to restore the competitiveposition of the country and to bring output and employmentback to equilibrium. This solution, price flexibility, is theone which is most closely related to the nominal exchangerate since both prices and the nominal exchange rate influ-ence the real exchange rate. Secondly, factor mobility, andmore in particular labour mobility,20 may solve the problemthrough migration towards another region or country, no-tably if the shock is region-specific or country-specific. Obvi-ously, the stronger factor adjustment capacities are, thelower the costs of EM U are in terms of transitory unemploy-ment.

This section will not discuss the adjustment capacity ofthe factor capital, since the competitive pressure on X-inefficiencies and on profit margins, notably monopoly pro-fits, was already extensively discussed in the context of theinternal market. 2I As part of EMU, the flexibility of profitmargins may be assumed to be further enhanced. In whatfollows, the present state of labour market flexibility in theCommunity is briefly examined, with reference to wageflexibility and labour mobility, and the effects of EMU onboth are analysed. 22 It is concluded that EMU will bring

In practice, if the shock is sector-specific, occupational mobility mayalso be of help if the unemployed find employment in another sector.This argument, and other elements of structural labour market policies,are, however, less related to the loss of the exchange rate in EMU, eventhough the Commission programme for the implementation of theSocial Charter (see Commission of the European Communities (1989)}may be expected to improve occupational mobility, vocational training,etc.Commission of the European Communities (1988).Concerning labour mobility, only regional mobility will be treated, sinceother aspects of labour mobility such as occupational mobility andflexible working-time are less directly related to the loss of the exchangerate instrument.

147


GRAPH 6.7: Relative unit labour cost and relative unemployment rates inside Germany and the Community

(a) Loader of the FRG

Relative ULC

(b) Initial ERM countries

Change in relative ULC

20

10

20

10

-10

-20 r

-10

-20

11

-10-8-6-4-2 0 2 4 6 8 10 12 14Relative unemployment rate

- period 1975-87 -a- period 1979-87

-10-8-6-4-2 0 2 4 6 8 10 12 14Relative unemployment rate

- period 1975-89 -e- period 1979-89

(c) Non-ERM countries

Change in relative ULC

-20

-10-8-6 -4-2 0 2 4 6 8 10 12 14Relative unemployment rate

— period 1975-89 -e- period 1979-89

This graph plots the relationship between rcLtivc unit l.ibour cost (on the y-axis) and relative unemployment rates (on [he x-axis) in (a) the German federation. <b> the initial ERMcountries and (c) the non-ERM countries (Oro-te and Portugal which remain outside the ERM, and Spain and the United Kingdom which joined in 1989 and 1990, respectively). A steeprelationship belween relative unit labour cos'-s and relative unemployment implies thai unit labour cost is relatively flexible in response -O unemployment. The regression lines for Gennanyand the ERM members are significant, and much sleeper for the ERM than for Germany, indicating that unit labour cost flexibf: [y among the ERM countries is stronger than betweenthe German Loader. Moreover, the steepness is increasing over time for Ihe ERM members, but decreasing inside Germany It is clear that the non-ERM members have less discipline.signified by a non-significant regression line Tacrc are however signs that it is becoming steeper and more signiricant.

For the Community countries, the y-a»s gives first differences Of relative unit labour costs, centred around zero. For Germany, the levels are given, also centred around zero (forcomparable estimates, see Appendix 6.6). Fct Germany, th; unit labour cosls and the unemployment rate are measured relative to Ihe average for Germany; for Ihe ERM and non-ERMthis is relative to the Community average. J

148


some automatic improvement in labour market flexibility,but that much depends on changes in behaviour of economicagents. It is argued that additional policies may be conduciveto these required changes in behaviour.

6.3.1. The impact of EMU on real wageflexibility

Real wage flexibility does not apply only to the effects onthe labour market situation of the loss of the exchange rateinstrument, but is important for the situation on the labourmarket in general. This implies that real wages have notonly to adjust to shocks in international competitivenessor foreign demand, but also to domestic shocks such asproductivity changes or movements of other factor prices.

A comparison of existing measures of real wage flexibilitywas presented in Section 6.2. While this is important inrelation to the asymmetric effects of shocks, the role of wageflexibility in relation to the exchange rate instrument inEMU should rather focus on the response of relative labourcosts to differences in unemployment performance. The roleof wages as a substitute for the nominal exchange rate willthen appear more clearly, since the effects of absolute wagelevels on the functioning of domestic labour markets arethereby disregarded.23

A comparison of the relative performance of the Communitycountries with that of the Lander of the Federal Republicof Germany (see Appendix 6.6 and Graph 6.7) shows thatrelative unit labour costs tend to decrease in response to apositive unemployment differential inside the German Fed-eration, but that this response is weaker than that of initialERM countries inside the Community. Non-ERM countrieshave a response which lies in between these two cases, butit is not statistically significant. Moreover, the flexibility ofrelative unit labour costs inside Germany is weakening overtime, while that of Community countries improves.

The preceding results provide an illustration of two moregeneral arguments likely to create a higher degree of wageflexibility in EMU.

(i) The discipline of the exchange rate regime appears whencomparing the results for ERM and non-ERM countries.EMU with a single currency may be seen as the most crediblecommitment to a system of fixed exchange rates. In a lesscredible setting, the temptation for the government to pursuea devaluation policy may tend to induce inflationary wage

increases.24 In a situation of disequilibrium, this may slowdown the adjustment process of the economy to equilibrium.It may therefore be expected that EMU will return faster toequilibrium than, say, the early EMS.25 This argument maybe illustrated with the results of a simulation with the IMFMultimod model (Graph 6.8), which shows the response ofFrench inflation to a domestic price shock in regimes ofpure float, the EMS with a realignment rule, and EMU. Inthe pure float regime, depreciation of the currency generatesa devaluation-price spiral which only dies out slowly. In theEMU regime, the absence of depreciation against othercurrencies of the union ensures that inflation quickly diesout. In the EMS regime, the inflation differential with Ger-many triggers off a devaluation in year 3, which prolongsthe inflationary effects somewhat compared to the EMUregime (see also Annex E).

(ii) A second argument concerns the impact of fiscal disci-pline on the speed of adjustment to equilibrium in the caseof temporary shocks. If fiscal policy is perceived to be lesslikely to 'bail out' a region lacking wage adjustment, wageadjustments are likely to come faster than otherwise. This isdemonstrated by the German results compared to those forthe Community countries. As seen in Section 6.7, Germanyhas a high degree of fiscal equalization.

(iii) A third argument why real wage flexibility is likely tobe enhanced in EMU relates to the effects of the completionof the internal market. In a more highly integrated productmarket the degree of monopoly of individual suppliers de-creases and product demand is more price elastic, and so isthe derived demand for labour. If labour unions take intoaccount the derived demand for labour in setting their wagedemands and the unemployment/real wage trade-off be-comes steeper, wage setting will become more responsive tolabour market conditions: increased competition in productmarkets may result in an increased responsiveness of wagesto unemployment.26

There are also factors, however, whose impact on wageflexibility in EMU is less certain a priori.

(iv) Considerations concerning social cohesion in EMUmight put a limit on too large deviations in income levels.To the extent that this translates into 'wage norms' of anykind, it has been demonstrated by Abraham (1989), that an

If relative unit labour costs are used, this has as additional advantagethat differences in labour productivity are taken into account.

See Horn and Persson (1988).See Begg (1990). For a rather pessimistic account of the Irish experiencein the EMS, see Dornbusch (1989).See Marsdenf 1989).

149


GRAPH 6.8: The disciplinary effect of EML on inflation after a price shockInflation. deviat;on from baseline

Source. Mullimod simulations.

This graph displays the effect on inflation of a 5"fa pnee 'hock in France, simulated with the Multimod model under thr;« regimes 'free float", 'EMS' and 'EMU'. Inflationis measured relative to the baseline, nhich is assumuj to be the equilibrium value. The disinflation process is fastest u~der the EMU regime due to the fact that the purefloat and the EMS allow competitiveness '.o be restored through depreciations or devaluations. The negative inflation deviation from baseline after year 3 for the "EMS' and'EMU' regimes are due to technical assumptions Tor the simulations on the desirable long-run price level, the speed o" adjustment of the regimes being independent of theseassumptions (see also Annex E(.

'absolute norm' aggravates unemployment problems in thepoorer regions. This might, for instance, explain the weakresponse of German unit labour costs to unemploymentdifferences. As proposed by Van Rompuy et al. (1990), a'relative norm' would be more desirable in this case, e.g. anunemployment benefit defined as a percentage of the wagerather than an absolute amount.

(v) An important factor influencing the flexibility of wagesmay be the wage determination process itself. A view whichseems to emerge holds that real wage rigidity may be rela-tively low when there is either highly centralized wage bar-gaining (allowing macroeconomic externalities to be takeninto account) or very decentralized wage bargaining (givingleeway to market forces and therefore resulting in sociallyefficient outcomes). In this view, rigidity would be the highestunder intermediate forms of centralization, since the organiz-ations bargaining at that level would be large enough tocause disruptions, but not sufficiently large to bear thecosts of these disruptions. 27 Although this hump-shaped

relationship between the degree of centralization of wagebargaining and real wage rigidity is not a uniformly acceptedview,28 it appears that the Community countries presentlyare often characterized by an intermediate degree of central-ization, thus pointing to room for improvement in the wagebargaining process in EMU. This does not impose anyparticular wage bargaining model such as extreme centraliza-tion or decentralization, as long as there is a system orprocedure which guarantees that the externalities implied bythe bargaining result are taken into account.29

Taken together, these arguments indicate that the effects ofEMU on wage flexibility are largely dependent, either di-

27 See Calmfors and DrifTtll (1988).

150

An alternative view posits that there exists a monotonic relationshipbetween the degree of 'corporatism' and real wage flexibility. As pointedout by Calmfors and Driffill (1988), the definition of corporatism is notvery well described, but is generally taken to denote the extent to which'broader interests' enter into wage bargaining, e.g. through governmentinvolvement.At the Community level, information and consultation procedures re-garding the externalities of wage bargaining results could take place inthe context of the multilateral surveillance and through a dialoguebetween the social partners at Community level.

sduquenoy


rectly or indirectly, on changes in the behaviour of economicagents determining and influencing wage behaviour. Forgovernments this concerns their reactions to temporaryshocks, whereas the participants in the wage bargainingprocess would have to take account of the implications ofEMU in the field of inflation expectations and its effects onreal exchange rates. Conditional on such considerations,wage flexibility may be able to substitute for the cushioningeffect of the nominal exchange rate instrument.

6.3.2. Regional mobility

In theory, regional mobility could substitute for real wageadjustment to absorb a regional shock. Unemployed workerscould migrate to another region, add to its labour force,increase its income and demand, and would therefore undothe effects of the shock. However, it has been recognizedthat large-scale labour mobility in the Community is neitherfeasible, at least not across language barriers, nor perhapsdesirable.

A comparison with existing federal States may serve as anupper ceiling for the extent to which migration may contrib-ute to the absorption of regional shocks. Table 6.2 showsthat regional net migration in the Community as a percent-age of that in the United States declined to 25 % in the firsthalf of the 1980s down from 50 % in the preceding decade.

Table 6.2Regional net migration in the EC, the USA and Sweden

(average rates p.a., % population)

1970-79 1980-85

EC (64 regions)USA (50 states + DC)Sweden (24 counties)

0,40,8—

0,2(1980-85)0,7(1980-85)0,4 (1985)

Noie: Numbers represent total net migration movements across regional boundaries, and thusinclude movements to or from regions from other Member Stales and third countries as wellas movements between regions within a country. The figure shown for each country is theaverage of the absolute values of the net migration balance for its regions.Sources: EC: Commission of the EC (1987), 'Third periodic report on the social and economicsituation and development of the regions of the Community'. Tab. 2.2.2-B-2, p. 74; USA; U.S.Dept. of Commerce and Bureau of the Census, 'Statistical Abstract of the United States 1987",Tab. 27 p. 24; Sweden: Statistics Centralbyran, •Statistical Abstract of Sweden 1990'. Tab. 37,p. 43.

In addition, since migration would tend to equalize regionalunemployment rates, a comparison of dispersion measuresfor unemployment rates inside federations would be anotheruseful indicator. Eichengreen (1990), for instance, comparedthe dispersion of unemployment rates of nine regions in the

GRAPH 6.9: The dispersion of regional unemployment rates

(a) Standard deviation

EUR 12

FRG

1975 1980 1985 1989

Years

(b) Standard deviation divided by mean

EUR 12 , r -

EUR 9

. , , FRG

1975 1980 1985 1989

Years

The graph compare* the dispersion of unemployment fates among Communitycountries to that among I he II Landvr m Germany Tuo measures are used, thesiand.ird deviation and the it^ndard deviation divided b> the mean (coefficient oflanationj According to the latter criterion, tht diNpemon of unemployment ratesamong tLR 9 countries li.e excluding Greece. Portugal and Spain! is close to thaiin.sidv Germany

151


United States to those of the Community (excluding Greece,Portugal and Spain). He concludes that the measures ofdispersion for the Community are 50 to 100% higher in theCommunity than in the United States.30 A similar compari-son between the German Ldnder and the Community (seeGraph 6.9) shows that for the same nine Community mem-bers the degrees of dispersion are only slightly above thosefor Germany. This result is mainly determined by a combi-nation of cultural barriers and levels of social security. Ifthe contribution of migration to the decrease in dispersionof unemployment rates would have to increase substantiallyin the Community, this would either require a strong de-crease in cultural barriers or a strong decrease in the levelof social security in the Community.

Concerning the cultural barriers, the completion of the in-ternal market will assure the complete freedom of movementof persons in the Community. In practice, Community resi-dents are already allowed to stay in another Communitycountry for a period of three months in order to find a job.The main effect of the internal market resides in the mutualrecognition of professional qualifications, which will there-fore primarily benefit the mobility of qualified workers.Since this is a relatively small layer of the labour force,3I

the additional effect on labour mobility will be commensur-ate. In any case, the evidence on regional mobility in existingfederations and even inside Member States does not suggestthat it would be large enough to bear a significant proportionof the adjustment to a regional-specific shock. In the Com-munity, cultural barriers and language differences seem toexclude that the degree of regional mobility of existing feder-ations will ever be achieved.32 Similarly, it is unlikely thatthe level of social security would decrease substantially. Onthe whole, and quite apart from its desirability, the role ofregional mobility in EMU as an adjustment instrument willtherefore remain small, except in higher income bracketsand certain border regions.33

Using regressions Eichengreen (1990) also finds, however, that the speedof adjustment to the (US or Community) average unemployment rateis only 25 % higher in the USA than in Europe. Furthermore, he findsthat regional unemployment rates are weakly related to each other notonly in the Community but also in the USA. This suggests that thetheoretical role for migration to smooth differences in unemploymentrates, even in a country which is renowned for its regional labourmobility, is fairly limited.As an illustration, 6,8 % of the population between 5 and 24 years wasenrolled in full-time tertiary education in the Community in 1985/86,see Eurostat (1989).Molle and van Mourik (1988) found that the responsiveness of migrationin the Community with respect to wage differentials is strong but that,on the other hand, cultural differences are a strong impediment.For example between Ireland and the United Kingdom, where thelanguage problem does not play a role. In addition, there is the theoreti-cal argument that if exchange rate stability increases real income varia-bility, the option value of not migrating is high, which should reducemigrations, see Bertola (1988).

Conclusion

In addition to the effect on profit margins, labour marketflexibility, and more in particular wage flexibility, is thesingle most important adjustment instrument in the absenceof the nominal exchange rate instrument. Since nominalexchange rates do not determine real exchange rates in thelong run, an adjustment in the latter would have to comefrom labour costs, with the nominal exchange rate delayingthe adjustment. There is theoretical and empirical evidencesuggesting that wage flexibility in the Community under aregime of increasingly fixed exchange rates is improving.Conditional on the behaviour of governments (in their stabi-lization efforts) and the social partners (in their wage bar-gaining processes), this improvement is likely to be strength-ened in EMU. Regional mobility may add to labour flexi-bility to a minor extent. Labour cost-increasing aspects ofsocial cohesion, on the other hand, may exert a dampeninginfluence on labour cost flexibility, but may be desirable onequity grounds. In such a case, relative rather than absolutenorms should be introduced to enhance the effect of auto-matic stabilization.

6.4. Shocks, adjustment and macroeconomicstability

The macroeconomic effects of EMU discussed so far in thischapter concerned not so much the levels of variables suchas output and inflation, as their variability. Random shocksdue to temporary changes in the behaviour of economicagents or originating abroad generally frustrate the attemptsof the authorities to stabilize the economy around desiredlevels for macroeconomic variables. As argued in Chapter 2,an increase in the variability of these variables is thereforewelfare-reducing. Since fixed exchange rates imply that na-tional monetary policy can no longer be used to stabilizethe consequences of asymmetric shocks, the variability ofnational macroeconomic variables might increase in EMU.In this sense, the loss of the exchange rate instrument mightbe considered to have a welfare cost.

This cost is conditional on the presence of asymmetric shocksand on the absence of instantaneous adjustments of wagesand prices to disequilibria on the labour and goods markets.Some attenuation of the cost may take place due to the factthat EMU could reduce these asymmetries and enhancewage and price discipline (as argued in the previous sections).There are two additional factors, however, which indepen-dently will tend to reduce macroeconomic variability inEMU and may therefore offset at least part of the cost.

152

sduquenoy


(i) The first factor relates to the fact that flexible exchangerates tend to change unexpectedly on exchange markets evenif policies or underlying economic factors ('fundamentals')do not change. These unpredictable disturbances in exchangerates increase real and nominal macroeconomic variabilitythrough their effects on real exchange rates and prices ofimported goods. By fixing intra-Community exchange rates,this source of exogenous asymmetric exchange rate shockswill disappear as a source of instability, representing a grossgain from EMU.

(ii) Secondly, there are gains related to the disappearanceof suboptimal or non-cooperative exchange rate or monetarypolicies. For instance, when faced with a common inflation-ary shock, individual countries might try to export inflationthrough competitive appreciations, whereas they could dobetter in terms of macroeconomic stability to coordinatetheir exchange rate policies.34

Giving up the exchange rate instrument therefore does notonly entail costs for macroeconomic stabilization, but alsosome gains. Whether the net effect of all these factors ispositive or negative depends on the precise alternative (the'baseline') with which to compare EMU, and on the empiri-cal magnitudes which are involved.

As discussed in Chapter 2, the baseline will not be the samefor all Community countries, not only because they presentlydo not all share the same exchange rate regime, but alsobecause, even if they all participate in the EMS in Stage I,this would not necessarily be a stable regime. Instead oftrying to work with one single baseliqe, therefore, EMUcould also be compared to different regimes. For instance,EMU would represent a larger gain from the disappearanceof asymmetric exchange rate shocks compared to the hypo-thetical situation of a free float inside the Community ratherthan with respect to the EMS; and similarly, also with respectto the effects on wage and price discipline. The EMS alsoremoves non-cooperative exchange rate policies as theymight exist in a floating exchange rate regime, since realign-ments are usually negotiated. On the other hand, the EMSintroduces an additional cost due to the asymmetry of mon-etary policy in the system. Since monetary policy is mainlydetermined by German domestic policy objectives, monetarypolicy in other EMS members can hardly react to domesticasymmetric shocks. This will reduce macroeconomic stability

for the EMS as a whole since German macroeconomicvariables represent less than half of the EMS total.35

In order to obtain a global idea of the size and the directionof the net effects on macroeconomic stability, a simulationexercise (see Box 6.4) was performed to compare a modelrepresentation of EMU, defined to consist of Germany,France, Italy and the United Kingdom, with three hypotheti-cal baseline regimes. These regimes, chosen not so much fortheir degree of realism, as to clarify analytically specificproperties of EMU,36 were:

(i) 'Free float': inside the Community, each country pur-sues the same monetary policy (interest rate reactionfunction) with respect to its domestic objectives foroutput and inflation; all countries freely float with re-spect to the dollar on the basis of uncovered interestrate parity, i.e. with expected exchange rate changesequal to interest rate differentials.

(ii) 'EMS': this stylized regime resembles most the EMS ofthe mid-1980s. It is assumed that Germany indepen-dently sets its monetary policy and is floating againstthe dollar and that the other countries pursue their ownmonetary policy only to the extent that their exchangerate remains inside a band with respect to the DM. Itdiffers from the observed EMS since there is an auto-matic realignment rule based on price differentials rela-tive to Germany which offsets a price differential of8 % with a devaluation of 4 %.

(iii) 'Asymmetric EMU': this is the same regime as the'EMS', but without realignment rule and zero bandwidth. All countries therefore follow German monetarypolicy. This regime might be interpreted as an inter-mediate stage of EMU with fixed exchange rates butno common monetary policy.

(iv) The EMU regime itself ('Symmetric EMU') is definedas a system of fixed exchange rates with the samemonetary policy (interest rate reaction function) as be-fore, but with average Community output and inflationas macroeconomic policy objectives. This could be inter-preted as the final stage of EMU with a single currency.

34 See Van der Ploeg (1990).

This analysis is not in contradiction with the credibility other EMSmembers may derive from German monetary policy in the face of acommon inflationary shock. Recent developments in the EMS haveshown, moreover, that there is a tendency for monetary policy to becomemore symmetric. This points once more to the difficulty of choosing anevolutionary system such as the EMS as a baseline.For this reason, the United Kingdom was assumed to participate in theEMS regime, ahead of its entry in the ERM on 8 October 1990, butassuming narrow margins rather than wide margins.

153


Naturally, the results of this analysis are surrounded withuncertainty due to the technique used and factors whichcould cause underestimation or overestimation of the re-ductions or increases in variability which are obtained. Sub-ject to this important caveat, the following results emerge(see also Graph 6.10).

GRAPH 6.10: Macroeconomic stability of EMU

120

no

100 f-

90 h

Inflation variability

Float

•; EMS (of mid-1980s)

EMS (cvolv ng)

80 EMU (fixed exchange rales)

EMU (single currency)

6080 90 100 110 120

Output variability

130 140

Indices EC average, free float = 100

This graph plois the combinations of t .inability of output <GDP) and inflation forIhe Community average in index form at rtlulling from the stochastic simulations.The position of each of the four regimes ("fro; fo.il'. "EMS", 'asymmetric EMIT, andE M U ] corresponds to an intersection bet wet i a reojmêpendenl output-inflationtrade-off curve and a shifting prefereno; cjn_-Source Stochastic simulations with the Mtistimod mode of the IMF under theresponsibility of the Commission senlcts dDP is measured us i percentage deviationfrom its baseline value, inflation is meaMirtd in percentage Tom I differences withrespect to baseline inflation rates. The indi-x- used in Ihe graph are obtained byaveraging, first, the squares of the deviations fo- <11 simulations over the penod I MO-99 and by taking the square root. Dinditp -j the rooi rne-n-squared deviations forthe free float regime jnd multiplication b, 100 then gut,-, the indices

Comparing the 'EMS' regime to the 'free float' regime, theCommunity reduces its inflation variability but increases itsoutput variability:

(i) The reduction in inflation variability is the consequenceof the reduction in asymmetric intra-Community ex-change rate shocks and wage and price discipline effects.Only for Italy inflation variability increases ratherstrongly. This may be due to the nature of the historicalshocks and backward-looking wage behaviour in Italy.

(ii) The increase in output variability is due to the constrainton monetary policy from the peg to the DM for thecountries other than Germany. The wage and price disci-pline effects on factor adjustment and reduction in asym-metric exchange rate shocks are present, but apparentlynot large enough to compensate for the reduced roomfor manoeuvre in monetary policy, except in Germanywhich retains its independent monetary policy.

The move from 'EMS' to 'asymmetric EMU' brings strongreductions to both output and inflation variability in theCommunity:(i) The output effect is due, firstly, to the disciplinary effects

on factor adjustment in returning output to equilibriumwhich are now much stronger in all countries pegging tothe DM due to the complete elimination of the remainingband width for exchange rates and the absence of re-alignments. In addition, there is the effect of the disap-pearance of intra-Community exchange rate shocks. Theeffect on output in Germany remains virtually the samedue to its monetary leadership.

(ii) The decrease in inflation variability is stronger in Franceand Italy, followed by the United Kingdom and Ger-many. This order is partly determined by the fact thatthe former three countries lose the possibility for realign-ment relative to the DM.

For the Community, the regime change from 'asymmetricEMU' to 'symmetric EMU1 would bring a further large gainfor inflation and a small further gain in reducing outputvariability:(i) The inflation gain outside Germany is considerable, and

may be ascribed to the fact that all countries' inflationobjectives are now taken into account in the Communitymonetary stance, removing suboptimal asymmetric mon-etary policies. Consequently, German inflation experi-ences a slight increase in variability which does notexceed that of'EMS', however.

(ii) The small effect on output is the net result of sizeableimprovements for France, Italy and to a minor extentthe United Kingdom, and a loss for Germany. Again thisis due to the disappearance of suboptimal asymmetricpolicies.

Although there are some arguments which give grounds foruncertainty about these results (see Box 6.4), these results

154


provide an indication that for the Community as a whole theEMU regime contains features which could to an importantextent offset the costs in terms of macroeconomic stabilityincurred from the loss of the exchange rate instrument.The disappearance of asymmetric exogenous exchange rateshocks and asymmetric suboptimal monetary policies couldreduce inflation and output variability considerably. Alsoeffects of exchange rate discipline on wage and price behav-iour are present.

Due to differences in behaviour and the asymmetric func-tioning of the 'EMS', the effects are not completely uniformacross countries. Compared to the 'EMS' as defined here,i.e. a semi-crawling peg system with German monetary lead-ership, the effects are different for Germany than for theother Community countries. The latter could be expected togain considerably relative to the 'EMS', both in terms ofoutput and inflation, whereas Germany's inflation varia-bility would remain unchanged and its output variabilitycould increase. With a high preference for inflation stabilityand relatively low historical output variability, the loss fromgiving up monetary policy sovereignty by Germany hastherefore to be judged small, certainly since no accountwas taken of the role of fiscal policy and the reduction ofasymmetric shocks.

A final issue is whether the expected improvements in macro-economic stability in EMU will lead! to higher levels ofoutput or lower levels of inflation. The empirical resultsobtained here might add to those of Chapter 4, which showedthe existence of positive correlations between the level ofinflation and the standard deviations of both inflation andGDP growth. Combined with the empirical results, thiswould suggest that EMU could contribute to a decrease ininflation via an improvement in macroeconomic stability.Since there is no conclusive evidence on a possible negativelink between GDP variability and GDP growth37 a similarcase is more difficult to make for output. As suggested inChapter 4, there is, however, an indirect positive effect onreal income per capita via lower inflation.

The relationship may be positive or negative depending on the measureof variability used:Growth = 0,023 + 0,607 * SDEV

(0,009) (0,270)Growth = 0,054 - 0,022 * CFVAR

(0,009) (0,008)where growth is the average GDP growth for 24 OECD countries over1960-88, SDEV the corresponding standard deviation and CFVAR thecorresponding coefficient of variation.

R2 = 0,187 24 observations

R2 = 0,244 24 observations

Box 6.4: Empirical evaluation of macroeconomic stability in EMU

This box presents an empirical illustration of the effects of fixedintra-Community exchange rates in EMU on the variability ofoutput and inflation using the technique of stochastic simula-tions. In other words, it is not the impact of EMU on levels ofoutput and inflation which is analysed, but the impact on thevariation around these levels. The technique used, which in-volves the repeated use of random shocks to the economy, givesan opportunity to obtain an average idea of how the variationwill change in EMU.

Expected effects

A priori, the absence of intra-Community exchange rates exertsboth positive and negative effects on the variability of outputand inflation:

(i) There will be a positive disciplinary effect on wage andprice formation, since economic agents will incorporate intheir price expectations that inflationary intra-Community

'devaluations are excluded; this will dampen wage-pricespirals and increase the speed of adjustment of output toequilibrium through the adjustment of factor prices.

(ii) When a common adverse supply shock occurs (e.g. an oilprice shock), EMU members can no longer try to exportinflation through futile competitive intra-Community ap-

preciations; the elimination of such non-cooperative policiesremoves a source of inflation variability.

(iii) Asymmetric exchange rate shocks on exchange marketswhich are not policy-induced will also disappear, removinganother source of instability of inflation and output.

(iv) With respect to the EMS, asymmetric policies which mainlytake the German output and inflation variability as objec-tives disappear, since the common monetary policy focuseson average EMU output and inflation; since Germanyrepresents less than half of the Community, this shouldimprove the average stabilization record.

(v) Asymmetric shocks other than exchange rate shocks arelikely to increase output variability due to the absence ofthe monetary policy instrument at national level to stabilizeoutput, although, as discussed in Section 6.2, the occurrenceof asymmetric shocks is likely to diminish (this possiblereduction was not taken into account).

(vi) Fiscal policy could play an important role in stabilization,but this effect was left out of consideration in order toisolate the effects of factor adjustment.

Given that there are both positive and negative factors at work,the net impact of EMU on output and inflation variability isultimately an empirical matter. Broadly speaking, it may be saidto depend on the effects of the EMU regime change on shocks,

155


as well as on economic behaviour in reaction to these shocks.This could be analysed by regarding the effects of individualshocks. But in order to obtain a global idea of the order ofmagnitude of the net impact, stochastic simulations were under-taken whereby shocks were given to all residuals from thebehavioural equations simultaneously.

Methodology

The stochastic simulations were performed with the Multimodmodel of the IMF, slightly adapted for the purposes of thepresent study.38 Multimod is a small-sized annual model incor-porating both backward-looking and forward-looking features,linking country models for all G7 countries and the rest of theworld, which therefore contains models for four Communitycountries. The method consists of making repeated draws fromthe observed statistical distribution of shocks, the latter beingbased on the historical residuals of the behavioural equationsof the model, as measured over the past 15 years. The draws forthe shocks are introduced as a disturbance in the correspondingequations. This implies that the shocks are unexpected whenthey occur, but that their effects are known afterwards due tothe forward-looking nature of the model. In reacting to thesedisturbances (keeping fiscal policy unchanged), the authoritiesare assumed to use the short-term interest rate to stabilize,around their equilibrium values, a certain combination of out-put, inflation and possibly the exchange rate. The equilibriumvalues are taken to be the baseline values of the model. Aneffective policy is therefore one which is able to obtain outputand inflation values close to the baseline values.

Due to the problems with the identification of a proper baselinewith which to compare the EMU regime (see Chapter 2), thisexercise is repeated for four different intra-Community exchangerate regimes. These regimes are not so much modelled for theirrealism, but rather to clarify analytically certain specific issues:'Free float': All Community currencies are assumed to floatfreely against the dollar, and therefore against each other. It isassumed that there is perfect capital mobility, perfect assetsubstitutability and perfect foresight. Exchange rates thereforefollow uncovered interest rate parity (UIP); each country hasthe same interest rate reaction function focusing on deviationsof domestic output and inflation from equilibrium.39 This func-tion is in accordance with observed behaviour of monetarytargeting.'EMS': Germany freely floats against the dollar on the basis ofUIP and the same interest rate reaction function as above; theother Community countries use their interest rates to peg theirexchange rate to the DM within a band, allowing them only

38 See Masson el a!. (1988. 1990) for a description of the Multimod model and Anne* Efor the modifications.

39 It is assumed that a one-percentage point increase of inflation increases the short-terminterest rate by two percentage points and vice versa. A 1 % increase of output relativeto baseline increases the short-term interest rate by 0.4 percentage points and viceversa. This reaction function was derived from the money demand equation of themodel. Half of the interest rate etTeci is assumed to take place in the current year, andhair in the next year.

very partially to stabilize their domestic output and inflation.Moreover, there is a stylized realignment rule derived from theEMS experience over the period 1985-87, saying that when anaccumulated price differential with Germany of 8 % is reached,the DM parity is devalued by 4 %.

'Asymmetric EMU': Germany freely floats against the dollaron the basis of UIP and has the same interest rate reactionfunction as above. The other Community countries follow Ger-man interest rate movements completely; this pegs their ex-change rates fully to the DM, and does not take account oftheir domestic output and inflation objectives.

'EMU': All Community countries float together against thedollar on the basis of UIP. The Community interest rate followsthe same reaction function as before, but the output and inflationobjectives now concern deviations from equilibrium for theaverage Community output level and inflation rate. This sym-metric system is taken to be the final stage of EMU.

As between the four regimes, the only equations which changeare the interest rate reaction function and, directly or indirectly,the determination of the exchange rate. All other equations areleft unchanged. Disciplinary effects of EMU therefore comeabout entirely through the forward-looking features of themodel, not through changes in model coefficients. This holdsnotably for future inflation, which determines real interest ratesand appears in the wage-price block.

A second aspect which changes between the four regimes is thenature of shocks. In order to avoid arbitrary assumptions,exchange rate shocks and shocks to other variables are assumedto be independent from each other in all four regimes. The sizeof, and dependence between, the non-exchange rate shocks isalso kept the same between the regimes. The exchange rateshocks themselves have to be modified, however. On the basisof observed shocks over the 1979-88 period, this involves increas-ing the average correlation among shocks to Community dollarexchange rates from 0,7 in the free float regime, via 0,9 in theEMS regime, to 1 in the EMU regimes. Furthermore, the sizeof the exchange rate shocks is allowed to differ, in the lightof historical data (see Annex E), between Germany (standarddeviation of shocks of 10,9%) and the other Communitycountries (standard deviation of 9,6%) in the free float regime,whereas it is assumed to be equal in the three other regimes(standard deviation of 10,9%).

Assessment

Analysing the effect of EMU on output and inflation variabilitywith stochastic simulations in the way it was done for thepurpose of this study has its strong, but also its weak points.Consequently, the results have to be interpreted with caution.

(i) Economic issues

The exercise only covers Germany, France, Italy and the UnitedKingdom. If all Community countries had been included, shocksmight have been more asymmetric on the whole and the re-duction in variability smaller. This would have introduced the

156

sduquenoy


problem of how to deal with the effects of the transition onthe nature of the shocks, however, notably for the peripheralcountries. After the transition, price shocks could decrease, forinstance.

The EMS was modelled in a stylized way as a semi-crawling pegmuch like the situation in 1985-87, but assuming free capitalmobility, completely anticipated realignments and participationof the United Kingdom. Since the EMS is an evolutionarysystem, any way of modelling it is bound to be ad-hoc. With theparticipation of the United Kingdom as of 8 October 1990, thepresent EMS presumably lies in between this semi-crawling pegand the asymmetric EMU regime. This should be taken intoaccount when evaluating EMU relative to the EMS baseline.

Another issue related to the question of baseline is the use ofthe same monetary reaction functions in the free float regimefor all countries. The variability in the free float regime wouldpresumably be reduced if there were asymmetric interest ratereactions to common shocks.

Similarly, the results depend on the assumption that fiscal policywas kept unchanged. As argued in Section 6.6, the use of fiscalpolicy with temporary asymmetric shocks may delay adjustmentand increase output variability.

Another issue concerns the size of the exchange rate shocks.Ceteris paribus, the higher the shocks, the stronger will bethe reduction of output and inflation variability due to thedisappearance of asymmetric exchange rate shocks. Althoughshocks in dollar exchange rates of some 10 % on an annualbasis seem not unreasonable for the 1980s, these could changedepending on future international monetary arrangements.

(ii) Technical issues

An argument frequently used against policy evaluation witheconometric models (the so-called Lucas critique) is that they

would not be able to deal with the effects of changes in policyregimes. The forward-looking aspects of the model go a longway in meeting this objection since once a regime (free float,EMS, EMU) is in place, the model behaves as if all agents knewits macroeconomic features. At the same time, this also impliesthat the results are conditional on changes in actual behaviour,notably concerning wage and price discipline.

The reduction of asymmetries in shocks was only introducedfor exchange rate shocks and not for shocks to other variables,even though it was argued elsewhere in this chapter that for thelatter also such a decrease could be expected. This could be asource of underestimation of the possible reduction in varia-bility. Moreover, the shocks were drawn on the basis of residualsfrom equations which, with some exceptions, were imposed tohave the same coefficients among countries. On the one hand,this may have introduced additional asymmetries in the residualswhich would not have been observed otherwise, but on the otherhand it may have reduced behavioural asymmetries.

Although the policy reaction function used for interest rates isconsistent with the estimated money demand equation of themodel, it was not derived from full optimization. This raisesthe issue of possible suboptimality of monetary policy. Sinceobserved behaviour may not be optimal either, this avoids mak-ing arbitrary assumptions.

From the methodological point of view, finally, the techniqueof stochastic simulations in evaluating policy regimes is stillunder development.40 Some technical aspects are discussed inAnnex E. Compared to the evaluation of regimes using singleshocks whose effects may go in opposite directions, stochasticsimulations have the merit that they provide an overall evalu-ation of stability, even though the shocks on which this is basedare historical. But this problem is not much different for singleshocks.

40 See Frenke] el al. (1989) or ihe Brookings conference'Empirical evalualion oFaltema-live policy regimes', 8 and 9 March, 1990.

157

sduquenoy


Results of stochastic simulationsRoot mean-squared deviations, free float = 100

Freefloat

EMS AsymmetricEMU

SymmetricEMU

Historicalvariability1

1973-88<PM)

EC2

Germany

France

Italy

United Kingdom

GDP3Inflation4

GDPInflation

GDP .Inflation

GDPInflation

GDPInflation

100100

100100

100100

100100100100

10996

9486

13598

1'31125

9886

9381

9683

10472

114104

8175

9073

10886

9360

10390

8068

1,4-2,2-

1,92,2

1,43,7

2,45,2

2,55,6

3,56,0

1 Standard deviation of inflation and growth rale of GDP at annual rates.1 Variables measured as unweighted average of Germany. France, llaly and the United Kingdom; for variability: minimum and maximum of all Community countries,3 Measured as percentage difference with respect to baseline.4 Measured as difference with respect to baseline in percentage points.Source' Simulations with the IMF Multimod model under the responsibility of the Commission services. The numbers in the table are root mean-squared deviations for annual data withrespect to baseline (= equilibrium) for 43 simulations over the period 1990-99. thus representing a draw of 430 shocks for all 95 behavioural variables.

6.5. Adjustment through external financing

With a single currency, the current account constraint vis-a-vis the other members of the union disappears. In a Tin-bergen-type target-instrument framework with external bal-ance as one of the objectives and the exchange rate one ofthe instruments, the disappearance of the exchange rateinstrument would be exactly compensated by the eliminationof the external balance target. In that context, there wouldbe no loss from the exchange rate instrument. There are,however, several reasons to pay separate attention to theexternal constraint. Firstly, since there are usually moretargets than instruments, the exchange rate loss can neverbe completely compensated by the disappearance of theexternal constraint. Secondly, the extent of attenuation ofthe exchange rate loss through this mechanism can onlybe assessed by analysing the current state of the externalconstraint ('baseline'). Thirdly, the external balance can keepa signalling function in EMU. Fourthly, as argued below,in a world of free capital mobility, there is still the constraintof long-run solvency. Sustainability conditions require thatthe trade balance cannot stay in deficit forever. An externalequilibrium over the medium run may be interpreted as

a sufficient, though not necessary, condition for long-runequilibrium, in this context.41

There are two important ways in which the absence ofconstraints on the balance of the current account in EMUmay (partially) substitute for the exchange rate instrument.Both cases rest upon the interpretation of the current ac-count surplus as the excess of national saving over invest-ment.

The first case is when there is a temporary adverse shock,e.g. a decrease in external demand, causing a decrease inincome and therefore in consumption. If there is no possi-bility of modifying the real exchange rate, the same level ofconsumption can be maintained by temporarily decreasingthe savings ratio, which will then deteriorate the currentaccount balance further. In a financially integrated area withfree capital mobility, the shock is therefore absorbed throughtemporary borrowing abroad.

41 See also Oliveira-Martins and Plihon (1990).

158

sduquenoy


A second case arises when a country for some reason expectsa future increase in productivity but has insufficient savingsto bring about the international equalization of real rates ofreturn implied by this productivity differential, as may bethe case with the poorer countries of the Community. A realdepreciation could generate a surplus on the current accountwhich could be used to finance the investment. Alternatively,with free capital mobility the investment could be financedby incurring a deficit on the current account through in-creased borrowing from abroad. When the borrowing hasbeen transformed into productive capital, the faster growthallows the foreign debt to be repaid.42

These two cases provide clear examples of how the currentaccount may take over the role of the nominal exchangerate. Nevertheless, they hinge crucially on assumptions aboutthe free mobility of capital, or, more generally, on the con-straints on the current account or 'external constraint'. Itis therefore important to analyse the role of the externalconstraint in order to assess to what extent its removalcompensates for the loss of the exchange rate instrument.Major issues in this respect are:(i) the operation of the external constraint;

(ii) the implications of EMU for the external constraint.

6.5.1. The external constraint

The external constraint may be defined as the extent towhich, for a particular country or economic region, thereexist limitations on the net acquisition of foreign financialassets and liabilities. These limitations may be describedunder several headings:(a) Barriers to capital market integration: these may consist

of capital controls, transaction costs, information costs,discriminating tax laws or the risk of future capitalcontrols. These barriers may be measured by the so-called country premium,43 i.e. the deviation from coveredinterest rate parity.

(b) Real exchange rate uncertainty: even in the absenceof barriers to capital market integration (zero countrypremium), real exchange rate uncertainty may preventthe equalization of real rates of return and therefore theoptimal international allocation of capital. A countrywith good investment opportunities but high exchangerisk may thus be deprived of the flow of capital neededto exploit the existing opportunities.44 With zero countrypremium, the deviation from real interest rate parity maybe due to (i) a nominal exchange risk premium, measuredby the deviation from uncovered interest parity, i.e. thedifference between the forward discount and expectednominal depreciation; and (ii) the deviation from relativepurchasing power parity (PPP).

(c) Correlated savings and investment: if there is a commondeterminant for the savings rate and the investment rateother than the internationally equalized real interest rate,changes in the domestic savings rate may crowd out theinvestment rate and therefore interfere with the optimalinternational allocation of capital. In practice, this maytherefore operate as a limitation to capital mobility. Ifthere are no limitations of this sort and if there is realinterest rate parity, the savings rate and the investmentrate should be uncorrelated. This would typically beexpected in countries small enough not to influence theworld real interest rate. This is the definition of perfectcapital mobility first presented by Feldstein and Horioka(1980). Government intervention and limited substituta-bility between bonds, equity and physical capital (e.g.due to sunk costs) may be the basic reason why theFeldstein-Horioka definition is violated.

The difference between the two approaches lies in intertemporal welfareconsiderations. The trend real depreciation implies a substitution ofcurrent consumption by saving, and is therefore inferior from the pointof view of consumption smoothing. See Blanchard and Fischer (1989)or Frenkel and Razin (1987),This terminology was developed by Frankel (1989). See also Box 6.5. See Baldwin (1990).

159


Box 6.5: Measuring capital mobility in the Community

The table, calculated on the basis of Frankel( 1989), decomposesthe mean deviation from (ex-post) real interest rate parity intoa country premium, exchange risk premium and deviation from(ex-post) relative purchasing power parity for the Communitycountries relative to Germany over the period September 1982to April 1988.45 Ideally, each of these factors should be zero forperfect capital mobility to hold.

The table allows the following conclusions to be drawn:

(i) The least stringent condition by which to judge the absenceof the external constraint is the country premium, i.e. the devi-ation from covered interest rate parity, which should be zero. Ifit is negative, it means that domestic interest rates are artificiallylow compared to the DM interest rate. This may be due tocapital controls, as is clearly the case with Greece and Portugal(but also the other countries having maintained capital controlsover the period), or due to transaction costs. For the threecountries with the least capital controls over the period, theUnited Kingdom, the Netherlands and Belgium, the country

Measures of capital mobility relative to Germany, 1982-88

Countrypremium

(1)

Exchangerisk

premium(2)

RelativePPP

deviation (3)

Currencypremium

(4) = (2)+ (3)

Real interestparity

BelgiumDenmarkGreeceSpainFranceIrelandItalyNetherlandsPortugalUnited Kingdom

-0,23-3,88-9,74-2,75-2,09-1,14-0,75-0,14-8,28-0,49

3,403,39

-0,474,873,353,164,660,267,16

-0,34

-1,34-1,59

4,530,340,110,50

-1,660,26

-1,773,51

2,081,802,494,783,014,803,080,746,602,27

1,82-2,13-7,93

1,820,812,822,300,58

-2,611,75

(1): i - i* - fd12): fd - E(3) ;E - (P - E")14): fd - (P - P*)( S ) : i - P - (i* - P')with fd = ihree-momh forward discount with respeci 10 the DM,wiih E = observed deprecialion wilh respect 10 DM (proxy for expected value), three-month percentage change a! annuwilh P = observed inflation (proxy for expected value) over three-month period al annual rale.with i = money market interest rate over three-month period.with * = refers to German variables.Now. The identities (5) = (l)" t"(4) and (4) = (2)+ (3) are not always respected due lo inconsistencies in the original data material.Source: Calculated from Frankel (1989), averages of monthly observations over the period September 1982 lo April 1988.

lal rate

premium is not more than minus 50 basis points, reflectingprobably only transaction costs.

(ii) Due to the proxies used in its measurement, the exchangerisk premium (deviation from uncovered interest rate parity) islinked to the variability of observed rather than expected localcurrency/ DM exchange rates. Of the countries participatingin the exchange rate mechanism (ERM) of the EMS in themeasurement period, the premium for the Netherlands is thesmallest, followed at some distance by the other narrow-bandcountries and finally Italy. Outside the ERM, the premiums forSpain and Portugal are both larger than for Italy.46

(iii) The ex-post real depreciation, i.e. deviation from relativePPP, is surrounded with even more measurement errors thanthe previous criterion, since not only expected depreciation butalso expected relative price changes are approximated by ob-served values. Nevertheless, on the basis of this criterion, itseems that this source of limitation to capital mobility does notplay a significant role.47

How may the Feldstein-Horioka hypothesis for the Communitycountries be analysed? Feldstein and Horioka (1980) had to

45 Price expectations and exchange rate expectations have been proxied by their observedvalues.

46 The results for Greece and the United Kingdom may seem anomalous at first sighi.implying negative risk premiums. This may be due. however, to the fact that ex-postinstead of ex-antf deprecialion has been used. If there is ex-post more deprecialionthan expected, which does not seem an unreasonable assumpiion in these two cases,this reduces the observed risk premium.

The exceptions being, as for the previous criterion, Greece and the United Kingdom.The stronger depreciation than presumably expected shows up with a mirror image.As a consequence, this measurement error cancels out if the exchange risk premiumand the ex-posi deviation from relative PPP are added to fonn the so-called currencypremium; see Frenkel (1989).

160


reject the hypothesis of uncorrelated savings and investmentusing data for most OECD countries over the period 1960-74.In so far as Community countries were included in the sample,the implied hypothesis that financial markets would not beintegrated was jointly rejected for them as well. Various authorshave advanced arguments which could explain the particularrejection of the hypothesis on the basis of cross-section esti-mates.48 The two most important aspects from the point of viewof EMU in this respect are the influence on the Feldstein-Horioka result of capital controls and exchange rate risk.

One way to proceed is to analyse periods and/or areas for whichit is known that there were no capital controls and exchangerisk. If savings and investment are uncorrelated for such regimes,it may be inferred that capital controls and exchange rate uncer-tainty play an important role in impeding perfect capital mo-

48 Feldstein and Horioka themselves already took account or the Tact that savings andinvestment could both behave pro-cyclically by averaging data over the cycle. Theyalso took account that both could be influenced simultaneously by demographicfactors. Similarly, they also analysed whether a breakdown of savings by economicagent and investment by component influenced ihe results. Finally, they tested forspurious correlation by taking the changes in the savings and investment rates between1970-74 and 1960-69. In all cases, their basic conclusion was not affected. Feldsteinand Bacchetu (1989) found that [he correlation had decreased, but was still signifi-cantly different from zero, over the period 1980-86 compared to earlier periods, aresult confirmed by Frankel (1989) using time-series regressions for ihe USA. Artisand Bayoumi (1989) estimated policy reaction functions showing that curren [-accounttargeting played a role in government behaviour in the 1970s which diminished in the1980s, thus highlighting Ihe role of endogenous government behaviour interfering withmovements of savings and investment.

bility. In this vein, Bayoumi (1989) ran several cross-sectionregressions among savings and investment rates over the GoldStandard period 1880-1913, finding them to be uncorrelated.Given that the Gold Standard was a period with fixed exchangerates, no capital controls and limited government intervention,this suggests evidence in favour of the Feldstein-Horioka hy-pothesis. Bayoumi and Rose (1989) obtained a comparableresult using regional data for the United Kingdom over theperiod 1971-85.

A second possibility, more focused on the EMU perspective,would be to single out the ERM countries in regressions coveringthe EMS period and to test whether the ERM countries displaycorrelation between savings and investment. This line of researchwas pursued by Bhandari and Mayer (1990). Their cross-sectionestimates indicate that the ERM countries (plus Austria) showno significant correlation between savings and investment forthe periods 1979-82 or 1983-87, thus suggesting that the ERMhas contributed positively to the mobility of capital. The presentcurrent account imbalances inside the Community remain, how-ever, much lower than those observed in federal States or monet-ary unions. Among Community members current account pos-itions ranged from a surplus of 5,8 % of GDP for Germany todeficits of 4,1 % and 3,4 % for the UK and Greece, respectively.In contrast, Luxembourg, which forms a monetary union withBelgium, had a current account surplus of 14,9 % of GDP andthe estimates of regional accounts in the MacDougall report49

also contain a number of examples of current account imbal-ances of over 10 % of GDP.

49 See Commission of the European Communities (1977).

Data on these criteria for the 1980s (see Box 6.5) seem toindicate for the measures related to real interest rate parity(country premium, exchange risk premium and deviationfrom relative PPP) that in the last decade the necessaryconditions for perfect capital mobility were met inside theCommunity to a limited extent. This was more the casefor the narrow-band ERM countries which have liberalizedcapital flows than for other Community countries. But eventhere, the most simple criterion, i.e. almost zero countrypremium relative to Germany, was only met for Belgiumand the Netherlands. Outside the ERM in the measurementperiod, the only exception concerns the country premium inthe United Kingdom which is low, as could be expected dueto its lack of capital controls over the period concerned. Thecurrency premium (sum of risk premium and deviation fromrelative PPP) is only below 1 % for the Netherlands, andvaries between 2 and 7 % for the other countries. Lookingat the real interest parity deviation, which is an aggregateof the three measures used, perfect capital mobility couldonly be said to exist between Germany and the Netherlands.Naturally, this is based on imperfect measures. Furthermore,the present situation is likely to have improved with ongoingcapital liberalization and low inflation differentials.

Concerning the correlation between savings and investment,there is strong historical evidence of the absence of such acorrelation in regimes with fixed exchange rates, no capitalcontrols and limited government intervention. This is cor-roborated by regression results reported in Box 6.5 for ERMmembers relative to other Community countries, demon-strating the positive impact of fixed exchange rates for capi-tal mobility.

6.5.2. Effects of EMU on theexternal constraint

The previous evidence indicates that the absence of capitalcontrols and exchange rate risk are important factors inenhancing the international mobility of capital. These twocornerstones of EM U may therefore be expected to stimulatepositively the optimal allocation of capital inside the Com-munity. For instance, it may be expected that the completeliberalization of capital flows as agreed for eight Communitycountries by 1 July 1990 and the other countries by theend of 1992,50 will reduce the country premium to figures

Portugal and Greece may delay the liberalization until the end of 1995.

161


comparable to those for Belgium, the Netherlands and theUnited Kingdom.

In principle, the only remaining limitation in EMU consistsin a long-run solvency constraint of companies, householdsand governments. The current account may for a long timebe in disequilibrium as long as there is the expectation thatin the end the foreign debt will be repaid without issuingnew liabilities.51 Since there is one single currency, the con-straint is not much different from that facing domesticborrowers vis-a-vis domestic lenders, and becomes less bind-ing in the short run.

This conclusion needs to be qualified, however, since itdoes not mean that the current account completely loses itsmeaning. For one thing, there is the problem that agents arenot able to observe the intertemporal solvency constraint.They may therefore still want to analyse the origins ofcurrent account imbalances. An imbalance due to net im-ports of capital goods might be viewed differently from animbalance due to net imports of consumption goods. Foranother, governments might still want to use the currentaccount as an indicator of possibly inflationary policies.Moreover, as argued by Artis and Bayoumi (1989), thegovernment might still want to target the current accountfor several reasons, such as the fear for the influence offoreign capital on the domestic economy, or the differencebetween social and private benefit, e.g. due to the fact thattaxes paid in the home country or abroad may have thesame welfare implications for an individual, but not for thehome country.

The conclusion is, therefore, that the complete liberalizationof capital flows and the irrevocable fixing of exchange ratesor a single currency would imply that the disappearance ofthe intra-Community external constraint would be at leastpartially a substitute for nominal exchange rate adjustmentin EMU.

such as private borrowing abroad or a transfer or loan fromother members of the union. The importance of budgetarypolicy in this context derives from the assumption that thegovernment cares about both internal and external equilib-rium, although the role of the latter changes in EMU, asdiscussed in the previous section. These two policy objectiveswill in general require two instruments, and the loss of theexchange rate instrument could imply the need to rely moreon an alternative instrument such as budgetary policy.

The two main issues in this respect are:

(i) when to assign budgetary policy rather than other instru-ments;

(ii) the effectiveness of budgetary policy relative to the ex-change rate instrument through monetary policy.

Throughout the discussion, it will be assumed that shockswhich occur disturb external equilibrium, such as a shockto external demand or competitiveness (real exchange rate).In other words, the shocks are such that the optimal policyassignment in a system of flexible exchange rates and nomi-nal rigidity would have been the exchange rate. Chapter 5has shown the need for budgetary policy in the case ofshocks of domestic origin.

6.6.1. The need for budgetary policy

The assignment of budgetary policy as an alternative to theexchange rate instrument in the face of a shock dependsbasically on three elements:

(i) policy objectives,(ii) adjustment mechanisms in relation to objectives,

(iii) the nature of the shocks.

6.6. Budgetary policy as an alternative to theexchange rate instrument

In relation to the loss of the exchange rate instrument inEMU, budgetary policy may have a role as one form offinancing to compensate temporarily for the loss of theexchange rate instrument, next to other forms of financing

As indicated above, this can only happen under certain circumstances,such as a temporary adverse output shock or an expected future increasein productivity. On the other hand, and in line with the theory ofintertemporal optimization, a permanent adverse shock to output shoulddecrease the permanent income level and therefore not lead to increasedborrowing from abroad as with a temporary shock.

Starting from a shock which disturbs equilibrium, it is as-sumed for the discussion here52 that the policy objectivesare internal equilibrium (output equal to production ca-pacity) and external equilibrium (equilibrium on the currentaccount).

The adjustment mechanisms which may return the economyto equilibrium are usually not the same in the short run asin the medium run. Moreover, their effects on the policyobjectives depend on the view taken regarding the workingof the economy.

This abstracts from other policy objectives such as inflation stabilization.

162

sduquenoy

sduquenoy


In the short run budgetary policy may have a direct impacton demand. Monetary policy influences domestic demandindirectly via interest rates, and the current account viathe exchange rate, given that exchange rates are flexible.Assuming wage and price rigidity in the short run, thisimplies that budgetary policy and monetary policy — viathe exchange rate — are the two instruments to be assignedto the two policy objectives in the short run.53

In the medium run, monetary policy loses much of its impactbecause the effect of the nominal exchange rate on the realexchange rate disappears through its effect on domesticprices. Budgetary policy may also lose some impact, forinstance because future tax changes to offset the budgetarypolicy are expected. Wage and price adjustments thereforeinevitably have to supplement government policies, notablyin order to maintain the effect on real exchange rates. Thereare still two objectives, but more than two mechanisms toestablish equilibrium since the mechanisms may lose part oftheir effect.

How do policy objectives and instruments change in EMU?As discussed in the previous section, external equilibriumdisappears as a constraint in the short run, although itremains a valid objective for the long run. This lengthensthe period over which the current account can stay in disequi-librium before it needs to be returned to balance. Thisimplies a significant compensation for the loss of the nominalexchange rate and independent monetary policy.

The assignment of alternative instruments is less neededdue to the short-run loosening of the external equilibriumobjective. To the extent that it still has to take place, theimpact on domestic demand which is lost with monetarypolicy can be taken over by budgetary policy. Wage andprice adjustments should mainly be geared to influence thereal exchange rate.54 In the medium run, budgetary policy,wages and prices are therefore the remaining instruments.

The precise assignment of these instruments is related to thenature of the shocks which cause the disequilibria in the firstplace.

A temporary shock to domestic (relative to foreign) demandor the real exchange rate could usually be corrected by

wage and price adjustments. But with nominal rigidity, thecorrection of internal disequilibrium which such a shockimplies may be temporarily taken over by budgetary policy,although this risks delaying the medium-run adjustment toexternal equilibrium since economic agents know that thegovernment will try to dampen fluctuations.55

A permanent shock to either relative demand or competitive-ness requires inevitably that factor markets adjust throughchanges in wages and prices, since budgetary policy cannotbe changed permanently without affecting government debt.So again, budgetary policy may help temporarily, but couldcause the factor adjustment to be sluggish.

Summing up, the assignment of budgetary policy as analternative to the exchange rate instrument is only partiallyneeded due to the virtual disappearance of the objective ofexternal equilibrium. If needed, it may be of some use in thecase of temporary shocks when there are nominal rigidities.Permanent shocks require the adjustment of factor markets,but should not be delayed by relying on budgetary policy.

6.6.2. The effectiveness of budgetary policy

The effectiveness of budgetary policy in general is the subjectof a long-standing debate among different schools ofthought, which will not be discussed here.56 What is import-ant, however, is the time framework to which these dis-cussions apply. With nominal wage rigidity, a severe adverseshock causing unemployment may be countered effectivelyin the short run through an increase in government spending.In the medium run, there should be an adjustment of thenominal wage to restore equilibrium on the labour market.Once this has happened, the government does not have tomaintain its deficit and can start repaying its debt longbefore the intertemporal budget constraint starts influencingprivate saving behaviour.

In addition to these general arguments, there are a numberof arguments that may have a particular influence on theeffectiveness of the instrument of budgetary policy in EMU.

The short run is assumed to cover several years, so that the feasibilityor desirability of fiscal fine-tuning is not an issue. Alternatively, theasymmetric shock may be assumed to be severe enough to warrantactive policy action.Alternatively, Begg (1990) assigns a more important role to wages andprices in obtaining internal equilibrium. This approach rests strongly onthe rather theoretical assumption that real balance effects are important.

See Begg (1990).The 'classical' and 'Keynesian' visions are opposed in their assumptionson the flexibility of money wages. For an analysis, see Sargent (1987).This analysis is essentially comparative-static.. In an intertemporalframework, another controversy concerns the impact of the deficitincurred by the government on its intertemporal budget constraint.People may see through the action of the government and anticipatefuture taxes by increasing private saving, thus rendering budgetarypolicy completely ineffective. This is called Ricardian equivalence, seeFrenkel and Razin (1987).

163


In the first place, there is the familiar Mundell-Flemingproposition that in a world of free capital mobility budgetarypolicy is more effective in a regime of fixed exchange ratesthan in a regime of flexible exchange rates. The reason isthat a fiscal expansion in a flexible regime will, throughan increase in interest rates, appreciate the currency andtherefore exert a negative effect on competitiveness which isabsent in a system of fixed exchange rates such as EMU.The counterpart in EMU is that the increase in interest rateswill spill over to other member countries and reduce theiroutput, which raises coordination problems (see Chapter 5).

The differences in impact on GDP of a fiscal expansion areillustrated, in this case for France, on the basis of simulationswith the IMF Multimod model57 in Graph 6.11. As expectedon theoretical grounds, the effect is stronger in EMU thanin the case of a free float, with the EMS as intermediatecase.

For a description of the treatment of the float, EMS and EMU exchangerate regimes, see Annex E.

GRAPH 6.11: The effectiveness of budgetar} policy in EMU

-0,4 L.

Source Muttimod simulation.

Effect on GDP % deviation from baseline

Float FraiceYear

-—(-- - EMS France - EMU -ranee

This graph shows the impact on GDP rel-iinc to j baseline of a simulation of an increase in government expenditure The result we~e obtained from a simulation for France with theMultimod model under three regimes, 'trix flo-i:'. 'LMS' and 'EML" The traditional Mundell-Hcmmg result thjt fiscal polity is most effective in a system of fixed exchange rules appearsfor the first five yean after the shock.

164


Secondly, in addition to the requirements on budgetarypolicy following from spill-over considerations, budgetarypolicy will be confronted, in EMU, with more constraintsthan would otherwise be the case:(i) the need for fiscal discipline;

(ii) a possible contribution to the Community policy mix.

These constraints, discussed in Chapter 5, may restrict theleeway for an independent budgetary policy, although theparticular country situation might of course be taken intoconsideration in the coordination procedure or as part ofthe Community policy mix.

Thirdly, as shown by Begg (1990), there is the point that inEMU (but also in alternative regimes) fiscal stabilizationpolicy in the case of supply shocks may be counterproductiveto real wage adjustment. 58 In a model with perfect foresight,households will adjust their wages less quickly to restoreequilibrium on the labour market if they know that they are— at least temporarily — bailed out by the government. Thisreduces the speed of adjustment of the economy towardsequilibrium. On the other hand, given that there is a gain inthe speed of adjustment to equilibrium in EMU comparedto more flexible exchange rate regimes such as the EMS (seeSection 6.3), this creates some room for fiscal stabilization,assuming the government is prepared to trade off the latteragainst the former. The trade-off between real wage adjust-ment and budgetary policy is illustrated, on the basis ofsimulations with the Commission's Quest model for France,in Graph 6.12. The graph plots different combinations ofpresent values of the real wage and the government budgetdeficit in the case of a negative export shock, each combi-nation representing a different policy alternative to a devalu-ation. 59 The loss of the exchange rate instrument is rep-resented by the distance between the devaluation point andthe trade-off curve. Obviously, wage moderation causes thelargest real wage loss, whereas an increase in governmentinvestment is the most desirable from this point of view.

Conclusion

In EMU, the objective of external equilibrium becomes lessimportant. This reduces the need to replace the exchangerate instrument by an alternative policy such as budgetary

Historically, an extreme case is that of the Mezzogiorno in Italy, dis-cussed in Chapter 9. See also OECD (1990).The size of the policy instruments were chosen so as to obtain the samepresent value for the loss of real GDP over a seven-year period as forthe devaluation. The size of the devaluation was chosen so as to returnto baseline real GDP after seven years. For a complete description, seeAnnex D.

policy. Nevertheless, free capital mobility renders budgetarypolicy more efficient in a system of fixed exchange rates suchas EMU than in more flexible exchange rate regimes suchas the early EMS. If external equilibrium is of concern,budgetary policy can therefore play a useful role. This holdsmore in the case of temporary shocks than in the case ofpermanent shocks, where factor adjustment inevitably hasto take place. More generally, there is a trade-off between theuse of budgetary policy and the speed of factor adjustment.

GRAPH 6.12: The trade-off between budgetary policy andwage adjustment

10Real wage rate decrease (present values)

Devaluation

Wage moderation

\\ Social security

Government investment

0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 5,5 6

Government expenditure, % GDP

Source Quest simulations.

This graph displays different adjusunenl policies in the wake of a negative exportshock, in Ihis case simulated for France with ihe Quest mode!. The y-axis shows thepresent value over a seven-year period of the real wage rate decrease which followsfrom the shock. The x-axis gives Ihe present value over Ihe same period of Ihe effecton government expenditure. The seven-year period corresponds lo the period whichis needed to bring real GDP back to baseline (equilibrium) if the negative export shockis followed immediately by an offsetting devaluation. Nexl. three other instruments areused as an alternative lo the devaluation, which result in exactly the same presentvalue of the output loss as the devaluation (there is an output loss due to the initialnegative effect on output before it returns lo equilibrium). Wage moderation has themost negative impact on the real income loss, as could be expected. A decrease inemployers social security contributions reduces the real income loss strongly, but hasa budgetary cost. An increase in government expenditure results in even slightly lessreal income loss, but comes at a greater budgetary cost. The cosl of the loss of theexchange rate instrument in lerms of teal wage decrease or government expenditureincrease is measured by the vertical or horizontal dislanoe from the devaluation poinlto the irade-off curve for the other instruments.

165


Moreover, budgetary policy should take account of theimplications for spill-over effects on other EMU membersand the constraints arising from fiscal discipline.

6.7. The role of central public finance in assistingregional adjustment

There are several ways in which central public finance in afederal system can contribute to the cushioning of country-specific shocks. The two most important areas are those ofinterregional and interpersonal redistribution. By taking therole of central public finance in the case of existing feder-ations as a baseline, this section tries to draw inferences forthe role of public finance in EMU. It is argued that EMUis a unique case, but that the trade-off between fiscal auton-omy and interregional fiscal equalization, which is a charac-teristic of existing federations, nevertheless applies.

6.7.1. Central public finance in existingfederations

Central public finance has an important place in existingfederations (see Table 6.3). This importance may be judgedfrom several indicators.

Federal expenditure as a percentage of total governmentspending is a composite indicator since it comprises severalfunctions of public finance. Even in the more decentralizedfederations like Switzerland, federal expenditure represents

at least 30 % of total spending. Excluding social expenditure,this figure attains a maximum of 54 % in the United States.Federal expenditure in existing federations includes majorexpenditure functions in the provision of public goods, butmay also cover part of the interpersonal grants reducingpersonal income losses due to economic shocks.

On the revenue side, federations usually have a considerabledegree of tax harmonization, and the highest level of govern-ment is normally the dominant tax authority in the sensethat few federations (e.g. USA and Switzerland) attributeexclusive tax competences to regional governments.

Federal grants, which give an indication of the size of interre-gional distribution, constitute between 15 and 30% of re-gional revenue.60 In terms of GDP, this amounts to figuresbetween 3 % (USA) and 7 % (Australia) of GDP.

As a consequence, primary income differentials betweenregions in existing federations are reduced by 30 to 40 %through the workings of central public finance.61 Roughlyhalf of this is automatic through taxation and direct govern-ment expenditure, the other half is discretionary and inten-tional in the form of specific and general purpose grants.Central public finance also provides for automatic stabiliza-tion of regional income in the case of economic shocks. Ithas been estimated in the MacDougall report62 that forunitary States such as France and the United Kingdom, one

See Lamfalussy (1989).See the MacDougall report. Commission of the European Communities(1977).See Commission of the European Communities (1977).

Table 6.3Central public finance in five major federations

Expenditure (net federal spendingas % of total consolidated spending')

Revenue(degree of State autonomy2)

Grants(%ofGDPJ)

SwitzerlandUSACanadaGermanyAustralia

2854373140

5663456756

7978561636

3,62,74,23,47,0

government. General government includes social security lunds.Data for 1980-81. Measured as the percentage of exclusive and competing Uses and non-fiscal income in total Slate revenue.Data for 1987. Source: calculated from OECD (1989b).

Source: Van Rompuy and Heylen (1986) and Commission services.

166


Box 6.6: Fiscal autonomy and interregional equalization

This box examines the relationship between fiscal autonomy andinterregional equalization in five major federations, Switzerland,USA, Canada, Germany and Australia. In order to determinethe degree of fiscal autonomy of a region (state) in a federation,three aspects are considered: revenue sources, policy com-petences, and the associated volume of expenditure and thedegree of independence with respect to borrowing and levels ofbudget deficits. For interregional equalization the results of theMacDougall report63 are referred to in view of the lack of morerecent comparable empirical evidence.

Fiscal autonomy

The degree of autonomy in terms of revenue sources is evaluatedby looking at the percentage of exclusive and competing taxesand non-fiscal income in total regional revenue (Table 6.3, thirdcolumn). For Switzerland and the USA the percentage is be-tween 75 and 80 %, for Canada it is in the order of 50 % whereasfor Germany and Australia the figures are much lower; inAustralia this is due to the fact that a region's revenue isdetermined for about one-third by shared taxes and about one-third by federal grants, while in Germany shared taxes represent70 % of the Ldnder revenue.

The degree of expenditure competence is evaluated by lookingat the share of state and local spending in total governmentspending, excluding social security for reasons of comparability(Table 6.3, column (a)). In Germany and Switzerland, the sub-federal level has a dominant expenditure competence of around70 %; in Canada and Australia this is around 60 %, while it islowest in the USA with about 45 %.

As regards the degree of independence with respect to borrowingand budget deficit levels, the only example of permanent federalcontrols over state borrowing or debt is found in Australia. Inthe USA, the state constitutions impose their own constraints,usually in some form of balanced budget rules. The GermanLander are constitutionally bound to the golden rule of limitingborrowing to the financing of investment expenditure; underexceptional circumstances, this rule can be deviated from. In

Ranking of fiscal autonomy

Revenue Expenditure Independence


1/2/31/2/31/2/345

addition the government can temporarily limit recourse of theLdnder to loans. Deficit spending and borrowing by Canadianprovinces and Swiss cantons are not subject to any legal con-straints.

On the basis of these three criteria, a rather clear ranking offiscal autonomy emerges. In Switzerland the regions have thehighest degree of fiscal autonomy; they are followed closely byCanadian provinces. On the other hand, Australian states havevery little fiscal autonomy. In Germany and the USA, thestates have a similar degree of fiscal autonomy with perhaps asomewhat greater autonomy in the USA: whereas in the USAstates have limited spending responsibilities but considerablerevenue autonomy, the German situation is the other wayaround, i.e. the Lander have a considerable degree of spendingresponsibilities but very limited taxation powers.

Interregional equalization

The MacDougall report64 provides two measures of inter-regional income equalization through federal public finances.On average, Australia is clearly leading with a reduction ofinterregional income disparities due to public finance of morethan 50 %, followed by Germany with about 35 %, Canadawith about 30%, the USA with about 25% and Switzerlandwith a somewhat incomplete figure of 15 %.

Commission of the European Communities (1977).64 Commission of the European Communities (1977. p. 30).

167


half to two thirds of the short-term loss in primary income,due to a shock, may be offset through central public finance.More recently, it has been estimated for the United Statesthat about 35 % of the loss is compensated for through thefederal budget.65

In addition, a typical characteristic of existing federationsis the. trade-off between fiscal autonomy and interregionaltransfers through central public finance. This is illustratedin Table 6.4, which ranks existing federations according tothe two criteria. The ranking of Australia, Germany, Can-ada, the United States and Switzerland in terms of the overallbudgetary autonomy of States/regions is clearly inverselyrelated to the degree with which interregional transfers re-duce income differences and thus exert a dampening effecton the adverse consequences of economic shocks.

Table 6.4Fiscal autonomy and fiscal equalization in existing federations


Ranking of fiscalautonomy

13245

Ranking of interregionalincome equalization

54321

Source: see Box 6.6.

6.7.2. Central public finance in EMU

According to any indicator, whether it is expenditure, rev-enue or the size of its interregional grants, the Communityin its present state is characterized by a very high degree ofnational fiscal autonomy.

This is combined with a very low degree of interregionalfiscal equalization. Concerning the latter, the Communityprovides some degree of cushioning of regional shocksthrough the Community's grant and loan instruments, inparticular the structural Funds and balance-of-paymentsloans, but this is incomparably smaller than what can beobserved in federal States. For the whole of the Communitybudget, Eichengreen (1990), for instance, estimates that notmore than 1 % of an income loss in a member country wouldbe compensated by lower taxes paid to the Community.

From the point of view of public finance, the Communityis therefore a unique case compared to existing federations,combining high fiscal autonomy with low fiscal equalization.There are three arguments which plead in favour of somerecalibration of the present role of central public finance inEMU.

The first argument arises from equity considerations, andconcerns the impact of EMU on regional disparities.w Thisimpact will result from the combined effect of the shockswith which national economies are faced and the adjustmentcapacities with which they are able to respond to theseshocks. If this impact is negative, collective welfare consider-ations in EMU could lead to the decision to reduce it byenhancing the role of central public finance.

If the latter does not take place, regional disparities mightbecome such that countries may have an incentive to with-draw from EMU. This does not only concern these countries,but may also weaken the credibility of EMU as a whole forthose who stay in. Enhanced central public finance maytherefore serve as an insurance for credibility of the system.

The third argument derives from the role of public financeitself in the provision of public goods. It says that in orderto maintain the same level of public goods, a reduction infiscal autonomy has to be compensated by an increase inthe role of central public finance. In other words, there is atrade-off between the two. Van der Ploeg (1990), for in-stance, enumerates five reasons why the size of the publicsector could be reduced in EMU:67

(i) the decrease in seigniorage revenue;

(ii) the possibility that national government spending be-comes more and more like a European public good,whose supply will be sub-optimal;

(iii) the downward effect on government revenue from taxcompetition;

(iv) fiscal policies pursued to obtain a real exchange rateappreciation in order to boost real income will be tighterthan in a cooperative situation due to spill-over effectson other EMU members;

(v) as markets become more integrated, a budgetary expan-sion is likely to be a locomotive policy, which willdiscourage a fiscal expansion in the face of commonadverse supply shocks.

6S See Sachs and Sala-i-Martin (1989).

168

This issue is taken up in more detail in Chapter 9.See also Chapter 5.

sduquenoy


There are several ways in which the role of central publicfinance could be enhanced. The structural Funds in generalsupport the adjustment capacity of regions. The negativeconsequences of shocks could be countered through Com-munity shock-absorption mechanisms. Van Rompuy et al.(1990) emphasize the need for interregional distribution bythe Community in support of vital public functions in thepoorer regions, such as education, transportation and hous-ing. This would therefore imply a widening of the eligibilitycriteria of the structural Funds. If horizontal transfers areestablished in the case of adverse shocks, they suggest linkingit to close surveillance of economic performance, whichintroduces an element of conditionality in the shock absorp-tion mechanism. Both Van der Ploeg (1990) and Wyplosz(1990) suggest interpersonal redistribution schemes to copewith the unemployment effects in depressed regions of ad-verse shocks. Van der Ploeg proposes to set up a budget-neutral European Federal Transfer Scheme (EFTS) whichwould distribute transfers on the basis of regional differencesin unemployment. To avoid problems of moral hazard, thetransfers should be conditional, for instance in the form ofmatching grants for training and schooling. Wyplosz goesfurther by proposing an explicit insurance scheme, also con-sisting of conditional matching grants. Participation shouldbe compulsory to avoid the problem of 'adverse selection'.

Conclusion: a trade-off between fiscalautonomy and fiscal equalization

The adjustment to adverse shocks in EMU can to someextent be borne through labour market flexibility. Politicalconsiderations (in the case of regional migration) or welfareconsiderations (in the case of wage flexibility) may howeverput a certain limit on this adjustment instrument. The re-maining adjustment will have to come from national orcentral public finance. Existing federations each have theirown mixture between federal and sub-federal spending, butin any case there is a trade-off between fiscal autonomy andinterregional fiscal equalization.

This line of argument leads to the conclusion that in orderto minimize the costs of EMU, either the role of centralpublic finance in EMU has to be strengthened or MemberStates/regions need to be left with a very high degree offiscal autonomy. Subject to the required amount of fiscaldiscipline, the Community should strike a balance betweenthe two. If the role of central public finance is enhanced, thestructural Funds may improve the adjustment capacity ofregions, while — possibly conditional — shock-absorption

mechanisms could provide insurance against the remainingburden of shocks.

Appendix 6.1: Do nominal exchange ratesdetermine real exchange rates in the long run?

If nominal exchange rates are able to determine real ex-change rates in the long run, this implies that their levelsshould move together in the long run. This is the hypothesisof so-called 'co-integration' of nominal and real exchangerates. For testing, the null hypothesis is usually that of noco-integration. A co-integration test between two variablesis performed in two stages (see also Engle and Granger(1987)). In the first stage, it is tested whether the levels ofthe two variables are each non-stationary, i.e. that they haveno tendency to revert to a given mean (this property is alsocalled random walk or unit root). If they revert to givenmeans, further testing makes no sense since they are thenco-integrated by definition. Similarly, if one of the variablesis non-stationary and the other is stationary, they are non-co-integrated, also by definition. When the variables are eachnon-stationary in levels but stationary in first differences, thesecond step of the test consists of testing whether the residualof the regression of the level of one of the two variables ona constant and the level of the other variable is itself non-stationary. If this hypothesis is rejected, the two variablesare said to be co-integrated. Intuitively this is clear: if onevariable is linearly related to another variable in the longrun, the residual of that relationship should have a tendencyto revert to zero.

The table overleaf presents the co-integration tests for thelogarithms of quarterly nominal and real effective exchangerates with respect to Community partners over the period1980 Q3 to 1989 Q4. From columns 1 and 2 it appears thatfor Belgium, the Netherlands and Portugal the hypothesisof non-stationarity has to be rejected for the real exchangerate but not for the nominal exchange rate, so that they arenot co-integrated by definition. For the remaining countries,the co-integration tests in columns 3 and 4 never reject thehypothesis of no co-integration at a significance level of5 %. The overall conclusion is therefore that since nominalexchange rates and real exchange rates are not co-integratedin the long run, the nominal exchange rate cannot a fortioribe considered as determining real exchange rates in the longrun.

169

sduquenoy

sduquenoy

sduquenoy


Co-integration tests for nominal and real effective exchange rates, 1980 Q3 -1989 Q4

Belgium/LuxembourgDenmarkGermanyGreeceSpainFranceIrelandItalyNetherlandsPortugalUnited Kingdom

Step 1:

Nominalexchange rale

2,076 yes1,706 yes1,171 yes0,002 yes1,891 yes2,508 yes2,059 yes2,424 yes1,074 yes1,486 yes1,180 yes

Non-stationarily?

Realexchange rale

3,217** no0,925 yes0,933 yes2,597 yes1,545 yes0,654 yes0,698 yes1 ,425 yes2,792** no4,173***no1,433 yes

Step 2:

Nominal on realexchange rate

1,769 no2,279 no2,967 no0,450 noU07no0,406 no2,227 no

2, 146 no

Co-integration?

Real on nominalexchange rate

2,490 no2,330 no0,145 no0,484 no2,563 no2,401 no2, 175 no

1,739 no

*,","•: significance at 10%, 5% and 1%.The values shown in Ihe table are T-sialistics of the coefficient b in Ihe regressionln(u,/u,.|) = a + blndi,.,) + c^niu,.,/^) + c^nfu^/u^,) + Cjlnlu^/u,,,) + cjnfu.^/u,.,).If e is the quarterly nominal effective exchange rale with respect to Community partners and r ihe corresponding real exchange rate, the variable u is defined per column as:Column 1: u = eColumn 2: u = rColumn 3: u = residual from regression ln(r) = const + f ln(e).Column 4: u = residual from regression ln(e) = const + g ln(r).The tests are augmented Dickey-Fuller tests. The critical values for columns 1 and 2 are from Fuller (1976), and have the values 2.60 for a 10% significance level, 2,93 for 5% and 3,58 for 1% with50 observations. The critical values for columns 3 and 4 are from Engle and Granger (1987) and have the values 3,17 for 5% and 3,71 for 1% with 100 observations. The regressions for columns 3and 4 excluded a constant. The real exchange rate was calculated using unit labour costs in manufacturing.

Appendix 6.2: Real wage rigidity

On the basis of recent econometric estimates the OECDcombined the short-run elasticity of nominal wages withrespect to consumer prices and the elasticity of nominalwages with respect to unemployment to a measure of realwage rigidity. Those Member States for which estimatesexist appear clustered together at high degrees of rigidity(very high for the United Kingdom, the Netherlands, Ger-many and France, lower for Denmark and Belgium, withItaly in between) far above the rigidity observed in the USAor Japan (see Graph 6.5). While quite intriguing at firstglance, closer examination reveals a high degree of uncer-tainty about these measures.

The first element, the short-run elasticity of nominal wagegrowth with regard to inflation, indicates very differentdegrees of temporary absorption of price shocks by moder-ation in nominal wage claims between the Member States(see Table 6.5).

The estimates for single Member States vary widely due todifferent estimation periods, equation specifications, choiceof variables, etc. According to the most recent estimates bythe OECD (1989a), the immediate (i.e. first half-year) pass

through of inflation to nominal wages is highest for Ger-many and rather low for Belgium, Denmark, Spain and theUnited Kingdom. For Belgium the respite is only brief,however, due to full ex-post indexation (shown by the Snees-sens estimate based on annual data). In comparison withthe OECD estimates of Coe (1985), the drop in Italianprice elasticity of wages corresponds well with the reducedcoverage of the 'scala mobile1; the reasons for the sharpincrease in indexation in Germany are less obvious. But inGermany, and to a lesser extent in Belgium, Denmark andSpain, money wages are found to react accommodatingly toshort-term productivity changes which provides for ad-ditional shock absorption.

The other major macroeconomic component of labour mar-ket flexibility is the impact of labour market tensions onwage claims (the slope of the short-term Phillips curve).Again estimates (see Table 6.6) are not robust to changes inspecification, etc. The OECD (1989a) estimates imply that— ceteris paribus —- increases in unemployment between2,5% points (Italy) and 10% points (Denmark) would berequired to slow money wage growth by one percentagepoint — as compared to 1,3 percentage points in the USAand 0,5 percentage points in Japan. In addition severalestimates for the United Kingdom find unemployment hys-teresis, i.e. the dampening effects of labour market slack onwage growth are diminishing over time.

170


Table 6.5Short-term price elasticity of money wages

BelgiumDenmarkGermanySpainFranceItalyNetherlandsUnited Kingdom

For comparison :United StatesJapan

Sources :1 Coe (1985).1 For Belgium: Sneessens and Drew (1986); for Italy:

OECD(l989a)

0,250,250,750,250,500,600,500,33

0,140,66

Zenezini (1989).

OECD1

0,44

0,470,960,47

0,04/0,12

0,220,93

Schul(ze(l985)

0,53/0,61

0,65/0,790,3/0,57

0,60/0,66

Others'

1,31/0,91

0,41

Table 6.6Elasticity of money wages with regard to unemployment

BelgiumDenmarkGermanySpainFranceItalyNetherlandsUnited Kingdom

For comparison :United StatesJapan

OECD (1989a)

-0,25-0,10-0,11-0,20-0,29-0,39-0,27-0,15

-0,61-1,87

OECD1

-0,25

-0,31-0,65-0,44-0,17

-0,33-3,31

Others3 Quest' 88

-0,11/-0,43

-0,12

-0,10-0,09

-0,09

-0,30

Quest' 90

-0,10

-0,13

-0,01

-0,19

Sources :1 Coe (1985).1 For Belgium: Sneessens and Drez£(l986); Tor Quest; Bekx el al. (1989) and Commission services.

171


Appendix 6.3: Country asymmetries versusmodel asymmetries: the aggregate supply curveThe results in the table are calculated from non-linked simu-lations of an increase in real government expenditure. The

slope of the aggregate supply curve is calculated by dividingthe three-year effect on the GDP deflator by the three-yeareffect on real GDP, both expressed as percentage deviationsfrom baseline values. Averages are arithmetic.

HermesQuestInterlinkMimosa

AverageMean absolute deviation

Germany

0,002,280,730,90

0,980,65

Franc*

0,481,260,380,06

0,550,36

Italy

1,092,180,300,52

1,020,61

UnitedKingdom

0,282,821,500,69

1,320,84

Average

0,462,140,730,54

0,970,62

Mean absolutedeviation

0,320,440,390,25

0,35

Source: For Interlink; Richardson (1987); Quesi and Hermes: Commission services; Mimosa: CEPll and OFCE (1990)

Appendix 6.4: GDP growth rates relative toCommunity average and trend real effectiveexchange rate changes

The differences in GDP growth rates are at annual rates;the trend real effective exchange rate (REER) is at annual

rates and based on a trend regression (*, **, *** = signifi-cant at 10%, 5% or 1 %) with as dependent variable theREER with respect to Community countries using unit lab-our costs in manufacturing industry, a plus-sign denoting adepreciation.


GDP1981-73

Trend REER1981-73

GDP1988-81

Trend REER1988-81

GDP1988-73

Trend REER1988-73

Belgium/ LuxembourgDenmarkGermanyGreeceSpainFranceIrelandItalyNetherlandsPortugalUnited Kingdom

0,1-0,7

0,01,1

-0,10,62,40,70,11,1

-1,2

1,42,4"1,0

-3,8***-2,9***-0,2

. 0,62,2***2,0**4,0*

-5,6**

-0,5-0,1-0,4-0,7

0,6-0,4-0,2

0,1-0,6

0,00,7

0,3-3,9**-2,9**

2,61,20,63,8***

-0,70,21,5***5,4***

-0,2-0,4-0,2

0,30,20,11,20,4

-0,20,6

-0,3

2,5***0,6

-0,5-3,0***-0,0

0,01,2***0,31,6***2,6***

-2,1***

Appendix 6.5: Elasticities and equilibrium on thetrade balanceIf exports are a function of foreign real income and the realexchange rate, and if imports are a function of domestic realincome and also the real exchange rate, the equilibriumcondition for the trade balance is equal to (see Krugman(1989)):R = a.(Xy.Y*-My.Y),where R = growth rate of the real exchange rate

X = elasticity of exports with respect to foreignreal income

Y* = growth rate of foreign real income

M = elasticity of imports with respect to domesticreal income

Y = growth rate of domestic real income

a = constant with negative sign if the Marshall-Lerner condition holds.

172


If X = M and the constant a satisfies the Marshall-Lernercondition, stronger domestic growth (Y > Y*) implies thatR should increase (i.e. depreciate) at a constant rate tomaintain equilibrium.

The Marshall-Lerner condition for the elasticities with re-spect to the real exchange rate is usually assumed to be valid.This is less so for the condition on elasticities with respectto real income. In a two-country world, the elasticity ofimports with respect to domestic real income for one countryis the same as the elasticity of exports of the other countrywith respect to foreign real income. In a world with morethan two countries, such a condition is only globally valid,i.e. the weighted sum of income elasticities of imports shouldbe equal to the weighted sum of foreign income elasticitiesof exports. Moreover, as observed by Krugman (1989), theratio of the foreign income elasticity for a country's exportscompared to the domestic income elasticity for its importstends to be positively correlated with the ratio of long-termdomestic growth compared to long-term foreign growth. Interms of the formula, this implies that Xy/M is proportionalto Y/Y*. This empirical regularity has considerable impli-

cations for the role of the real exchange rate in equalizingthe trade balance. It implies that part or all of a long-termgrowth differential is absorbed by differences in elasticitieswith respect to real income. This explains why countries mayexperience strong growth while at the same time appreciatingin real terms, as happened to Japan over the last decades.As a consequence, the real exchange rate is deprived of itsrole as adjustment instrument.

Appendix 6.6: Wage flexibility in theCommunity compared to the German federation

This appendix presents regression results which comparethe responsiveness of relative unit labour costs to relativeunemployment rates in the Community to that inside theGerman federation. For the Community, estimates are pre-sented for the initial ERM members, the non-ERM membersand the Community as a whole (see table below).

Time series cross-section regression results for relative unit labour costs on relative unemployment rates, 1975-89

ULCj, = aj + b*[URit - UR]

Period Germany

1 For Germany, [he period ends in 1987.Legend:i = Germany: II Loader; ERM: BLEU. D. DK, F. I. IRL. NL; Non-ERM: GR, E, P. UK.ULC: Unit labour cost in manufacturing in country i in year t/average unit labour cost (Germany or EUR 12).URj.: Unemployment rate in country i in year t — average unemployment rate (Germany or EUR 12).a,: Country-specific constant, not reported.b: Semi-elasticity or relative ULC with respect to relative unemployment (standard error between brackets)sdev; Estimated standard error of the regression.mean: Sample mean of the dependent variable.

Non-ERM EUR 12

1975-89'

1979-89'

H I P — T U P —aUlA-j, ULA, i t_! 3j

Period

1976-89'

1980-89'

b

sdev/meanb

sdev/mean

+ b*[URjt - URt]

b

sdev/meanb

sdev/mean

-0,394(0,108)

1,4%- 0,276

(0,121)1,2%

Germany

-0,020(0,078)

1,0%-0,002(0,095)1,0%

-1,767(0,537)9,1%

-2,533(0,490)7,2%

ERM

-1,110(0,274)4,2%

-1,401(0,301)4,1%

1,017(0,831)14,3%0,090

(0,974)10,2%

Non-ERM

-0,823(0,548)9,7%

-1,013(0,876)10,1%

-0,148(0,438)11,9%- 1,420(0,497)9,0%

EUR 12

-0,947(0,286)6,7%

- 1,236(0,387)6,9%

173


The results are presented for two models, one with the levelof the relative unit labour cost as dependent variable andanother with the first difference of the relative unit labourcost. Each estimation was performed for two periods, oneincluding the pre-EMS period and a second on the EMSperiod alone. The model in levels shows that the responseof unit labour costs to differences in unemployment is signifi-cant both in Germany and the ERM, but that it has de-creased in Germany and increased in the ERM during theEMS period. For the non-ERM countries the response hasa positive sign and is not significant, although it has de-creased in the EMS period. As a result, the responsivenessof the Community as a whole has become significant for the

EMS period compared to the period before. The model infirst differences provides the same conclusions concerningthe direction of changes in coefficients, but gives non-signifi-cant coefficients in the cases of Germany and the non-ERMmembers, and significant coefficients for the Community asa whole.

These results suggest that: (i) wage flexibility is highest inthe initial ERM countries, somewhat weaker in Germanyand still low in the non-ERM countries; (ii) wage flexibilityhas improved in the EMS period for all Communitycountries, but has deteriorated inside Germany.

174


References

Abraham, F. (1989), 'Wage norms and Europe's single mar-ket', International Economics Research Paper 63, Centrumvoor Economische Studien, Katholieke Universiteit Leuven.

Artis, M., Bayoumi, T. (1989), 'Saving, investment, financialintegration and the balance of payments', IMF WorkingPaper 89/102, International Monetary Fund, WashingtonDC, 14 December.

Baldwin, R. (1990), 'On the microeconomics of the Europeanmonetary union', in European Economy (1990).

Bayoumi, T. (1989), 'Saving-investment correlations: immo-bile capital, government policy or endogenous behaviour?',IMF Working Paper 89/66, International Monetary Fund,Washington DC, 22 August.

Bayoumi, T. A., and Rose, A.K. (1989), 'Domestic savingsand intra-national capital flows', mimeo, December.

Begg, D. (1990), 'Alternative exchange rate regimes: the roleof the exchange rate and the implications for wage-priceadjustment', in European Economy — 'The economics ofEMU1, special issue, 1990.

Bekx, P., Bucher, A. Italianer, A. and Mors, M. (1989),The Quest model (version 1988)', Economic Papers, No 75,Commission of the European Communities, Directorate-General for Economic and Financial Affairs, March.

Bertola, G., (1988), 'Factor flexibility, uncertainty and ex-change rate regimes', in De Ceccho, M., Giovannini, A.A European Central Bank?, Cambridge University Press,Cambridge, pp. 95-119.

Bhandari, J., Mayer, T. (1990), 'Saving-investment corre-lation in the EMS', mimeo, January.

Blanchard, O. J., Fischer, S. (1989), Lectures on macro-economics, MIT Press, Cambridge Massachusetts/London.

Calmfors, L., Driffill, J. (1988), 'Centralization of wagebargaining', Economic Policy No 6, April, pp. 13-61.

CEPII and OFCE (1990), 'Mimosa, une moderation deI'economie mondiale', Observations et diagnostics econ-omiques No 30, Observatoire francais des conjonctures econ-omiques, Paris, January, pp. 137-197.

Coe, D. T. (1985), 'Nominal wages, the NAIRU and wageflexibility', OECD Economic Studies No 5, autumn, pp. 87-126.

Cohen, D., Wyplosz, C. (1989), 'The European monetaryunion: an agnostic evaluation', in Bryant, R. C., Currie,D.A., Frenkel, J.A., Masson, P.R., Portes, R. (eds), Macro-economic policies in an interdependent world, InternationalMonetary Fund, Washington DC, pp. 311-337.

Commission of the European Communities (1977), Reportof the study group on the role of public finance in Europeanintegration, collection of studies, Economic and financialseries Nos A13/B13, Commission of the European Com-munities, Brussels/Luxembourg, April.

Commission of the European Communities (1987), Thirdperiodic report on the social and economic situation and devel-opment of the regions of the Community, Commission of theEuropean Communities, Luxembourg.

Commission of the European Communities (1988), Theeconomics of 1992', European Economy No 35, Commissionof the European Communities, Luxembourg, March.

Commission of the European Communities (1989), Com-munication from the Commission concerning its action pro-gramme relating to the implementation of the CommunityCharter of basic social rights for workers, COM(89) 568final, Brussels, 29 November.

Commission of the European Communities (1990), Theimpact of the internal market by industrial sector: the chal-lenge for the Member States', European EconomyjSocialEurope, special issue, forthcoming.

Dornbusch, R. (1980), Open economy macroeconomics, BasicBooks, New York.

Dornbusch, R. (1989), 'Credibility, debt and unemployment:Ireland's failed stabilization', Economic Policy No 8, pp.173-209.

Eichengreen, B. (1990), 'One money for Europe? Lessonsfrom the US currency union', Economic Policy No 10, April,pp. 117-187.

Engle, R. F., Granger, C.W.J. (1987), 'Co-integration anderror-correction: representation, estimation and testing1,Econometrica Vol. 55, No 2, March, pp. 251-276-

European Economy (1990) — The Economics of EMU',special issue.

175


Eurostat (1989), Basic statistics of the Community, Com-mission of the European Communities, Luxembourg.

Fair, R. C. (1988), 'Sources of economic fluctuations in theUnited States', Quarterly Journal of Economics, May, pp.313-332,

Feldstein, M,, Horioka, C. (1980), 'Domestic saving andinternational capital flows', Economic Journal No 90, June,pp. 314-329.

Feldstein, M., Bacchetta, P. (1989), 'National saving andinternational investment', National Bureau of EconomicResearch paper presented at conference on savings.

Frankel, J. A. (1989), 'Quantifying international capital mo-bility in the 1980s', NBER Working Paper No 2856, Na-tional Bureau of Economic Research, Cambridge, Massa-chusetts, February.

Frenkel, J. A., Razin, A. (1987), Fiscal policies and the worldeconomy, MIT Press, Cambridge Massachusetts/London.

Frenkel, J. A., Goldstein, M., Masson, P. R. (1989), 'Simu-lating the effects of some simple coordinated versus uncoor-dinated policy rules', in Bryant, R. C., Currie, D.A., Frenkel,J.A., Masson, P.R., Portes, R. (eds), Macroeconomicpoliciesin an interdependent world, International Monetary Fund,Washington DC, pp. 203-239.

Fuller, W. A. (1976), Introduction to statistical time series,John Wiley, New York.

Gandolfo, G. (1987), International economics, Springer Ver-lag, Berlin/Heidelberg.

Giavazzi, F., Giovannini, A. (1987), 'Exchange rates andprices in Europe', Welwirtschaftliches Archiv, 124, No. 4, pp.592-604.

Greenaway, D., Milner, C. (1986), The economics of intra-industry trade, Basil Blackwell, Oxford/New York.

Horn, H., Persson, T. (1988), 'Exchange rate policy, wageformation and credibility', European Economic Review, 32,No 8, October, pp. 1621-1636.

Jacquemin, A., Sapir, A. (1988), 'International trade andintegration of the European Community: An econometricanalysis', European Economic Review, 32, No 7, September,pp. 1439-1449.

Krugman, P. (1989), 'Differences in income elasticities andtrends in real exchange rates', European Economic Review,33, pp. 1031-1054.

Lamfalussy, A. (1989), 'Macro-coordination of fiscal policiesin an economic and monetary union in Europe', in Reporton economic and monetary union in the European Community,Committee for the study of economic and monetary union,Commission of the European Communities, Luxembourg,pp. 91-125.

Marsden, D. (1989), 'Occupations: The influence of theunemployment situation', in Molle, W. T. M., van Mourik,A. (eds), Wage differentials in the European Community,Avebury/Gower, Aldershot, pp. 105-139.

Masson, P., Melitz, J. (1990),'Fiscal policy independence ina European monetary union', paper prepared for a confer-ence on 'Exchange rate regimes and currency unions', Deut-sche Bundesbank, 21 to 23 February.

Masson, P., Symansky, S., Haas, R., Dooley, M. (1988),'Multimod — A multiregional econometric model', Worldeconomic and financial surveys, International MonetaryFund, Washington DC., April, pp. 50-104.

Masson P., Symansky, S., Meredity G. (1990), 'MultimodMark II: a revised and extended model', Occasional PaperNo 71, International Monetary Fund, Washington DC, July.

Molle, W. T. M., van Mourik, A. (1988), 'Internationalmigration of labour under conditions of economic inte-gration: the case of Western Europe', Journal of CommonMarket Studies 26, pp. 369-394.

Mussa, M. (1986), 'Nominal exchange rate regimes and thebehaviour of real exchange rates: evidence and implications',Carnegie-Rochester Conference Series on Public Policy, 25,pp. 117-224.

OECD (1989a), Economies in transition, OECD, Paris.

OECD (1989b), National accounts, Volume II, 1975-87,OECD, Department of Economics and Statistics, Paris.

OECD (1990), Italy, OECD economic surveys, Paris.

Oliveira-Martins, J., Plihon, D. (1990), 'Transferts internationaux d'epargne et integration fmanciere', Economie etStatistique, forthcoming.

176

sduquenoy


Richardson, P. (1987), 'A review of the simulation propertiesof OECD's Interlink model', OECD Working Paper No 47,OECD, Economics and Statistics Department.

Sachs, J., Sala-i-Martin, X. (1989), 'Federal fiscal policy andoptimum currency areas', mimeo, Harvard University.

Sargent, T. J. (1987), Macroeconomic theory (2nd edition),Academic Press, London.

Schultze, C. L. (1985), 'Real wages, real wage aspirations,and unemployment in Europe', in Lawrence, R. Z., Schultze,C. L. (eds), Barriers to European growth. The BrookingsInstitution, Washington DC, pp. 230-291.

Sneessens, H. R., Dreze, J. H. (1986), 'A discussion ofBelgian unemployment', Economica, 53, pp. S89-S119.

Tootell, G. M. B. (1990), 'Central bank flexibility and thedrawbacks to currency unification', New England EconomicReview, Federal Reserve Bank of Boston, May/June, pp. 3-18.


Van Rompuy, P., Abraham, F., Heremans, D. (1990), 'Econ-omic federalism and the EMU', in European Economy (1990).

Van Rompuy, V., Heylen, E. (1986), Openbare fmancien inde deelgebieden van federate landen (Public finance in theregions of federal countries), Acco, Leuven.

Weber, A. A. (1990), 'Asymmetries and adjustment prob-lems: some empirical evidence', in European Economy(1990).

Wyplosz, C. (1990), 'Monetary union and fiscal policy disci-pline', in European Economy (1990).

Zenezini, M. (1989), 'Wages and unemployment in Italy',Labour 3, No 2, pp. 57-99.

177

sduquenoy


Chapter 7

External dimensionsThe primary economic aim of EMU is to strengthen theintegration of the Community and to improve its economicperformance. However, due to the Community's weight EMUwill also have far-reaching implications for the world economy.

Two types of effects are considered in this chapter. First, EMUwill tend to redistribute responsibilities within the internationalmonetary system in line with the relative weights of the mainregions, increasing the Community's influence. Changes wouldmainly arise in the international role of different currencies(Section 7.1) and in the field of macroeconomic policy coordi-nation (Section 7.2). Secondly, it could foster a more funda-mental change of monetary regime, from a still asymmetricsystem to a multi-polar one (Section 7.3).

Most of the benefits and costs discussed in this chapter willonly arise with the adoption of the ecu as the single Europeancurrency. Throughout this chapter, it wilt be assumed thatStage III of EMU implies a single currency.

The main conclusions are the following:(i) The ecu will emerge as a competitor to the dollar as an

international currency. However, neither dramatic norinstant changes should be expected since for some func-tions (trade invoicing, asset holdings) the role of thedollar has already been significantly diminished, whilefor others (use in exchange markets), technical reasonsfavour a single standard, giving the dollar an initialadvantage.

(it) Expansion of the ecu as a vehicle currency will yieldsome small microeconomic efficiency gains for the ECeconomy, by reducing transaction costs on the exchangemarket for trade with non-EC countries (up to 0,05 %of Community GDP), by reducing exchange rate risksdue to the development of ecu invoicing (which mightincrease by about 10% ofEC trade), and also by givingto European banks enlarged opportunities to work intheir own currency.

(Hi) Regarding the official sec tor, EMU would allow a savingon the exchange reserves of Community Member States,amounting perhaps to USD 200 billion. It would alsobe beneficial to partner countries, especially in EasternEurope, who could choose to peg their currency to theecu. Additional gains would come from revenues ofpublic seigniorage arising from foreign cash holdings ofecus, the importance of which (0,045 % of CommunityGDP annually, corresponding to about USD 35 billion

of foreign holding) is however likely to be much moremodest than sometimes argued.

(iv) As the European currency will become a vehicle fortrade, an increase in the demand for ecu assets can alsobe expected in financial markets. This effect is likelyto be of a relatively small size (about 5% of totalinternational markets) since international portfolios arealready well diversified. This would increase the ex-posure of the European monetary policy to externalshifts in preferences or in the amount of ecu borrowingby non-residents, but this exposure would remain morelimited than it would be for Germany if the DM wereto develop further as an international currency. Whetheror not this would also lead to a temporary appreciationof the ecu cannot be assessed with certainty. However,the exchange rate policy of the Community should beready to react to an exchange rate shock.

(v) EMU will strengthen the Community as an economicpolicy pole within the world economy because adoptionof a common monetary policy under the responsibilityof EuroFed will enhance the Community's identity andweight in international policy cooperation. This will be

felt in macroeconomic policy coordination within the G7.Monetary coordination at this level can be expected tobecome easier, provided the sharing of responsibilitiesfor exchange rate policy between EuroFed and theCouncil ensures an efficient handling of this policy.To some extent, fiscal policy coordination could alsoimprove, although there is a risk that coordination mightbe restricted to monetary policy.

(vi) As the Community becomes a policy pole, spill-overeffects of domestic policies and therefore the need forcoordination at the global level increase. Since the re-duction in the number of policy actors would also makeit easier to reap coordination gains, EMU could act usan incentive to tighter policy coordination at the globallevel.

(vii) EMU could finally be a decisive building block for amore stable multi-polar monetary regime. Monetarycooperation among the G7 countries still falls short ofan adequate monetary system. The establishment of sucha system, that would remove all remaining asymmetriesand provide the public good of monetary and exchangerate stability, would be a major benefit for all countriesparticipating in world trade and finance.

7.1. Recasting of international currencies

Although the present monetary system is characterized bythe predominant role of the US dollar, a trend towards amore symmetric multi-currency regime is already present.

178

sduquenoy

Chapter 7. External dimensions

The emergence of a genuine European currency would accel-erate these changes. Absorbing the already internationalizedEuropean currencies, the ecu would become a major inter-national currency alongside the dollar and the yen. However,both the magnitude and the effects of these changes have tobe qualified:

(i) Assuming that in a multi-polar monetary regime thedistribution of roles among the three major currencies wouldbecome roughly proportional to the economic weight ofeach region, many of the changes to be considered wouldbe incremental since the dominance of the US dollar hasalready been significantly reduced, and since the UnitedStates would in any case remain a major player on the worldeconomic scene. In addition, since the quality of a currencyas a vehicle basically depends on its use by other agents,former dominant currencies tend to retain a large part oftheir role for a significant period of time.'

(ii) Being the issuer of an international currency is frequentlyconsidered as a privilege yielding significant economic ben-efits. However, a large part of these benefits do not hold.The issue of an official reserve currency does not provideadditional room for manoeuvre in a world of open capitalmarkets, and it does not yield any significant direct revenueeither. The gains a country can draw from the issue ofan international currency arise either from borrowing andtrading in its own currency, which means lower transactioncosts and the possibility of shifting part of the burden ofexchange rate risks to the rest of the world, or from seignior-age revenues derived from private holding of its banknotes,which even for the United States are small. These benefits,however, come with a cost due to a relaxation of the controla country has over the use of its currency by third countriesin trade and finance. As the experience of the UnitedKingdom and the United States has shown, this implies agreater exposure of monetary policy and the exchange rateto shocks unrelated to any domestic development. Sincethe exchange rate and interest rates of each internationalcurrency are set on the world market, demand and supplyfor assets denominated in this currency can change indepen-dently of the balance of payments of its issuer, and evenindependently of any macroeconomic developments amongthe issuers of the international currencies.2

In order to assess the extent of changes that could be broughtabout by the emergence of the ecu and their possible econ-omic effects, the different kinds of shifts in the use of inter-national currencies that might result from the completion of

EMU will be considered in turn, assuming that no changesoccur in the international monetary system itself. The effectsto be considered are those corresponding to the standardfunctions of an international currency as unit of account,means of payment and store of value. Table 7.1 presents themajor uses of international currencies with respect to thesecategories.

Table 7.1Functions of international currencies

Private sector Official sector

Unit of account Vehicle for trade in- Definition of parities,voicing and quotation or target zonesof merchandise

Means of payment Vehicle currency in Intervention on ex-foreign exchange mar- change markets, bal-kets ance of payments fi-

nancingStore of value Assets and liabilities Reserve holdings

Source. adapted from Kenen (1983).

While the above presentation draws clear distinctions be-tween different functions of money, there are in practiceclose relationships between, for example, the use of a givencurrency as vehicle in foreign exchange markets and fortrade invoicing. The sections below are therefore based ona slightly different categorization: Section 7.1.1 presents theuse of the ecu as a vehicle for trade and payments in exchangemarkets; Section 7.1.2 discusses the role of the ecu as anofficial currency and Section 7.1.3 international seigniorageeffects; finally, Section 7.1.4 is devoted to the portfolio andexchange rate effects of EMU.

7.1.1. The ecu as a vehicle currency

The functions of vehicle for trade invoicing, quotation ofmerchandise like oil and raw materials, and foreign exchangetransactions are analytically distinct.3 In each case, however,the same issues arise in the choice of a currency as a vehicle:for a currency to be chosen it has to be widely used, competi-tive in terms of transaction and hedging costs, and backedby a financial market which is substantially free of controls

1 This point is underlined by Alogoskoufis and Fortes, (1990).2 For example, an increase in the dollar debt of LDCs increases the supply

of US dollars independently of US macroeconomic policy.

For example, as mentioned by Krugman (1984), until 1974 smallerPersian Gulf nations used to require payment in sterling although theprice of their oil was set in dollars.

179


and of sufficient breadth (i.e. large assortment of instru-ments) and depth (i.e. existence of well-developed secondarymarkets). Also the use of a currency as a unit of accountdetermines to a large extent its use as a means of payment.4

The present situation

In spite of the move to flexible exchange rates, the US dollarremains by far the dominant vehicle currency for tradeinvoicing and, overwhelmingly, for foreign exchange trans-actions.

The first panel of Table 7.2 gives an aggregate picture of therole of different currencies in trade invoicing. As a rule, largeindustrialized exporters invoice their exports in their owncurrency, because this saves the cost of acquiring infor-mation about and techniques of foreign exchange.5 It also

4 See, for example, Tavlas (1990)5 Figures in Table 7.2 should be considered as indicative only since statisti-

cal information regarding the use of different currencies is incomplete.The same applies to the other tables of this section.

saves transaction and hedging costs if the buyer is indifferentto the currency denomination of its imports. Smaller ex-porters, however, generally invoice in a vehicle currency or inthe currency of the importer. Hence, the picture of Table 7.2partly reflects relative weights in world trade. However,about 15 % of the exports of European countries and asmuch as 60 % of Japanese exports are invoiced in dollars.6Therefore, the share of currencies in trade invoicing stilldiffers significantly from trade shares (given in the secondpanel of Table 7.2). Only for Germany is the share of thecurrency in line with the trade share.

For the major industrial countries, the weight of the dollaris greater in imports, due to the fact that trade in oil andother primary commodities is mostly invoiced in this cur-rency as is natural for goods whose prices are set on a worldmarket. This implies also more variability since trade in fuels

Exports of manufactures by the LDCs, which are also mostly invoicedin dollars, do not appear in Table 7.2. Therefore, the share of the USdollar is somewhat underestimated. According to Chevassus (1989), fora set of countries representing some 72 % of world exports, the dollarshare was 55% in 1980 and 44% in 1986; for imports, shares wererespectively 53 % and 49 %.

Table 7.2Trade invoicing currencies of the six major industrialized countries

A. Currency breakdown of foreign trade invoicing1

Dollar

B. Shares in total trade of the six countries

FF. UKL. LIT Total

Exports: 19801987

Imports: 19801987

44,641,7

63,649,5

4,67,5

1,22,9

25,526,7

16,718,3

25,324,1

18,522,9

100100

100100

USA Japan Germany France, UK, Italy Tou!

Exports: 19801987

Imports: 19801987

26,021,6

26,133,2

15,319,9

15,312,3

22,825,3

20,418,7

35,933,2

38,235,8

100100

100100

Sources: A. Calculated from Black (1989), B. Eurostat.1 Merchandise trade calculated on 95,5% and 93,4% respectively of exports in 1980 and 1987, and on 91,5% of imports in 1980 and 1987, the resi being invoiced in other cunencies.

180

sduquenoy


fell from 23 % of world merchandise trade in 1980 to 11 %in 1987. Thus, the oil price fall could explain most of thedecrease in the dollar share for import invoicing in the late1980s.7

Graph 7.1 gives an aggregate picture of the role of the majorcurrencies in foreign exchange markets. The first diagramgives the role of each currency in local trading, i.e. trans-actions involving the home currency on one side. The secondconcerns non-local trading, i.e. for market transactions be-tween third currencies. Both types of transactions amountapproximately to the same volume. The role of the dollar isemphasized in both cases, but overwhelmingly in the second.This predominance can be explained by lower transactioncosts related to the size of the market, and almost indepen-dent from the variability of exchange rates.8 In fact, trans-actions via the dollar between two other currencies are oftenless costly than a direct bilateral exchange. Since it is usedas a vehicle currency in the interbank market for foreignexchange, most interbank transactions involve the dollar onone side.9

7 See Chevassus( 1989).8 See Boyd, Gielens, and Gros (1990).

Possible changes

For an assessment of the consequences of EMU, a distinctionhas to be made between those functions of the vehicle cur-rency which can be shared (between the dollar, the ecu andpossibly the yen) and those which are linked to the use of astandard and can only shift entirely from a currency toanother.(i) Regarding trade invoicing, effects of the first type can beexpected, since the ecu would be the single money of thelargest world exporter and of the biggest monetary andfinancial markets, with monetary policy oriented towardsstability. But these changes would be incremental. Assumingas a benchmark that the share of the ecu in total tradeinvoicing of the six major industrial countries of Table 7.2were to reach the level of the trade share of the Communityas is presently the case for Germany (i.e. that the figures inthe top and bottom panel of Table 7.2 were equalized), some13 % of EC exports and 24 % of imports would shift frominvoicing in US dollars to ecu invoicing.10 These figures,

Sec Black (1985) for a survey, and Alogoskoufis and Fortes (1990).

10 According to Tavlas (1990), 7,4 % of German exports were denominatedin dollars in 1987, but the proportion was 12,3% for France, 16,6%for the UK and 23,8 % for Italy. The share of dollar denominatedexports tends to be higher for smaller countries.

GRAPH 7 . 1 : Currency breakdown of transactions in foreign exchange markets, 1989

Local currency business Non-local currency business

21.7

89.4

dollar DM UKL others UKL, DM. others

Local currency business refers to all exchange transactions involving the home currency Non-local currency business refers lo transactions among third currencies. In both cases, eachtransaclion is oojy counted once For example, the K9.4 •-« share of the doll.ir in the diagram on the nght-hand side means thai 89.4 % of total transactions involve the dollar on one side.Note th.il the share of the DM is underestimated due lo the non-pjrlicipalion of Germany in the survey, so that the German transactions are only captured to the extent thai institution!,in these countnes conducl operations with parties participating in the surveySourer BIS. Survey of foreign exchange markel activity, f-ebruary 1990

181


however, should be taken as an upper bound which wouldprobably not be reached in the medium term due to con-servatism in invoicing practices, and the likely persistence ofdollar invoicing for oil and raw materials which representedabout 15 % of EC imports in 1987. H A reasonable assump-tion is therefore that some 10 % of EC foreign trade (includ-ing intra-Community trade) could shift from dollar to ecuinvoicing. Furthermore, the use of the ecu could also extendto trade between countries linked to the Community, es-pecially those of EFTA and Eastern Europe.(ii) Regarding quotation of oil and other raw materials, onlya shift from one standard to another can be envisaged. Thiswould happen only if the ecu were to replace the dollar asthe predominant world currency. Such a hypothesis wouldbe inherently speculative since the weight of the present ECin the world economy is similar to that of the United Statesof America.

However, bilateral ecu contracts could be arranged for oil and rawmaterial imports from African, Middle Eastern or East Europeancountries.

(iii) Regarding the function of vehicle in the exchange mar-ket, EMU would eliminate intra-European transactions, themajority of which involve the dollar. It would also lead todirect dealing in exchange markets between the ecu and thirdcurrencies, instead of going through the dollar because oflower transaction costs. These changes, combined with theunification of the European money market, which will be-come the largest in the world, could make the ecu a moreserious competitor to the dollar in exchange markets. How-ever, hysteresis is strong in this field.12

Krugman (1980) shows that even if fundamental trade and investmentpatterns do not exhibit a strong dominance of a particular country,economies of scale can lead to making all foreign exchange transactionswith a single vehicle, A schematic example may help to understand theconsequences: suppose currencies A, B and C have equal weight, butday-to-day operations in the interbank market are conducted with A asa vehicle. Then, every transaction involves A on one side. This obviouslyis only valid for the interbank market. But this market represents aboutthree quarters of the total exchange rate market turnover.

Box 7.1: EMU and transaction costs in trade with third countriesor in third currencies

The gains from eliminating transaction costs evaluated inChapter 3 only relate for each country to intra-EC paymentswhich involve a foreign Community currency. Similar gains arisein intra-EC transactions in third currencies (mainly the dollar)and in transactions with the rest of the world.

The gains brought by EMU in that respect can be best illustratedby an example. Consider the case of a Portuguese trader (i.e.exporter or importer) facing a given payment (receipt or expendi-ture). Four different cases arise.

(i) The payment is denominated in escudo; with EMU, as theecu replaces the escudo as the domestic currency, no changeoccurs from the point of view of the trader.

(ii) The payment is denominated in another Community cur-rency, say UKL; then, if it is an intra-Community payment,transaction costs are eliminated as evaluated in Chapter 3;the same is true for a payment with third countries, but thecorresponding volume is very small, so no evaluation iswarranted.

(iii) The payment is denominated in a third currency, say theUSD, and remains so. Then, EMU yields a saving becauseUSD/ecu transaction costs are bound to be significantlylower than USD/escudo costs due to the breadth of the ecumarket.

(iv) The payment is presently denominated in USD, but shiftsto the ecu. This coutd happen because for most partners,except the USA, the choice of the ecu instead of the dollarwould be neutral since both would be major vehicles.

Therefore, the saving would amount to the USD/escudotransaction cost. If the partner is not indifferent to thereplacement of the USD by the ecu, which would be obvi-ously the case for the USA, part of the gain would be offsetby price decreases, possibly reducing the saving to the levelof that of (iii).

The above classification can be used as a basis for a roughevaluation of transaction costs savings.

Assuming dollar invoicing remains unaffected, savings oftype (iii) provide a lower bound of the EMU effects. Accordingto Annex A, total transaction costs for current payments amounton average to about 0,4 % of the flow, with a range from 0,3 %to 0,8%. Since dollar invoicing is much more frequent forsmaller EC countries whose financial markets are less efficient,associated transaction costs can be supposed to be in the upperrange of that bracket, say 0,6%. Since USD/ecu transactioncosts would be minimal, say 0,3 %, the saving would amountto (0,6 % - 0,3 %) times the share of the dollar in EC currenttransactions, which is about one fifth. With trade in goods andnon-factor services amounting to about 26,5 % of EC GDP,this would give savings of the order of 0,03 % of EC GDP.

Assuming alternatively that one half of EC dollar trade, i.e.10% of Community trade, were to shift from dollar to ecuinvoicing, this would eliminate transaction costs altogether forthis fraction of trade. Total savings would then amount to0,05% of EC GDP.'3

Thus, overall savings on transaction costs in trade with thirdcountries or in third currencies should amount to 0,03 % to0,05 % of Community GDP.11 Under the assumption that dollar invoicing presently concerns only trade with third

countries. The figure would be slightly lower, due to double counting, for dollar tradewithin the EC.

182


Associated costs and benefits

What would be the consequences of changes of that nature?Three types of effects could be (i) a reduction in transactioncosts; (ii) a reduction in short-term exchange-rate variabilityand risk; (iii) wider opportunities for Community banks.

(i) The emergence of the ecu would, in addition to eliminat-ing intra-Community transaction costs, reduce the costsarising from transactions between European and non-Euro-pean currencies. This would be a gain of microeconomicefficiency of the same nature as those analysed in Chapter 3.Rough calculations show however this gain would onlyamount to 0,03 % to 0,05 % of Community GDP (Box 7.1).

(ii) As a larger share of EC trade would be denominated inecus instead of dollars, the exposure of the Communityto short-term exchange rate variability would be reduced.Invoicing in ecus by producers or manufactures from thirdcountries (like NICs and more generally LDCs) would elim-inate the effects of dollar exchange rate variability, and theassociated risk. This would be to the benefit of both partnerssince both would gain in stability (although obviously thebilateral exchange-rate risk would remain, and be supportedby the non-European trader). Effects of the same kind wouldarise if the price of oil and other raw materials were denomi-nated in ecus. The elimination of exchange rate risk wouldbe an unqualified benefit for the EC, and also a benefit forthe oil exporters whose trade is more oriented to the ECthan to the dollar zone. In addition, the ecu price of oilwould become more stable because it would not reflect ecu/dollar exchange-rate movements any more. Although in themedium to long term the currency denomination of pricesdoes not affect the real price of oil, this could have lastingeffects due to the short-run rigidity of nominal oil prices.However, such a hypothesis remains speculative.

(iii) European banks will benefit from the enlarged oppor-tunities to work in their own currency (less risk exposuresto hedge, better access to their lender of last resort) and tosee an increase in the transaction balances held in ecus. Thedevelopment of international ecu markets would provide tothese banks a significant opportunity to compete in worldfinancial markets.

Due to the increased role of the ecu in trade invoicing, anadditional effect would be that economic agents engagedin foreign trade would increase their demand for ecus fortransaction purposes. Although it is not known with pre-cision, estimates suggest that the size of transaction balancesmay be about 20 % of all current account transactions.14

On this basis, an increase in the use of the ecu for tradeinvoicing of 10% of the EC's current transactions wouldimply an additional demand of USD 60 billion of ecu-denominated assets (1988 figures), which would add to theportfolio effects discussed in 7.1.4.

7.1.2. EMU and the official monetary sector

EMU will substantially change the official monetary land-scape. For Member States, the elimination of intra-Com-munity exchange rates will reduce the need for reserve hold-ings. To third countries, EMU will offer the opportunity ofan alternative anchor and the ecu will become a majorreserve asset.

The rationale for holding reserves lies in the need to interveneon exchange markets in order to stabilize the exchange rateand to counter runs on the currency. A generally acceptedassumption is that for a given exchange rate system theoptimal amount of reserves depends on the volume of foreignexchange transactions. On this basis, it can be assumed thatEMU would reduce Community reserve holdings by thesame percentage as the share of intra-Community trans-actions in total foreign transactions. This would free aboutUSD 230 billion in reserves15 of a total of about USD 400billion. An alternative hypothesis would be to assume forthe Community the same ratio of total reserves/imports asin other industrial countries. Savings would then amount toUSD 200 billion. One can therefore consider that savingson reserves could amount to approximately 4 % of GDP.

However, these savings should not be taken directly aseconomic benefits since reserves already yield revenues. Theyshould be considered as capital that could be redirected toother uses by the Community or the Member States, whichwould give them more room for manoeuvre. Monetaryauthorities would also have to take care of undesirableexchange rate effects of a reduction of their dollar reserves.Indeed, they would probably have to retain these reservesin the initial period of EMU, since a tendency towards theappreciation of the ecu would already exist.

The above considerations only hold on a ceteris paribusbasis. Further savings could arise from the move towards atri-polar monetary system, as discussed in Section 7.3, sincein such a system foreign exchange reserves could technicallybe replaced by swap agreements among the major monetarypolicy centres.16

This figure is taken from Brender, Gaye and Kessler (1986), which usedata for a sample of LDCs to estimate the size of transaction balances,since indebted LDCs presumably only hold hard currency balances forpayment purposes.

Including USD 85 billion of gold reserves, evaluated at market price.This only concerns the use of reserves as an instrument for exchange ratestabilization, not their disciplinary function in an asymmetric monetarysystem.

183


The ecu would also be used as an official currency by thirdcountries. Since the collapse of the Bretton Woods system,a progressive diversification of currencies used inter-nationally as the numeraire for exchange rate policies canbe observed. Table 7.3 shows this decrease in the unit ofaccount function of the dollar. However this fall is somewhatexaggerated because of the remaining role of the dollar insome of the basket peg practices and flexible arrangements.Nevertheless, a clear trend of diversification appears, exem-plified by the EMS, the creation of baskets adapted to thespecific interests of individual countries, and the DM pegoption chosen by Austria.

Table 7.3Shares of different types of exchange rate regimes in 1975, 1981 and1989

(by number of countries, %)

1981 1989

I. PeggingDollar pegSterling pegFrench franc pegOther currency pegSDRpegOwn basket peg

50,87,8

10,23,1

—10,2

29,50,7

10,12,2

10,816,5

25,5—9,22,65,2

20,3

II. Limited flexibilitySnake/EMSSDK peg with marginsOther more flexible

III. Other including float

(Number of countries)

4,7

13,2

(128)

5,0

2,9

5,22,63,3

22,3 26,1

(139) (153)

Source: 'Exchange arrangements and exchange restrictions' IMF Annual Reports 1975, 1981,1989.

The role of means of payment and store of value of inter-national currencies can be assessed by the distribution offoreign exchange interventions, when available, and by thedistribution of reserve stocks. The trend towards a tri-polarmonetary world is confirmed by the currency breakdown ofofficial reserves. Holdings of reserve assets denominated ina particular currency are related to transaction needs and tothe ease of liquidating assets to make payments, as someempirical studies have shown.17 The three main determi-nants of these needs are the kind of exchange rate arrange-

ments of a country, the share of its trade with a particularreserve-currency country and the currency denomination ofits debt service.18 Since the trade structure has an impacton the two other determinants, the importance of a reservecentre as a trading partner is the basic long-term determinantof the official use of a reserve currency.

Table 7.4 gives the evolution in currency composition offoreign exchange reserves, by group of countries.

For all countries combined, the share of reserves denomi-nated in dollars has significantly declined. This fall has beenmatched by a rise in the shares held as DM and yen. Forthe industrial countries, the fall of the dollar share hasbeen more pronounced since 1976. Paradoxically, countriesparticipating in exchange rate cooperative arrangements notbased on the dollar, like the EMS, tend to hold relativelyhigher shares of dollar reserves, since the dollar has beenfor long the main intervention currency. For developingcountries, the lower share of the dollar, despite the import-ance of their dollar indebtness, is explained by the moreexclusive transaction role of their reserves. Thus, the actualand optimal portfolios are more similar for developingcountries.19

EMU can be expected to accelerate the trend towards diver-sification, since it will offer to third countries the opportunityto link their currency to that of a major economic polewhose monetary policy would be oriented towards stability.

A first, almost mechanical, effect would be the substitutionby official sectors in some non-EC countries of Europeannational currencies for the ecu to peg or to target the externalvalue of their currencies, to intervene on exchange markets,and to hold foreign reserves. This would especially apply tosome EFTA countries like Austria and the Nordic countries,but also to African or Mediterranean developing countries,some of which already link their currency to that of aCommunity Member State.

As the role of the ecu as a vehicle increases, EMU willattract more countries to link with the EC for commercial orfinancial reasons. This trend will be reinforced by the de-

Heller and Knight (1978) explain variations in the proportions of acountry's foreign exchange reserves held as assets denominated in theinternational currencies.

Dooley, Lizondo and Mathieson (1989).See Braga de Macedo, Goldstein and Meerschwan (1984). The differencein behaviour between LDCs and industrial countries is due to thedifferent needs to pursue an active exchange rate policy (increasing withconvertibility, with financial role of a currency, with managed float andthe EMS), and to the financial surplus aspect most pronounced for thereserves of the industrialized countries. The result is for this group anexcess of reserves denominated in dollars in comparison with the optimalportfolio structure.

184

sduquenoy


Table 7.4

Currency composition of official reserves

Currency 1981 1988

A. All countries

US dollarMajor European currencies(of which DM)Yen

B. Industrial countries

US dollarMajor European currencies(of which DM)Yen

C. Developing countriesUS dollarMajor European currencies(of which DM)Yen

78,413,2(5,5)

87,36,7

(2,6)

55,230,3

(13,2)0,2

79,610,4(7,0)0,7

86,94,9

(3,8)0,4

72,715,7

(10,1)1,1

71,517,3

(12,8)4,0

78,714,7

(12,8)3,7

64,119,9

(12,8)4,4

66,020,0

(14,9)7,6

68,420,8(7,5)8,2

62,118,8

(10,8)6,8

63,322,1

(16,2)7,2

67,422,9

(18,3)6,4

53,920,5

(11,9)9,0

Source: 'Exchange arrangements and exchange restrictions', IMF Annual Reports 1975, 1981, 1989.Note: Major European currencies are DM, UKL, FF, HFL. Shares of national currencies in total identified official holdings of foreign exchange, at current exchange rales. Identified holdingsamount to about 85 % of total holdings.

crease in transaction costs due to the development of anintegrated money market in Europe and to the increase inthe transaction needs for third countries to make paymentsin ecus. However, this does not necessarily imply a returnto fixed exchange rates, since countries might choose moreflexible arrangements. It means in any case the use of theecu to target, to intervene and to accumulate reserves.

The importance of the opportunity EMU would provide toEastern European countries must be underlined. The ecuwould be a natural choice as an international vehicularcurrency for countries engaged in a redefinition of theirexternal economic relations in the direction of stronger linkswith the Community. It would also be a natural choice asan anchor for exchange rate policy.20 Without the EMU,an alternative choice would be the DM. However, the cre-ation of a, DM-zone in Eastern Europe might increase thedegree of asymmetry among the currencies of the Com-munity and place an excessive burden on German monetarypolicy.

7.1.3. Seigniorage effects

It is often argued that the current role of the US dollar as themajor vehicle currency carries significant, if not exorbitant,privileges for the United States in the form of involuntaryand interest-free lending by the rest of the world. This relatesto what is usually called international seigniorage (implicitincome received by the authorities from home currencyassets held internationally but which bear no interest, oryield interest income below market rates) and is of the samenature as the domestic seigniorage discussed in Chapter 5.21

International seigniorage revenues may be derived by monet-ary authorities from three types of central bank liabilities:official reserves of foreign central banks, required depositsof commercial banks with the central bank, and cash heldby private non-residents.

Bofinger (1990) discusses the exchange rate policy options of EasternEuropean countries and the possibility of using the ecu as an anchor.

We do not consider private seigniorage revenues such as those derivedby banks from issuing traveller's cheques or demand deposits sincethey correspond to bank intermediation margins in open competitivemarkets.

185


Since official reserves are mainly held in the form of interest-bearing assets like Treasury bills, their holding cannot giverise to direct seigniorage revenues. Holdings of governmentsecurities by foreign central banks indirectly lower theinterest rate on public debt because they reduce the supplyof bonds to the public, but although theoretically correct,this argument is empirically unimportant. For example, the'above-normal' holdings of dollars in foreign central bankreserves (assuming the 'normal share' of the dollar corre-sponded to the weight of the US in the total G7 GNP) wasabout USD 300 billion in 1988. This has to be comparedto a total US federal debt outstanding of about USD2 700 billion or, alternatively, to a total dollar portfolio ofnon-residents of about USD 2 500 billion. Hence, officialholdings do not represent more than 12 % of either aggre-gate. For some years, purchases of US Treasury securitiesby foreign central banks have been much higher, but thisrelates to exchange rate interventions, not to seignioragerevenues.

Since neither Eurodollar deposits nor deposits by non-resi-dents through the New York International Banking Facili-

ties (IBF) are subject to compulsory reserves, the secondcategory of holdings is also empirically irrelevant for the US.Thus, only seigniorage revenues arising from cash holdingsoutside the territory have to be considered. Except for a fewmarginal cases of official use of a foreign currency (e.g. thedollar in Panama and Liberia), the use of cash by non-residents is a phenomenon mostly related to economic andpolitical disorders such as the 'dollarization' of economiesexperiencing very high inflation, political turmoils, and partof illegal transactions in drugs, arms, etc. Anecdotal evidenceindicates wide use of US dollars for hoarding and transactionpurposes in Latin America, Eastern Europe and the MiddleEast.

Although estimates of both present dollar holdings outsidethe US and the potential shift towards ecus are extremelyweak, computations based on reasonable assumptions sug-gest that the potential stock of international cash holdings ofecus should not exceed some USD 35 billion. The associatedpermanent revenue would then be limited to about 0,045 %of Community GDP (Box 7.2).

Box 7.2: Estimating potential shifts in international cash holdingsand associated seigniorage revenues

Most international cash seigniorage accrues today from dollarholdings outside the US, since international holdings of Swissfrancs, Deutschmarks, Pounds sterling and French francs arerelatively minor.

According to surveys conducted by the US Federal Reserve,only some 15 % of the dollar stock outstanding is explained bystandard domestic household behaviour. Research in progressat the US Federal Reserve indicates that holdings by the businesssector, including estimated holdings for illegal and undergroundactivities, do not explain more than 50 % of total dollar holdings,even when statistical problems are taken into account. Thus,holdings of dollars outside the US might represent as much ashalf of the outstanding total. Further evidence is provided bythe fact that per capita cash outstanding of Canadian dollars

(which are not held outside Canada) is only about half ofthat of the US. Hence, a reasonable estimate of the stock ofinternational dollar holdings is 50 % of the total outstanding incirculation, i.e. USD 100 billion in 1988.

Assessment of the share of these USD 100 billion that couldshift to ecus should be based on the geographical breakdown ofpresent dollar holdings, since Latin American holders wouldsurely not react to the emergence of the ecu in the same way asMiddle Eastern or Eastern European holders. Due to the lackof reliable information, crude methods have to be used. Sincethe weight of the EC in OECD GDP is around one-third, thisshare can be used as a benchmark. Thus, maximum potentialshifts in currency use might lead to a once and for all stockadjustment of USD 35 billion (0,6 % of Community GDP) infavour of the ecu. Assuming a 7 % nominal interest rate, theassociated annual seigniorage revenue in the steady state would(using the same definition as in Chapter 5) amount to amaximum of USD 2,5 billion, i.e. 0,045 % of Community GDP.

7.1.4. Portfolio adjustments and exchange rateeffects

It has been argued above that EMU should lead to a shiftof private transaction balances and official reserves fromdollar to ecu holdings. Both effects would mean a portfolioadjustment in favour of the ecu or, equivalently, an upward

drift in the demand for ecu assets since they would ariseindependently of any change in expected returns.

This relates to the more general issue of the private portfolioadjustments that could occur as a consequence of EMU.Due to the segmentation and the relative thinness of Euro-pean and Japanese financial markets, private financial port-

186


folios are biased in favour of the US dollar. In the presentcontext of rapid financial globalization and deregulation,the emergence of the ecu would offer to private asset holdersa better alternative to the dollar than national Europeancurrencies, leading therefore also to an upward drift of thedemand for ecu assets.

Such a drift could theoretically be accommodated eitherthrough price changes, i.e. exchange rate and interest ratechanges, or through a parallel shift in supply. Empirically,price adjustments are likely to occur in the short term dueto the inelasticity of supply, whereas in the long term thesupply of ecu denominated assets can be expected to adjust,mainly through EC balance of payments deficits and devel-opment of ecu borrowing by the non-EC private sector.Whether EMU would have significant exchange rate effectsin the medium term depends basically on the degree ofsubstitutability between assets denominated in different cur-rencies and on the relative speed of those demand and supplydrifts.

A shift in the demand for ecus

Due to both methodological and statistical problems, noprecise assessment can be given of the currency compositionof private financial portfolios. Only rough figures can beobtained, which rest on a number of a priori assumptions.Table 7.5, which presents the result of an attempt to drawan overall picture, is based on assumptions detailed in theappendix to this Chapter. It is important to keep in mindthat it aims at representing the currency composition of

gross financial wealth denominated in foreign currencies,not that of total financial wealth. Data on the currencydenomination of assets are only available for internationalbonds and bank deposits (either at foreign banks or athome banks, but in foreign currencies). For 'other financialsecurities', i.e. stocks and bonds held by non-residents, theaverage currency composition of the three other categorieshas been used.

Two major conclusions emerge from these data. First, takingas a rough benchmark shares in world GNP, the financialrole of the dollar still exceeds the economic weight of theUS, the reverse being true for Europe and especially Japan.For Europe, the remaining gap is significant, but not aslarge as frequently believed if the share of the EC in OECDGDP is taken as a ceiling for the share of the ecu. Thisresult, which is based on currency breakdown data coveringmore than two-thirds of the total portfolio, can be con-sidered relatively robust. The second conclusion is moretentative: although no comparable global picture can begiven for previous years due to a lack of data, a trendtowards diversification is already apparent22 in the data forinternational bonds and deposits at foreign banks: the shareof the dollar has decreased to the benefit of Europeancurrencies and the yen. This result has however to be takenwith caution because of the insufficient coverage of the 1981data. Currency breakdown is only available for about 60%of the total portfolio. In so far as it can be relied upon, this

As explained in the appendix, the figure for 1981 is based on incompletedata. However, partial figures for bonds also suggest a decline of thedollar.

Table 7.5Size and currency composition of world financial wealth, 1988 and 1981

(billion US dollars and %}

World reference portofolioInternational bondsOther financial securitiesDeposits at foreign banksForeign currency deposits ofresidents

Billion USD

408610861063

964

973

L9S8of which in %:

USD

50,343,3n.a.

56,5

65,8

ECU

26,625,1n.a.

28,7

18,7

Yen

7,912,2n.a.3,6

4,9

Others

15,219,4n.a.

11,2

10,6

Billion USD

1652194641572

245

1981of which in % :

USD

n.a.52,6n.a.

71,7

68,9

ECU

n.a.20,2n.a.

16,4

n.a.

Yen

n.a.6,9n.a.1,6

n.a.

Others

n.a.20,3n.a.

10,3

n.a.

Benchmark dataOECD GNP (current prices) 14 084 34,41 33,79 20,20 11,58Ratio: share in portfolio/share inGNP 1,59 0,71 0,34 1,2

7804 34,39 39,30 13,62 12,66

1,94 0,44 0,21 1,01

Source: Evaluation by Commission services on ihe basis of BIS data. Sec appendix to this chapter.

187


result is illuminating because it shows that diversificationhas been under way for several years in spite of an importantsupply of dollar assets due to the US current accountdeficit.23

The creation of EMU would give rise to a specific effect: itwould lead private agents to increase their demand for ecu-denominated assets in order to reweight their transactionbalances and to achieve a more balanced distribution oftheir portfolios between the major currencies of the threemajor economic zones. It is often argued by financial marketparticipants that no real alternative to the dollar exists atpresent since the size, liquidity and depth of European andJapanese markets are still very inferior to those of the USmarket. As this would no longer hold as EMU with a singlecurrency triggers the unification of European markets, amove towards the ecu can be expected. The anti-inflationarystance of EuroFed can only reinforce this effect.

An opposite effect has nevertheless to be taken into accountwhich stems from risk-aversion considerations. The behav-iour of European and non-European residents towards theecu would not be identical. For third-country holders outsidethe US, the emergence of the ecu will make European assetsmore attractive as compared to assets in the US. But for anItalian resident, for example, ecu assets will be at the sametime substitutes for lira, DM and assets denominated inother European currencies. Hence, EMU will mechanicallyreduce this exposure to exchange rate risk. He might there-fore be led to increase his demand for dollar or yen assetsin order to keep the same balance as before between riskand yield. However, as exchange rate variability is alreadylow inside the Community, this diversification effect wouldonly partly offset the pressures towards higher demand forecus.On the basis of the above considerations, room for a specificEMU effect appears to be rather limited. Economic weightconsiderations lead to a shift in a range of 5 to 10 percentagepoints. Taking into account the trend that already exists andthe European diversification effect, a shift of 5 percentagepoints in favour of the ecu appears to be a reasonableconjecture for the EMU effect.

demands to impact on the exchange rate is that investors donot regard assets denominated in different currencies asperfect substitutes. This hypothesis has been tested by alarge number of researchers under the assumption of rationalexpectations, and it has been consistently rejected by thedata.24 Therefore, it is legitimate to assume that a shift inthe demand for ecu assets would impact on the exchangerate. However, there is no consensus on the degree of assetsubstitutability, i.e. on the magnitude of the associated ex-change rate. A number of studies find quantitatively smalleffects, while others suggest larger effects.25 Therefore, thediscussion below will remain qualitative.

In the long run, the required adjustment to the upward driftof the demand curve is bound to be a corresponding increasein the supply of ecu-denominated assets. This would be ofthe nature of a stock adjustment, resulting in a new equilib-rium after a temporarily higher flow of ecu assets. In thisnew equilibrium, agents would hold a larger share of theirportfolio in ecu-denominated assets, but neither the ex-change rate nor the interest rate would be affected.26 Realexchange rates would also be stable in the steady state, andreal interest rates would be equalized across countries, soassets denominated in different currencies would yield thesame rate of return.

The main consequence of this shift would therefore arisefrom a greater use of the ecu by non-European residents,which implies that external shifts in preferences for assets ofdifferent currency denomination or changes in the amountof ecu borrowing by non-Europeans would have a greaterimpact on the market for foreign exchange. In so far asmonetary authorities care about the exchange rate, this alsoimplies a greater exposure of monetary policy to externalinfluences. As mentioned, this is the kind of exposure alreadyexperienced to a high degree first by the United Kingdom,and later by the United States, and it has been until recentlya major reason for German reservations against the inter-nationalization of the DM. However, this exposure wouldbe much smaller for a major zone like the Communitythan for a medium-sized economy. Internationalization is

Macroeconomic consequences

What would be the effects of such a shift? A necessarycondition for ex-ante changes in relative asset supplies or

Barenco (1990), which presents data on the net dollar position of thenon-US private sector, seems to reach an opposite conclusion. However,he focuses on net dollar positions while the present study focuses ongross portfolios. The increase in the gross dollar position of the non-US private sector, as computed from Table 7.5, amounts to USD 1 275billion between 1981 and 1988, which estimated by Barenco is morethan twice the increase of the net dollar position of the non-US privatesector.

See Edison (1990), who surveys the recent literature, and the referencestherein. It should be noted that the assumption of perfect asset substitu-tability has not been tested in isolation, but rather jointly with thehypothesis that current market prices perfectly reflect future prices assuggested by rational expectations.See e.g. Blundell-Wignall and Masson(1985), Brender, Gaye and Kessler(1986), Frankel (1982, 1986).Except for effects resulting from the higher interest payments by theCommunity to the rest of the world or from hysteresis after a temporaryappreciation. These are, however, second-order effects.

188


problematic when it gives to a currency a role out of pro-portion with the economic weight of the issuing country, asfor the UK in the interwar period. In that respect, EMUwould indeed be beneficial for Germany as it would reducethe risks of destabilizing shocks arising from the inter-nationalization of the DM. For an economic zone like theEC in a multi-polar world economy, this risk would be muchreduced as there is no a priori reason to see the ecu takingthe role of a monetary hegemon.27

Turning now to short- and medium-run effects, the basicissues are through which channels the increase in supply ofecu-denominated assets would arise, and whether and towhat extent it could take place without exchange rate andinterest rate changes.

Since changes in reserve holdings by the official sector wouldnot offset the private portfolio effects, but rather add to theportfolio shift, the standard channel through which theincrease in demand for ecus could be matched would be theaccumulation of current account deficits for the Communityas a whole. Per se, this should not be seen as negative sincein this case, the deficit would not signal any weakening ofthe competitive position of the Community, but rather be theconsequence of capital account developments. A temporarycurrent deficit arising, for example, from a lasting increasein productive investment in the context of faster growth as aresponse to EMU, combined with the need for East Germanreconstruction, should therefore not be taken as risky butrather as a welcome development which would limit the needfor exchange rate appreciation. Indeed, some adverse effectscould appear if those deficits were to result from a realappreciation of the ecu, for this could have undesirableeffects on the traded goods sector, whose competitivenesswould be temporarily reduced.

Two important additional channels of supply of ecu assetsare however also worth considering. These are capital ac-count deficits arising for example from a higher flow ofCommunity direct investment abroad, and ecu borrowingby official and non-official agents from the rest of the world.Both would have exactly the same direct effect on the marketfor ecu assets as a current deficit of the Community. Bothcould also be relevant in the context of the 1990s since thecapital needs for reconstructing Eastern European econom-ies as well as of developing countries could be financed eitherthrough Community direct investment or through ecu loansand bonds.

A 5 percentage-points increase in bank liabilities to non-residents de-nominated in home currency would imply for the EC a rise in the ratioof these liabilities to the monetary base from 59 % to 71 %. This wouldstill be substantially lower than present ratios for the US (185%) orJapan (85%).

To the extent that the drift of the demand for ecu assetsremained smooth, it could therefore probably be accommo-dated by parallel changes in supply of one of the abovetypes, without significant exchange rate effects. This couldhappen if agents were willing to adjust progressively thestructure of their portfolios. However, a more sudden shiftcannot be ruled out because the creation of EMU and theestablishment of a credible EuroFed would represent a majorsignal to the markets. In the present context of protractedUS deficits, and given that the safe haven argument in favourof the dollar has weakened in recent years, a strong tendencytowards appreciation of the ecu could appear. Recent monet-ary history indeed provides examples of large exchange rateswings.

From a policy viewpoint, an ex-ante ecu shock cannot be apriori unambiguously qualified as either a cost or a benefit.Monetary authorities would face a modified choice betweenappreciation of the currency and reduction in interest rates.Whether this would be an advantage or a loss depends onthe baseline situation. However, if this shock were to leadto a persistent real exchange rate misalignment it should beconsidered as a cost. Therefore, the practical consequencethat can be derived from this discussion is that since suchshocks could already arise by anticipatory behaviour inprivate markets as soon as the EMU perspective was judgedto be credible, provisions for common exchange rate policyin the transitional stages may be of some importance.

7.2. The Community's part in internationalcooperation

The developments envisaged so far are primarily directconsequences of the emergence of the ecu as a major inter-national currency. This section is devoted to another typeof international effects of EMU, which would arise from thestrengthening of the Community as an economic policy polewith a single monetary policy and a close coordination offiscal policies.

These effects cannot be described as yielding direct economicbenefits of the same nature as those analysed in the previouschapters. Gains could arise, however, from a change in theconditions of international economic cooperation with theUnited States and Japan. This means that the discussion ofthese gains has to be based on a prior assessment of thechanges in the cooperation practices that could be broughtabout by EMU. This is the purpose of subsection 7.2.1below, whereas subsection 7.2.2 discusses possible coordi-nation gains. Throughout this section, it is assumed that nomajor change in the international monetary regime occurs.More fundamental effects of EMU are discussed in Section7.3.

189

sduquenoy


7.2.1. EMU and international cooperation

The G7 is presently the key institution for macroeconomicand especially exchange rate policy coordination, in its tworelated forms of summits of Heads of State and more econ-omic policy-oriented meetings of finance ministers and cen-tral bank governors.28 EMU would have a major impact onthis policy forum since the unification of monetary andexchange rate policies of the EC would de facto reduce thenumber of players in this field from seven to four (US,Japan, EC—whatever its representation—and Canada). Thiscould presumably facilitate exchanges of information amongpolicy-makers and also, when necessary, make coordinatedpolicy changes easier to negotiate and to enforce.

This characterization obviously goes with important qualifi-cations. First, as fiscal policy would not be centralized atCommunity level, the fiscal side of policy coordinationwould spontaneously be less changed than the monetaryside. Secondly, as the responsibility for exchange rate policywould be shared between the Council and EuroFed, theinstitutional setting of exchange rate policy coordinationwithin the G7 would depend on the relative role of the twoCommunity bodies.(i) Fiscal policy would remain the responsibility of nationalgovernments. Hence, one should not expect a representativeof the Community to replace the ministers of finance in G7meetings.29 Cooperation will probably have to deal withfour monetary and seven fiscal authorities. This could re-inforce the already existent tendency to restrict the formalcoordination exercise to the monetary side, with fiscal policylagging behind. Such an outcome would not be desirable,since an excessive emphasis on monetary coordination at theexpense of fiscal coordination might be suboptimal, if notcounterprod ucti ve.

However, as discussed in Chapter 5, EMU would also fostergreater intra-EC coordination of fiscal policy with a view toexternal variables like the current account and the exchangerate, since a large part of the spill-overs across MemberStates would precisely be felt through these variables. Onecan therefore expect fiscal coordination procedures withinthe EMU to be tailored in order to ensure an adequateaggregate policy mix of the EC in the world economy.30 In

this respect, EMU could well foster fiscal policy coordinationwithin the G7.(ii) Regarding the exchange rate policy, the responsibilityfor it would belong to the Community, but would be sharedas in all major countries between the institution in chargeof general economic policy, which would be responsible forthe definition of the policy, and the central bank, whichwould be in charge of its management. The frontier betweenthese two fields, however, is not clear-cut in the presentcontext of managed floating among the major currencies,and, indeed, differs from one country to another.31

For international policy coordination at global level to be-come more efficient, it is of paramount importance that thedefinition of responsibilities ensures an efficient handling ofthis policy. The two major requirements for the Communityin that respect are to be able to speak with one voice inexchange rate policy discussion at G7 level, and to ensureconsistency between its exchange rate and monetary policyobjectives. In what follows, it is assumed that both con-ditions are fulfilled.

Bearing those qualifications in mind, EMU could havestrong effects on the practice of policy cooperation. Obvi-ously, some coordination of EC positions within the G7already exists. However, according to political scientistsCommunity members are far from speaking with one singlevoice in G7 meetings.32

7.2.2. Global coordination gains from EMU

Economic policy coordination has been for several years atopic of extensive research, at both the theoretical and theempirical levels. The purpose of this section is to draw onthis literature (actually, on the most basic models) to examinehow EMU would affect the gains from coordination andtheir distribution. This issue should neither be neglectednor overestimated: empirical evaluations tend to show thatpotential welfare gains from coordination for the G7countries as a whole should be in a range of 0,5 % to 1,5 %of their GDP.33

A more detailed discussion of the impact of EM U on the major economicpolicy forums can be found in Alogoskoufis and Fortes (1990).Representatives of the Council and the Commission might howeverparticipate in the meetings.The new convergence decision already includes a provision for fiscalcoordination in the presence of external shocks. See Council of Ministers(1990).

At one extreme, exchange rate policy definition is limited to the choiceof a legal regime (i.e. fix versus floating). At the other end of thespectrum, the management of interventions by the central bank is doneunder specific instructions of the Treasury (although the central bankis free to decide whether or not to sterilize the intervention). Germanyand the United States are approximately at the two ends of that spec-trum.See Putnam and Baynes (1987), Funabashi (1988).Gains measured in terms of GDP by using a macroeconomic welfarefunction, with respect to a full information non-cooperative case. See,for a recent survey, Currie, Holtham and Hughes Hallett (1989).

190


EMU can be expected to bring three types of effects:(i) As the spill-over effects of Community policies on the

rest of the world would be larger than those of individ-ual Member States, the need for, and the benefits ofpolicy coordination would increase.

(ii) As the number of actors would be reduced, some of theusual obstacles to coordination would be alleviated.

(iii) A greater bargaining power for the EC, which wouldhave a single voice, might also affect the distribution ofgains between the Community and the rest of the world.

The economic consequences of these changes will be exam-ined in turn.(i) The basic rationale for explicit policy coordination isthat coordination through the market is inefficient when onecountry's policy decisions significantly affect its neighbours.This is because by making economic policy decisions separ-ately, governments do not take into account the welfare oftheir economic partners, and can therefore make suboptimalchoices while reacting to shocks. Coordination allowsgovernments to take full account of these externalities andto maximize the collective welfare of the participatingcountries.34 Coordination gains therefore increase with thedegree of cross-country spill-overs: they can be weighty forlarge interdependent economies.

As long as Community governments and central banks settheir policy in a non-coordinated way, Europe appears as acollection of medium-sized policy centres facing two majorpoles, the US and Japan. Spill-over effects of individualEuropean policy decisions on non-European countries aresmall, while those of the US on Europe are several timeslarger. This asymmetry, which arises from relative sizes anddegrees of openness, implies that ceteris paribus the UnitedStates has presently less to gain than Europe in transatlanticcoordination. Since coordination always involves risks, be-cause of imperfect information, and costs, at least thosewhich arise from domestic political considerations, the incen-tive for the US to engage in such an exercise is weak. Tosome extent, the United States can exploit this asymmetryby making its policy choices in a non-coordinated fashionwithout suffering much from a similar behaviour of Euro-pean nations.35

In technical terms, the first outcome is labelled non-cooperative equilib-rium and the second one cooperative equilibrium. It can be shown thatcoordination leads to a Pareto-superior outcome, i.e. that it can increasethe welfare of all partners.Size and degree of openness are obviously not the only factors affectingcoordination gains. Differences in size and behaviour, e.g. wage andprice behaviour, also impact on the magnitude and the distribution ofcoordination gains. See Oudiz and Sachs (1984), Hughes Hallett (1986)and, for a recent survey, Currie, Holtham and Hughes Hallett (1989).

The effect of EMU would be to aggregate 12 economies intoa single major block whose degree of interdependence withthe US, Japan and the rest of the world would be meaningful.Therefore, a lack of coordination among the three majorpolicy centres would cause more welfare losses than at pre-sent, and the benefits from global or at least transatlanticcoordination would be increased accordingly. This wouldespecially affect the Community's partner countries since ascoordination improves within the Community, some gainscould be reaped for the Member States whilst the cost of alack of global coordination would be increased for the USand Japan. Roughly speaking, as the sizes and degrees ofopenness of the EC and the US would become close to eachother, the welfare costs of uncoordinated policies wouldbecome less unevenly distributed. This is illustrated inGraph 7.2 in Box 7.3 below.36

(ii) The above considerations relate to the gains from fullcoordination. However, it has been frequently argued thatpractical obstacles make full coordination very difficult toachieve. In practice, coordination is often restricted to infor-mation exchange and partial bargains over specific targets.Another, different impact of EMU would be to change theconditions of information exchange and partial coordi-nation.

A first result, which directly derives from the previous con-siderations, is that as coordination gains would be largerand more evenly distributed, all partners could become morewilling to cooperate in order to reap these benefits. Increasedspill-overs might in particular make the United States moreconcious of the limits of independent policy-making, andthis would be a benefit to its partners. This illustrates themore general point that EMU could not only result inabsolute gains from coordination, but also in relative gainsarising from a better distribution of policy changes amongcountries.

Moreover, according to recent research, information prob-lems (regarding either the economic situation, the economicmechanisms or the policy preferences of the partnercountries) are a key obstacle to coordination.37 Disagree-

This only holds in so far as the economies of the Member States are notcharacterized by strong asymmetries. Were the national economies ofthe Community very different in structure and behaviour, one couldimagine that their aggregation in a single EMU would have mixedeffects on the intensity of spill-overs on the US because opposite effectsof national policies would offset each other. Here, we suppose that theeconomies of the Community are to a large extent symmetric (seeChapter 6 for a discussion), and we focus on the pure aggregation effect,leaving aside the issue of differences in behaviour between the US andEurope as a group.See, for example, Feldstein (1988), Frankel and Rockett (1988).

191

sduquenoy


ments often arise more from differences in the perception ofreality than from genuine distributional issues. Indeed, re-cent research has come to the conclusion that significantgains can be reaped through information exchange withoutentering a costly formal bargaining over the whole range ofpolicy instruments. This is called partial coordination andis in fact to some degree permanent among policy-makers.

The reduction in the number of actors could make an import-ant difference in that field as information problems growrapidly with the number of players. In particular, EMUcould force European countries to reveal their collectivepreferences more clearly. Hence, partial coordinationthrough information exchange would allow to secure a largerpart of the potential benefits, and monitoring of the com-pliance of more ambitious agreements would be facilitated.Bargaining over specific policy targets would also become

easier and the risks arising from free-rider behaviour or fromunsustainability of the agreement would be less critical.

This would obviously not remove the most frequent objec-tions to coordination but probably make it more feasible.Although no precise estimation of the associated gains canbe confidently made, indications arising from coordinationmodels suggest that this could be a significant improvement(see Box 7.3).38 This gain would not especially accrue to theCommunity, but to all nations taking part in coordination.

One could object that coordination is already highly centralized withthe G7 being the key institution and in some instances the US, Japanand Germany acting as a de facto G3. However, this structure alsocarries costs since participating authorities cannot confidently act onbehalf of non-participating ones, and because information problemsarise regarding the policies of the smaller countries.

Box 7.3: EMU and macroeconomic policy coordination

The purpose of this box is to illustrate how the aggregation ofCommunity countries within a single EMU would affect the sizeand distribution of coordination gains. For the sake of sim-plicity, it is assumed that prior to EMU, there is no intra-ECcoordination whatsoever (but possibly global coordination, inwhich all EC countries take part), and that after EMU there isonly one macroeconomic policy for the Community as a whole,with a single Community authority taking part in global coordi-nation. This policy is supposed to be optimal in the sensethat no EC country suffers welfare losses arising from intra-Community coordination failures. As developed in the maintext, these are obviously simplifying assumptions whose onlypurpose is to help to clarify the mechanisms at work.

Only two 'countries' are therefore considered, which are rep-resented in Graph 7.2: the Community and the US, which areassumed to be similar in size, structure, degree of openness andpolicy preferences. The criteria for evaluating coordination gainsare the welfare losses of both partners, which measure thedeviation from target of a few macroeconomic variables.

/. Full versus partial coordination

These gains can be discussed by using the following benchmarkcases:39

(i) Degree zero (not depicted on Graph 7.2): small countrycase. Each government acts independently taking all itsexternal environment as given.

(ii) Degree one (D1 on Graph 7.2): non-cooperative isolationistpolicies. Each government knows the structure of its neigh-bours' economies, but not the preferences of the othergovernments. It can take cross-country spill-overs into ac-count, but not the policy reactions of its partners.

(iii) Degree two (D2): optimal non-cooperative policies withfull information exchange. By exchanging information,governments can know the reactions of their partners totheir own policies; however, they do not engage in bargain-ing as each country maximizes its welfare separately, takingas given the actions of the other countries.

(iv) Degree three (D3): full coordination. Governments jointlyset their instruments in order to maximize welfare. Thelocus of possible outcomes is represented by the FF curve.The choice of an optimal point on this curve depends onthe weights attached to the respective welfares of Europeand the US, which in turn depend on their respectivebargaining powers.

Like that of Oudiz and Sachs (1984), most empirical studiesfocus on a full coordination involving a bargain over the wholerange of target variables (real GDP, inflation, the current ac-count, etc.), i.e. on the FF schedule. For this purpose, nationalwelfare losses (with respect to the objective function) areweighted with coefficients reflecting the weight of each countryin the bargaining. No universally accepted procedure exists toestimate these weights, which are often chosen by judgment.

However, evaluations with several models tend to show thatlarger gains arise while moving from degree zero to degreeone and from degree one to degree two. In comparison, gainsprovided by the move from degree two to degree three aremuch smaller.40 This is because important gains can be achieved

This classification is adapted from Brandsma and Pijpers (1985).See Brandsma and Pijpers (1985), Hughes Hallell (1986), Brandsma and HughesHal1ett<1989).

192

sduquenoy


through better information. Full coordination remains a theor-etical case, which is only approximated in rare circumstances(the Bonn Summit of 1978 being the classic example). Coordi-nation in the real world is a mix of incomplete informationexchange and partial bargaining, which can be represented by apoint like A between Dl and D2.

2. The EMU effect

Prior to EMU, the small size of spill-over effects of individualEuropean policies leads the welfare cost of uncoordinated poli-cies to be larger for Europe than for the US. This is representedby Dl. As EMU brings more symmetry, the non-cooperativesolution moves to Dl (EMU), a gain for the EC and a loss forthe US. The same holds for the Nash full-information case D2,but associated gains or losses are smaller.'"

As to full coordination, the same outcome can be achieved withor without EMU,42 but the distribution of gains depends on thebargaining weights of the players. Here it is assumed that theweight of the EC also increases with EMU (illustrated by themove from D3 to D3 (EMU)). However this appears to be arelatively minor issue: since large gains can be achieved throughinformation exchange and without formal bargaining, it immedi-ately follows that the distribution of weights is not crucial. Therange of possible outcomes is also bounded because no countrywould agree to engage in coordination without the perspectiveof a gain as compared to the full-information case D2.

11 This illustrates a welt-known result of game theory: that a coalition among a subgroupof players is welfare-reducing for the other players. Sec Van der Ploeg (1990).

12 This neglects the fact that EMU would reduce the number of available policy instru-ments.

Important gains can reasonably be supposed to be achievedthrough information exchange. As the number of actors isreduced, the outcome from real coordination comes closer tothe full information case, as illustrated by A (EMU). As the USbecomes more willing to engage in coordination, some furthereffects can also arise. What the figure illustrates is that dependingon the relative magnitude of spill-over effects and informationproblems, the global effect of EMU, which is unambiguously again for Europe, can be either harmful or beneficial to the US.

GRAPH 7.2: EMU and global coordination: an illustration

7 -

6 H

5 -1

4 H

3 H

2\i

I J

0 *

na8.e

UJ

1 3 5

US welfare losses

A caveat should however be added: partial'coordinationfrequently tends to focus on specific targets or instruments;in some instances this can be very suboptimal. A furtherfocusing of international coordination on exchange rate andmonetary matters would not depart from recent trends, butcould well be counterproductive if fiscal policies were leftentirely uncoordinated. For example, exchange rate target-ing with monetary policy being used as the correspondinginstrument can become counterproductive if exchange ratetargets are inconsistent with fiscal policies.43 Although thisis clearly a risk, EMU could facilitate fiscal coordination ifintra-EC procedures are adequately tailored.

This point has been made by Feldstein (1988).

(iii) A third, purely distributional effect of EMU on policycoordination has to be discussed: would a more unitedexpression of the Community change the distribution of thecoordination gains between the US, Japan and Europe?

The issue might be of a less importance than appears atfirst sight. First, the distribution of coordination gains isinfluenced both by economic determinants like size inter-national linkages, and behaviour, and by the bargainingpower of the players. Formal models indicate that it is morelinked to the former than to the latter (see Box 7.3). As togenuine bargaining issues, economic theory has little to sayon the factors determining the weighting of each country'swelfare in cooperative bargaining. On the one hand, one

193


could argue that a united EC would represent all MemberStates (instead of four) and could discuss on an equal footingwith the USA and Japan, and hence that its weight wouldbe increased. On the other hand, four EC countries partici-pate in G7 meetings, and although there is no formal voting,scholars of coordination indicate this has given them someweight in discussions.44 There is no strong evidence thatEurope's interests are presently underweighted in coordi-nation when European governments agree on the policyprescription.

Summing up, the most important effects of EMU would beto increase the need for global coordination, especially fromthe point of view of partner countries, and to make thecorresponding gains easier to achieve. As both would bebeneficial to the Community, this could be a significant gain,but the net effect for the rest of the world could be alsopositive.

7.3. Towards a better international monetaryregime

A common feature of the changes discussed in the previoussections is that EMU would bring more symmetry in inter-national economic and monetary relations. This holds bothfor the international role of currencies and for policy coordi-nation. Such an evolution would not be a consequence ofEMU per se, but EMU would amplify a trend arising fromchanges in the world economy towards a more balanceddistribution of wealth and power. As these changes takeplace, genuine systemic effects can be expected to appear, asopposed to the mere incremental effects within a given sys-tem which have been analysed so far. The present section isdevoted to these systemic issues, i.e. to the internationalregime that could emerge at the beginning of the next cen-tury.

Although most of this chapter has to rely on a priori assump-tions, economic analysis can provide assessments of thepossible impacts of EMU. This is not true for systemiceffects since (i) there is no experience of genuine multi-polarmonetary regimes, (ii) opportunities for the creation of amore balanced and more stable world economic system willappear, but there are also risks of instability; and (iii) thefinal outcome will depend on the ability of the major econ-omic powers to profit from these opportunities. Hence thediscussion in this field necessarily has a somewhat speculativecharacter.

Therefore, only general issues will be debated here, since itwould not be relevant for the present study to enter specificdiscussions regarding the design of the future internationalregime. For similar reasons, the discussion will also remainentirely qualitative. Section 7.3.1 briefly presents the limitsof the present monetary regime, while Section 7.3.2 discusseswhether a more symmetric regime is indeed desirable.

7.3.1. The limits of the present regime

The international monetary regime is presently of a hybridnature. Since 1973, both floating and various degrees ofmanaged floating have been experienced. The experiencewith floating rates is widely considered disappointing.45 Asalready discussed in Chapter 2, floating rates have led tohigh exchange rate variability and, what is more, long-lasting real misalignments. Consequences have been both adegradation of the quality of price signals, and disruptivereal effects on manufacturing and other traded goods sectors,with the associated welfare losses. By aggravating tradefrictions, exchange rate misalignments have in some in-stances posed threats to the maintenance of an open multilat-eral trading system. The record of floating rates has notbeen better regarding policy discipline since pressure fromthe exchange markets to make policy adjustments has fre-quently been insufficient, delayed or even misleading. Fi-nally, permanent asymmetries regarding the burden of ad-justment have persisted in spite of the formally symmetriccharacter of the regime because of the special internationalsignificance of the dollar exchange rate and of the US interestrate.

Since the Plaza-Louvre agreements of 1985-87, the regimehas de facto evolved in the direction of a target zone system:central bank intervention on the exchange markets, targetrates set by G7 ministers and (limited) domestic policy ad-justments in the context of multilateral surveillance havebeen used with a view to influencing the evolution of ex-change rates. However, cooperation remains extremely prag-matic and short of the more precisely defined systems pro-posed by various economists or officials. This ad hoc charac-ter has a number of disadvantages, since the permanence ofthe system rests on the persistence of a political consensusinstead of being rooted in commonly agreed rules. As longas monetary cooperation is not based upon firmer grounds,it is bound to lack the robustness that is required to ensurecredibility. The lack of rules also means lack of clarity asregards the responsibility for overall monetary stability(N-l problem) and the distribution of adjustment across

44 See Putman and Bayne (1987), and Funabashi (1988).

194

45 See. for example, Williamson (1985), Krugman (1989a).


countries. Finally, excessive reliance on announcements ofexchange rate targets and on monetary instruments (inter-ventions, frequently sterilized, and domestic policy changes)does not do enough to foster fiscal policy changes.

Nevertheless, the cooperative experience of the last five yearsprovides a basis upon which a more structured regime mightbe designed. A number of proposals have been made whosetechnical features are beyond the scope of the presentstudy.46 However, a distinctive effect of EMU would be togive a decisive boost to the search for a genuine multi-polarregime. By leading to the emergence of a unified Europeanmonetary pole, EMU would be an important building blockfor a tri-polar structure in which Japan, the USA and Europewould have equal weight and responsibility.

However, these considerations might be challenged on twogrounds: first one can argue that a multi-polar monetarysystem is prone to instability; second and more generally,regional arrangements can be challenged because it is notclear whether they increase the aggregate welfare of theworld.

7.3.2. Is a multi-polar system desirable?

It has been argued for long by some scholars that for aregime to be stable it has to be hegemonic.47 Examinationof the arguments for the hegemonic stability thesis leads todistinguish between three different aspects of a monetarysystem: genesis, operation and dynamic evolution.48

(i) Regarding genesis, history gives some support to thehegemony thesis since a single country has frequentlyplayed a leading role in the establishment of a newregime. However, this is of limited relevance for thepresent discussion, for no country is in the situationof taking over the role of hegemon. Moreover, theCommunity experience shows that hegemony is not aprecondition for the design of a new system.

(ii) Regarding operation, the very purpose of a regime isto set rules which apply to all participating countriesand which can be maintained without hegemony. Themost forceful case for a hegemonic regime, made by

See e.g. Lebegue (1985), WiUiamson and Miller (1987), McKinnon(1988).The theory of hegemonic stability is rooted in the analysis by Kindel-berger (1973) of the modern monetary systems. It has been developedby Keohane (1984) and other political scientists. For a recent discussionand empirical examination, see Eichengreen (1989).The following paragraphs draw on Eichengreen (1989) and Keohane(1984).

Kindelberger (1973) focuses on the lender-of-last-resortfunction in crisis situations. However, Eichengreen(1989) points out that for this function to be effective,the market power of the hegemon has to exceed by aconsiderable margin that of the other countries. Sinceeconomic weight is now more evenly distributed, thecase for hegemony appears to be weakened. This doesnot mean that the monetary regime could not exhibit ade facto asymmetry in the conduct of monetary policy,as is presently the case within the EMS. Indeed, theleadership issue arises in any monetary regime, becausethe overall stance of monetary policy has to be set by apolicy centre. However, this kind of de facto asymmetrywithin a formally symmetric regime, which does notdetermine a priori which country should be the anchorof the system, is very different from the structural asym-metry of for example the Bretton Woods system whoserules gave the leadership to a particular country inde-pendently of the quality of its policy. Moreover, it canbe considered desirable that the operation of the systemrewards performance by linking effective leadership toreputation.

(iii) As far as dynamic evolution is concerned, the built-inproblem of a hegemonic system is precisely that it can-not evolve over time with economic fundamentals andcan even accelerate the relative decline of the hegemon.Although the presence of a hegemon may be beneficialas long as it remains the anchor of the system, it is nolonger so when it ceases to provide stability.

Concerns for stability have also been raised in a more narrowsense:49 it has been argued that asymmetry is indeed desir-able because if one economy is much larger and more closedthan the others, it may act as a leader, while smaller countriesdevote their monetary policy to exchange rate targeting vis-a-vis the large country, and that this kind of arrangementmight be more stable than a symmetric system. However,the present situation is already very far from this kind ofasymmetry. Moreover, the advantages of such an asymmet-ric system can also be challenged on systemic grounds. First,a multi-polar system does not have to rely permanently onthe policy of a single, specific country to provide the nominalanchor of the system. This should be considered a benefitsince, if the international monetary responsibilities of a coun-try are out of proportion with its economic capacity, aconflict of objectives is bound to arise. Second, it does notraise the kind of free-rider issues which are frequent instructurally asymmetric systems, where countries face thetemptation to benefit from the public goods (e.g. stability,

This paragraph follows the discussion of Giavazzi and Giovannini (1989)by Alogoskoufis and Fortes (1990).

195


liquidity) provided by the anchor country without participat-ing in the required disciplines.

The hegemonic stability argument does not therefore appearvery convincing in the present context. The optimality of asystem consisting of three major monetary zones may how-ever also be challenged on welfare grounds. Indeed, it is notself-evident whether this particular arrangement is superiorto for example a genuine multilateral arrangement in whichEuropean countries would participate individually, or a sys-tem with a single world currency as proposed by somescholars.50 While this issue may appear highly theoretical,it deserves a short discussion because it is sometimes arguedthat among all possible systems the worst is that whichconsists of a few blocks.

Although precise results only apply to specific policy fields(e.g. trade) and are model-dependent, the more generalreasoning behind this argument is that a trade-off is involvedin the choice of regional arrangements within the worldeconomy: on the one hand, such arrangements could bewelfare-improving for their members but on the other hand,they could be welfare-reducing for the rest of the world. Thisreasoning can be applied to the creation of a free-tradearea,51 but also to monetary arrangements. In both cases,it can be argued that the first-best solution would be a singlezone, i.e. free trade or one money, and the second best anatomistic world. Therefore, an arrangement consisting of afew large blocks might be undesirable.

However, such arguments are based upon two conditionswhich do not hold. First, the world has to be homogeneous,i.e. there should be no natural trading or currency zone,which in practice is a highly questionable hypothesis. Second,the suboptimality of regional trade or exchange rate arrange-ments always rests on the assumption that the zone pursuesnon-cooperative policies vis-a-vis the rest of the world,through, for example, beggar-thy-neighbour tariffs or mon-etary policies. Once this assumption is relaxed, the subopti-mality of a tri-polar arrangement does not arise anymore.52

Indeed, one of the major purposes of a tri-polar systemwould be to foster cooperative behaviour among the threeblocks.

The costs and benefits of a tri-polar monetary system haveto be assessed with respect to realistic alternatives. A firstalternative would be returning to the floating regime, with

See, for example, Cooper (1984), The McKinnon proposal also comesnear to such an arrangement.This point is explicitly made by Krugman (1989b).This can be linked to the above discussion on coordination gains.

the associated efficiency and stability losses. A second couldbe the maintenance of the present ad hoc arrangements,whose robustness remains subject to the persistence of apolitical consensus. A third could be a US-Japanese duopoly,which is not a very desirable perspective from a Europeanpoint of view. A fourth would be a tri-polar system with theDeutschmark being the third pole. As argued above, thiscould be harmful for Germany itself, because it would carryajnonetary role out of proportion with its economic weight.

A tri-polar monetary regime offers therefore better perspec-tives. In so far as EMU would be a building block for sucha regime, this could be a major benefit accruing not only tothe Community, but to all countries participating in worldtrade and finance.

Appendix to Chapter 7: Estimating a referenceportfolio for the evaluation of potential shifts inthe private demand for ecu assets

Potential shifts in the demand for assets denominated indifferent currencies may concern a wide range of assets:mainly bonds denominated in home or foreign currency,shares, bank deposits. A first step is therefore to establishthe size and the structure of the reference portfolio. Thisraises both methodological and statistical questions.

Since one is interested in assessing potential shifts in thecurrency composition of private wealth, the aim is to evalu-ate a gross world portfolio. This distinguishes the presentapproach from that of for example Barenco (1990), whoseaim is to assess the net dollar position of the non-US privatesector.

Theoretically any type of financial (and to a certain extentreal) wealth of private agents may be subject to shifts inpreferences for assets denominated in other currencies. Inpractice, simplifying hypotheses have to be made regardingboth the holders of the portfolio and the nature of the assets.Table 7.5 is based on the following;

(i) Separability hypothesis: due to aversion towards riskand preferred habitat behaviour, the choice by privateagents among assets denominated in different currenciescan be split between two separate choices between, first,home and foreign currencies, and second, different for-eign currencies. Thus, the reference portfolio is restrictedfor each country of residence to assets denominated inforeign currencies.

1%


(ii) Liquidity hypothesis: only financial wealth is con-sidered, as real wealth and notably direct investment ismore subject to other determinants like real exchange-rate level, access to market, industrial strategy, etc.Hence, main categories in the reference portfolio areinternational bonds, national bonds and shares (otherfinancial securities) held by non-residents, deposits atforeign banks (net of some interbank positions), andforeign currency deposits of residents.

Both hypotheses may be questioned: diversifiable wealthobviously encompasses both assets denominated in the homecurrency and part of non-financial wealth. In addition, therecan be no specific supply for this kind of portfolio sincesecurities are not primarily issued for non-residents only.However, this portfolio is valid as long as it is used fordescriptive purposes. Alternative hypotheses are even morequestionable.53

Table 7.5 presents an estimated world reference financialportfolio compounded of assets denominated in foreign cur-rencies. As detailed below, the construction of this portfoliofor 1988 is mainly based on data published by the BIS.

Aggregate portfolio

1. The point of departure is the recently published BISestimate for the world external liabilities (USD 8 505 billionat the end of 1988), estimated through the addition of totalprivate and public foreign liabilities of each country.54

2. Total deposits from residents denominated in foreigncurrencies as currently published by the BIS55 have to beadded (USD 798 billion). This gives the world internationalportfolio (USD 9 303 billion) which is a gross total includingdouble counting due to some interbank positions. Therefore,interbank assets corresponding to such double counting(USD 3115 billion) are eliminated, following the method ofthe BIS to estimate a total net of the international bankingmarket.56

Frequently used alternatives are gross or net financial wealth of privateagents. Gross wealth is obviously biased by dominant monetary habitatbehaviour. Net wealth is a too restrictive definition of the diversifiableportfolio.Appendix Table 2, p. 36 in M. Delatry and J. Van 't dack (1989).International Banking and Financial Developments Quarterly, Table 3 d.Guide to the B/S statistics on international banking. Basle, February1988, pp, 54-56. International Banking and Financial Market Develop-ments, May 1989, p. 1.

3. The next adjustment is to subtract from this portfolio theworldwide stock of direct investment (USD 1 077 billion atyear-end 1987,57 liability side, plus an estimated USD 150billion corresponding to 1988 balance of payment flows58 =USD 1 227 billion at year-end 1988).

4. Finally, world official reserves (USD 875 billion59) aresubtracted since their purpose is quite different from privateportfolios. This gives an estimate of the size of the worldinternational financial portfolio (USD 4 086 billion at year-end 1988).

Decomposition by category of asset

5. This global amount can be split into four main categories,using mainly BIS statistics for three of them, the last one(other financial securities) being estimated by difference.However, only international bonds issued and correctedfor redemptions and repurchases are known with precision(USD 1 086 billion).60 Corrections for the other categoriesare as follows.

6. Deposits at foreign banks are obtained by addition ofbank liabilities to the non-bank sector (USD 817 billion61)and liabilities resulting from placement of trustee fundschannelled via banks in Switzerland (USD 146,5 billion62)since the placements were deducted as interbank assets bythe reporting banks. This adjustment leads to an amount ofUSD 963,5 billion for the net deposits at foreign banks.

7. For the deposits of residents in foreign currencies, theBIS statistics are less exhaustive than for cross-border busi-ness; in particular, domestic interbank assets in foreign cur-rency between different branches of the same banks areexcluded. Identified non-bank sector deposits of residents63

US industrial outlook 1990, Department of Commerce.Balance of payments statistics. Yearbook 1989, IMF.Delatry and Van 't dack, (1989).Table 7, International Banking and Financial Market Developments, May1989.Table 3b International Banking and Financial Market Developments, May1989, revised series.Swiss National Bank.Tables 3d, 5b and 7 of International Banking and Financial MarketDevelopments. The total recorded deposits amount to USD 137.5 billion.The USD 190 billion figure includes an estimate of foreign currencydeposits of US and Japanese residents, which are not recorded in BISdata.

197


only amount to USD 190 billion, while total loans to resi-dents made in foreign currencies amount to USD 1 020billion. This means that the banks themselves provide foreigncurrency loans with money taken from national markets.From these liabilities, following this methodology, only USD47 billion (USD 3 115 - USD 3 068 billion) are consideredas double counting in bank activities with residents. Al-though there is some risk that part of these liabilities couldhave a counterpart in other items already registered,Table 7.5 considers that these liabilities behave as non-bankdeposits from the point of view of currency managementof foreign banks. So, this item reaches USD 973 billion(USD 1 020 - USD 47).

8. Other financial securities, estimated by difference, am-ounted to USD 1 063 billion at the end of 1988.

Decomposition by currency

9. The currency breakdown is available for bonds. Fordeposits, BIS data exist but have to be taken with caution

since the coverage is incomplete: the breakdown for depositsof non-residents in domestic currencies is not published forthe non-dollar currencies, and the breakdown for the de-posits of banks in off-shore centres does not exist. Thispartial subset of data has to be extrapolated for the other(unidentified) financial assets, for which no statistics exist.

Table 7.6Construction of the world reference portfolio

(billion VS dollars)1988 1981

World external liabilities (= assets) 8 505 3 745+ Deposits of residents in foreign currency 798 312= World international portfolio 9 303 4 057— Interbank assets 3 115 1 190— Direct investment 1 227 700— Official reserves 875 515= World reference portfolio 4 086 1 652

198

sduquenoy


References

Alogoskoufis, G., Fortes, R. (1990), 'International costs andbenefits from EMU', in European Economy (1990).

Bank for International Settlements (1990), 'Survey of foreignexchange market activity', Basle, February.

Barenco, B. (1990), 'The dollar position of the non-USprivate sector, portfolio effects, and the exchange rate of thedollar', OECD Working Paper No 76, February.

Black, S. (1985), 'International money and internationalmonetary arrangements', in Handbook of International Econ-omics, Volume II, edited by R.W. Jones and P.B. Kenen,Elsevier Science Publishers BV.

Black, S. (1989), 'Transaction costs and vehicle currencies',IMF Working Paper No 89/96, November.

Blundell-Wignall A., Masson, P. (1985), 'Exchange rate dy-namics and intervention rules', IMF Staff Papers, Vol. 32,pp 132-159, March.

Bofinger, P. (1990) 'Economic reform in Eastern Europe:Implications for the ecu, the EMS, and European monetaryunion', paper for the Conference 'Vers L'Union Economiqueet Monetaire', organized by the Ministere de TEconomie,des Finances et du Budget, Paris, June.

Boyd, C, Gielens, G., Gros, D. (1990), 'Bid/ask spreads inthe foreign exchange markets', unpublished.

Braga de Macedo, J., Goldstein, M. and Meerschwan (1984),'International portfolio diversification1, in Exchange ratetheory and practice, edited by Bilson and Marston.

Brandsma, A., Hughes Hallett, A. (1989), The design ofinterdependent policies with incomplete information', Econ-omic Modelling, July.

Brandsma, A., Pijpers, J. R. (1985), 'Coordinated strategiesfor economic cooperation between Europe and the UnitedStates', Weltwirtschaftliches Archiv, Heft 2.

Brender, A., Gaye P., Kessler, V. (1986), 'L'apres-dollar',Economica, Paris.

Canzoneri, M., Gray, J. A. (1985), 'Monetary policy gamesand the consequences of non-cooperative behaviour', Inter-national Economic Review, Vol. 26, No 3, October.

Chevassus, E. (1989), 'Le choix de la monnaie de factura-tion', Ph.D. thesis, Universite de Paris X Nanterre.

Cooper, R. (1984), 'A monetary system for the future',Foreign Affairs, autumn.

Council of Ministers (1990), 'Decision du Conseil relative ala realisation d'une convergence progressive des politiqueset des performances economiques pendant la premiere etapede 1'Union economique et monetaire1, in European Economy,Supplement A, No 3, March.

Currie, D., Holtham, G., Hughes Hallett, A. (1989), Thetheory and practice of international policy coordination:Does coordination pay?1, in Macroeconomic policies in aninterdependent world, CEPR/IMF and Brookings Institution.

Delatry, M., Van 't dack, J. (1989), The US external deficitand associated shifts in international portfolios', BIS Econ-omic Papers, No 25, September.

Dooley, M., Lizondo, Mathieson, D.J. (1989), The currencycomposition of foreign exchange reserves', IMF Staff Papers,Vol. 36, No 2, June.

Edison, H. J. (1990), 'Foreign currency operations: an anno-tated bibliography1, International Finance Discussion Paper,No 380, Board of Governors of the Federal Reserve System,May.

Eichengreen, B. (1989), 'Hegemonic stability theories of theinternational monetary system', in Can nations agree? Issuesin International Economic Cooperation, edited by R. Cooper,B. Eichengreen, C. Randall, G. Holtham and R. Putnam,The Brookings Institution, Washington.


The economics of EMU',

Feldstein, M. (1988), 'Distinguished lecture on economics ingovernment: Thinking about international economic coordi-nation', Journal of Economic Perspectives, Vol. 12, No 2,spring.

Frankel, J. (1982), 'In search of the exchange rate premium:a six currency test assuming mean variance optimization1.Journal of International Money and Finance, Vol. 1, pp. 255-274, December.

Frankel, J. (1986), The implications of mean-variance opti-mization for four questions in international macro-economies'. Journal of International Money and Finance,Vol. 5, March.

199


Frankel, J. (1988), 'Obstacles to international macro-economic policy coordination', Princeton Studies in Inter-national Finance, University of Princeton.

Frankel, J., Rockett, K. (1988), 'International macro-economic policy coordination when policymakers do notagree on the true model', American Economic Review,Vol. 78, June.

Funabashi, Y. (1988), "Managing the dollar: From the Plazato the Lonne', Institute for International Economics, Wash-ington.

Giavazzi F., Giovannini, A. (1989), 'Monetary policy inter-actions under managed exchange rates', Economica 56, pp.199-213.

Heller and Knight (1978), 'Reserve-currency preferences ofcentral banks', essay in International Finance, No 131, Prin-ceton University.

Heller and Knight (1986), 'An analysis of the managementof the currency composition of reserve assets'.

Hughes Hallet, A. (1986), 'Autonomy and the choice ofpolicy in asymmetrically dependent economies', OxfordEconomic Papers, Vol. 38.

Kenen, P. (1983), 'The role of the dollar as an internationalcurrency', The Group of Thirty, Occasional Papers 13, NewYork.

Keohane, R, O. (1984), After hegemony, Princeton Univer-sity Press.

Kindelberger, Ch., (1973), The world in depression, 1929-39,University of California Press, Berkeley.

Kiyotaki, N., Wright, R. (1989), 'On money as a mediumof exchange', Journal of Political Economy, Volume 97,No 4, August.

Krugman, P. (1980), 'Vehicle currencies and the structureof international exchange', Journal of Money, Credit andBanking, August.

Krugman, P. (1984), 'The international role of the dollar:theory and prospect', in J. F. O. Bilsen and R. C. Marston(editors), Exchange rate theory and practice, University ofChicago Press/NBER.

Krugman, P. (1989a), Exchange rate instability, The MITPress.

Krugman, P. (1989b), 'Is bilateralism bad?', NBER WorkingPaper No 2972, May.

Lebegue, D. (1985), 'Pour une reforme du systeme monetaireinternational', Economic Prospective Internationale, No 24,fourth quarter 1985.

McKinnon, R. (1988), 'Monetary and exchange rate policiesfor international financial stability: a proposal', Journal ofEconomic Perspectives, Volume 2, No 1, winter 1988.

Oudtz, G., Sachs, J. (1984), 'Macroeconomic policy coordi-nation among the industrial economies', Brookings Paperson economic activity, No 1.

Putnam, R., Bayne, N. (1987), Hanging together: cooperationand conflict in the seven power summits, Harvard UniversityPress.

Tavlas, G.S. (1990), 'On the international use of currencies:the case of the Deutschmark', IMF Working Paper WP/90/3, January.


Williamson, J. (1985), The exchange rate system, Institutefor International Economics, Washington.

Williamson, J., Miller, M. (1987), "Targets and indicators: ablueprint for the international coordination of economicpolicy', Institute for International Economics, Washington.

200

PartC

The impact through time and space

sduquenoy

Chapter 8. Transitional issues

Chapter 8

Transitional issuesThis chapter discusses the economic mechanisms that operateduring the transition towards EMU and thus determine thedesirable speed of transition. The main consideration concern-ing the speed is the extent to which different costs and benefitsarise already at different intermediate stages and whetherthese intermediate stages are stable. These two issues arediscussed in the following two sections. The third sectionthen deals with the issue of convergence, i.e. to what extentconvergence has to be achieved before exchange rates are fixedand how to minimize the cost of reducing inflation.

The main findings of this chapter are:(i) Stage I yields already substantial benefits in terms of

exchange rate and price stability and also implies themain cost of EMU, i.e. the loss of the exchange rate asan adjustment instrument.

(ii) However, the full benefits from the elimination of ex-change transaction costs and from the greater role of theCommunity in the global monetary system arise only inthe final stage of EMU. Since going beyond Stage Iwould not involve any costs this implies that beyondStage I there are only benefits.

(Hi) The preceding, considerations imply that the transitionshould be fast. The main factor limiting the desirablespeed for the transition might be the cost of too rapid orinsufficient convergence.

(iv) Under certain circumstances the transitional Stages Iand II may not be stable because of the potential forspeculative attacks. It would therefore seem prudent toprepare for a rapid passage to Stage III so that thisstep could be taken without further delay if instabilitymanifests itself.

(v) The need for convergence in other areas, especially exter-nal current account and public deficits is more difficultto assess. A stable and credibly anti-inflationary EMUrequires, however, that the explosive growth of publicdebt of some countries be stopped.

(vi) Convergence towards low inflation would be made easierthrough a credible exchange rate commitment, as shownby the experience with the EMS. The extreme form ofan exchange rate commitment would be the adoption ofthe single currency. For some of the high inflationcountries the cost of disinflation might therefore be sub-stantially reduced and the transition period much shorterif they adhered to the single currency once the lowinflation countries decide to take this step.

8.1. Benefits and costs by stage

The economic benefits and costs identified in Part B of thisvolume do not only arise in the final stage of full EMU.This section therefore discusses to what extent different costsand benefits arise already during the intermediate stages ofthe transition. In doing so it is convenient to discuss separ-ately the different costs and benefits described in Chapters3 to 7 of Part B.

The different stages that are compared here are Stages I andIII of the Delors report since Stage II does not introducesignificant economic effects. It is, however, difficult to assessthe costs and benefits that arise in Stage I because Stage Iis defined in the Delors report mainly by the elimination ofcapital controls which has already happened. However, forthe purpose of this chapter it is assumed that Stage I com-prises not only full capital mobility, but also infrequentrealignments that are limited in size to the overlappingbands. The benchmark to which this Stage I is comparedhas to be assumed arbitrarily since it is impossible to saywhat would happen if capital controls are not liberalized.For the purpose of evaluating the benefits of Stage I thebenchmark and starting point is therefore taken to be theyears preceding the formation of the EMS.

The only officially proposed alternative to the stages ad-vanced in the Delors report is that of the British Govern-ment, which in July 1990 proposed an alternative Stage II.This would be based on a revision of the definition of theecu such that it would never be devalued in relation toany EMS currency, thus becoming a 'hard ecu'. Executivefunctions related to the 'hard ecu' would be entrusted to aEuropean Monetary Fund. As in the case of the Stage II ofthe Delors report, this proposal is intended to be transitional,and its overall economic implications are hard to evaluateas distinct from a fully developed Stage I (see Chapter 2).

8.1.1. Price stability

The original narrow band ERM members have alreadyachieved a considerable degree of convergence and pricestability. This suggests that by the end of Stage I, whenrealignments would no longer be available to offset differ-ences in inflation, the goal of price stability will have beenrealized to a large extent.

As underlined in Chapter 4 the benefits of price stability areconditional since they depend on the policy of EuroFed. Thefull benefits will therefore be available only if EuroFed hasestablished its credibility for a consistent anti-inflationary

203

Part C — The impact through time and space

policy. However, this can occur only in Stage III, whenEurofed obtains exclusive responsibility for the commonmonetary policy. Stage III would therefore yield additionalbenefits in terms of price stability, as well as in terms ofminimizing the cost of price stability.

8.1.2. Efficiency gains

Chapter 3 identified four main sources of efficiency gainswhich arise at different stages of the transition and aretherefore discussed separately.

Exchange rate variability

Exchange rate variability has already been reduced consider-ably by the EMS as shown in Chapter 3 above.' For example,among the three major EMS countries intra-ERM exchangerate variability has already been reduced to one quarter ofits pre-EMS level. Taking the present exchange rate varia-bility as a good approximation of Stage I this implies thatthe elimination of exchange variability is three quartersachieved in Stage 1 so that Stages II and III add only onequarter.

Moreover, it was argued in Chapter 3 that the marginal costof exchange rate variability is an increasing function of thedegree of variability. This implies that even more than threequarters of the benefits from reduced exchange rate varia-bility should arise already in Stage I.

Although Stage I should reduce actual exchange rate varia-bility it might not reduce uncertainty about exchange rateadjustments to the same extent. As argued in Chapter 3 theinterest rate differential between the Dutch guilder and theDeutschmark shows that financial markets can perceive ex-change rate uncertainty even if actual exchange rates do notmove for quite some time. This residual uncertainty mightpersist in Stage II and even in the beginning of Stage III,when exchange rates are supposed to be irrevocably fixed.In this sense only, Stage HI and in particular the introductionof a single currency would completely eliminate exchangerate uncertainty.

Transaction costs

Exchange rate related transaction costs, such as bid-askspreads and other commissions, are not eliminated by fixingexchange rates. Indeed, no significant reduction of these

1 See Chapter 3 for a discussion of the related concepts of variability anduncertainty.

transaction costs can be expected in Stages I and II (andpossibly Stage HI as long as national currencies continue toexist) since they do not appear to depend strongly on ex-change rate variability, but rather on the economy of scaleproperties of dominant international currencies. The savingsin transaction costs would therefore arise only very late inthe transition towards monetary union, probably only withthe introduction of a single currency which constitutes theonly certain way to eliminate all transaction costs.

Foreign exchange transactions could, conceivably, also bealmost completely eliminated even before the introductionof a single currency by a system of 'par clearing', i.e. asystem in which all foreign exchange transactions (amongCommunity currencies) that go through financial intermedi-aries are executed at one central rate, without a bid-askspread and without any allowance for foreign exchangecommissions. However, such a system would be acceptableto financial intermediaries only if exchange rates are irrevo-cably fixed.

While transaction costs would therefore be eliminated onlylate in the transition towards monetary union, i.e. in StageIII, it is apparent that these costs should be significantlyreduced through the internal market programme that willincrease the efficiency of the financial sector,2 Part of thesavings in transaction costs should therefore' arise alreadyduring the transition towards economic union.

Building on 1992 and dynamic effects

The gains from building on 1992 and the dynamic effectsmight arise in a more continuous way throughout the tran-sition towards EMU.

The dynamic effects are caused in principle mainly by thecomplete integration of financial markets that comes froma single currency. However, as can be observed at presentwith the 1992 internal market programme, they could al-ready be triggered by the anticipation that EMU will becompleted soon. It is therefore probable that they will notarise only at the end of Stage III. Through the anticipationeffect they might therefore come about in a continuous way,but only if there is no doubt that the transition would leadto full EMU, and notably within a fixed time horizon thatis sufficiently short to be relevant to entrepreneurial decision-making.2 The potential importance of the internal market programme in this area

is demonstrated by the study on the cost of non-Europe, which showsthat the cost of certain types of foreign exchange transactions shouldfall by as much as 10 to 15% in some countries through the effect ofgreater competition in the banking industry (see Commission of theEuropean Communities (1988) and Price Waterhouse (1988) for furtherdetails).

204


A credible commitment to the definitive EMU is thereforeimportant to achieving dynamic gains at an early stage,and thus offsetting in some degree the transitional costs ofadjustment. The example of 1992 in fact suggests that thisoffset may be more than complete. In its study on theeconomics of 1992 the Commission hypothesized a 'J curve'pattern in the evolution of the employment effect, with somenet employment losses early in the process. In fact there hasbeen continuous and substantial employment expansion inthe Community since the late 1980s when the credibility ofthe 1992 programme was established.

8.1.3. Public finance effectsThe main benefits of sounder public finances can be har-vested only once the disinflation process has established itscredibility because only at that point can interest paymentson public debt diminish in those countries that at presentstill have higher than average inflation rates and wherethe interest rate on public debt therefore incorporates anadditional risk premium. It is difficult to establish at whatpoint this risk premium would be eliminated. Experiencesuggests that it declines gradually over time if inflationremains low. This suggests that to some extent the riskpremium on public debt should decrease significantly alreadyin the course of Stage I. The experience of the Netherlandsshows, however, that even after a long period of stableexchange rates and low inflation a small risk premium canpersist. This benefit would therefore be fully available onlyin Stage HI.

The reduction in seigniorage revenue for some countrieswould also occur to a large extent already in Stage I sinceduring that period inflation rates would already convergeconsiderably and the increasing degree of financial marketintegration would lead to pressures to lower required re-serves on commercial banks.

8.1.4. Adjusting without the nominal exchangerate

The nominal exchange rate is, of course, no longer an adjust-ment instrument once exchange rates are irrevocably fixed.As discussed in Chapters 2 and 6 this can be considered themain cost of a monetary union because an adjustment ofthe nominal exchange rate might at times constitute a moreefficient response to a country-specific shock than an adjust-ment in wages and prices, which are often not very responsiveto economic conditions in the short run. However, as arguedin Begg (1990) this cost arises to a significant extent alreadyin Stage I since at this point realignments should be in-frequent and are limited in size to the overlapping bands.

Moreover, Begg (1990) also finds that the nature of theadjustment should change with the approach towards EMU.With exchange rates totally fixed the adjustment of wagesand prices is considerably faster than in the EMS becauseeconomic agents realize that they have to react faster to agiven shock if they cannot rely on the authorities to adjustthe exchange rate.

It is a matter for political choice how far the mechanisms ofthe EC budget are developed in each stage, in order tocompensate in some degree for the loss of the nominalexchange rate instrument. It is likely that such mechanismswould be more amply developed in Stage III, and in particu-lar under the single currency, given the definitive loss of thenational exchange rate and enhancement of the union'sresponsibilities.

8.1.5. External effects

As argued in more detail in Chapter 7 EMU will have animpact on the international monetary system mainly throughthe single currency which has the potential to become aninternational vehicle currency as important as the dollar.The seigniorage benefits and the portfolio shifts that can beexpected from EMU should therefore arise only when thesingle currency has been introduced. In addition gains froma more effective role by the Community in internationalcoordination are likely to be easier to harvest when theCommunity's representation is completely unified on themonetary side.

8.2. Stability in the transition

The previous section discussed what costs and benefits arisein the different stages of the transition towards EMU. How-ever, this assumes that the intermediate stages are not in-herently unstable. If they did prove to be unstable it wouldmean that these stages were harbouring serious problems(and therefore in some sense costs). Indeed, it has beenargued that the transitional Stages I and II are likely to beinherently unstable.3 It would follow from this view thatEMU would be the only alternative to fluctuating exchangerates. This section therefore discusses the systemic stabilityof the transitional Stages I and II.

Stages I and II of the Delors plan are essentially a 'fixed butadjustable' exchange rate system without capital controls.The main danger for the stability of such a system comes

3 See Padoa-Schioppa (1987) and (1990).

205


from the absence of capital controls which creates the poten-tial for (i) speculative attacks, and, (ii) currency substitution.

8.2.1. Speculative attacks

It has been argued that any fixed exchange rate system isinherently unstable because the authorities always have onlylimited reserves whereas speculators in the foreign exchangemarkets can demand to convert unlimited amounts of assetsinto foreign currency, thus forcing the authorities to aban-don the fixed exchange rate commitment. If the authoritiesfollow a policy of full (or more than full) accommodation,that is if they ratify any increase in prices by a devaluationor if they link the money supply to the exchange rate apotential for self-fulfilling attacks arises. Such a self-fulfillingattack might then arise if agents in financial markets havedoubts about the exchange rate. If enough speculators de-mand foreign exchange for their assets they can exhaust theforeign exchange reserves thus forcing the authorities toabandon the fixed exchange rate and let the currency float.According to this line of argument the assumed policy offull (or more than full) accommodation then implies thatthe monetary authorities would increase the money supplyafter the speculative attack so that the floating exchangerate will be at a lower level than the previously fixed rate andthe doubts of the speculators would have been confirmed.4

It is apparent that a policy of at least full accommodationis essential for this to happen. If the authorities react to thespeculative attack by lowering the money supply, the floatingexchange rate that results after the speculative attack wouldbe above the pre-attack level. In this case the initial expec-tations of a devaluation would not be confirmed and specu-lators would lose money in participating in the attack. Sincethis would be anticipated in the markets no speculativeattacks of this sort could therefore arise if the authoritiesare known not to accommodate them.

A potential for speculative attacks might also arise in theabsence of an accommodating monetary policy stance if theauthorities of a country with a very large public debt areperceived as not being willing to raise interest rates to defendtheir currency because this would have undesirable conse-quences for their public finances via higher interest pay-ments.5 The mere perception of this reluctance in financialmarkets might induce many speculators to exchange theirdomestic currency assets into foreign currency and this capi-

tal flight would require the domestic central bank to inter-vene to support its own currency. However, in doing so itwould lose reserves and this loss of reserves might worsenthe confidence crisis. Without capital controls the capitalflight could then rapidly increase and become so large (ifthe loss of confidence is strong enough) that the amountsconverted by speculators exceed the foreign exchange re-serves the domestic authorities have at hand. The latterwould then no longer be able to maintain the exchange ratesat the intervention margin and the entire system mightcollapse.

It is important to note that this process can be set in motiononly if a large proportion of the public debt needs to berefinanced at the time the doubts arise about the authorities'willingness to pay higher interest rates. It is apparent that ifmost of the public debt is in foreign currency the problemdoes not arise because in this case a devaluation does notlower the value of the debt. Nor can it arise if publicdebt has, on average, a long maturity because in this caseturbulences in financial markets could affect only the interestrate on a small proportion of the total debt.

The Basle/Nyborg agreement which increases the availabilityof the 'very short-term facility' may be viewed as an attemptto lower the probability that a speculative attack could notbe contained by increasing the amount central banks canmobilize immediately for the defence of their exchange rate.

A factor that could further reduce the danger of speculativeattacks in the intermediate stages is the binding proceduresfor budgetary policy that may be established in Stage II.These may require or induce member governments to takecorrective action well before their financial situation be-comes unreasonable in the eyes of the markets. In this senseone could argue that constraints to prevent excessive debts/deficits might be more useful in the intermediate stages thanin the final stage, for which they are foreseen in the Delorsreport.

The basis for the argument that a fixed exchange rate systemis inherently unstable, because in the absence of capitalcontrols there exists a potential for these self-fulfilling specu-lative attacks, is, however, that financial markets have somereason to doubt the commitment of the authorities to defendthe exchange rates. If no such doubts exist a fixed exchangerate system might be stable even without capital controls assuggested by the Dutch experience.

4 See Obstfeldf 1988)5 See Giavazzi and Pagano (1989).

The Dutch guilder has since 1983 been pegged much moreclosely to the German mark than required by the EMS rules,which allow for margins of ± 2,25%. Indeed it has nevermoved outside a corridor of about 1 %, i.e. it has behaved

206

sduquenoy


as if the allowed margins were ± 0,5 %. Despite full capitalmobility, continuing large public deficits, and a growingpublic debt that is now close to 80 % of GDP, there havenever been any speculative attacks on the Dutch guilder/German mark exchange rate nor has the potential for suchan attack been noted by the Dutch authorities.

The Dutch-German experience (as also the linkage of theAustrian schilling to the German mark outside the EMS)represents, of course, a special case since these two countriesare even more highly integrated in terms of trade and finan-cial flows than the members of the Community in general.The differences in their approaches to economic policy mak-ing are also smaller than among Community members ingeneral. Moreover, the Dutch-German monetary union rep-resents an 'asymmetric1 union to a degree that would not bethe case for the entire Community in Stages II and III. Othercases have been similarly 'asymmetric1, for example thepegging of the Irish pound to that of the UK until thecreation of the EMS, and currently the pegging of the HongKong dollar on that of the USA. Indeed some observers6

have argued that historically symmetric monetary unionshave never been stable. In many instances countries havepreferred to reimpose capital controls or to devalue, ratherthan accept the discipline coming from a fixed exchange rateand open capital markets.

A related argument why the system of fixed exchange ratesin Stages I and II might not be stable relates to the obser-vation that most fixed exchange rate systems contain acountry that provides effective leadership either because ofits economic weight, or because of its superior price stabilityrecord.7 Up to now Germany has ppovided this leadershipbecause it had a superior price stability record and the markwas the only major international reserve currency in thesystem. However, as the price stability record of some othercurrencies gets close to that of the mark and as capitalmarket liberalization gives the other currencies a potentiallybigger international role the predominant position of theBundesbank in the system may indeed be weakened.

Tighter voluntary cooperation in Stage I and the establish-ment of EuroFed in Stage II might overcome this potentialleadership vacuum. However, it remains to be seen how wellthese arrangements can provide for an effective leadership,as opposed to the mere resolution of conflicts about thethrust of national monetary policies.

* For example Carli (1989) and Giovannini (1990a).7 See Matches (1988).

Recent events in Eastern Europe, in particular the unifi-cation of Germany, have further complicated the issue ofthe leadership within the EMS. It is not possible to say atthis point what the overall effects of these developments willbe, but the reaction of financial markets suggests that theoutlook for price stability in Germany has become moreuncertain. One indicator of this is that in the spring of 1990France and other EMS countries were able to reduce theirinterest rate differentials vis-a-vis Germany to a historicallyvery low level.

The danger of speculative attacks is generally assumed toarise mainly in the intermediate Stages I and II, when realign-ments are still possible. In Stage III exchange rates aresupposed to be irrevocably fixed and there should thereforebe no danger of a lack of credibility of the exchange ratecommitment. However, it is not clear how this 'irrevocable'commitment never to change exchange rates can be madecredible. Economic theory has little to say about this exceptthat the market will implicitly judge the strength of thepolitical will to maintain parities unchanged.

Adoption of a single currency is the only sure way to over-come this credibility issue. This can be illustrated with theexperience of the monetary union between Belgium andLuxembourg. Although this union, at the rate of one Belgianfranc to one Luxembourg franc, has existed now for over50 years, the mere rumours that Luxembourg was consider-ing not to follow Belgium during the last EMS realignmentswere enough to induce financial markets to differentiatebetween these two currencies. At present forward coveragainst changes in the Belgian franc/Luxembourg franc par-ity costs about 25 basis points, which indicates only a lowprobability of a change actually occurring, but it does indi-cate that the market does not accept even the BLEU as an'irrevocable' commitment. This suggests that at the Euro-pean level it will be difficult to convince markets that ex-change rates are 'irrevocably' fixed as long as nationalmoneys continue to exist.

8,2.2. Currency substitution

This represents another potential source of instability forthe intermediate stages. Even if exchange rates are expectednot to change (and therefore interest rates have converged)certain currencies might become more attractive than othersbecause they are more widely used in intra-Communitytrade.8 The possibility therefore arises that especially largefirms may concentrate their holdings of monetary balanceson the larger EMS currencies and choose to use less thesmaller, less important currencies. Although one would not8 See Giovannini (1990b) and HM Treasury (1989).

207


expect such a movement to be sudden it might still provokelarge shifts in money demand over the short to medium run.Such shifts would make it more difficult for national centralbanks to interpret their own national aggregates. As long asnational currencies continue to exist and national centralbanks retain some margins of manoeuvre, these shifts indemand across currencies constitute a potential factor ofinstability. However, since these shifts should net out for thesystem as a whole they should not jeopardize the capacityof EuroFed to maintain an anti-inflationary overall stance.

It is not clear, however, whether the fixing of exchangerates would increase or diminish the potential for currencysubstitution. The main reason why it is feared that currencysubstitution might become widespread is not the fixing ofexchange rates, but the increasing integration of Europeanmarkets in general and the financial markets in particular.The reduction or elimination of the residual exchange ratevariability might therefore not be a decisive factor and thedanger of currency substitution might therefore be viewed aside effect of the internal market programme or the economicunion aspect rather than a specific consequence of the tran-sition towards monetary union in Stages I and II.

The argument that the transitional Stages I and II could beunstable implies that it might be preferable to keep them asshort as possible and to take the final step of irrevocablylocking parities even before full convergence has beenachieved. However, if exchange rates are irrevocably lockedbefore inflation rates have converged a different type oftransitional problem might arise which is discussed in thenext section.9

8.3. How much convergence remains to beachieved?

How much convergence needs to be established beforeundertaking the step of irrevocably fixing parities is anunresolved question. This section therefore describes whatproblems might arise when parities are locked before theseconditions are met.

The degree of convergence that needs to be achieved variesfrom country to country. The country-specific issues arediscussed in more detail in Chapter 10. This section discussesonly the broad orders of magnitude of the adjustment thatwould be necessary in certain countries to enable them toparticipate in the transitional stages.

Regarding inflation, three broad groups of countries mightbe distinguished:(i) The original narrow band members are already charac-

terized by a high degree of convergence towards lowinflation since inside this group inflation differentialsare only about 1 %, not far from what is required byEMU as suggested by the examples examined above.

(ii) More adjustment in terms of inflation is needed in Italy,Spain and the UK where inflation has now stabilizedat a level that is about 3 to 5 percentage points abovethe best performance in the EC. However, for this groupparticipation in Stage I should be feasible provided theadjustment is seen to continue.

(iii) In Portugal and Greece inflation is still above 10 % sothat even participation in Stage I clearly needs moretime.

As discussed in Chapter 2, there are, however, reasons toexpect that non-negligible inflation differentials can arise ifproductivity growth differs across countries. Full conver-gence in inflation can therefore be expected only once allmember countries have reached a similar productivity level.

Regarding the need to establish sound public finances, it isagain possible to identify three groups of countries (seeChapter 5 for a further analysis of the concept of a sustain-able fiscal position):

(i) In Denmark, Ireland and the United Kingdom the levelof the public debt/GDP ratio is declining and Franceand Germany are very close to this situation so that inthis group of countries budget policy should be undercontrol.

(ii) In Belgium, Spain and Portugal, the public debt/GDPratio has not yet been stabilized, but this objective seemsto be in reach within the near future since it requiresonly an adjustment in the primary balance of about onepercentage point of GDP.

(iii) In Greece, Italy and the Netherlands present trends inbudgetary policy would lead to a rapid deterioration inthe public debt/GDP ratio and must therefore be ur-gently rectified.

The examples of Italy and Belgium show, however, thatthese imbalances do not necessarily need to be an obstacleto participation in Stage I. However, they would need tobe addressed during that stage to achieve the necessaryconvergence for the following stages.

9 See Giavazzi and Spaventa (1989).

208

Convergence to sound public Finances is necessary for astable EMU in the long run. However, lack of convergence


in short run fiscal policy can also create problems evenif the debt to GDP ratio does not explode because largedifferences in the stance of fiscal policy can create tensionsin the exchange market and slow down convergence in otherareas. For example, a combination of a tight monetarypolicy (to keep the nominal exchange rate within the EMSmargins) with an expansionary fiscal policy can create asituation in which, temporarily, inflation convergence isstopped and a large current account deficit arises. Since sucha policy cannot be sustained forever, it will lead to a needfor faster convergence later. To avoid the adjustment prob-lems that arise when the policy has to be reversed someconvergence in the stance of fiscal policy is therefore desir-able even during the transition towards EMU. It is apparentthat, to the extent that exchange rates become more andmore fixed, no individual member country will be able todetermine its own monetary policy and therefore a mix of atight monetary and a lax fiscal policy should be avoided.

The need for convergence towards a sustainable externalequilibrium is less apparent since the external current ac-count disequilibria that exist at present might be financedwithout difficulty once capital markets are completely inte-grated. As discussed above, further convergence in externalcurrent accounts is desirable from a policy point of viewonly if the current disequilibria reflect excessive public dissa-vings. It is therefore not evident that further policy actionwould be necessary to correct current account imbalancesonce sound public finances have been achieved in all membercountries.

8.3.1. How to achieve convergenceWhile it is clear that a stable EMU needs a considerabledegree of convergence the important question might be howbest to achieve this degree of convergence to minimize themacroeconomic costs of disinflation. As discussed in Chap-ter 3, modern macroeconomic theory argues that these costsarise for two reasons: wage rigidity and lack of credibilityin the adjustment process. Both of them affect the desirablemode and speed of the transition.

One extreme position concerning the debate about conver-gence is the so-called 'coronation theory' according to whichthe passage to a higher stage of monetary integration shouldonly ratify or 'crown' the convergence that has already beenachieved. At the other extreme the so-called 'monetarist'school holds that convergence could always be achievedthrough exchange rate commitments, which could thereforebe taken without any prior convergence.10

The experience of the EMS so far suggests that neither ofthe two extremes should be followed. On the one hand itis widely acknowledged that the EMS has been a useful'disciplinary device* for more inflation-prone countries." Inthis view the EMS was established to enforce convergencethrough the ERM. When the ERM was established theconvergence which some observers regarded as a necessaryprecondition did not exist. Instead it was brought about bythe ERM only later and this convergence to low inflation isconsidered the main achievement of the EMS. On the otherhand, the strains the EMS experienced in the first years ofits existence also illustrate the difficulties that can arise whenconvergence is insufficient.

The influence of the EMS in reducing the cost of disinflationin the 1980s has been analysed quite extensively in theeconomics literature.12 This analysis suggests that the EMSdid not have a strong impact on the credibility of disin-flationary policies in the years up to 1983 because there werefrequent realignments that were used to validate ex-postinflationary tendencies in order to keep real exchange ratesapproximately constant. However, there is a clear breakafter 1983 which suggests that after that date the EMS hasbeen influential in reducing the cost of disinflation. Thisdate coincides with a change in emphasis inside the EMSsince after 1983 realignments were less frequent and did notcompensate fully for past inflation differentials.

The experience of the EMS suggests therefore an intermedi-ate position: without any exchange rate commitment orpolicy coordination full convergence might never comeabout and the goal of monetary union might not be achieved.It might therefore be useful to proceed with Stages II andIII even without full convergence in terms of price stability.Convergence would then be achieved through market forcesprovided there is a credible commitment of the authoritiesnot to slide back. Such a credible commitment requires, ofcourse, that the initial situation is not such as to imply anunrealistically fast convergence.

The rigidities in the wage-setting process that create a short-run trade-off between inflation and output are, of course,not directly affected by the exchange rate regime. However,even in the case of the most important type of wage rigidity,namely backward-looking wage indexation schemes, a cred-ible commitment to a fixed exchange rate with a stablecurrency could provide the necessary degree of confidence

This debate concerns especially convergence of inflation rates; the issuesraised by the adjustment in fiscal policy are discussed in Chapter 5.

11 Of the large literature on this subject, see for instance Giavazzi andPagano(1988).

12 See De Grauwe (1989). Giavazzi and Giovannini (1988) for furtherreferences.

209


that inflation will remain low to induce the social partnersto abandon this instrument, or reform it on a forward-looking basis during the transitional period.

The experience with the EMS has shown that the credibilityof the exchange rate commitment can be an important factorin reducing the cost of disinflation. At its extreme this ideaimplies that by exchanging the national currency for anotherone, in this case the single European currency, the macro-economic cost13 of disinflation could be eliminated alto-gether. One could therefore argue that less than full conver-gence by all members is needed before the single currency isintroduced. As long as there is a stable core of countrieswith a reputation for price stability which ensures a credibleanti-inflationary policy for the single currency some periph-eral countries could therefore join in the final stage of EMUeven if convergence in inflation is not complete. This radicalstep would ensure immediate convergence, provided, ofcourse, that the initial level of real wages is appropriate andthat all links with the past in the form of backwards-lookingwage indexation are eliminated.

8.4. Speed of transition

The foregoing analysis has implications for the question ofthe speed with which the transition should be effected.

The discussion of the costs and benefits by stages suggeststhat the main costs in terms of achieving price and exchangerate stability arise early in the transition towards EMU. The

13 The need for a fiscal adjustment would not be affected by this strategy.

move to the final stage of EMU, i.e. the introduction ofthe single currency, should therefore imply no significanteconomic cost and it would yield the benefits in terms ofeliminating transaction costs and induced dynamic effects.This suggests that the transition should not be too long.

The danger of instability of the transitional Stages I and IIprovides another argument for proceeding with the sub-sequent steps towards Stage III as soon as possible. Even ifthe probability of speculative attacks and large scale cur-rency substitution is only minor, it would still be prudent toprepare for this potential instability by making the necessarypreparations for the passage to Stages II and HI. Theseadditional steps could then be taken without further delayshould the instability become a real danger for the system.

The cost-benefit analysis therefore suggests that the tran-sition should be kept relatively short. However, it is moredifficult to describe how this could be achieved. On the onehand irrevocably Fixing exchange rates before full conver-gence is achieved involves costs; but, on the other hand,waiting for full convergence before taking this step mightlengthen the transition because in the absence of an exchangerate constraint convergence might slow down.

The main factor limiting the speed in the transition mighttherefore be the costs that arise during the process of disin-flation and fiscal adjustment. These costs are of a strictlytransitory nature (as opposed to the benefits of EMU whichwould be permanent), but they might increase if the adjust-ment is too fast. They should be reduced by a credibleframework that clearly establishes the path for convergence.The transition should therefore not only be fast, but alsoprovide clear signals about the need for convergence.

210

sduquenoy

sduquenoy


References

Alesina, A. (1989), 'Politics and business cycles in industrialdemocracies', Economic Policy, April, pp. 57-89.

Begg, D. (1990), 'Alternative exchange rate regimes: Therole of the exchange rate and the implications for wage-priceadjustment', in European Economy — The economics ofEMU', special issue.

Bhandari, J., Mayer, T. (1990), 'Saving-investment corre-lations in the EMS', mimeo, International Monetary Fund,Washington DC.

Carli, G. (1989), The evolution towards economic and mon-etary union: a response to HM Treasury paper', mimeo,Italian Ministry of Treasury, December 1989.

Commission of the European Communities (1988), Theeconomics of 1992', European Economy No 35, Commissionof the European Communities, March.

Commission of the European Communities (1977), Publicfinance in European integration (the "MacDougall report'),Office for Official Publications of the European Communi-ties, Luxembourg.

Committee for the Study of Economic and Monetary Union(1989), 'Report on economic and monetary union in theEuropean Community' (the 'Delors report1), Office for Of-ficial Publications of the European Communities, Luxem-bourg.

De Grauwe, P. (1989), 'The cost of disinflation and theEuropean Monetary System', International Economics Re-search Paper No 60, Katholische Universiteit Leuven.


The economics of EMU',

Giavazzi, F., Giovannini, A. (1988), The role of the ex-change rate regime in disinflation: empirical evidence onthe European Monetary System', in Giavazzi, Micossi andMiller (eds), The European Monetary System, CambridgeUniversity Press.

Giavazzi, F., Pagano, M. (1988), The advantage of tyingone's hands', European Economic Review 32, pp. 1055-82.

Giavazzi, F., Pagano M. (1989), 'Confidence crises andpublic debt management', CEPR discussion paper No 318.

Giavazzi, F., Spaventa L. (1989), The new EMS', CEPRpaper No 369, January.

Giovannini, A. (1990a), 'Currency substitution and monet-ary policy', paper prepared for the conference on 'Financialregulation and monetary arrangements after 1992', Mantanaand Gothenberg, 20 to 24 May 1990.

Giovannini, A. (I990b), 'Money demand and monetary con-trol in an integrated European economy', in European Econ-omy (1990).

HM Treasury (1989), 'An evolutionary approach to econ-omic and monetary union', London, November.

Matthes, H. (1988), 'Entwicklung des EWS mit Blick auf1992', in Duwendag, Professor D. (ed.), Europa Banking,Baden-Baden.

Obstfeld, M. (1988), 'Competitiveness, realignment andspeculation. The role of financial markets', in Giavazzi, F.,Micossi S., Miller M. (eds), The European Monetary System,Cambridge University Press, pp. 232-246.

Padoa-Schioppa, T. (1990), Towards a European centralbank; fiscal compatibility and monetary constitution', paperpresented at the Bank of Israel and David Horrowitz Insti-tute, conference on aspects, of central bank policy making,Tel Aviv, January 1990.

Padoa-Schioppa, T. (1987), Efficiency, stability and equity,Oxford University Press, Oxford.

Poloz, S. (1990), 'Real exchange rate adjustments betweenregions in a common currency area', mimeo, Bank of Can-ada, February.

Price Waterhouse (1988), The cost of non-Europe in financialservices. Commission of the European Communities.

211


Chapter 9

Spatial aspectsThe purpose of this chapter is to complement the previousdiscussion of the overall costs and benefits of economic andmonetary union by examining in the light of economic theoryand empirical evidence its regional impact, especially on theCommunity's peripheral regions. The following main con-clusions are possible:

(i) Concern with the distribution of benefits (and losses)from economic integration has been expressed at everystage of development of the EC: its creation in the 1950s,the northern enlargement in the 1970s, the southernenlargements in the 1980s and more recently the 1992single market. After an initial reduction in the 1960s,there has been no clear long-term trend in regional dis-parities. In the period up to the first oil crisis, there wasan overall convergence of living standards. However, thiswas followed by a slight increase of disparities in thesecond half of the 1970s and early 1980s. Since then,regional disparities have remained at roughly at the samelevel.

(H) Economic literature has been dominated by a debatebetween the presumption that economic integration, perse, aggravates the problems of peripheral regions andthe view that convergence is a more likely outcome.Earlier attempts to model the effects of economic inte-gration upon regional imbalances were based almostexclusively on the consideration of transport costs andscale and agglomeration economies. Recently attentionhas shifted towards the importance of structural disad-vantages which limit the competitive strengths affirmsin lagging regions. Business surveys are viewed as auseful tool for providing a detailed assessment of factorsshaping the competitiveness of regions.

(Hi) Empirical evidence from Commission studies suggest thatreduction of regional disparities coincides with periodsof rapid growth whereas the problems of poorer anddisadvantaged regions are accentuated in periods of re-cession. Also, convergence occurs more spontaneouslyamong countries and regions that have already reacheda relatively mature stage of economic development.

(iv) The adjustment of less favoured regions to the 1992single market is being assisted by the Community struc-tural Funds which were significantly enhanced for thatpurpose. The financial envelopes for these Communitypolicies are settled up to 1993. Monetary union addsto .the regional consequences of the single market and

"" economic union in that the nominal exchange rate instru-

ment can be more important to economies undergoingdeep structural change. The transition to monetary unioncalls therefore for adequate support from Communitypolicies.

(v) The key to the catching-up of the less favoured regionslies in obtaining synergies between Community and na-tional policies. Together with policies geared at reducinglocational disadvantages and enhancing market ef-ficiency, the fixing of clear and credible policy objectivessuch as for the single market and EMU are also highlyrelevant. Indeed, given the ambitious nature of the catch-ing-up process, the appropriate concept seeems to be thatof a credible and comprehensive regime change for thecountries and regions concerned.

9.1. Principles

9.1.1. Concepts: regions and regional problems

Although the principal tools of economic analysis ignore thespatial dimension, this becomes important once it is realizedthat:(a) the impact of economic measures and policies is in fact

variable throughout a continuum of regions that differ interms of natural characteristics and economic structures;and

(b) such measures and policies interact with the specificobjectives of policies designed to tackle localized regionalproblems or overall imbalances in the spatial structureof the economy.

Therefore, the consideration of the spatial dimension or theimplications for regional problems has become an importantpart of economic policy analysis.

The concept of'region' may vary according to the purpose:it can be defined in geographical, political or economicterms. In the latter case, regions can be delimited on ahomogeneity criterion, or by trying to capture the highestdegree of interdependence within each region. In our case,the most important level is that of political regions: theMember States of the Community. Their importance comesfrom the fact that the centre of political power is concen-trated at that level, and that they play the main role inthe bargaining process inside the Community. The variousstages of economic integration concern mainly the relin-quishing of national policy instruments — not regional ones— and thus its effects are felt in all regions of a country. Asshown below, regional disparities in the enlarged EC areincreasingly determined' by inter-country differences. The

212

sduquenoy

Chapter 9. Spatial aspects

intra-country regional level nevertheless continues to bequite relevant for the purposes of this analysis.

There is no simple definition of what constitutes a regionalproblem. The following goals, which are not equivalent,might be considered in alternative or complementary ways:

(i) a spatially balanced distribution of population andeconomic activities — in order to avoid negative exter-nalities and long-run social and environmental prob-lems;

(ii) an adequate level of provision and accessibility of publicgoods and local services to all the populations;

(iii) equalization of levels of GDP per capita;(iv) equalization of levels of personal disposable income, or

per capita consumption, or welfare — not the same asthe previous due, inter alia, to taxes, transfers and pricechanges affecting the consumer surplus.

As examples, relocation of industry towards more peripheralregions has coexisted historically with the migration of wor-kers towards more central regions, where job creation in theservices sector is faster; economic decline of some regions interms of relative GDP can also coexist with an increase inper capita consumption and better provision of basic servicesby the welfare state in the region. The first objective —spatial balance of population — is in a sense more demand-ing than the fourth, since welfare equalization may beachieved through emigration and large-scale subsidizationof a region. GDP per capita as an indicator combines severalelements: to satisfy this criterion the region must have aneconomic basis to provide its inhabitants with a high stan-dard of living, assuming certain distributional conditions arefulfilled.

It is not the purpose of this chapter to select one or anothercriterion, but one should remember that all of them may beimplicit in the discussion. In particular regional GDP percapita is a criterion used in Community structural inter-vention.

9.1.2. Geographical effects of economicintegration

One traditional view of the effects of economic integration,known as the 'convergence school', derives from the neo-classical theory of international trade, and holds that thefree movement of goods and services will, under strongassumptions, equalize factor returns and living standards.At regional level, under somewhat different assumptions,

mobility of the factors of production is emphasized as theforce equalizing income levels among regions (see Borts,1960). According to neo-classical theory, the subsistence ofregional disparities would be mainly due to the time-lagsinherent to the process of integration and to imperfect factormobility.

Another school of authors preferred to stress the mechan-isms that work towards greater divergence instead of conver-gence. This is in the tradition of Myrdal (1957) but it canalso be found in the early works of Giersch (1949), Bye(1958) and Scitovski (1958). Regional problems might arisein a customs union, as the abolition of restrictions on tradeand factor movement increase the attractiveness of highlyindustrialized centres for the location of new activities. Thisconclusion was based on the role of agglomeration econom-ies of scale in the observed patterns of regional developmentwithin industrialized countries. A similar conclusion isreached by Perroux (1959), whose views on development arebased on the concept of a growth pole, offering a goodinfrastructure as well as external economies. To the extentthat some countries or regions do not possess such poles,they will lose from integration, at least in relative terms. Theemphasis of this literature was on the real side of the econ-omy, and the centripetal effects of a monetary union, asopposed to a customs union, were generally considered tobe rather small (see Stahl, 1974, p. 220).

Balassa (1961) defends the convergence school by claimingthat these arguments put too much emphasis on agglomer-ation economies of scale, and disregard the increased oppor-tunities for taking advantage of lower costs in backwardregions. Frontier regions in particular are prone to benefitfrom greater integration. Other positive spread effects couldalso benefit the relatively poor regions, given the tendencyof industries to become less bound by natural resources,local external economies and transport costs, as a result oftechnological developments. That view prevailed during theconception and earlier years of the Community (see Section9.2).

An instrument of the divergence school is the concept of'economic potential' of C. Clark (see Clark et al., 1969)representing the availability of inputs and, especially, theproximity of markets to the region. The economic potentialof a region is defined as the summation of its income andthose of the surrounding regions, divided by the transportcosts between them. Depending on the measure of incomeand cost, economic potential will no doubt vary, but almostby definition economic integration will increase the econ-omic potential of the central — and already developed —areas while increasing the disadvantage of peripheral areas.

213


The traditional arguments about convergence or divergencein an integrated area thus try to balance two types of effects:(i) the allocative benefits of interregional trade and better

exploitation of comparative cost advantages, and(ii) the centripetal forces resulting from agglomeration econ-

omies of scale, externalities and transport costs

The basic insight is correct but its loose analytical underpin-ning makes it difficult to reach a consensus on the basis ofthis type of consideration. Without a proper model, the signand nature of externalities, and the way they extend throughspace, are not a priori clear. Similarly, the temptation tofind the determinants of comparative advantage and special-ization among regions is often frustrated as these cannot beidentified by an economic model. In the case of developingregions or countries it is hazardous to attempt to distinguishtheoretically or empirically where ex ante comparative ad-vantages lie and to predict what pattern of specializationmight evolve. Who is poised to benefit most from integrationis therefore impossible to ascertain with any degree of pre-cision. Policies which will hinder the realization of thesegains are, however, easier to identify.

This may be why the geographical aspects of economicactivity have not played a very prominent role in economictheory. Stressing the issue of method, Krugman (1989) ob-serves that the traditional model of competitive generalequilibrium does not provide answers to the more pertinentissues raised by economic geography. In particular, thatmodel cannot take into consideration facts such as econom-ies of scale and externalities which currently are accepted toplay an important role in the location of production factorsand, a fortiori, of economic activity. The analysis of thespatial effects of economic integration is no exception to this.In a nutshell, economic theory does not currently providedefinite and well-established results on the issue. Naturally,the empirical analysis of economic geography is affected bythis situation.

In order to provide new approaches to the theory of econ-omic geography and in the line of recent contributions tothe theory of international trade, Krugman (1989) presents amodel of specialization among regions, with factor mobility,which does not depend on resource-based comparative ad-vantage, nor on vaguely defined externalities (technology,etc.). Instead, he uses a model of imperfect competition, witheconomies of scale and transport costs in the manufacturingsector. There are two regions, with labour mobility betweenthem, and two sectors, agriculture (immobile) and manufac-turing (mobile). The question is whether manufacturing ac-tivities, and the corresponding workers, will tend to concen-trate themselves in one of the regions, or whether fromarbitrary initial conditions, an equilibrium will be reachedwhere manufacturing is distributed between the two regions.

The answer depends on three key parameters of the model,which bring greater precision to the centripetal forcesmentioned in the traditional discussion. The forces workingtowards and against regional convergence in a union arethen:

(i) Economies of scale. The more important they are, thegreater the tendency to concentrate manufacturing inone region — even though they may be purely internalto each firm.

(ii) Transport costs. The higher those costs, the more un-likely concentration becomes. This suggests' that theopening of trade, by lowering the costs on the movementof goods, works towards divergence.

(iii) The size of the (footloose) manufacturing sector. Sinceonly in this sector can circular causation set in, thelarger it is, the more likely divergence becomes.

The second point can be further qualified by the followingreasoning due to Krugman and Venables (1990): the com-mon assumption is that the elimination of barriers (whichcan be assimilated to a reduction of transport costs betweencountries) will bring about a concentration of production inthe central location, in order to take advantage of economiesof scale and easy access to markets. If, however, one takesthis reduction of barriers to the extreme and they becomeinsignificant, economies of scale may instead be achieved bylocation on the periphery, where variable costs are generallylower. In other words, the usual argument that the reductionof barriers benefits the centre assumes that some kind ofbarriers or costs (transport) remain so that the advantage ofgeographical proximity to the markets (the centre) super-sedes the lower production costs at the periphery. Thisimplies a U-shaped curve in the relocation of certain indus-tries when trade barriers are progressively reduced. Evenwhen totally free trade would foster economic activity in theperiphery, partial liberalization might be the worst optionfrom the point of view of those economies. The patternapplies also to the relative wages of the periphery, withimplications for the welfare situation mentioned earlier.

Another element typically absent from the traditional modelsof circular causation of regional disparities is the existenceof multiple economic centres among, and even within,countries. If a small country integrates with a larger one andboth have an established centre, it may happen that thecentre of the smaller country benefits at the expense of thelarger country's by recapturing a part of its natural hinter-land — see Krugman (1989). Even when integration doesresult into a net relocation of industries in favour of thecentre (larger market) it remains true that a smaller countryis likely to reap more important welfare benefits from inte-

214

sduquenoy


gration. With barriers to trade, a small country is unable toachieve economies of scale and, furthermore, is characterizedby a low level of domestic competition. These two effects,which add to the more conventional gains from comparativeadvantage, are highlighted by recent models of internationaltrade with economies of scale and oligopoly (see Krugmanand Venables, 1990) and constitute an important asymmetryfavouring, this time, small countries.

To summarize, the traditional view of the spatial effects ofeconomic integration, which emphasizes external economiesof agglomerations, overlooks a number of elements thathave recently been focused on:(i) the interaction between economies of scale and trans-

port or transaction costs;(ii) imperfectly competitive market structures;(iii) the larger welfare (static) benefits for small countries;(iv) the multipolar structure of the real world.

The addition of these aspects will result in a more complexpattern of locational changes and implies that the overalleffect of market integration will differ from one laggingregion to another.

9.2. Basic facts and trends

9.2.1. Empirical evidence

Overall, the existence of the European Community has mar-ked the period of strongest convergence of national wealthlevels of its members in the last 150 years or so (see Molle,1989). But the convergence trend was interrupted in the1970s. As Table 9.1 shows, it is too early to judge whetherthe more favourable growth environment of the late 1980shas resulted in a sustained resumption of the catching-upprocess.

Global measures of the intensity of regional problems, likethat of Table 9.1, are plagued by many problems: the defi-nition of 'regions', which are more determined by statistical,administrative and political convenience than by economiccriteria;' the lack of reliable regional data; and the sensitivityof the results to the measure of dispersion used (like theTheil coefficient in the table). Even if it was possible to havea clear picture of the evolution of regional disparities withinthe Community of 6, 9 or 12, this does not amount topassing a judgment on the regional impact of economicintegration. For that, one would have to be able to observethe anti-monde where the relevant countries would stay out-side the Community and find out whether there would bemore or less convergence of economic performances.

International comparisons (also suffering from the problemsmentioned above) show, however, that regional disparitieswithin the Community, with respect to a variety of socio-economic indicators, are important. For example, regionalincome dispersion is about twice as great as in the UnitedStates. As far as structural backwardness is concerned thereis within the Community of 12 a clear centre-to-peripherypattern, as well as a northeast-southwest gradient. A studysponsored by the Commission (see Keeble et al., 1988) com-puted an index of peripheral! ty for EC regions,2 and groupedthem accordingly into central, intermediate and peripheralregions. Box 9.1 contains a map from that study, togetherwith a comparison between the characteristics of peripheral,intermediate and central regions, using regional data relatingto demographic trends, economic activity and the labourmarket. These indicators give a clear illustration of theeconomic dimension of peripherally: in the outermost ring

1 In the Eurostat NUTS II list of regions — the one most often used byanalysts — the smallest region has 404 km2 and the largest 94 200 km2.

2 The study used 1983 data for 166 Community regions of NUTS level II.See Box 9.1. for details.

Table 9.1

Disparities in per capita GDP in EUR 12,1950-87 (Theil coefficient)

1980 1981 1986 1987

Average 10 weakest regionsAverage 10 strongest regionsWeighted coefficient of variation

4714526,1

4614626,5

4614726,8

4514927

4514927,2

4515027,5

4515127,9

4515127,5

4515127,5


215


of peripheral areas of the Community, representing 40 % ofits surface, live about 20 % of its population and only 13 %of GDP is produced.3 Those areas also represent more thanhalf of the agricultural population of the Community andhave significantly higher rates of unemployment than theaverage.

A glance at the map also shows that the degree of peripheral-ity is broadly in line with the boundaries of what has beencalled, in the reform of the structural Funds, the Objective 1regions (see Section 9.4.1), i.e. Ireland, Portugal, Greece andparts of Italy and Spain. That is not surprising, given thecorrelation between peripherality and regional GDP perhead, the criterion used in the definition of Objective 1. TheGerman Democratic Republic, although not covered by theabovementioned study nor (yet) by the reform of the Funds,would also qualify as a major problem region of the Com-munity.

In a broader context, Williamson (1965) examined the long-run evolution of interregional income disparities in a numberof countries and proposed a two-stage model of regionaldevelopment. In the first stage, economic development seemsto take root in some sectors and regions of a country; abovea certain threshold, it begins spilling to the rest of theeconomy. Thus, in the first stage of growth there is regionaldivergence, in the second, regional convergence. However,many questions remain unanswered, namely when and howthe divergent trend is reversed.

Many studies have tried to relate integration-specific vari-ables such as trade flows, migration and foreign investmentto the convergence performance of the national economies.In general, no clear relationship emerges which is valid forall countries at all times. However, one conclusion at leastcan be drawn from the EC experience and all the relatedempirical studies: convergence tends to dominate in periodsof strong growth, and to recede in periods of stagnation.This may be explained by two reasons:(i) Weaker regions tend to contain a higher proportion of

sensitive sectors and small and marginally efficient firms(see Vanhove and Klaassen, 1980). These are the first tobe affected by a recession, leading to a relative declineof the whole region;

(ii) Those regions also receive a relatively higher proportionof public investment and new — as opposed to replace-ment investment — from the private sector. Empiricalevidence and past experience show those categories arethe most responsive to the economic climate.

Since the figure is in PPS, it probably underestimates the imbalances inthe location of economic activities among the regions.

Historically, the reduction of regional disparities in somefederations, like the United States, has been ascribed to twomain factors:

(i) financial transfers through the central governmentbudget, and

(ii) mobility of the factors of production.

This may explain the higher degree of regional differencesin income — as well as in rates of unemployment — in theCommunity compared to the United States despite the muchgreater geographic extension and natural diversity of thelatter (see Box 9.2). Whereas the income differences maypoint to different stages of economic development, the dis-parities in unemployment rates may be a sign of weakeradjustment capacities to shocks than in existing federalStates. The weaker adjustment capacity may in turn belinked to the stage of economic development, underliningthe balance between convergence and divergence that em-erges from the recent theoretical literature.

9.2.2. The response in EC policies

In the late 1950s and early 1960s, there was considerableanxiety about the impact of the integration process andincreased competition upon the Community's problem re-gions. In the original EC, these regions consisted predomi-nantly of localized problems of industrial decline (like in thecoal and steel and a number of other traditional sectors).The creation of the EC seemed somewhat in conflict withthe tendency at the time in European countries to answerthose sectoral problems, as well as promote the industrializa-tion of backward regions, through the use of strong andinterventionist regional policy tools. Within the fast growthof the EC's early years, however, those regions performedin general quite well, being able to attract modern investmentand share in the overall rising prosperity of the Community.

There were also some regional safeguards in the articles ofthe EEC Treaty covering the agriculture, transport andcompetition policies. It was considered that the dynamiceffects of the creation of the Community would spreadstrongly enough to the problem regions so as to promoteconvergence and that any further measures should be left tonational governments. Nevertheless, a Community insti-tution was created, the European Investment Bank, whosemain task was promoting regional development. Specialregional protocols concerning southern Italy and the intra-German border regions were also agreed. Thirty years later,the Mezzogiorno and East Germany remain special cases ofthe regional problem in the Community.

216

sduquenoy


In 1970 the Werner report on economic and monetary unionbrought to the discussion the need for accompanying struc-tural measures in the integration process. The first enlarge-ment brought in a number of new problem regions, and the1973 oil shock, to which Member States reacted in verydifferent ways, made regional problems worse: the declineof manufacturing accelerated and spread to new sectors andregions. Unemployment also increased sharply, stemmingthe traditional migration flows among regions and countries.Profitability and investment also experienced a strong de-cline, with a more marked effect on lagging regions, for thereasons mentioned above (9.2.1).

All these events marked a turning point in the acceptanceby the Community of a larger responsibility for structuraland regional problems. The justification for this has beentwo-fold:(i) on economic grounds, closer forms of integration are

inhibited by an excessive disparity of structures andincome levels;

(ii) on political grounds, the pursuit of internal cohesionand equity which is characteristic of national Statesshould gradually be extended to the Community levelas integration progresses.

Within the envisaged EMU, also, the transfer of an increas-ing number of policies to the Community level and thesuppression of important national policy instruments shouldbe counterbalanced by more active Community structuralpolicies. This is more so if one believes that the naturalconsequence of integrating the national economies is toincrease regional problems, as was the prevailing view.

The first movement towards a regional policy at Communitylevel was the adoption in 1971 of the principles of coordi-nation of regional aid regimes, which aimed at minimizingthe distortions in competition within the Community whileprotecting national regional development objectives andavoiding an overbidding in the level of aid. 1975 saw thecreation of the European Regional Development Fund. TheERDF was thereafter reinforced, especially after the EC'ssouthern enlargements, and an effort was made to includea regional dimension in the most important Communitypolicies. That is the case of the Guidance Section of theEuropean Agricultural Guidance and Guarantee Fund, in-troduced in the early 1970s in the aftermath of the MansholtPlan, and the European Social Fund which, although presentsince the founding of the Community, was significantlyreformed in 1972 and 1977 and had the regional objectiveexplicitly included. Finally, the importance of regional andinter-country disparities and the prospects of a deepeningof the integration process after the Single European Actprompted a major overhaul of the Community's structuralpolicies in 1988, as will be described more in detail in Section9.4.1. Table 9.2 gives an idea of the increasing importanceassumed by the structural Funds in the EC budget since the1970s.

9.3. Businesses' perceptions of regionalhandicaps and competitiveness

One important source of empirical information in assessingthe effects of policy measures is provided by the surveysconducted for the EC Commission among industrial and

Table 9.2The structural Funds and the EC budget, 1970-90

1970 1975 1980 1985 1990

MECU MECU MECU MECU MECU

RegionalSocialAgriculture

Total funds

Total EC budget

75,3157,9158,8

64,0 1,2 392,0

5448,4 100,0 6213,6

1,22,52,6

6,3

100,0

1 126,41 014,2

624,7

2 765,3

16 057,5

7,06,33,9

17,2

100,0

2 495,32 188,5

852,9

5 536,6

28 223,0

8,87,83,0

19,6

100,0

4 704,53321,91 449,0

9 475,4

46 808,7

10,13,13,1

20,2

100,0

Source: European Economy No 42; Court of Auditors' Annual Report, several issues.

217

sduquenoy


business services firms. Those surveys show that positiveexpectations as to the effects of the European single marketand EMU are widespread in the countries of the Community.But the positive expected effects are not evenly spread overthe regions. Graph 9.1 shows the results of a 1989 survey(IFO, 1989) where businessmen were asked to assess thelikely impact of the completion of the internal market upontheir companies and the regions they are located in. Thissurvey covered about 9 000 companies situated in threetypes of regions: lagging, declining industrial regions andprosperous (control) regions (see Box 9.3).

The companies located in the control (central and prosper-ous) regions see their own prospects as well as those of theirregion, after the completion of the internal market, morepositively than the companies in problem regions. The pros-pects for declining regions are seen, moreover, as less favour-able than those for lagging ones, although there is a consider-able degree of subjectivism in this assessment.

Similarly, prospects for the declining areas are viewed as lessfavourable than for the control regions in Belgium andGermany.4 However, for the Netherlands and the UK onecannot say the same, since in declining regions the firmsevaluate the impact upon both region and company morepositively than in control regions. In Spain, on the otherhand, firms have more optimistic views for the impact uponlagging regions than for the control regions. These resultsconfirm the suggestion of Section 9.1.2 that no uniformconclusions are likely to be valid for the spatial distributionof the economic benefits of EMU.

Further disaggregation of the survey data shows that theexpected benefits increase nearly always with firm size, par-ticularly in the lagging regions; the orientation of the firmtowards foreign markets — measured by the percentage ofturnover from exports — unsurprisingly produces a similareffect; and firms producing investment goods and, especially,business services, have better expectations than those pro-ducing intermediate and consumer goods. The latter resulthas strong regional implications and probably accounts formost of the differentiated expectations between lagging, de-clining and prosperous regions: the most dynamic sectorstend to be concentrated on the latter.

Behind the attitudes of enterprises towards the effects ofeconomic integration lies their assessment of a number of

factors that influence their ability to compete in a more openinternational environment. The IFO survey also provides adetailed insight on the relative importance of factors affect-ing firm and regional competitiveness, as evaluated by firmmanagers. Both factors which are country-specific and thosewhich are region-specific were listed, with the results sum-marized in Box 9.3.

Graph 9.1 represents the difference between the evaluationof those factors in lagging and declining regions and theirevaluation in the control regions. The exchange rate as afactor of competitiveness does not seem to affect firms situ-ated in problem regions more critically than those situatedin the control regions. This suggests a relatively neutralregional impact of monetary union, at intra-country level.The cost of credit and availability of risk capital, on theother hand, work to the disadvantage of the lagging regionswhereas the cost of labour works to their advantage, butnot so much to that of the declining regions. The impact ofEMU will be to improve the efficiency of financial markets,especially in the lagging countries. However, in Section 9.4.2it is also shown that financial integration may bring aboutsome disruption of the local credit markets, where smallentrepreneurs are totally dependent on the banking systemand that this effect may be a disadvantage to set off againstthe gain of greater overall monetary stability and efficiencyin the Financial sector.

Among the regional factors, it is interesting to note that thephysical distance to markets (consumers and suppliers) isnot perceived as a serious disadvantage in the lagging, andgenerally peripheral, regions. Instead, the lack of certaintypes of infrastructure (transport, communications, socialfacilities, education and training centres) are seen as makinga more important difference. In short, geographical disad-vantage is less important than 'economic' — and amendable— backwardness.

In this respect, the main difference between lagging regionsand the industrial declining ones is the availability of econ-omic and social infrastructure. These are at the head ofthe regional factors negatively affecting the lagging regions.Problem industrial regions, however, have inherited a muchmore developed infrastructure even if it is often outdatedand badly maintained. The absence of such established facili-ties, as mentioned in Section 9.1.2, is an obstacle to thecatching-up of lagging regions.

Note that in Belgium, Germany and the Netherlands there are no lagging(Objective 1) regions.

Finally, services to enterprises (legal, consulting, etc.) alsofigure highly among the relative factors of disadvantage inlagging regions and also to some extent in declining regions.

218


GRAPH 9.1: Enterprise assessment of factors of regional competitiveness

National factors

Cost of creditEconomic growth

Availability of risk capitalSector's outlookIndustrial policy

Other macroeconomic factorsExchange rate policy

Administrative proceduresIncome/corporate taxationLabour market regulation

Legal regulationsWage costs

Indirect labour costs

I Lagging regions f r i Declining regions

60 -40 -20 20 40 60

Regional factors

Proximity of trainingTransport network

Supply and cost of energySchool facilities

Servicing machineryCultural and social facilitiesProximity of high education

Leisure facilitiesCommunication systemsBanks/insurance/lawyers

Advertising and consultingBusiness culture

Social climateProximity of suppliers

Waste disposalCooperation of local authorities

Proximity of customersOther regional factors.

Cost of housingRegional policy incentives

Industrial sitesSupply of qualified labour

Local taxesSupply of unqualified labour

-60

fioie Figures represent the net balance of respondents, in percentages, making negative or positive assessments for the lagging or declining regions.

219

sduquenoy


9.4. Wider strategic considerations

9.4.1. The less favoured regions and economicunion

As the previous sections showed, there are no unambiguousresults for the EC less favoured regions as regards the deep-ening of the European integration process and EMU inparticular. The outcome will very likely depend on the com-bination of the policies implemented with the 'initial con-ditions' of each region. Policies should be geared to theenhancement of the economic and social cohesion within theCommunity which means, for the EC lagging regions, topromote their catching-up to the EC levels. Both the Com-munity and the Member States are committed to followpolicies aiming at economic and social cohesion: MemberStates gear domestic economic policy towards cohesion andthe Community supports its implementation through theintervention of the structural Funds, the EIB and otherfinancial instruments. This arrangement of goals, policiesand levels of decision was defined at the outset of the singlemarket programme, and is effectively aiming at a successfulconstruction of economic union within the Community.Indeed, this is the message of Article !30 (a) to (e) of theEEC Treaty.

The 'initial conditions' embrace a wide spectrum of factors,ranging from endowments to political attitudes. As regardsendowments, the initial conditions of the EC lagging regionsmay be synthetically described by a few quantifiable factors,the first of which is the simple indicator of GDP per headalready mentioned. Second, as the survey results suggest,come infrastructure and human capital. A third dimensionis industrial structure, including for some purposes financialstructure as well.

Less favoured regions have lower levels of GDP per headthan the core. Within the EC, GDP per head (in purchasingpower standards) in Greece, Ireland, Portugal and the GDR,as well as in most of the Spanish regions and the south ofItaly, lies under 70 % of the Community average. Amongthese regions there are — and will remain for a while — twodistinct groups: the first constituted by Ireland and thelagging regions of Italy and Spain, which are not far fromthat threshold, and the second by Portugal and Greece,who are well below and not much above one half of theCommunity average.5

Recent studies of these countries and regions can be found in Bliss andBraga de Macedo (1990), chapters by Vinals, Katseli and Braga deMacedo on Spain, Greece and Portugal. On Italy see Faini (1990). i

Although having in common lower levels of income perhead, the EC lagging regions have not displayed identicalgrowth performances. Taking the year of adhesion to theEC as a benchmark, the evolution of GDP per head varieswidely among Ireland, Greece, Spain and Portugal, the fourpoorer Member States. Convergence has not occurred uni-formly, neither through time nor space, suggesting that theadequacy of domestic economic policies play a crucial role.

The Italian Mezzogiorno was experiencing a catching-upprocess towards northern Italy's level of income when theTreaty of Rome was signed. The promotion of developmentin the Mezzogiorno had been made an explicit policy objec-tive for the Italian authorities at the end of World War IIand significant resources (both institutional and financial)had been allocated to it. This effort was continued andreinforced in the new framework brought about with thecreation of the EC. However, the catching-up process cameto a halt in the aftermath of the first oil shock and up to1987 the income gap relative to northern Italy actuallywidened. Explanations for the break in the convergenceprocess generally point to a slackening of the investmenteffort targeted at southern Italy. This was related to thedeterioration of profitability conditions due, inter alia, tothe reduction in wage cost differentials with the north. TheMezzogiorno was in risk of becoming a declining rather thana lagging region, without ever in recent times having beenprosperous. This reversal shows the narrow limits of struc-tural interventions which are not supported by a viablemicroeconomic environment.

In Ireland, a sound catching-up process has been evidentonly since 1985, after the successful implementation of afiscal adjustment programme. Until then, the growth per-formance of the Irish economy was very disappointing withGDP per capita relative to the Community average fluctuat-ing around 65%. The contrast between the recent periodand the first 12 years of membership is thus clearly apparent.Given the marked progress in nominal convergence whichoccurred in recent years and the growth momentum cur-rently experienced, the Irish economy seems quite favourablypositioned to face the challenges of the EMU process.

Accession to the EC did not yet prompt any durable catch-ing-up process in Greece. In the course of the 1980s theincome gap between Greece and the EC widened: GDP percapita (in purchasing power standards) declined from 58 %of the Community average in 1980 to 54% in 1989. Inthe most recent years the economic situation in Greecedeteriorated sharply, with the imbalances in the fields ofpublic finance, inflation and external accounts reaching un-sustainable levels. The 'initial conditions' of Greece vis-a-visEMU thus seem weak, calling for a period of rigorous

220


domestic stabilization in order to create the basis for a soundcatching-up process.

As regards the new members, both Spain and Portugalseem to have well exploited the opportunities offered by fullmembership of the EC. In the first four years after accession(1986-89) both countries experienced steady growth wellabove the Community average. The prospects are for thecontinuation of the catching-up process. Nevertheless, un-employment in Spain and inflation in Portugal continue tobe abnormally high. In Spain there is room for a moreeffective regional policy, especially in making the labourmarket less rigid. In Portugal, virtual full employment hasbeen preserved, but structural adjustment away from wide-spread State intervention has only begun in the last few years.In both countries, and in Portugal in particular, domesticeconomic policies have to secure effective progress in nomi-nal convergence so that the real convergence process not beput in jeopardy.

As seen above, the less favoured regions of the EC lie mainlyin its periphery. The concept of peripherality is first of all anotion of distance. In economic terms, this means that forthose EC regions both transport and communication costsdo matter. An extra time-distance penalty is suffered by theless favoured regions because of factors such as bordercrossings (intra and extra-Community), lack of direct motor-way links, sea crossings, etc. This may be calculated as afactor of proportionality between travel time and geographi-cal distance, with respect to the centre. The penalty variesby a factor ranging from 1.9 for Athens to 1.8 for Dublin,1.6 for Rome and Lisbon and 1.4 for Madrid. (Thus Athens'time-distance from the centre is 1.9 times its kilometre dis-tance). In addition there are the problems of peripheralitywithin countries; the figures given are for capital cities andtake no account of time taken to travel from, say, Galwayto Dublin, or Seville to Madrid which will add considerablyto the above penalties.

Transport and communication costs can be reduced (in somecases very significantly, namely in the field of telecommuni-cations) through the establishment of adequate infrastructu-ral facilities. As mentioned, poor countries and regions areinadequately endowed with basic infrastructure. The numberof kilometres of motorway per square kilometre in Greeceand Portugal is below one tenth and one fifth, respectively,of the EC average. As another example, the number oftelephone lines per thousand inhabitants in the less favouredregions of Spain is less then half of the EC figure while inthe GDR it stands below one fourth. The enhancement ofthe infrastructural capital stock helps attenuate geographicaldisadvantage, both directly and by increasing the competi-tiveness of the countries and regions.

The less favoured regions also have a low qualified humancapital stock largely reflecting the poor performance of thecountries' education systems. In Greece, total public expen-diture in education amounted to 2,9% of GDP in 1986while for the EC as a whole the figure is 4,8 %. Spendingper pupil in Greece is only 28 % of the EC average while inPortugal and Spain it amounts to 40 % and 51 % respect-ively. Compulsory education in the five EC less developedcountries ends earlier than in the core countries where therange is from the age of 15j in France to 18 in Belgium.The possibilities for increasing skills in these countries aretherefore very large. The upgrading of manpower means notonly an increase in the stock of knowledge (which willincrease the marginal efficiency of labour) but also makingpeople more open to innovation and to modern technologies.Labour productivity can thus increase significantly in thelagging countries.

The lagging countries and regions of the EC have a longtradition of migration from the south, west and east towardsthe central Member States. However, the situation haschanged significantly in the last two decades and the processof European integration has not been accompanied by anylarge-scale movement of labour, despite the wide regionaldisparities. Ireland is the only less developed country of theEC where net emigration is still taking place, mostly towardsthe UK. As regards the new members, Spain and Portugal,freedom of circulation and establishment for people willonly take place from 1993 onwards. Whether substantialmigration to the EC will materialize after 1993 is not clear.If the experience with Italy and Greece is followed, nosignificant outflows of labour from Portugal and Spain areto be expected. Moreover, there is no tradition of migrationfrom Portugal to Spain, unlike the relation between the Irishand the UK labour markets.

The determinants of migration are several. While traditionalviews emphasized the role of wage differentials as a causefor migration, current views stress that the absolute wagelevel of the potential migrant is decisive. Even in the presenceof an important wage differential an adequate level of thelocal wage would discourage emigration. Moreover, the soc-io-economic situation of rural areas in Europe, which werethe main source of emigration, has improved significantly,not least as a consequence of the implementation of Com-munity policies, the CAP in particular. Therefore, thereasons that were behind the rural exodus of the 1950s and1960s no longer prevail.

The situation of the GDR, as regards emigration, is a par-ticular one. The large outflows to the FRG which occurredin the final part of 1989 and beginning of 1990 reflectedboth the exploitation of the breach opened in the repressed

221


economy of the GDR and the uncertainties then prevailingregarding its future. Outflows began to decline when theprospects of German unification became clear. The eventualcontinuation of emigration from the GDR to the FRG willto a large extent depend on the success of the adjustment ofits economy which, in turn, will be shaped by the terms ofthe State Treaty. Given that a single currency now appliesin the German monetary union, the evolution of nominalwage differentials between the GDR and the FRG will becritical to the prospects for employment in the GDR.

Being small and open economies, the increase in welfare inthe EC lagging regions has to come from external trade. Theindustrial structure of those regions will be decisive for theirtrade pattern vis-a-vis the EC. A recent study carried out bythe Commission services6 analysed the different sectoralpositioning of each country vis-a-vis the completion of the1992 single market. As theory would predict, it is observedthat intra-industry trade is less developed in the Europeancountries with a lower level of income per head. In fact, intra-industry trade is less important in Portugal, and especially inGreece, than in other member countries. Spain and Ireland,and in some respect Italy as well, stand in an intermediatesituation.7

Greece clearly displays an inter-industry foreign trade pat-tern and relies on exploiting the comparative advantagestemming from low wage costs. As far as Portugal is con-cerned, inter-industry trade is concentrated on the strongersectors, but some degree of intra-industry trade is alreadytaking place in sectors which have a high technologicalcontent, and which host the export-oriented branches offoreign enterprises. Spain differs from the other southerncountries in that the strong points of its external trade donot belong exclusively to the traditional industries but alsoinclude highly capital-intensive sectors utilizing highly quali-fied manpower. Considered as a whole, Spanish industry ismuch more engaged in intra-industry trade with the otherEuropean countries than Portugal or Greece.

The trade pattern of Ireland is clearly different and betterpositioned as far as modern sectors are concerned. In fact,this country is in a relatively favourable position in high-technology sectors such as informatics, telecommunicationsequipment, precision material and pharmaceuticals. How-ever, these- sectors are largely dominated by foreign firms,

'Les £tats membres face aux enjeux sectoriels du tnarche interieur', inEuropean Economy No 44 and Social Europe.A similar conclusion is presented by Neven (1990) and by the countrystudies on Spain, Greece and Portugal contained in Bliss and Braga deMacedo{1990).

which renders Irish industry rather dualistic and thus morevulnerable to the changes in the strategies of foreign inves-tors.

It is possible to identify two scenarios for the dynamicadjustment of these countries' industrial structure to thesingle market. A first scenario would consist in the specializa-tion in inter-industry trade in those sectors where the lessdeveloped regions currently have comparative advantages;in the second scenario the development of industry in thoseregions would follow the pattern of the more advanced ECcountries with specialization in intra-industry trade. Nat-urally, a combination of these two scenarios is also possible.There are no grounds to conclude that the intra-industryscenario is uniformly better than the inter-industry one, orvice versa. Nevertheless, specialization in traditional indus-trial sectors (e.g. textiles) must be accompanied by an 'up-grading' in quality to avoid increased exposure to the compe-tition of newly industrializing countries from outside theCommunity. Moreover intra-industry specialization is morecompatible with EMU (see Chapter 6) since it decreases thelikelihood of asymmetric shocks and adjustment problems.Foreign investment will be called to play a crucial role inboth scenarios. In order to secure inflows of foreign capital,peripheral regions have to provide conditions for the econ-omic efficiency of investment. In this respect, the perform-

9.2 : Spain and Portugal — trade and investment flowswith EUR 10

1963

—— Spain

1968 1973 1978 1983 1988

Portugal Spain Portugal

222

sduquenoy


ance of the newest EC members — Spain and Portugal— is quite striking. Since 1986 one has observed not onlyan acceleration and diversification of the trade flowsbetween the EC and Spain and Portugal but, especially,a boom of foreign direct investment in the two countries(see Graph 9.2).

These initial conditions relative to GDP per head, infrastruc-ture, human capital and industrial structure are the factorsthrough which the economic impact of the completion ofthe single market will be channelled. Not surprisingly, theCommunity policies addressed to the adjustment of thelagging regions to the single market, notably the structuralFunds, are directed to these factors. Indeed, the existingCommunity structural policies were designed in the contextof the single market programme, and so are largely gearedto the achievement of economic union. The Communitystructural policies were significantly enhanced with the adop-tion of the Single European Act and the 1992 single marketprogramme. The structural Funds were reformed in 1988and their financial allocations doubled in real terms from1987 up to 1993. The interventions of the structural Fundswere concentrated on five objectives, of which three have aregional dimension: Objective 1, promoting the catching-up of the regions whose development is lagging behind,Objective 2, for the reconversion of industrial zones in de-cline, and Objective 5b, promoting rural development. Theless developed regions of the EC, excepting (for the timebeing) the GDR, are fully covered by these objectives, andare therefore the largest beneficiaries. Objective 1 regions —principally Greece, Ireland, Portugal, 70 % of Spain, theItalian Mezzogiorno and Northern Ireland — account forabout 60 % of the total financial amount allocated to thestructural Funds for the period 1989-93 (ECU 60,3 billion,1989 prices). The reform of the Funds began to be im-plemented in 1989 with the adoption of the Communitysupport frameworks within which the interventions of thestructural Funds for the period 1989-93 were largely allo-cated (see Box 9.4). According to the 1988 reform of theFunds, decisions about the post-1993 period are to be basedon a global evaluation of the reform to be carried out in1991.

The Community structural policies are thus already set forthe period up to 1993 which overlaps with Stage I of EMU.The rationale for these policies, as reflected in the Com-munity support frameworks, was primarily to allow theenhancement in the adjustment of the less favoured regionsto the new economic environment that the 1992 single mar-ket will bring about. As a consequence, these policies shouldbe continued along with the completion of full economicunion within the EC.

9.4.2. The less favoured regions and monetaryunion

The preceding section dealt, in a concise way, with the issueof the Community's less favoured countries and regions inrelation to the construction of economic union. The econom-ies of these countries may be hurt before they are helpedaccording to the U-shaped curve mentioned in Section 9.1.To contribute to leading those regions to the upward slopingpart of the curve may involve a significant lowering of tradebarriers. It may also require Community accompanying poli-cies

The question now arises whether the superimposition ofmonetary union will add any spatial effects to those ofeconomic union. Are there any particular effects on laggingcountries and regions which directly derive from monetaryunion? Even if the answer were clear-cut for the case whereeffects of currency areas take for granted that an economicunion is already performing smoothly, that is not the rel-evant situation. As far as the EC is concerned, the construc-tion of economic union and monetary union will be parallelprocesses. Will monetary unification amplify the impact onthe less favoured regions of the real shocks stemming fromthe internal market? Will they be attenuated? Will there beany specific monetary shocks? Two domains where this issueis relevant have been analysed in Chapters 5 and 6, namelythe abandoning of the nominal exchange rate and the elimin-ation of hidden forms of taxation for the financing of publicdeficits.

Both domains are not particular to lagging regions but theymay be felt there with more intensity. In fact, besides ahigher possibility of being hit by an asymmetrical externalshock, those regions have to adjust to the real shocks stem-ming from the completion of economic union; this adjust-ment may call for the utilization of the nominal exchange-rate instrument. Such a viewpoint relies on the acceptancethat nominal instruments are adequate to cope with realshocks which is not accepted without dispute. Besides, theactive management of the exchange rate as a policy instru-ment has been, de facto, relinquished by most of the ECperipheral countries. It is perhaps the perception of thissituation that lies behind the fact that firms do not view theexchange rate as a major constraint for the development oflagging regions, as the survey quoted in Section 9.3 abovedisplays.

However, as argued in Chapter 6, abandoning the nominalexchange rate is least costly for countries that are initiallyin equilibrium (including the exchange rate) conditions andonly have random shocks to absorb. The situation ofcountries whose economies are initially very distorted is

223

sduquenoy


quite different and these may need structural adjustments tothe exchange rate as part of their general policy strategy.Greece may be now in this latter category whereas the restof the EC is in the former group. Portugal may find itselfin an ambiguous position in between.

The issue of abandoning the nominal exchange rate effec-tively concerns mainly three countries which are eligiblefor structural interventions, namely Greece, Ireland andPortugal. The less-favoured regions of Spain and the ItalianMezzogiorno are already deprived of an autonomous ex-change rate policy given that they belong to a (national)currency area. In an identical situation is now the GDR,given its monetary union with the FRG (see Annex B).

The adoption of a hard-currency policy with eventuallyfixing the exchange rate requires sound conditions of dom-estic monetary stability which, in turn, are associated withthe public finance situation. Ireland is a member of theexchange-rate mechanism (ERM) of the EMS since its cre-ation and a successful adjustment programme of its publicfinances has helped make it possible to avoid any unilateraldevaluation of the Irish punt since 1986. Recent develop-ments as well as the forecast of Ireland's external accountsdo not warrant any departure from the existing exchange-rate policy. Spain joined the EMS in mid-1989; althoughbenefiting from the wider band of fluctuation, the pesetahas been close to the upper band due to capital inflowsreflecting the fact that domestic policies continue to begeared to monetary and financial stability. Ireland and Spainare therefore at least half-way to locking their exchangerates, with the adjustments in relative prices having to rely,more and more, on wage developments. With the Germanmonetary union, the economy of the GDR is covered by theDeutschmark exchange rate policy; as regards the adjust-ment of the GDR within an united Germany, the role of theexchange rate was confined to the bilateral parity adoptedfor monetary unification (see Annex B).

Portugal and Greece still follow autonomous exchange-ratepolicies. Portugal has adhered, for more than a decade, to apre-announced crawling-peg regime reinforced by infrequentdiscrete devaluations, the last one dating from mid-1983.This policy, given the low degree of indexation of the econ-omy, especially wages, has aimed at the maintenance ofcompetitivity and allowed adjustment to domestic and exter-nal shocks. The balance of payments situation is nowcomfortable, and the country is faced with a similar dilemmato Spain's in the setting of the appropriate long-term levelof the exchange rate in the prospect of ERM or EMUmembership. The current account deficit is financed bystrong capital inflows that respond to the tight internalmonetary conditions as well as to the investment opportunit-

ies brought about by Portugal's integration in the EC. Mean-while, the persistence of domestic inflation leads to a realappreciation of the escudo. It is hard to assess the sustain-ability of these trends, but the adoption of a firm pluri-annual fiscal adjustment programme would provide thenecessary credibility for an early exchange-rate commitment.

Greece has been following a policy of managed floating, andis the peripheral country that has experienced the strongestoverall devaluation since 1979. It is also the country withthe highest level of inflation and public deficit, though thepublic debt to GDP ratio is not as high as Italy's, forexample. Some arguments have been put forward against anearly entry into the EMS: for example, strong disparitiesbetween Greece's economic and trade structure and those ofits EC partners increase the possibility of asymmetric shocksthat would require the exchange-rate instrument. Like Ire-land, Portugal and Spain, Greece is committed to eliminatingrestrictions on capital movements by 1992.

Just like abandoning of the exchange-rate instrument, theissue of renouncing hidden or implicit forms of taxation isnot limited to the less favoured countries and regions, aswas stressed in Chapters 5 and 6. Nevertheless, the southernStates of the Community are those where this phenomenontakes higher proportions (see Graph 5.5). On the other hand,the lagging countries are more expansionary-prone given thechallenges of the catching-up. Budgetary policy is less flex-ible in the less developed regions with the consequence thatthe temptation for surprise inflation as well as for otherforms of implicit taxation is higher. Finally, it should beborne in mind that, in their present form, the transfers ofthe Community's structural Funds have a direct impacton the national budgets, given that they co-finance publicexpenditure, and capital expenditure in particular. Havingthe nature of 'matching-grants1 and having to respect theadditionality principle, the transfers from the structuralFunds are a constraint for the national (including regional)budgets' management. Such a constraint may have signifi-cant proportions in the national (and regional) budgets ofthe lagging countries given that they are the largest recipientsof financial support from the Community and given therelative magnitude of the received transfers (in the order of3 % of GDP, see Box 9.4). The importance of this constraintfor the management of national public expenditure wouldbe amplified in a framework of an enhanced role of centralpublic finance in the Community (see Section 6.7). This factgives a certain specificity to the lagging regions as far as therenouncement of implicit taxes is concerned, calling for theimplementation of flexible mechanisms to encourage thenational effort to reduce public deficits, e.g. through thewidening of the eligibility criteria for structural Fund sup-port.

224


Aside from abandoning nominal exchange rate changes andeliminating hidden forms of taxation, a specific effect ofmonetary union on less favoured regions and countries maybe stressed. This is related to the underdevelopment of thefinancial market that often accompanies excessive publicdeficits and the resort to hidden taxes. One may refer to itas the crowding-out of weaker local borrowers. What is atstake, again, is the domestic financial sector and the waysof financing public-sector deficits. Very often, the monetaryauthorities of countries with less performing financial sys-tems have to resort to direct methods of monetary control(credit rationing, administrative setting of interest rates, etc.)as the sole alternative for the management of domesticliquidity growth. Such direct controls impinge upon theprivate sector which is crowded out from the credit market.Recent theories and empirical studies (see Branson, 1990)have shown that local financial intermediaries are able toenjoy some monopoly power due to information asymmet-ries and sunken costs (e.g. lasting credit relationships withborrowers). In such a hierarchical structure of the financialsystem, local banks stand between international credit mar-kets and domestic borrowers, impeding direct contact be-tween them. Local lenders have an advantage over foreignbanks because they trade instruments with costly local know-ledge (e.g. the monitoring of clients), which can be con-sidered non-traded internationally. With full monetaryunion local banks will loose this monopoly power assumingthat borrowers will have direct access to foreign banks, eitherlocally established or not, after the opening of the domesticfinancial market. Beside this credit availability effect, theborrowers of peripheral countries will also benefit from thelevel of interest rates which very likely will stay below those,prevailing in the region before monetary unification. Thesetwo effects — availability and lower price — will representa clear benefit for the borrowers of lagging regions.

However, local banks may be disrupted because foreignbanks will rapidly seize the best segment of the marketleaving the local banks with the less performing borrowers.Besides, the new and open financial market will develop ina context of tighter financial discipline and stricter rulesfor credit granting which may crowd out the (marginal)borrowers of the lagging regions. The net effect is ambigu-ous. For instance, it is not clear that a small enterprise in,say, the north of Portugal, which currently can only dealwith its local bank to obtain credit, will be better off whenit will have access to, say, a large British bank as well. Withthe local bank the small enterprise would eventually getsome credit, but with the large foreign bank it may notbecause it does not meet the 'eligibility' standards of thebank. Naturally, the situation will disappear once the localbank has recovered and adapted to the new conditions. Theeffect of monetary union for the credit market of a laggingregion could well be as ambiguous as was the case with

economic union. The possible crowding out of weaker localborrowers calls for the setting of adequate Community poli-cies addressed to the adjustment of the local financial sector.Such policies would aim, inter alia, at securing credit avail-ability for the 'marginalized1 borrowers of lagging regions(e.g. funding of regional development societies, risk capitalsocieties, etc). In the absence of such policies, the nationalauthorities of the lagging countries would be tempted tooppose the opening of the financial markets through themaintenance of domestic regulatory practices. Such practiceswould go counter to the process of monetary union, notallowing the benefits of EMU to be fully reaped.

9.4.3. Regime change and regime structure

The bottom-line issue for the lagging countries and regionsis how to overcome their development bottlenecks and catchup as fast as possible with the more prosperous parts of theCommunity. This objective translates into a need to sustain,for a long period, higher-than-average rates of real GDPgrowth. This in turn calls for a stable, though dynamic,economic environment, and the avoidance of unrealisticgoals and stop-and-go policies.

Regime change. Given the ambitious nature of the catching-up objective, the appropriate concept seems to be that of acredible and comprehensive regime change.

A comprehensive regime change is a dynamic process wheremultiple synergies are at work between expectations andreal economic variables, and which may radically alter thefeasible dynamic path of the economy. For example, theinsertion of a country into an economic space as the Euro-pean single market can have such an effect upon the expec-tations and strategies of domestic and international investorsthat investment levels are achieved which otherwise mighthave required the maintenance of relatively much lower realwages. This effect is in fact currently observed in Spain andPortugal.

If a comprehensive and credible regime change is the keyfor the convergence of lagging countries and regions to ECstandards, how far can one characterize the strategic aspectsof that change, in terms of policies at the EC and nationallevels?

There is a widespread consensus on some elements, such asa unified and open European single market; national policiesof enterprise reform, aimed at the introduction of marketdiscipline mechanisms in all sectors of the>economy; labourmarket regulatory reform, aimed at more flexibility, ef-ficiency and employment; modernization of tax systems; and

225

sduquenoy


stability-oriented and transparent monetary and budgetarypolicies. As far as the latter are concerned, an importantquestion is of course the distribution of responsibilities be-tween the EC and national level but, more generally, allthese aspects must be considered within the perspective ofEMU.

In the previous section the abandoning of the nominal ex-change rate as a policy instrument as well as the eliminationof hidden forms of taxation for the financing of publicdeficits were already stressed. The peculiarity of underdevel-oped local financial markets was also mentioned. In all cases,it was apparent that the prescription of regime change wouldnot differ greatly between the countries and regions in ques-tion.

The most clear-cut and extreme examples of regime changeare currently observed in Central and Eastern Europeancountries. By comparison, the scale of the changes still tobe made by Spain, Portugal or Greece is, of course, muchsmaller. These Community countries have already substan-tially changed their political and economic systems in thelast 15 years or so. But credibility is an important elementof any regime change, which is why a binding integrationinto an international structure like the EC can play a vitalrole. This was certainly appreciated by those three countrieswhen they decided, in the 1970s, to apply for membership.Without the influence of this 'external liberalization con-straint' upon expectations, desirable reforms might not havebeen feasible and the policy-maker's room for manoeuvremight in fact have been reduced by a lack of domesticcredibility.

Regime structure. Important as these common desirable el-ements of regime change may be, there are also some keyfeatures of regime structure that separate the different lag-ging countries and regions of the Community. Since theseinvolve Community policies, it is necessary to understandthe patterns underlying these differences in regime structure.

The three issues in question, in fact interdependent, are:(i) labour migration: how much should be expected and

considered desirable?

(ii) budgetary redistribution: how much interregional redis-tribution is appropriate?

(iii) the exchange rate: at what point should flexibility bereduced and, eventually, abandoned in EMU?

The initial conditions of the lagging regions of the Com-munity differ most clearly according to whether they aresub-regions of nations (like parts of Italy and Spain, andnow also East Germany) or nation-regions like Greece,Portugal or Ireland. The former group experience high in-ternal migration and fiscal redistribution and zero exchangerate flexibility between regions; the latter group experiencerelatively little external migration and (EC) redistribution,while retaining at least some exchange rate flexibility.

With the progressive integration of the Community, theseinitial conditions gradually change, and the dynamic aspectsof regime structure should also be taken into account. Thisis portrayed schematically in Table 9.3, with identificationof four successive regimes: international relations, the ECof the 1980s, representing a transitional stage of integration,the EC under EMU conditions, and a mature federal State(notably one with considerable unity of language, cultureand history).

From the standpoint of the public authorities a key differ-ence between the three variables is that labour migration isa behavioural factor, assuming that the legal freedom tomigrate is granted. The budgetary and exchange rate vari-ables are policy instruments that may be adjusted, in reactionto the behaviour of the economy, or addressed to specificpolicy objectives.

Normatively all three variables have to be viewed with cir-cumspection. None can be considered simply desirable orundesirable; each have their value or utility but also theirhazards.

Table 9.3Evolution of aspects of regime structure with the degree of integration

Variables affectinginterregional relations

Labour migrationBudget redistributionExchange-rate flexibility

Type of regime

Internationalrelations

littlenone

considerable

EC of the1980s

somesomesome

EC with EMU

moremorenone

Maturefederal State

muchmuchnone

226

sduquenoy


As regards migration, its value for enhancing economicefficiency and welfare and as an adjustment variable forattaining economic equilibrium is widely acknowledged.However, at some point it can become socially disruptiveand inefficient, notably where local regimes perform sopoorly that mass emigration is a feature of its collapse (asEast Germany has just illustrated). The propensity to mi-grate clearly is greater when there are no language frontiers.Within language communities migration can quite easily betriggered by differences in local fiscal capacities, which isan important reason for the budgetary redistribution andequalization mechanisms in mature federations. These con-siderations are in fact observed in the reunification of Ger-many, with the need to combat 'catastrophic migration'emerging with the extension of the normal mechanisms offiscal federalism. They are also observed in the interregionalredistribution mechanisms of Spain and Italy, which, lessdramatically of course than in Germany today, have longbeen related to concern over migration and demographictrends as between regions.

The absence of these factors in the migration propensity ofthe nation-regions of the Community is an important reasonwhy even in EMU it would not be expected that the Com-munity's redistribution function replicate that observed inmature federations. That is at least the case at present,although the categories in question are by no means absoluteor fixed. Thus Irish emigration is quite significant, facilitatedby language. It is also not so many years ago that Portugaland Greece experienced structural emigration of consider-able proportions. These three small nation-regions in fact

benefit from redistribution from the Community on a scalethat is qualitatively greater than for any other Member State.

While budgetary redistribution can thus serve efficiency aswell as equity objectives, there are also hazards to beavoided. These are not in principle different from the hazardsof any redistribution policy, and concern the risks of weaken-ing incentives for production activity and creating a Statewelfare dependency. At the level of the region such hazardscan operate either at the level of the wage bargaining system,which may be inclined to neglect regional productivity differ-entials, or at the level of local government, where the avail-ability of ample external finance may fail to combine withefficient resource allocation. Community structural policyhas recognized the existence of such hazards, and assistanceis therefore directed towards promoting productive invest-ment, human capital formation and infrastructures directlylinked to economic development.

The exchange rate instrument, similarly, has its uses andhazards. For relatively underdeveloped economies, and areasin the course of eliminating major market distortions, adegree of exchange rate flexibility may be particularly im-portant. The development process must not be blocked bya structurally overvalued exchange rate. On the other hand,repeated recourse to devaluation becomes itself no morethan an illusion in relation to economic realities, and usuallyjust a means of accommodating a high rate of inflation. Inparticular, as Chapter 6 has warned, persistent real exchangerate depreciation does not appear, in the EC at least, to buyfaster longer-term growth.

227

sduquenoy


Box 9.1: Peripherality in the Community of 12

Peripherally is related to both geographic distance and thevolume of economic activity in different regions. The latter ismeant to be a proxy for market opportunities, the availabilityof inputs and the various types of external economies.

A study published by the EC Commission in 1988 used thefollowing formula:

_

P. = 2^

DJJ is the distance between regions i and j, and is represented bythe shortest road link between the largest cities or towns ineach region. Special treatment is given to islands, taking intoconsideration ferry and shipping costs

I

This ind

)- is defined as - i

ex has been calculate!

k / area of region (sq km)V 71

d with 1983 data for 166 NUTS IIK4 regions of the Community of 12, and its distribution is shown

J +ij

_ 1 in the annexed map,Dii mediate

with regions classifiedand peripheral. The

into central, inter-adjacent countries which

jî members of the Community were also takenare pot

into account, afterallowing for customs barriers.

where M- is the level of economic activity inmeasured by regional GDP in ecus.

region i, and is

Shares of central, intermediate and peripheral regions,1 1983

Centralinnerouter

Intermediate

Peripheralinnerouter

EUR 12

1 See map.

Number of

442519

62

601941

166

Area

10,05,05,0

33,9

56,115,140,2

100,0

Popul.

31,021,79,3

35,7

33,313,020,3

100,0

GDP

Total

37,9 35,628,2 25,4

9,7 10,2

37,0 37,0

24,5 27,411,6 10,412,9 17,0

100,0 100,0

Employment

Agric.

11,66,35,3

31,656,812,244,6

100,0

Manuf.

37,924,713,2

40,0

22,19,2

12,9

100,0

Serv.

38,828,610,2

36,5

24,710,614,1

100,0

(EUR 12= 100)

Unemployment

Total

31,521,010,5

29,3

39,316,023,3

100,0

Young

25,816,69,2

29,4

44,817,727,2

100,0

Worn.

29209

3139,16,23,

533

2

312

100,0

Source .-Keeble el al. (1988).

228


EUR 12 regional peripherality indices, 1983Contours as % of EUR 12 average

> 150 inner central120-150 outer central70-120 intermediate60-70 inner peripheral< 60 outer peripheral

229

sduquenoy

sduquenoy

sduquenoy

sduquenoy


Box 9.2: Regional disparities in per capita GDP and unemploy-ment rates: comparisons between the EC, the USA and Canada

General remarks

Indicators of regional disparities are strongly influenced by thedegree of regional disaggregation: the greater the degree ofdisaggregation, the more pronounced are the regional dispar-ities. Comparisons between the EC and Canada should betreated with particular caution, because of the latter's muchsmaller population and the small number of regions into whichit is divided.

Regional disaggregation

The regional disaggregation used for this exercise is set outbelow. EC: 60 regions, Denmark, Greece, Ireland, Luxembourgand Portugal each being considered as one region. The otherMember States are divided into NUTS I regions: 11 each forGermany, Italy and UK, 3 for Belgium, 7 for Spain, 8 for Franceand 4 for the Netherlands. USA: 51 regions — 50 states plusthe District of Columbia. Canada: 11 regions — 10 provincesplus one region comprising the Yukon and North-West Territor-ies. Because of lack of data, the latter region is excluded fromthe comparison of unemployment rates.

GDP per capita

Ratio of top tobottom deciles'

Coefficient ofvariation2

Theilcoefficient

EC (60 regions, 1986 data in PPS3)EC (12 Member States)Germany (11 Lander)USA (50 states + D. of Columbia, 1986)4

Canada (10 provinces + 1 region, 1984)5

2,62,01.61.72,2

25,2%17,6%13,7%16,1%20,1%

0,01350,00710,00370,00510,0086

Deciles defined in terms of population. Averages for top and bottom deciles obtained by Hnear interpolation. In the EC the regions covering the top decile are: Nord-Ouest, Emilia-Romagna and Lombardia (Italy), Bremen and Hamburg (Germany), Noord-Nederland, Bruxelles and lie de France, The regions covering the bottom decile are: Portugal, Greece, Surand Centre (Spain) and Ireland.Weighted.Source: Eurostat, Statistical indicators for the reform of the structural Funds, November 1988.Source: US Department of Commerce, Survey of current business. May 1988.Source: Ministry of Supply and Services, Canada Year Book I9SS

Unemployment rates

Ratio of top tobottom deciles'

Coefficient ofvariation2

Theilcoefficient

EC (60 regions, averages 1986-88)3

EC (12 Member States, averages 1986-88)Germany (11 Lander, averages 1986-88)USA (50 states + D. of Columbia, averages 1984-86)4

Canada (10 provinces only, averages 1984-86)5

5,14,012,92,32,1

46,9%37,4%33,0%22,8%25,6%

0,04280,02760,02240,01110,0135

Deciles defined in terms of labour force. Averages for deciles obtained by linear interpolation. In the EC the regions covering the bottom (best) decile are: Rheinland-Pfalz, Hessen,Bayern and Baden-Wurttemberg (Germany) and Luxembourg (Grand Duchy). The regions covering the top (worst) decile are; Sur, Canarias, Este, Noreste, Centre and Madridf^nflinl Pamnnnin anH ^flivfoonn fl1alv\ anH NnrHpm Irplanii /I 1K\(Spain), Campania and Sardegna (Italy) and Nor hem Ireland (UK).

5 Weighted.3 Source: Eurostat, Statistical indicators for the reform of the structural Funds, November 1988.4 Source: US Department of Commerce and Census Bureau, Statistical Abstract of the United States 1988.s Source: Ministry of Supply and Services, Canada Year Book 1988.

230

sduquenoy


Box 9.3: Company view on the effects of the internal market

The Directorate-General XVI of the Commission of the Euro-pean Communities commissioned a study whose main purposewas to identify the factors that either limit or promote expansionin terms of employment and production, by firms in laggingregions and in industrial regions in decline. To carry out thistask, the IFO institute of Munich organized a large-scale surveycovering about 9 000 companies in industry and business servicesin 45 problem regions, and for the purpose of comparison alsoin 10 prosperous regions (control regions).

The questionnaire asked firms to assess the incidence of differentfactors of competitiveness (regional or national) upon their firmsas positive, negative or neutral. The same was asked aboutthe impact of the completion of the European internal market.Tables A, B and C are based on their replies, where an index of100 would correspond to a positive assessment by all firms, and- 100 a unanimous negative assessment.

Table ACompany assessment of the effect of the completion of the Euro-pean internal market on company1

Types of region

Lagging Declining Control

Belgium — 25,7 28,7Germany — 12,2 21,4Greece 1 1 ,2 — —Spain 15,1 -1.6 8,6France 21,8 39,1 53,0Ireland 37,8 — —Italy 27,9 — 36,5Netherlands — 22,0 10,9Portugal 2,3 — —United Kingdom 21,1 24,1 -23,4EEC (sample) 17,4 17,9 25,1

Sourre:!FO(l989).

Table BCompany assessment of the effect of the completion of the Euro-pean internal market on regions'

Types of region

Lagging Declining Control

Belgium — 5,0 22,4Germany — 6,8 30,0Greece 13,4 — —Spain 34,5 10,5 20,9France 34,8 35,2 43,1Ireland 14,2 — —Italy 7,1 — 19,3Netherlands — 22,0 11,9Portugal 12,3 — —United Kingdom 4,4 8,1 -19,2EEC (sample) 14,2 9,4 23,4

Source: IFO (1989).

Table C

Company assessment of factors of regional competitiveness1

National factorsExchange rate policyIncome/corporate taxationCost of creditAvailability of riskcapitalGeneral economicgrowthSector's outlookWage costsIndirect labour costsLabour market regu-lationIndustrial policyLegal regulationsAdministrative pro-ceduresOther macro factors

Regional factorsProximity of customersProximity of suppliersBusiness cultureBanks, insurance, law-yersAdvertising and con-sultingServicing machineryTransport networkSupply and cost of en-ergyCommunication sys-temsWaste disposalIndustrial sitesCultural and social fa-cilitiesLeisure facilitiesSocial climateCost of housingSchool facilitiesSupply of qualifiedlabourSupply of unqualifiedlabourProximity of trainingProximity of highereducationRegional policy incen-tivesCooperation of localauthoritiesLocal taxesOther regional factors

Source: IFO (1989).

Lagging

-21

-70-98

-18

4434

-42-79

-621

__ T

-36~9

-512210

32

149

16

-18

29-21-3

-13-13

17-18-3

-18

14-15

7

1

-17-32

3

Declining

-7

-64-32

-8

6855

-53-99

-670

-22

-31-5

684511

52

244270

18

74-3

10

1621211825

-29

1521

42

2

0-57

0

Control

-20

-76-45

5

7246

-79-120

78

10-27

-38-2

573631

60

404766

30

58-12-7

241936

-1638

-25

-237

40

3

-10-41

6

231

sduquenoy


Box 9.4: Less favoured regions and Community structural policies for the period 1989-93

Public expenditure considered in the Community support frameworks (CSF)

Regions

in MECU, 1989 prices

Greece (entire country)Ireland (entire country)Portugal (entire country)Spain (70% of the country)Italy (Mezzogiomo)

in% of regional GDP

Greece (entire country)Ireland (entire country)Portugal (entire country)Spain (70% of the country)Italy (Mezzogiomo)

Total StructuralFunds

12 995 7 1956126 3672

14026 736816507 977914062 7583

5,2 2,93,8 2,36,6 3,52,0 1,21,5 0,8

National finance Communityrequirement loans

(3) (EIB, ECSC)

5 802 1 4102 454 5606 658 2 8056 728 2 2066 479 1 475

2,31,53,10,80,7

Note: Furthermore, these regions will benefit front the programmes adopted under Community initiatives which are not included in ihe CSFs. The GDR will benefit from a specificprogramme of structural Fund assistance for a period starting with the German political unification and ending in 1993. From this date on the GDR will be considered with theother EC countries.

CSF public expenditure — Breakdown by main categories in %

InfrastructureAids to product, investmentAgricultureManpowerRegional programmesOthers

Total

Portugal Spain Italy

27,3 53,1 47,317,0 9,9 29,011,9 14,0 8,328,0 22,7 14,81 5,6 ' >0,2 0,3 0,6

100 100 100

Ireland Greece

17,1 31,326,5 7,024,5 13,031,2 13,7

1 34,50,6 0,4

100 1001 Included in the other categories.

232

sduquenoy


References

Balassa, B. (1961), The theory of economic integration, Rich-ard Irwin, New York.

Bliss, C. and Braga de Macedo, J. (1990), Unity with diversitywithin the European economy: the Community's southern fron-tier, Cambridge University Press.

Borts (1960), 'The equalization of returns and regional econ-omic growth', American Economic Review, 50, pp. 319-347.

Braga de Macedo, J. (1990), 'External liberalization withambiguous public response: the experience of Portugal', inBraga de Macedo and Bliss (eds), Unity with diversity withinthe European economy: the Community's southern frontier,Cambridge University Press.

Branson, W. (1990), 'Financial market integration, macro-economic policy and the EMS', in Braga de Macedo andBliss (eds), Unity with diversity within the European economy :the Community's southern frontier, Cambridge UniversityPress.

Bye, M. (1958), 'Localisation de 1'investissement et commu-naute economique europeenne', Revue economique.

Buigues, P., Ilzkovitz, F. and Lebrun, J. F. (1990), 'Les Etatsmembres face aux enjeux sectoriels du marche interieur',European Economy No 44 and Social Europe.

Clark, C., Wilson, F. and Bradley, J. (1969), 'Industriallocation and economic potential in Western Europe', Re-gional Studies, 2.

Cohen, D. and Wyploz, C. (1989), 'The European monetaryunion: an agnostic evaluation', CEPR Discussion PaperNo 306, London.

EC Commission (1990), 'The economic and financial situ-ation in Germany', Economic Papers, forthcoming.

Faini, R. (1990), 'Regional development and economic inte-gration: the case of southern Italy', paper presented at theconference 'Portugal and the internal market of the EC',Lisbon.

Giersch, H. (1949), 'Economic union between nations andthe location of industries', Review of Economic Studies, 17,pp. 87-97.

IFO-Institut fiir Wirtschaftsforschung (1989), An empiricalassessment of factors shaping regional competitiveness in prob-lem regions, Munich.

Katseli, L. (1990), 'Economic integration in the enlargedEuropean Community: structural adjustment of the Greekeconomy', in Braga de Macedo and Bliss (eds), Unity withdiversity within the European economy: the Community'ssouthern frontier, Cambridge University Press.

Keeble, D., Offord, J. and Walker, S. (1988), Peripheralregions in a Community of 12 Member States, Commissionof the European Communities, Brussels.

Krugman, P. (1989), 'Increasing returns and economicgeography', NBER Working Paper, Washington.

Krugman, P. and Venables, A. (1990), 'Integration and thecompetitiveness of peripheral industry', in Braga de Macedoand Bliss (eds), Unity with diversity within the Europeaneconomy: the Community's southern frontier, CambridgeUniversity Press.

Magnifico, G. (1973), European monetary unification, Mac-Millan, London.

Molle, W. (1989), 'Will the completion of the internal marketlead to regional divergence?', paper presented at the confer-ence 'The completion of the internal market', Kiel, June1989.

Myrdal, G. (1957), Economic theory and underdeveloped re-gions, London.

Neven, D. (1990), 'EEC integration towards 1992: somedistributional aspects', Economic Policy, April 1990

OECD (1989), Regional policy developments in OECDcountries, Paris.

Perroux, F. (1959), 'Les formes de concurrence dans lemarche commun', Revue d'economie politique, 1.

Stahl, H. M. (1974), Regionalpolitische Implikationen einerEWG-Wahrungsunion, Tubingen.

Scitovski, T. (1958), Economic theory and and Western Euro-pean integration, Stanford University Press.

Vanhove, N. and Klaassen, L. H. (1980), Regional policy: aEuropean approach, Saxon House.

233

sduquenoy

sduquenoy


Vinals, J. et al. (1990), 'Spain and the EC cum 1992 shock', Williamson, J. (1965), 'Regional inequality and the processin Macedo and Bliss (eds), Unity with diversity within the of national development: a description of the patterns', Econ-European economy: the Community's southern frontier, Cam- omic development and cultural change, 13, pp. 3-45.bridge University Press.

234

Chapter 10. National perspectives on the costs and benefits of EMU

Chapter 10

National perspectives on the costsand benefits of EMU// is understandable that the question is often put, for EMUas well as for the internal market, what the distribution ofbenefits and costs is likely to be by Member State.

The approach to answering this question taken in this study isto consider how the sources of benefits and costs (as listedabove, in Box 1 or Table 1.1) relate to the initial situation ofthe different Member States. This is done in the first part ofthe short country sections that follow below, whereas thesecond part of each section gives a view of the present stateof the national debate about EMU in political and opinion-forming circles. It will be evident, however, that these positionstaken in political debates and negotiation may well evolve overtime, and also become more precise when the Intergovern-mental Conference gets under way. The information recordedin this chapter may thus be viewed as a photograph of positionsat a certain point in time (October 1990). It has the none theless the important merit of recording the initial positions of theMember States, which, ultimately, will have to be adequatelyreconciled.

The potential economic impact of EMU is in the nature of apositive-sum game. It is not therefore, at least primarily, amatter of distribution of benefits and costs between countries,which would be the case if it were a zero-sum game. There are,certainly, inter-country distributional aspects which should notbe neglected (some of these were discussed in Chapter 9). Butthose aspects should not divert attention from the fundamentalpoint, which is this: the main sources of potential benefits fromEMU are several in number and heterogeneous in character.Individual countries can identify the major sources of potentialbenefit, and risks of costs, that are most relevant to its specificeconomic structure. The cocktails of benefits and costs ofindividual countries will often be quite different.

The grouping of countries below, by the extent of their partici-pation in the EMS, and by size, already indicates two import-ant parameters for assessing the benefits and costs by country.As pointed out in Chapter 8, countries that are already highlyconvergent within the EMS will have already undergone themajor adjustment costs of EMU, but have not yet reaped allthe benefits. The less convergent countries have bigger gainson offer, but larger transitional costs to overcome as well. Forsmall, very open economies the benefits of EMU are easiestto establish, especially those with currencies that are little usedinternationally and for which transaction costs are high.

Other important parameters include the extent to which thecredibility of national institutions for stable and sound monet-ary and budgetary policies would be improved by EMU. Thisis a source of potential gain for most countries, and themechanisms here lie in minimizing the transitional costs bothof disinflation and the reduction of public sector deficits. Onthese accounts Germany is the notable exception given itsanchor role in the EMS. The major potential benefits in thiscase are, therefore, of a different nature. As the largest ex-porter of goods and capital to other Community countries,Germany has an above-average interest in the stability andtransparency of the price and monetary mechanisms of thewhole of the union. Germany will be also most affected by thegeneral reshaping of the international system.

10.1. Currencies in the narrow band of theexchange-rate mechanism

10.1.1. Germany

Benefits and costs

Germany would gain from the perfection of market mechan-isms within the Community that would be achieved by EMUand, in particular, as a major exporter of capital, from theelimination of exchange-risk uncertainty for investors. Bycontrast, the reductions in transaction costs and uncertaintythat would be generated by a single currency would not beso important for Germany, a sizeable proportion of whosetrade is already invoiced in Deutschmarks.

For the past 40 years low inflation has been a key objectiveof German policy and has been achieved chiefly thanks tothe monetary policy pursued by the Bundesbank. There isconcern that EuroFed might have a weaker commitment toprice stability and to discipline in public finances. However,to the extent that EMU would reinforce the commitment ofothers to price stability and to sound public finances, itwould reduce the risk of imported inflation. In addition,Germany seeks to secure guarantees of price stabilitythrough the independence of EuroFed and the priority itwould give to stability.

During the transition to EMU the risks of the Deutschmarkbeing undervalued and, consequently, of some misallocationof resources to the export sector of the economy, mightincrease a little.

The replacement of the Deutschmark by the ecu as aninternational currency would relieve the German authorities

235


of the inconveniences of having to cope with exchange-ratepressures that emanate from the role of the Deutschmark asa major reserve currency rather than as the national cur-rency. This is also relevant in the context of the likelydemand in Eastern Europe for parallel currencies in the yearsahead: in the absence of the ecu as an adequate alternative, astrong demand for the Deutschmark would present a numberof problems.

German economic and monetary unification will have reper-cussions on the German position in relation to EMU. It isHkely to cause some increase in inflation and interest ratesfor a while and some reduction in the current-account sur-plus. This will make for closer convergence of economicperformance, albeit as a result of changes in Germany ratherthan in other economies.

Perspectives

The Government has given strong support to the EMUobjective and the broad approach proposed in the DelorsReport.

In its September 1990 'Statement on the establishment of aneconomic and monetary union in Europe', the Bundesbankremains very much in favour of the 'coronation theory', i.e.there should at first be a convergence of economic keyvariables and institutional safeguards, before competencesare transferred to EuroFed. In referring to 'substantial tran-sitional problems as a result of the intra-German unificationprocess, and when developments in Eastern Europe arestill unclear in many respects' the Bundesbank prefers topostpone further steps to EMU 'until such time as theeconomic situation in Germany as a whole and in the Euro-pean Community can be regarded as sufficiently consoli-dated'.

As regards the final stage of EMU, the Bundesbank enumer-ates several necessary pre-conditions related to the economicconvergence in Europe and to the structure and conduct ofmonetary and fiscal policies in all member countries and atthe European level which 'can be fulfilled only in the courseof a lengthy transitional process. During this process, noinstitutional changes which result in any curtailment of thefreedom of reaction of national monetary policy may bemade'.1

Academics are less enthusiastic about EMU than other sec-tors of opinion in Germany. The Board of Academic Ad-visers, a group of 30 academics which advises the EconomicsMinistry, set out its reaction to the Delors Report in a letter

sent to the Ministry in June 1989. This recognized that theReport was justified in rejecting the 'coronation theory',according to which monetary union can be introduced onlywhen virtually perfect economic convergence has alreadybeen achieved, but criticized the approach to monetary unionproposed in the Delors Report. The Board objected to theproposal that monetary union should be preceded by a two-stage transition period, because this involves the risk first,that the transitional arrangements. would turn out to bepermanent and second, that the Economic and FinancialAffairs Council would in fact acquire a weight incompatiblewith the autonomy of national central banks. It also doubtedthe need for formal ex-ante coordination and binding rulesfor fiscal policy.

The Council of Economic Advisers in their Annual Reportreleased in November 19892 commented that though theDelors Report provided a good basis for further discussion,serious objections could be raised to some of the positionsit adopted. While acknowledging the advantages that a singlecurrency would confer, the Council suggested that for thetime being the Member States' economies were too hetero-geneous for them to abandon the instrument of exchangerate adjustment without running grave risks. With the com-pletion of the internal market, greater nominal and realconvergence, and increasing integration, the conditions formonetary union would improve decisively, but for the timebeing it was necessary to retain the possibility of realigningexchange rates.

The Eurobarometer survey taken in autumn 1989 indicatedthat public opinion is mildly in favour of EMU. It is attractedby the advantages of a single currency for private trans-actions, but less keen than the EC average on economicintegration.

10.1.2. France

Benefits and costs

EMU should consolidate the advances France has made inrecent years towards much closer convergence with Germaneconomic performance in terms of price stability and soundpublic finance.

Pegging French monetary policy yet more firmly to the pricestability standard of Germany should help ward off anyremaining doubts as to the anti-inflation stance of the auth-

1 Deutsche Bundesbank (1990).

236

Sachverstandigenrat zur Begutachtung der gesamtwirtschaftlichenEntwicklung(1989).


orities. As the continuation of low inflation becomes morecredible and the residual degree of exchange risk perceivedby the markets is eliminated, interest rates will decline. Thisprogress is already in evidence, and will further assist publicfinances by alleviating the cost of servicing the public debt.Lower interest rates, together with savings in transactionscosts, will also encourage an expansion of private invest-ment.

France will thus strongly benefit from the effect of EMU onexpectations about future inflation and monetary accommo-dation. The unambiguous signalling of a commitment to astability-oriented EMU will help to persuade economic ag-ents to adapt their behaviour, for example in wage bargain-ing strategies. This will be important for overcoming thepersistent problem of relatively high unemployment.

The importance of the current-account constraint, a long-standing preoccupation for France, will diminish. Indeed, inan environment in which capital movements are alreadycompletely liberalized and the credibility of the commitmentto a fixed exchange rate is becoming established, the finan-cing constraint will be alleviated; without the exchange-raterisk only small interest rate differentials will be needed toproduce the appropriate financial flows. With the advent ofa single currency the financing constraint will be removedentirely.

Perspectives

The Government favours moving rapidly to EMU, for bothpolitical and economic reasons. Furthermore, it is widelyaccepted that devaluing the franc is not an efficient meansof resolving current-account problems. The transfer of sover-eignty is accepted, given that the French monetary authori-ties would exert influence on the common monetary policy.Some concern regarding budgetary policy has been voicedby the Minister of the Economy and Finance.3 Too rigid acoordination of fiscal policies, and the consequent loss ofsovereignty in this sphere, would create difficulties. Thesereservations apart, the strong support of the French authori-ties for EMU has been reiterated on several occasions. Fur-thermore, the French Government now supports the prop-osition of an independent central banking institution.

The Governor of the Banque de France in several speecheson EMU4 stressed the importance of directing a commonmonetary policy towards price stability and of ensuring that

See, for example, Beregovoy (1989).See, for example, de Larosiere (1989), and Les Echos (1990).

fiscal policy is also consistent with that objective. He alsosupported the idea of an independent central banking insti-tution.

Segments of the opposition political parties dissent from theproposal for a single currency in the third stage of the DelorsReport. For them, a common currency existing in parallelwith national currencies would suffice for the effective work-ing of the single market and would avoid the loss of sover-eignty of national governments in monetary policy-making.

In academic circles there is clear support for EMU. Surveysindicate that heads of large companies attach great import-ance to the setting up of EuroFed and the creation of a singlecurrency, since transaction costs and uncertainty would begreatly reduced. Public opinion appears from Eurobaro-meter polls to be strongly in favour of a single currency.

10.1.3. Italy

Benefits and costs

Italy's currency was until recently in the wide band of theEMS exchange-rate mechanism. Consequently, proceedingto EMU represents a more important change than for manyMember States. In recent years Italy's economic perform-ance has been in many respects highly positive. With respectto inflation and public finances, however, it still lags signifi-cantly behind the best standards. This implies costs in thetransition to EMU, but also potential gains if a crediblecommitment to EMU accelerates convergence. Not onlywould there be gain in terms of greater price stability andhealthier public finances. Once economic agents adaptedtheir expectations, there would be benefits too in the labourmarket and for employment.

The reduction in interest rates that will result from theelimination of exchange risk and inflation premiums will bea major gain for Italy because of its very high public debt(98 % of GDP in 1989) and the present wide margin betweenlira and Deutschmark rates. A decline of one percentagepoint in the treasury-bill rate is reckoned to save roughlyLIT 3 000 billion in the first year and LIT 7 000 billion inthe second.

None the less, though lower interest rates will significantlyreduce the debt-service component of the budget deficit, thisalone will not suffice to restore sound public finances. Aconcerted effort will still be necessary to bring the generalgovernment deficit (10,2 % of GDP in 1989) to a sustainablelevel. In addition, the establishment of a single financial

237


market will generate some losses in seigniorage revenue,resulting from the harmonization of marginal reserve re-quirements, and in fiscal revenue, resulting from the re-duction in the withholding tax rate on bank deposits, whichat present is 30 %. Although these losses will not be negli-gible, they will be smaller than the gains from lower interestrates and will be partly offset by the seigniorage gains ofEuroFed.

The effect of losing the exchange rate as an adjustmentinstrument will depend importantly on how far economicagents are convinced that this option no longer exists andadjust their behaviour accordingly. Since the inception ofthe EMS significant changes in behaviour have occurred inthe tradables sector (chiefly manufacturing industry) but farfewer in the non-tradables sector (chiefly services, public andprivate). With the removal of barriers to competition, marketforces should bring about the desired change in sectorsexposed to competition, e.g. financial services, but theGovernment will need to act directly on the public sector.

Perspectives

The Government is a forthright proponent of EMU, andstresses the need to bring the budget deficit under controlas a Community discipline necessitated by EMU.

The Banca d'ltalia supports both the principle of EMU andthe three-stage process outlined in the Delors Report. Itwelcomes the prospect of a common monetary policy di-rected at price stability and would like independent statusfor EuroFed. It also favours binding budgetary rules.5

Academic opinion is strongly positive towards EMU, whichit views as the culmination of the process of stabilizingexchange rates and reducing Italy's inflation to the lowerlevels of most of its EC partners. Some reservations areexpressed regarding the premature loss of seigniorage taxwhile the public debt remains excessive. There are warningstoo about the implications for the prospects of the less-developed regions of the country.

Public opinion broadly supports EMU. This includes thebusiness sector, despite recurrent complaints from industrial-ists about deteriorating competitiveness. The public assessesEMU, beyond its economic merits, as part of the process ofEuropean integration. Eurobarometer surveys show thatItalians favour the adoption of common policies, includinga single currency, by a wider margin than any other nation.

Smaller economies

The smaller countries in the exchange-rate mechanism havealready incurred, in the context of membership of the EMS,most of the costs of adjusting to EMU, in terms of reducinginflation to low levels and managing without an independentmonetary policy. They retain, however, some drawbacksof monetary independence, such as transaction costs andexchange-risk premiums on their interest rates. In addition,in EMU they will have a full part in determining the policyof EuroFed.

The consequences of a single currency in lowering trans-action costs and enabling the benefits of the single marketto be reaped to the full will be particularly advantageous tosmall open economies.

10.1.4. Belgium

Benefits and costs

The principal benefits to Belgium will be the efficiency gainsarising from the elimination of exchange-rate uncertaintyand transaction costs, and the discipline that EMU willimpose on public finances.

Commitment to EMU will reinforce the credibility of lowinflation. The consequent reduction in interest rates willcontribute significantly to lowering the budget deficit (6,3 %of GDP in 1989) because of the large sums that go toservicing Belgium's very high public debt (135 % of GDP in1989). This will not, however, obviate the need for morestringent control of public finances. Indeed, progresstowards EMU will entail both institutional and marketpressure on Belgium to reduce both the budget deficit andthe public debt faster than it is doing at present.

Perspectives

The Government has explicitly supported the Delors Report,including both the principle of EMU and the proposal toachieve it in three stages.6 The transfer of powers to asupranational authority that EMU involves is not viewed asa diminution of national sovereignty, since Belgium retainslittle de facto autonomy in the relevant areas. Oppositionparties accept the principle of EMU but question theGovernment's ability to deal satisfactorily with the problemsposed by the state of public finances.

5 See Banca d'ltalia (1990).

238

6 See Banque Nationale de Belgique (1990).


The governor of the central bank has stressed the impli-cations of EMU for the financing of budget deficits, notablythe abolition of monetary financing.7 This was resorted tofrequently in Belgium in the early 1980s but recently hasbeen used much less.

10.1.5. Netherlands

Benefits and costs

For a very open economy, such as that of the Netherlands,the elimination of transaction costs and exchange rate uncer-tainty in relation to the EC as a whole will be a significantbenefit.

A fixed parity has been maintained for several years betweenthe guilder and the Deutschmark, which has contributedlargely to achieving low inflation. Inflation is even slightlylower than in Germany, and the premium on Dutch interestrates over German rates is small and for long-term rates hasalmost been eliminated. As a result the main adjustmentsrequired for EMU have already been secured. (This experi-ence is analysed in detail in Annex B, given its importanceas a case study in gradual transition to effective monetaryunion.)

The Netherlands has a sizeable budget deficit (5,1 % of GDPin 1989), however, which partly reflects the interest rateburden of its high public debt (83 % of GDP in 1989). Bothneed to be reduced to ensure that the Netherlands has thefreedom of manoeuvre in fiscal policy that will be necessaryin EMU.

The Social and Economic Council, in an opinion issued inJune 1989,9 favoured an independent central bank, alongGerman lines, which it saw as offering a stronger guaranteeof price stability.

Academic opinion supports the idea of a single currency asoffering reduced transaction costs and a more useful unit ofaccount while also eliminating speculation between Com-munity currencies.

10.1.6. Luxembourg

Benefits and costs

In terms of its rate of inflation and financial situation theLuxembourg economy is already well adjusted to the pros-pect of EMU. Like other small open economies, it willbenefit particularly from the disappearance of transactioncosts and exchange rate uncertainty in relation to the countryas a whole.

The importance of the financial sector to the Luxembourgeconomy, notably with regard to employment and fiscalrevenue, has focused attention on the impact that EMU willhave on it. The liberalization of capital movements couldfavour the development of Luxembourg as a financial centre,though closer monetary integration among EMS memberscould erode some of the advantages that Luxembourg cur-rently enjoys. A fixed exchange rate or single currency andharmonization of banking legislation would tend to diminishthe volume of banking activity generated by foreign ex-change business, currency arbitrage and tax evasion.

Perspectives

In his government statement of November 1989 the PrimeMinister confirmed his government's support for EMU asreinforcing the benefits of the single market.8 He alsostressed the importance of Stage I of the Delors Reportand the necessity to liberalize capital markets. There is awidespread view that it is in the interest of the Netherlandsthat monetary decisions should be taken by a Communitybody and not become in effect the preserve of the largecountries. The Government shares this view and has alsorepeatedly emphasized the subsidiarity principle, insistingthat to the extent compatible with the objectives that havebeen set, Member States should retain responsibility forpolicy.

Perspectives

The Government supports the three-stage process towardsEMU outlined at the Madrid Council. None the less, it hassome reservations. These concern not so much the loss ofmonetary independence, which has already been effectivelyrelinquished, as the transfer of sovereignty in the context ofeconomic and fiscal policy.

The Ministry of Finance expects that as the Communityapproaches the final stage of EMU, coordination, harmon-ization and convergence among the 12 Member States willintensify.10 A specific fear is that revision of the Treaty of

7 Verplaalse(1989).8 Regeringsverklaring, Tweede Kamer (1989).

9 Commissie voor Economische Deskundigen van de Sociaa! Economi-sche Raad(1989).

10 Bausch(1989).

239


Rome will entail abandoning unanimous decision-making inthe realm of tax policy.11

In an interview in June 1989 the director-general of themonetary authority (Institut monetaire luxembourgeois)supported EMU but suggested that it should be achievedprudently and gradually in several stages.12

10.1.7. DenmarkBenefits and costs

Denmark might gain less than other small open economiesfrom the removal of transaction costs and exchange rateuncertainty, because a relatively large proportion of its tradeis with non-EC (Nordic) countries.

Through participation in the ERM the Danish economy hasmade substantial adjustment to the future disciplines ofEMU, and the rate of inflation is now among the lowest inthe ERM. The level of interest rates, however, still comprisesexchange-risk premiums because of Denmark's previous rec-ord of inflation in excess of German rates. By bringing aboutlower interest rates, EMU will have a positive impact onpublic finances. (The internal market puts pressure on Den-mark to reduce its very high indirect taxes, but EMU doesnot affect this.)

EMU will mean that the current-account constraint on Dan-ish economic policy, which has frequently been a matter ofgreat concern, will be substantially alleviated.

Perspectives

In a report on EMU published at the end of November1989, the Economics Ministry supported full participationby Denmark in the process of EMU as set out in the DelorsReport. It points up the macroeconomic advantages of EMUand of exchange rate stability for an economy in which smalland medium-sized companies predominate and agricultureis a major export sector. Although irrevocably fixed ex-change rates need not necessarily lead to a single currency,the Economics Ministry sees a number of advantages in this.It would like subsidiarity to be broadly applied, and suggestsvoluntary coordination of fiscal policy within guidelinesaimed at avoiding destabilizing deficits.13

The Central Bank favours gradual movement towards EMU.Its 1989 annual report emphasized that the growing interde-pendence among economies has substantially reduced theroom for manoeuvre of even medium-sized countries. Inview of this, participation in EuroFed is seen to offer Den-mark a better opportunity to exert influence on monetarypolicy in the Community.14

The Economic Council, a government-financed forum whichis broadly representative of the academic community, as-sesses the EMS very positively. It has underscored the im-portance for Denmark of broadening European monetarycooperation to encompass other Nordic countries.

There has recently been a marked change in the views of theopposition Social Democrat Party, which previously hadbeen critical of EMU. At the meeting of European Socialistsin Dublin in May 1990 they evinced a more positive ap-proach to the Community in general and to EMU.

10.1.8. Ireland

Benefits and costs

Ireland has made substantial progress towards economicconvergence in terms of price and wage inflation, reducedbudget deficits and a more favourable external balance.Hence much of the nominal adjustment required for passingsuccessfully to economic and monetary union has beenachieved. EMU, by reinforcing the credibility of these poli-cies, will help to ensure that they are sustained.

In the real economy, significant weaknesses remain in termsof low income per head, high unemployment and emigration,continued heavy reliance on the primary sector and inef-ficiencies in the productive system. An EMU with adequatepolicies to promote cohesion, however, would represent apositive, if challenging, framework within which to overcomethese. This implies, for example, an urgent need for changesin the structure of public finances and the economy ingeneral.

In EMU, fiscal policy will, along with incomes policy, as-sume greater importance as an internal instrument of stabili-zation and adjustment. Although its effectiveness should,in principle, be enhanced, the need for overall budgetarydiscipline will be accentuated in an EMU committed to pricestability. Further reduction of the high ratio of debt to GDP(104% in 1989) is essential, both to permit the structural

Hirsch(1989).d'Letzeburger Land(\9W).0konomiministeriet (1989). 14 Danmarks NatJonalbank (1990).

240

sduquenoy

sduquenoy


changes in the budget required for successful participationin EMU, and to create an adequate fiscal margin to respondto external shocks. Efforts in this regard would be facilitatedby a reduction in interest rates within the completed EMU.A key requirement, nevertheless, is continued tight controlon public expenditure.

Perspectives

The Government supports the principle of EMU, but itstresses the need for progress towards EMU to take fullaccount of the Community's objective, as set out in Article130(a) to (e) of the amended Treaty, to strengthen economicand social cohesion. This is also the position of the mainopposition parties. The particular concerns in political circlesare the problems likely to face a small, peripheral and lesstechnologically-advanced economy in competing in an inte-grated Europe, and the budgetary consequences of the fiscalharmonization required for this to come about. There is ageneral consensus in Ireland in support of EMU. A reportissued by the National Economic and Social Council in1989 broadly represented the combined viewpoint of theadministration, the trade unions, the employers, and agricul-tural interests. The NESC concluded that completion of theinternal market should not be expected to narrow the incomedisparities between regions, let alone bring about conver-gence; serious concern was expressed at the difficulties likelyto face the Irish economy, and industry in particular, in thisregard. The NESC argued for a strategic objective of creatingan advanced economic and monetary union, on the basis thata successful EMU required the formation of a substantialCommunity budgetary mechanism and extensive develop-ment of common Community policies, to ensure appropriateredistribution of the benefits of further integration.15

The Central Bank firmly supports monetary union in thebelief that fixed exchange rates will provide the best environ-ment for controlling inflation and promoting economicgrowth.16 Monetary union must, however, be accompaniedby greater coordination of other policies, particularlybudgetary. The Governor of the Central Bank, in a papersubmitted to the Delors Committee, argued that EMU willbe feasible only if all regions of the Community have reacheda 'broadly similar level of economic development', and rec-ommended the mobilization of Community policies to thatend and a greater degree of policy centralization as anelement of the integration process.

10.2. Currencies in the broad band of theexchange-rate mechanism

10.2.1. Spain

Benefits and costs

Having recently joined the exchange-rate mechanism withinthe wider bands, Spain presently finds itself in a situationsomewhat comparable to that of some other ERM partici-pants a few years ago. The immediate priority is to usethe ERM commitment as the framework within which todiscipline economic agents and to reduce the main disequili-bria: inflation and the external deficit. This also requires aredistribution of roles between monetary and fiscal policy.A fiscal policy adjustment to curb excess demand is required,given that monetary policy now has fewer degrees offreedom.

As has also been seen in Italy, this transition entails the riskof temporarily high nominal interest rates until inflation hasbeen sufficiently reduced.

But a credible commitment to EMU will reinforce the effec-tiveness of an economic policy aimed at price and exchange-rate stability, bringing substantial gains from the reductionin uncertainty. Ultimately the decline in interest rates willassist efforts to eliminate the general government deficit(2 % of GDP in 1989, with a target of 0 % for 1992) as wellas reduce the financing costs of the private sector. EMU, byproviding an environment of free capital movements andfixed exchange rates, will make it easier to finance current-account imbalances with only small interest rate differentials.This could be a major benefit to a country such as Spain,which needs large capital inflows to boost investment andgrowth. The combination of 1992 and EMU will also havea radical impact on modernizing financial markets, whichwill result in major reductions in the costs to the Spanisheconomy of capital and financial transactions.

Perspectives

The Government favours closer economic and monetaryintegration but would like this to include measures to assistthe less-developed regions of the Community. The Ministerof Economy and Finance sees no reason why accepting theexchange rate discipline of the EMS should put a brakeon medium-term economic growth, which would be betterbalanced if inflation slowed. Furthermore, long-term com-petitiveness could be improved.17

15 National Economic and Social Council (1989).16 Central Bank of Ireland (1990). 17 Solchaga(1989).

241


The Governor of the Central Bank, in a speech delivered inSeptember 1989, voiced broad agreement with monetaryunion and its implications.18

Business leaders assess the EMU process very positively,since they reckon that the advantages of exchange-rate stab-ility and economic discipline outweigh the disadvantages.They prefer some loss of national independence in economicpolicy-making to the fluctuations in the policy mix that haveoften occurred in the past. They also favour an independentcentral bank. Although in recent years the corporate sectorhas become more aware of the need to develop long-termcompetitiveness, there is concern that the Spanish economywill not adapt sufficiently rapidly to the new situation. Toaccelerate the process, most business leaders would like tosee a firm commitment made to monetary union.

Academic opinion views Spain's participation in the EMSand subsequently in EMU as reinforcing the credibility ofthe counter-inflation policy. Most believe that EMU willhave beneficial effects by obliging Spain to adopt a restrictivefiscal policy and to reintroduce an incomes policy, but addthat it should be complemented by appropriate measures ofregional policy. The independence of the central bank isregarded as important. Some academics are more pessimisticand believe that because of the gap that remains betweenSpain and the more developed EC economies, EMU couldhave negative effects, including impeding further catchingup by the Spanish economy.

Public support for EMU is quite strong and increasing.Spanish public opinion, as measured by Eurobarometer, issecond only to Italian in favouring a common monetarypolicy. Surveys also indicate that households' support forEMU is based on expectations of lower financing andtransaction costs.

10.2.2. United Kingdom

Benefits and costs

For the United Kingdom there are two specific issues inassessing the eventual benefits and costs of EMU: the extent

Rubio(l989).See, for example, Institute de Espana-Espasa-Calpe (1989).

to which these would arise from sterling's participation inthe ERM, which began on 8 October 1990; and the absenceof a clear acceptance by the Government of further stagesleading ultimately to the adoption of a single currency.

The serious aggravation of inflationary pressures during thepast two years and the consequent decline in the credibilityof the Government's own medium-term financial strategyfor controlling inflation increased acceptance of the relativemerits of participating in the ERM. The benefits of Stage Iof EMU, through ERM participation, are thus potentiallysubstantial for the UK. It should bring the disinflationadvantages that original ERM members have largely alreadyachieved through the discipline of maintaining an exchangerate objective. Significant reductions in nominal and realinterest rates and a lessening of uncertainty will also bebeneficial, particularly to the business environment.

The reduction in inflation expectations necessary for success-ful participation in the ERM, however, may entail substan-tial adjustment costs. Recent experience of UK labour mar-kets suggests that any deterioration in competitiveness re-sulting from wage inflation is expected to be restored bydepreciation of the nominal exchange rate. In the ERM thisoption will be severely limited. Wages and prices will haveto adjust to maintain competitiveness, particularly in themore competitive environment that is likely to obtain after1992. To ensure such adjustment during a 'learning phase'may entail output costs.

Full EMU, including a single currency and EuroFed, wouldlend additional credibility to the need for wage and pricediscipline. Other gains from a common currency includereductions in transaction and information costs and theelimination of concerns about financing external deficits.The UK could benefit too from an expansion of activity infinancial services in a fully integrated economic and monet-ary union.

Perspectives

The British position in relation to public debate about EMUhas developed rapidly since the Madrid European Councilin June 1989. Then the Government committed itself to theobjective of ultimate economic and monetary union and toStage I of the Delors Report, during which sterling wouldparticipate in the ERM. The Prime Minister also clarifiedthe conditions under which sterling would join the ERM,which included a reduction in UK inflation and the abolitionof capital controls in the rest of the Community.

242


Support for sterling joining the ERM was expressed by theLabour Party, the employers1 organization (CBI) and theTrades Union Congress (TUC).20 Academic opinion wasalso largely favourable. The principle of ERM membershipis opposed chiefly by academic economists of the monetaristpersuasion, who would prefer monetary policy to be free topursue purely domestic objectives. Such reservations weak-ened as the UK met with increasing difficulty in controllingits own domestic monetary aggregates.

The Government published an alternative set of proposalsto the Delors Report in November 1989.2I Essentially thistook Stage I to its limits by envisaging the removal of allrestrictions on the use of the different national currenciesand financial services throughout the Community, and com-petition among currencies to provide the non-inflationaryanchor in the EMS. Eventually the system could evolve intoone of fixed exchange rates, 'but that cannot and shouldnot be decided now'. Binding rules on budget deficits were'neither necessary nor desirable'. More recently, in June1990, the Government published further proposals for StageII, according to which a 'hard ecu' parallel currency wouldbe managed by a European Monetary Fund.22 This wasfollowed, on 8 October 1990, by entry of the pound into thebroader band of the ERM.

Regarding later stages of EMU, the Bank of England's view,as represented by its Governor, is that Stage I is likely tolast until the efficient operation of the single market hasbeen clearly demonstrated — probably the mid-1990s.23

The Labour Party and the TUC are chiefly concerned thatEMU should be buttressed by more generous regional fund-ing and that the democratic accountability of EuroFedshould be ensured.

Business, in the form of the CBI, broadly supports EMU,including a single currency, but sees this only as the culmi-nation of a successful process of economic convergence.24

Academic economists and independent commentators arecritical not so much of the goal of a single currency as ofthe proposals for binding budget guidelines in the DelorsReport, which is seen as unnecessarily centralized.25

See House of Lords (1989).HM Treasury (1989).Bank of England (I990a and b).Leigh Pemberton (1989).Confederation of British Industry (1989).See, for example, National Institute of Economic and Social Research(1989).

10.3. Countries outside the exchange-ratemechanism

Greece and Portugal, the two least developed economies inthe Community, both need to achieve substantial conver-gence with the rest of the Community, especially with respectto inflation and public finances, before joining the exchange-rate mechanism of the EMS. Portugal is considerably moreconvergent in these respects, however.The example of the EMS and the prospect of EMU are ofgreat importance to Greece and Portugal because they pro-vide a clear framework and strong motivation for the mod-ernization of their economies. While the 1992 process andthe structural Funds strongly affect most rf the productivesectors, the prospect of EMU emphasizes the need for stabili-zation and labour market reforms. EMU will be particularlyimportant in reinforcing the institutional credibility of amodern, stability-oriented policy strategy. EMU also offersto both countries particularly large gains from the elimin-ation of currency transaction costs, given their very highlevel and the importance of the tourist sector and emigrants'remittances.

10.3.1. GreeceBenefits and costs

Efforts to close the gap between the Greek economy and therest of the Community have so far met with only limitedsuccess. Inflation was close to 14% in 1989 and acceleratedto over 20 % in the first half of 1990, so the already largedifferential with the EMS countries widened further. Largebudget deficits (18 % of GDP in 1989) have caused a rapidincrease in the public debt, which was 105% of GDP in1989.

The economic policy programme of the new governmentformed in April 1990 has opted against a fully accommodat-ing exchange-rate policy. This, combined with the restrictivewage policy and, more importantly, the implementation ofa medium-term fiscal consolidation plan aimed at halvingthe budget deficit by 1993, should ensure that the processof disinflation is set in train and so pave the way for partici-pation in the exchange-rate mechanism.

Once inflation has slowed sufficiently to allow Greece tojoin the exchange-rate mechanism, its economic policy wouldbe set to gain further in credibility.

During the disinflation transition real interest rates willnecessarily be high to begin with. However, when conver-gence becomes highly credible these interest rate premiumswill decline, and ultimately be eliminated in EMU. Thisinterest rate effect could then offset the reduction in monet-ary seigniorage.

243


Perspectives Perspectives

Most academic, business and political opinion favours pro-gress towards EMU and Greece's eventual involvement.26

The new government has repeatedly expressed its determi-nation to implement its economic policy programme, whichenvisages Greece being able to join the ERM by 1993.

10.3.2. Portugal

Benefits and costs

The Government supports the principle of EMU and theproposals of the Delors Report. But it has also argued thatEMU risks widening again the gap between Portugal andthe more developed Member States. To minimize the costsand risks that EMU entails for Portugal, it should be ac-companied by a further increase in the Community's struc-tural Funds.27 The Portuguese authorities have recentlystated that the escudo should join the ERM relatively soon,but that the present large inflation differential between Portugal and the core ERM countries (9 %) prevents this hap-pening in the near future.

Since Portugal joined the Community in 1986 the economyhas made some progress towards convergence. Inflation hasdeclined from 19% in 1985 to 13% in 1989. The state ofits public finances remains weak, though it has improvedslightly. In 1989 the budget deficit was 5% of GDP, andthe public debt stood at 70 % of GDP. Further improvementin public finances will be necessary before Portugal canparticipate successfully in the ERM and EMU.

Participation in the ERM and the EMU should reduce (andultimately eliminate) both the exchange risk and the inflationpremium elements of domestic interest rates, and so help toalleviate the burden of servicing the public debt.

The process of catching up with the rest of the Communitywill take some time, during which current-account deficitsare likely to persist. With EMU, these would be less of amacroeconomic constraint and easier to finance.

Among politicians there is broad but qualified support forjoining the ERM and for EMU.28 In business circles fearshave been expressed about the possible deterioration ofcompetitiveness, particularly by industrialists in traditionalexport sectors, such as textiles and footwear, and about theloss of control of domestic firms to foreign capital.

The academic community tends to support membership ofthe exchange-rate mechanism, mainly because of the greaterprice stability and fiscal discipline that would be implied.Most take the view that the escudo should join only wheninflation draws closer to the EC average, but some believethat the escudo should join soon so that the effort to curbinflation can benefit from membership.

26 See, for example, Demopoulos (1990).27 See Ministerio las Financas (1990).28 See Rodrigues( 1990).

244

sduquenoy


Table 10.1Selected economic indicators for EC Member States, 1980-89

B DK

GDP per capita(EUR =100)1980 104,2 108,11989 102,3 106,5

Private consumption deflator(Annual percentage change)1980 6,4 10,71989 3,1 5,0

D

113,8112,7

5,83,1

GR

58,254,2

21,913,8

E

73,475,8

16,56,6

F

111,8108,8

13,33,5

IRL

64,167,1

18,64,1

i

102,6103,7

20,46,0

L

116,0125,8

7,63,4

NL

111,1103,0

6,91,1

p

55,155,0

21,612,8

UK

101,1106,3

16,25,8

EUR 12

100100

13,54,8

General government lending (+)/borrowing ( — )(As % of GDP)1980 -9,2 -3,31989 -6,3 -0,7

Intra-EC trade(as % of GDP)Imports1980 36,0 14,91989 45,4 14,1Exports1980 38,5 13,21989 46,3 13,5

-2,90,2

11,411,9

12,115,7

-17,6

10,915,2

6,46,9

-2,6-2,1

5,010,2

5,17,8

0,0-1,3

10,613,8

9,311,9

-12,7-3,1

43,536,6

33,547,3

-8,6-10,2

10,210,1

8,99,2

-0,43,3

36,045,4

38,546,3

-4,0-5,1

24,830,0

32,137,0

-5,0

17,127,0

10,819,5

-3,41,6

9,112,2

9,69,3

-2,8

12,214,2

12,314,2

Table 10.2Eurbarometer survey, autumn 1989

Unification of Western Europe'

For very muchFor to some extentAgainst to some extentAgainst very muchNo replyTotal

B

3154618

100

DK

243619147

100

D

423782

10100

GR

542833

12100

E

413542

18100

F

295072

12100

[RL

383343

22100

i

4442419

100

L

32421448

100

NL

26501158

100

p

472241

26100

UK

2742125

14100

EUR 12

3741

73

12100

Decision-making on currency^

CommunityNational

B

6728

DK

5438

D

5935

GR

4439

E

4639

F

7223

IRL

5735

i

6824

L

5633

NL

5338

p

2849

UK

4251

EUR 12

5735

Qaeadon*1 Attitude towards the unification of Western Europe

In general, are you for or against efforts being made to unify Western Europe? If FOR, are you very much for this, or only to some extent? If AGAINST, are you only lo some extent againstor very much against?

1 National or joint Community decision-makingSome people believe that certain areas of policy should be decided by (national) government, while other areas of policy should be decided jointly within the European Community. Intervieweeswere given a range of policy areas (including currency) and asked to express a view as to the appropriate forum for decision-making.

245


GRAPH 10.1: Variation of EC Member State currencies with respect to the Deutschmark(1980=100)

1201-

100

60

40

20

HFL

- — -v-.—......_ ._.._.._.-.c:-- FF

-- ESC3

DR_l—————[_

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990

246

sduquenoy


References

Banca d'ltalia (1990), 'Considerazione fmali' of the Gover-nor of the Bank of Italy in Relazione Annuale.

Bank of England (1990a), 'EMU beyond Stage I: the hardecu', paper published on 21 June.

Bank of England (1990b), 'The hard ecu in Stage II: oper-ational requirements', paper published on 21 June.

Banque Nationale de Belgique (1989), Revue de Presse, vari-ous issues.

Bausch, R. (1989), 'La fiscalite de 1'epargne', d'LetzeburgerLand, 8 June, p. 36.

Beregovoy, P. (1989), 'Intervention de M. Pierre Beregovoya la Chambre de Commerce et d'lndustrie de Francfort, le6 Novembre 1989', partially reprinted in Deutsche Bundes-bank, Auszuge aus Presseartikeln, No 88, 8 November.

Central Bank of Ireland (1990), Annual Report, summer.

Commissie voor Economische Deskundigen van de SociaalEconomische Raad (1989), Europese integrate en het sociaal-economisch beleid, 22 June.

Confederation of British Industry (1989), 'European monet-ary union: a business perspective', report of the CBI Euro-pean Monetary Union Working Group, November.

Danmarks Nationalbank (1990), Report and accounts for theyear 1989, Copenhagen.

de Larosiere, J. (1989), 'National monetary policy and theconstruction of European monetary union', in DeutscheBundesbank, Ausziige aus Presseartikeln, No 89, 10 No-vember.

Demopoulos, G. D. (1990), 'Financial liberalization, theEMS and the consequences for macroeconomic policy inGreece', forthcoming in The Greek economy: economic policyin the 1990s, London, Macmillan.

Deutsche Bundesbank (1990), 'Statement on the establish-ment of an economic and monetary union in Europe', Sep-tember.

Hirsch, M. (1989), 'Le Luxembourg et la Communaute euro-peenne', d'Letzeburger Land, 8 June, p. 13.

HM Treasury (1989), An evolutionary approach to economicand monetary union, November.

Institute de Espana-Espasa-Calpe (1989), El Sistema mone-tario europeo y elfuturo de la cooperacion en las CEE.

House of Lords (1989), "The Delors Committee Report',sound report of the Select Committee on the EuropeanCommunities of the House of Lords, session 1989-90,HMSO, London.

Leigh Pemberton, R. (1989), The future of monetary ar-rangements in Europe', speech by the Governor to the Insti-tute of Economic Affairs, 26 July, reprinted in Bank ofEngland Quarterly Bulletin, August.

Les Echos (1990), 'de Larosiere precise les contours de lafuture banque centrale europeenne', 28 March.

Ministerio las Financas (1990), Quadro de ajustamento na-tional para a transicao para a uniao economica e monetario(versao premiminar), Lisboa.

National Economic and Social Council (1989), 'Ireland inthe EC: performance, prospects and strategy', Report No 88,August.

National Institute of Economic and Social Research (1989),National Institute Economic Review, August.

0konomiministeriet (1989), Okonomisk og Monetar Union,Copenhagen.

Regeringsverklaring, Tweede Kamer (1989), Staatscourant,21 November, No 231, p. 4.

Rodrigues, F. (1990), 'O escudo no SME: potencialidades eexigencies', Seminario Economica, No 179, 16 June.

Rubio, M. (1989), 'La Economia espanola en la union mone-taria Europea', Boletin Economico, Banco de Espana, Sep-tember.

d'Letzeburger Land (1989), 'Europaische Wirtschafts- undWahrungsunion: Ein Optimum ist nicht garantiert', 8 June,p. 19.

Sachverstandigenrat zur Begutachtung der gesamtwirtschaftlichen Entwicklung (1989), Jahresgutachten 1989/90,Stuttgart and Mainz.

247


Solchaga, C. (1989), 'Comparecencia del Ministro de Econ- Verplaatse, A. (1989), interview in De Standaard, 3 July,omia y Hacienda ante el pleno del congreso de los Diputados Reprinted in Revue de Presse of the Banque Nationale depara informar sobre la integracion de la peseta en el SME', Belgique, No 126, and in Annual Report 1989 of the NationalInformation Comercial Espanola, Ministerio de Economia Bank of Belgium, p. xviii.y Hacienda, 3 to 9 July.

248

sduquenoy

Annexes

Annex A — Exchange transaction costs

Annex A 2. Financial transaction costs

Exchange transaction costs

1. Introduction

Firms and individuals buying, selling, working or investingin another Member State incur currency conversion costswhen their domestic money is not accepted as a medium ofexchange.

Exchange transaction costs caused by the absence of a singlecurrency in the Community can be split into two parts:

(i) on one hand, there are the 'financial' costs consisting ofthe bid-ask spreads and commission fees that householdsand non-bank enterprises pay to banks for foreign cur-rency conversion. Under competitive circumstancesthese financial expenses mirror the cost of resources inthe banking sector absorbed by the provision of foreignexchange;

(ii) on the other hand, there are the sometimes significantin-house costs companies operating internationally facein the form of resources tied up in departments likeaccounting and treasury to manage foreign exchange forintra-EC transaction needs, or in the form of paymentdelays or of poor return on cash management.

This annex is concerned with describing and quantifying thetransaction gains EC firms and households will enjoy whenpayments or receipts in EC currency other than the domesticmoney related to intra-EC current or capital account trans-actions have been eliminated upon the introduction of asingle currency. Other exchange transaction gains associatedwith a possible shift in payment denomination from dollarsto ecus, or with a reduction of the cost of buying or sellingdollars were dealt with in Chapter 7.

Apart from eliminating exchange transaction costs, a singlecurrency is an indispensable instrument for cutting the rela-tively high expenses and long delays cross-border paymentsin the Community currently suffer from.

Financial transaction costs can best be approached by seeingthem as resulting from two factors: the volume of foreignexchange transactions converting one EC currency into an-other, and the prices banks charge for these services.

2.1. The volume of foreign exchangetransactions in the Community

Turnover on the Community's foreign exchange markets

In April 1989, central banks of some 20 countries carried outin collaboration with the Bank for International Settlementscomprehensive surveys on the foreign exchange operationsof banks and other dealers in their markets. The surveycovered all EC Member States except Germany and Luxem-bourg.

The overall picture the survey provides is summarized inTables A.I, A.2 and A.3 in which the original data havebeen converted into ecus and expressed on an annual basis.

The net turnover reported in Table A.I indicates that thetotal volume of operations on 10 Community foreign ex-change markets amounted to ECU 61 340 billion in 1989.According to the BIS staff1 the net daily turnover in Ger-many and Luxembourg together may be put at USD 100billion per day, bringing the EC's overall yearly total toECU 84 067 billion. Half of this Community turnover isrealized on the UK market, the world's largest. Apart fromGermany, the other foreign exchange markets in the Com-munity are rather small, with hedging and trading operationsmainly conducted with banks in other countries rather thanwith local interbank counterparts.

The column 'business with customers' denotes the volumeof direct foreign exchange transactions between banks andnon-bank financial institutions, companies, or individualinvestors.2 It represents a fairly small part (15 %) of totalnet turnover. However the indirect importance of business

The quantifiable 'mechanical' savings a single currency willgenerate through the various channels identified in thisannex can be put at between ECU 13 and 19 billion, or 0,3to 0,4 % of the Community's GDP. The summary table atthe end of this annex assembles the components leading tothis overall estimate.

BIS (1990), p. 2.The distinction between interbank and customer business is not alwayseasy to make; it becomes less and less so as large non-bank firmsincreasingly participate directly in the foreign exchange market, not onlyto meet the foreign currency needs arising from their main business, butalso to engage in arbitrage or purely speculative operations.

251

sduquenoy

Annexes

Table A.ITotal foreign exchange market activity

(billion ECU. annual basis)

Country Gross of which:

Domesticinter bantoperations

Nei turnover1

Cross-borderinterbankoperations

of which:1

Businesscustomers

Net domesticinterbankoperations

UKFrance (95%)4

NetherlandsDenmark (90%)4

Belgium (90%)4

Italy (75%)4

SpainIrelandPortugalGreece

Total EUR 12 excluding Germanyand Luxembourg

54772727236363409272725001 3411 250

250159

77317

24 5453

25003

1 3863

79572736468215991

31 138

425003

59093

29543

2954227222721000I 182

204913

61 340

24 3183

34093

19323

227217271 81841821045

91463

37 181

5909136341295295318136234623

8636

12 2733

1 3643

70S3

4093641823416846234

15773

USAJapanSwitzerland (85%)

395453295415454

20 4543

138635000

293183

2613612954

16 1363

106828 182

227277272045

10 2273

70452500

1 Items do not always add up lo total net turnover because the classificacion is not exhaustive,2 Figures Tor individual countries indicate turnover net of double-counting arising from local interbank business.1 Based on estimates of domestic and cross-border interbank business arranged through brokers.* No adjustment was made for less than full coverage. Figures in parentheses indicate the estimated market coverage.Source: BIS.

Table A.2Foreign exchange market activity, by type of transaction: gross turnover

(billion ECU, annual basis!

Country Grossturnover

Spot Forward Options

Swaps Outright Total Bought Written

UKFranceNetherlandsDenmarkBelgiumItalySpainIrelandPortugalGreece

547332211

772272636409727500341250250159

35000431819541 6361 3862091

8181000

227136

190912111

727659772341409523250

2323

nana

1 52314321 250

na477nana0

nana

13634191na45nana23

na182

0—0

10—nana0

6824552345*230

————

nanananana

0————

nanananana

0————

Total EUR 12 excludingGermany and Luxembourg 77317 48568 27818 205

USAJapanSwitzerland

395453295415454

2500013 1818409

12045188636818

9545168185909

1 9561 956

773

636

0

2500955277

1 227nana

1 205nana

na: not available.— : (virtually) non-exislent.*: including futures.Source. BIS.

252


Table A.3Currency composition of foreign exchange market activity: gross turnover

1 Both currencies in individual transactions included.Source: BIS.

(billion ECU. annual basis)

Country

UKFranceNetherlandsDenmarkBelgiumItalySpainIrelandPortugalGreeceUSAJapanSwitzerland

Gross turnoverin all currency

54772727236363409272725001 3411 250

250159

395453295415454

Domesticcurrency

168183409227210681 11416141045

13611468

37954263639318

USD

4909052272500272722271 341

955750182114

3795431 36311 591

DM

159093182143210681 136

59147770511468

13 1823 1825000

Yen

95453869168

—23

—2300

1090926363

1 205

UKL

13611491

—9123

227230

568212271 182

ECU

909159322368

227—250

00

114—114

Other1

175001 818

9091 659

9321 114

182686823

11 13638642500

with customers is much larger because it exerts a 'multiplier'effect, generating a sizeable portion of interbank activity.For instance, forward orders from non-bank enterprises areusually carried out by way of two deals: a swap and acorresponding spot transaction. When it involves two ECcurrencies, the forward contract is likely to give rise toat least four supplementary transactions in the interbankmarket: each of the two EC moneys will be swapped andexchanged spot for US dollars, the latter fulfilling the vehiclecurrency role discussed in Chapter 7.3

Table A.2 classifies foreign exchange activity according tothe type of transaction. Spot transactions on EC marketsstill claim more than 60 % of turnover, compared to 36 %for forward contracts. No indication exists regarding theshare of forward contracts arising from hedging needs forcommercial purposes, the remainder being associated withpurely financial operations like covered interest arbitrage.Currency options occupy as yet a marginal position on theCommunity's markets. Most of the options traded concerntransactions relative to the dollar or yen. Given the lowvolatility of their bilateral exchange rates, options involvingtwo ERM currencies hardly exist.

More concretely, suppose a bank sells French francs forward to a cus-tomer against Dutch guilders. The bank will typically conduct thisoperation by two swaps (on the one hand, a forward purchase of Frenchfrancs and spot sale of French francs against US dollars; on the otherhand, a forward sale of guilders and spot purchase against US dollars)and two spot transactions (French franc purchase against dollars andguilder sale against dollars).

Table A.3. shows the currency composition of foreign ex-change transactions when both sides of every deal are coun-ted. It bears out the still overwhelming role of the dollar onEuropean markets, with about 90 % of all transactions onthe UK market involving the US currency (in 1986 it waseven 96 %). On other Community markets, foreign exchangein and out of the dollar claims on average about 70 % oftotal turnover.4

As a corollary, direct transactions between EC currencies arerelatively minor. Their precise volume is, however, unknownsince the survey does not disclose systematically turnoverper pair of currencies. The 'indirect' transactions may beequally if not more important but their magnitude cannotbe measured accurately either for want of evidence on whatpart of EC currency/dollar exchange transactions constitutesone step of a foreign exchange transaction between two ECmoneys, and what part concerns deals 'in their own right'.If foreign exchange deals on EC markets relating directly orindirectly to transactions between EC currencies are denotedby A and all other deals by B, and one assumes that half ofthe transactions between an EC currency and the dollar areintermediary exchanges linking one EC money to another,the way to measure properly the share of transactions be-tween EC currencies in total foreign exchange market ac-tivity (i.e., A/(A + B)) is given in the first column of Table

For the remainder of the analysis the hypothesis is adopted that half ofthe last column ('other') of Table A.3 is composed of EC currencies notaccounted for explicitly in the table.

253

sduquenoy

Annexes

A.4. However, in the absence of information on turnoveraccording to currency pairs, Table A.3 does not permit toapply this correct yardstick. Instead, two alternative compu-tations using the data of Table A.3 are proposed, each exhi-biting sources of over- and underestimation as signalledin the second and third column of Table A.4. The firstcomputation leaves out the figures contained in Table A.3for 'domestic currency' and half of'other', whereas all figuresare made use of in the second calculation. According to thefirst computation 34 % of all operations on the Community'sforeign .exchange markets concern transactions between ECcurrencies, whilst the second suggests it is 43 %. These fig-ures would seem to provide a lower-bound estimate in viewof the fact that the German foreign exchange market, wherethe position of the dollar and yen is probably much lesspredominant than in the UK, was excluded. More particu-larly, the EC currency component of foreign exchange oper-ations on behalf of non-bank clients is very likely to exceedthe indicated percentages since such operations are oftendriven by foreign exchange needs stemming from inter-national trade and direct investment.5

Foreign exchange turnover and the balance of payments

As is well known, the turnover on foreign exchange marketsis extremely high in comparison to the transactions recordedin countries' balance of payments, as shown in Tables A.5and A.6 below. The net turnover on EC foreign exchangemarkets equals 65 times the sum of intra- and extra-ECgoods trade. In fact, this disproportion is even larger sincean important part of extra-EC trade is settled in an ECcurrency and therefore does not give rise to a foreign ex-change transaction on the side of EC residents. Furthermore,not all foreign currency payments or receipts occasion ex-change transactions as enterprises 'net' out their foreignexchange needs.

However, the foreign exchange market and balance of pay-ments data become broadly reconcilable when looking onlyat the turnover in connection to.business with non-bankcustomers and bearing in mind that the capital accountfigures of the balance of payments relate to stock changes,which may only be a fraction of the gross capital flows, theactual variable of interest.6

P. Demarsin (1990), p. 9.This approximate compatibility shows up best in the case of countrieswhere, on account of capital controls, gross capital flows by the non-bank sector are unlikely to have been considerable. For instance, theannual estimate derived from the BIS survey of Italy's foreign exchangeturnover arising from business with non-bank customers amounts toECU 318 billion. Given that about 70 % of Italy's current account relatedinternational payments and receipts is denominated in foreign currencyand assuming this percentage to apply also to capital in- and outflows,the balance of payments data suggest Italy's (non-bank) internationaltransactions in foreign currency equalled around ECU 260 billion.

Table A.4

Measurement of the share of transactions between EC currencies intotal foreign exchange turnover

Possible foreign exchange transaction Correct Compulalion ComputationI 2

Domestic currency — other EC cur-rencyOther EC currency — USDOther EC currency — yenDomestic currency — USDDomestic currency — yenUSD — yenOther EC currency — non-EC cur-rency other than yen and USDUSD — non-EC currencyYen — non-EC currency

AA, B

BA,B

BB

BBB

AA, BA, B1

B2

BB, B2

_ iBB

A, A1

A, BA, B'A, BA,B'B, B2

A, B1

B, B2

B, B2

1 Source of overestimate,2 Source of underestimate.

Table A.5

Volume of EUR 12 goods trade, 1988(billion ECU)

Extra-EC

ImportsExports

530,3530,3

389,8367,0

Source.' Eurostat.

Foreign currency transactions by households

Because the minimum values set in the foreign exchangemarket survey as lower thresholds for reporting were ratherhigh, the figures contained in Table A.I normally do notinclude the foreign currency payments by way of euro-cheques, traveller's cheques, international credit cards, orbanknotes. The latter are typically the international paymentinstruments of the individual cross-border shopper ortourist.

254

sduquenoy


Table A.6Member States' current and capital account transactions

(1988, billion ECU!

BLEU

Denmark

Germany

Greece

Spain

France

Ireland

Italy

Netherlands

Portugal

UK

Imports (M)Exports (X)

MXMXMXMXMXMXMXMXMXMXMX

Goods trade

72,973,821,723,3

194,7260,1

10,15,0

48,733,5

142,5135,7

12,915,6

108,8108,3

76,483,413,49,1

152,8121,4

Currentaccount

transactions

118,3121,237,235,7

303,3344,4

13,518.766,363,2

237,4234,4

19,920,5

157,1152,6116,4120,8

17,216,7

275,4253,4

Liabilities (L)Assets (A)

LALALALALALALALALALALA

Capital accounttransactions by

non-bank sector1

27,331,38,46,3

14,562,3

1,9na

16,53,8

48,175,1

3,12,0

38,622,119,917,65,21,1

32,042,5

1 Sum of stock changes with regard to direct investment, portfolio investment, other long-term capital, short-term capita! other than by deposit banks.Source; Eurostat.

Around 50 million eurocheques with an average value ofECU 125 are used annually for international settlementsinside Europe. Since this figure also relates to EFTAcountries like Austria and Switzerland where they form acommon means of payment, eurocheques can be reasonablyestimated to attain a yearly volume of ECU 5 billion asinternational payment instruments inside the Community.

The annual volume of traveller's cheques denominated in anEC currency and sold in the Community can also be set atabout ECU 5 billion. Credit card payments by Communityresidents in foreign EC currency can be roughly estimatedto total ECU 10 billion per annum.

Not all central banks of the Community keep a systematicrecord of the volume and currency breakdown of sales andpurchases by the domestic bank sector of foreign banknotes.

The available information, covering five Member States thatare jointly responsible for about two thirds of intra-ECtrade, is presented in Table A.7.

The reported total for Germany needs to be revised upwardbecause the Bundesbank data pertain to banknote tradearising from travel only. Data from other countries suggesttourism and travel generate just over 30 % of all foreignbanknote sales and purchases. Accordingly, one can plausi-bly assume that the German bank sector turnover in foreignEC currency banknotes amounted to twice the reportedvolume, raising it to ECU 11 billion in 1988.7 For Italy, the

The added part was supposed to be exclusively in the money of ECMember States bordering on Germany {FF, BFR/LFR, HFL, DKR),the key for the distribution among these four currencies reflecting theirrelative importance in the travel-induced banknote transactions.

255

Annexes

Table A.7Currency breakdown of banknote transactions, 1988

(Purchase and sale of foreign banknotes in EC currency by domestic bank sector)(million ECU)

BLEU (%) Germany1 (%)

BFR — — 201 (3,6)DM 2563 (41,5) — —FFUKLLITPTA

670 (10,9) 1 339193 (3,1) 384170 (2,8) 1 127123 (2,0) —

HFL 2 270 (36,8) 1 357ESC1RLDRDKR

674

181 (2,9)3 I3

43385

(24,2)(6,9)

(20,3)—

(24,5)(12,2)

(0,2)(0,8)(6,9)

Italy1

74

28361 512

671—

448153

9—4731

(*

(1,3)(49,0)(26(H

,1),6)

(7,8)(2,3)

France

240937471471648798154

(%)

(7,0)(27,3)(13,7)(13,7)(18,9)(23,2)

(4,5)(0,2) 1

M57 (4,6)3(0,8) J(0,5) 29 (0,9)

Total 6170 (100) 5541 (100) 5787 (100) 3434 (100)

Banknote transactions associated with travel only.Banknote transactions by foreign exchange institutions other than banks are not included in these figures.Estimate.

total goes up to ECU 8 billion once the transactions byforeign exchange institutions other than banks are takeninto account.

These revisions bring EC currency banknote trade in thefive countries under consideration to ECU 28,5 billion in1988. The latter figure can be used to arrive at an estimatefor the Community in its entirety. An extrapolation basedpurely on countries' weight in intra-EC trade would not beappropriate, however, since it would suffer from a downwardbias for two reasons. First, in countries like Greece, Spainand Portugal, cash is still a much more predominant pay-ment instrument than elsewhere in the Community. Sec-ondly, and partly because of this very prevalence of cash asa medium of exchange, converting foreign banknotes inthe 'South' is clearly cheaper than in the 'North' , as isdemonstrated in Table A. 13 below. It follows that foreignbanknotes are likely to be much more actively traded in anumber of Member States outside the sample. For thesereasons, the EUR 12 volume of banknote conversions in-volving two EC currencies is assumed to have lain in 1988between twice and three times the ECU 28,5 billion turnoveridentified earlier, i.e. between ECU 57 and 85 billion.

Member State differences in foreign exchange exposure

The BIS data on foreign exchange turnover did not permitto determine accurately the part of foreign exchange trans-actions to vanish upon the establishment of a single Com-munity currency. An alternative method of arriving at thetransaction volume between EC currencies is to ascertaineach individual Member State's gross current and capitalaccount flows in foreign currency and to isolate the ECcomponent.

Unfortunately, most countries lack the detailed statisticalinformation necessary for such an exercise. This method istherefore unable to yield a meaningful assessment of theEC's aggregate amount of foreign exchange transactions.Additionally, it is liable to overestimation as it fails todistinguish between on one hand the volume of paymentsand receipts in foreign currency and on the other the volumeof exchange transactions, the latter volume falling short ofthe former because firms net out their foreign exchangeneeds. Despite these problems and the limited sample, thisapproach yields interesting results as it illustrates clearly thatthe relative significance of intra-EC transactions settled in

256

sduquenoy

sduquenoy


foreign currency, and hence the extent of foreign exchangeexposure, differs considerably between Member States. Thisis borne out by Table A.8, which provides the current and

capital account related receipts and payments in foreign ECcurrency arising from intra-EC transactions for the BLEU,Germany, Italy and France.

Table A.8Balance of payments transactions with other EC countries in EC currencies (exclusive of the domestic currency) in 1988'

Germany Italy France3

Absolute valuebillion ECU

% of nationalGDP

billion ECU % GDP billion ECU GDP billion ECU % GDP

1. Through means of paymentother than banknotes:

Goods tradeServicesTransfersInvestment incomeCurrent accountDirect investment

Other capital account trans-actions by non-bank sector

2. Purchases and sales ofbanknotes in EC currency

91,57,8

548,4

72,76,2

436

51,413,55,19,2

79,25,9

96,33

7,80,6

9,5

64,317,1

5,6

87,04,4

12,30,6

59,7

40,4

100,17,2

283,84' 5

12,50,9

35,4

by domestic bank sector

3. Total6,0

653,7

4,7

519,6

5,56

186,91,119

8,17 1,1 4,3

295,4

0,6

49,41 Partly own calculations on working hypotheses.1 Transactions in EC currency other than FF with rest of the worldj Long-term capital account transactions only (securities, investment certificates and fixed-interest bearing assets). For shon-term transactions only net stock changes on a monthly or yearly basis

are available.4 Figures for some important categories of short-term capital account transactions are not available in the form of gross flows. In that case, the slock changes on a monthly basis were included.' To derive the EC-currency component, the EC-currency share in worldwide investment income was taken (28,9%).6 Banknote transactions related to travel only.7 Banknote transactions by banks as well as other foreign exchange institutions.na - not available.

Although a cross-country comparison should desirably referto the balance of payments as a whole, the poor quality ofthe gross capital flow data for Germany and France as wellas the fact that 1988 was still an unrepresentative year forFrance's and Italy's trade in financial assets due to remainingcapital controls make it appropriate to focus on the currentaccount and direct investment. Whereas foreign EC currencypayments and receipts of the BLEU represented in 198879 % of its GDP, the corresponding figure for Germany,Italy and France amounted respectively to 8,4%, 12,9%and 13,4%. Tables A.9 and A. 10 show that the principaldeterminants of this striking disparity are the prominenceof the domestic currency as an international means of pay-ment and the degree of economic integration with otherMember States.

As indicated in Table A.9, Germany pays for about 60 %of its imports from other Member States with Deutschmarks;77 % of its exports are settled in Deutschmarks. France alsoenjoys the benefit of a national currency that is relativelywell accepted as an international payments instrument. Incontrast, 50 to 60 % of payments and receipts by the BLEU

or Italy are denominated in an EC currency other thandomestic money.

The intensity of trade links is the chief factor explaining thedifference between the BLEU and Italy as regards theirexposure to other EC currencies. This is illustrated by TableA. 10, where openness is measured as the sum of intra-ECimports and exports of goods relative to GDP. On thisdefinition, the BLEU is more than five times more integratedwith the rest of the Community than Italy and is thereforemore vulnerable to the problems the multitude of EC cur-rencies can give rise to.

Size distribution of payments and receipts in foreigncurrency

Foreign exchange transaction costs will be shown presentlyto decline in relative terms with the size of the transaction.An accurate evaluation of the transaction cost savings theintroduction of a single currency will permit therefore re-quires information on the size distribution of settlements inforeign currency.

257

sduquenoy

sduquenoy

Annexes

Table A.9Currency distribution (%) of payments' by (or to) domestic residents to (or from) EC Member States (current account transactions and foreigndirect investment)

Italy* France3' * Germany* Denmark6

Receipts Payments Receipts Payments Receipts Payments Receipts Payments Receipts Payments

BFRFFUKLHFLDMLITDKRESCPTADRIRL

USD

ECUYEN

Other

30,812,55,98,0

19,33,10,6

0,020,6

0,010,1

16,0

1,60,6

0,9

25,511,85,08,6

23,22,80,5

0,030,90,10,2

19,0

1,00,5

0,9

2,117,67,12,6

26,437,10,50,12,60,00,13,2

0,40,1

0,1

2,415,64,34,9

31,028,8

1,30,02,10,10,2

8,4

0,50,3

0,2

9,259,15,51,79,83,00,00,00,10,00,0

15,2

1,3other7

2,1

2,951,94,61,8

11,23,20,10,00,30,00,1

16,4

1,4other

6,5

6,34,71,6

77,63,60,50,00,10,00,12,6

0,00,1

1,2

1,47,55,12,9

59,73,40,60,00,30,00,2

10,7

0,20,27,8

otherother

7,02,0

14,0other

37,0otherotherotherother

21,0

otherother

19,0

otherother

4,03,0

17,0other33,0otherotherotherother

23,0

otherother

20,0

Payments executed through the banking sector. Banknote payments are as a general rule not included.Payments arising from intra-EC trade account transactions and technology transfers.Payments arising from current account transactions only.Own computations on the basis of trade statistics according to the country or destination or origin and of the currency distribution of global current account transactions. Figures shouldtherefore be seen as indicative, in particular as it proved hard to distinguish between the EC and the rest of OECD-Europe.Trade account transactions only.Global current account transaction.Subsumed under 'Oiher'.

Table A.10Sum of intra-EC imports and exports of goods, 1988

EUR 12

(% ofCDP)

BLEUDenmarkGermanyGreeceSpainFranceIrelandItalyNetherlandsPortugalUK

87,325,925,627,317,123,877,318,363,144,820,9

26,8

Source: European Economy, Statistical annex.

The limited evidence available to this effect is assembled inTable A. 11. It suggests that more than 50 % of the volumeof bank settlements connected with trade or current accountoperations within the EC takes the form of payments andreceipts in excess of ECU 100 000. More than 90 % of thevolume of cross-border capital flows appears to be generatedby transactions worth ECU 1 million or more.

2.2. The cost of foreign exchangeThe financial costs economic agents in the Community incurdue to the multitude of currencies in the Community canvary greatly, depending on the currency of exchange, thenature of the foreign exchange 'product' (spot, forward,swaps, options, etc.), the size of the transaction and theimportance of the bank customer. Typically, moving in andout of small-country or weak currencies proves more costlythan buying or selling Deutschmarks or French francs; pay-ments with foreign banknotes cost much more than by creditcards; exchange risk coverage is invariably more expensive

258


Table A.llSize distribution of trade account settlements through means of payments other than banknotes

ECUECUECUECUECUECUECUECUECU

BLEU Germany

(a) (b) (c)

1 lo 10000 1,0 0,1 1,01 000 to 5 000 4,4 0,9 4,45 000 to 10 000 4,2 0,5 4,310 000 to 50 000 16,5 0,9 22,750 000 to 100000 8,6 0,8 6,5100 000 to 500 000 19,9 1 4,0500 000 to 1 mio 7,3 (59 Q 3,8 % g 61 21 mio to 5 mio 17,6 ' 26,05 mio + 14,7 63,0

Italy Weighted averageof BLEU,

(d) Germany and Italy

0,3 0,81,0 3,23,1 3,9

39,3 26,114,7 10,018,8

n!o 42'° 53'96,4

(a) Current account transactions and foreign direct investment with other EC Member Stales.(b) Gross capital flows (other than direct investment) lo and from other Member States.(c) Trade account transactions with the rest of the world.(d) Trade account and technology transfer related transactions with the other EC Member States.

than transacting on the spot market; and due to the existenceof overhead expenses, foreign exchange costs become rela-tively less important with the size of the transaction.

The bid-ask spread on the interbank marketAt the lower end of the cost-range are foreign exchangeoperations in the interbank market. Table A. 12 illustratesthe size, expressed in percentage terms, of the bid-ask spreadprevailing on this wholesale type of market. The underlyingdata relate to the purchase and sales rates displayed by amajor British bank in London in November 1989.

Table A.12Interbank bid-ask spreads expressed in percentage terms(data collected in November 1989)

USD ECU

USDUKLDMECUFFHFLYENBFRLITDKRDRPTAIRLESC

X0,060,040,040,080,050,060,100,070,100,180,090,100,13

X0,110,110,150,100,140,180,130,170,250,160,190,20

X0,080,120,090,130,160,120,14————

X0,130,10—0,140,130,13————

X0,130,200,330,20—————

X0,130,160,20—————

The implicit cost for a bank of obtaining foreign currencyon the spot market is half this spread. It averages around0,05 to 0,10 % for deals between EC currencies. Differencesin spreads across pairs of EC currencies appear primarilydriven by market size. The economies of scale phenomenonexplains why the spread on intensely traded European cur-rencies like the Deutschmark or the pound is only aboutone third of the spread registered for moneys in whichinternational transactions are very limited like the Greekdrachma or the Portuguese escudo. The spread on the cross-rate between the latter two can easily exceed 0,30 %-

The figures in the table confirm the pivotal role of the USdollar on present foreign exchange markets: the bid-askspread is always lowest when moving in and out of thedollar, with the spread relative to the cross-rate between twoEC currencies equalling or exceeding the sum of the spreadson the exchange rate between each of the two and the dollar.For example, the spread on PTA/UKL was 0,16 %, whichcan be divided into PTA/USD (0,09%) and USD/UKL(0,06 %).8

Bank transfers

Payments between firms of different countries are carriedout by means of international bank transfers. The exchangecost of a transfer in foreign currency consists basically of

inurce: Telerate.

This should not be interpreted to mean that 0,16 % was the lowest PTA/UKL spread to be obtained on that day. One could do better throughgeographical arbitrage. For instance, the spread could be reduced byselling pesetas for dollars to a major bank in Madrid posting a spread ofonly 0,05 % and subsequently purchasing pounds for dollars in London.

259

Annexes

two parts. Aside from an explicit conversion fee, which islargely fixed, a margin is added to the interbank rate atwhich the paying enterprise's bank has bought or sold thecurrency in question. This margin is inversely related to theamount, with the importance of the client and the currencyof payment often playing a role as well. As a result ofcompetitive bidding among banks and the fact that largeenterprises increasingly have direct access to the foreignexchange markets, this margin is reduced to zero for verylarge amounts and top customers. For payments with a zeromargin, the interbank spread dealt with earlier applies ineffect.

As no published information exists on banks' charges forthe foreign currency conversion of large amounts, the Com-mission services submitted a confidential questionnaire tothis effect to a limited number of banks in each MemberState. The replies received suggest that the spot purchase ofanother EC currency against the local money for an equiva-lent of ECU 10000 costs about 0,5%, with reportedextremes ranging from 0,1 % to 2,5%. The same foreignexchange transaction for an equivalent of ECU 100 000would cost on average about 0,3 %. Although for amountsof the latter magnitude pricing differences are much smallerdue to competitive pressure, costs can still rise above 1 %for purchases of weak currencies that are barely used ininternational transactions, like the drachma or the escudo.

Banks located in the less developed Member States appearto charge clearly higher prices than their counterparts else-where in the Community, a finding in keeping with theevidence on financial services' costs reported in Price Water-house (1988).9

Hedging by way of forward or similar contracts is bound tobe more costly than transacting on the spot market becauseit always involves a spot transaction plus at least one otherforeign exchange or loan operation. The replies to the ques-tionnaire suggest that on average forward contracts costabout 0,2 to 0,3 percentage points more than correspondingspot deals for amounts equivalent to ECU 10 000. This costdifference narrows down with the value of the contract,diminishing to about 0,1 percentage points for transactionsworth ECU 100 000.10

Eurocheques and credit cards

Eurocheques are a convenient international payments instru-ment when they are written in the currency of the foreigncountry and when the amount involved does not exceed theguaranteed disbursement limit (about ECU 160).

Domestic use of eurocheques is costless in most EC countriesupon the payment of a fixed fee. For their international usethe issuing bank charges a commission fee of about 1,6 %of the cheque's value (with a minimum amount), to whichmust be added the exchange margin when the cheque iswritten in a currency other than that of the account to bedebited. The applied exchange margin is determined freelyby the cheque issuing bank, but will typically lie around 1 %.

Exchange transaction costs with international credit cardsare analogous to those incurred with eurocheques. Also freeof charge for domestic use, the fixed fee for making paymentsin foreign currency equals 1 %, with the exchange marginapproximating very closely the one charged by banks foreurocheques.

Traveller's cheques and foreign banknotes

The highest transaction costs are faced with the purchase orsale of traveller's cheques and cash.

Traveller's cheques are a form of quasi-cash in that they aredirectly and in principle costlessly convertible into localcurrency at banks.,11 They are also often accepted as directpayment at retail outlets in tourist areas. Traveller's chequesare guaranteed against loss or theft, for which asupplementary 1 % commission is charged. Price Water-house (1988) surveyed the cost of an ECU 100 traveller'scheque in seven EC countries. The lowest cost was observedin Luxembourg, Germany and the UK and amounted toECU 5; France turned out most expensive at ECU 7,5.

An overview of the cost of foreign cash is provided inTable A. 13 showing the differences expressed in percentageterms between banks' buying and selling prices of foreignbanknotes. Columns indicate the place of currency conver-sion and all figures relate to exchange operations involving

Price Waterhouse (1988), pp. 147-148.Prices of various hedging instruments like forward, future or optioncontracts can differ considerably due to their distinct insurance andexpected return properties as well as to domestic tax- or regulation-induced distortions.

Cashing-in a traveller's cheque costs nothing when it is denominated inthe local currency. However, in the event of a different denomination— which is unavoidably the case in Belgium-Luxembourg, Denmarkand Greece as traveller's cheques in the local currency do not exist —costs can rise steeply.

260

sduquenoy

sduquenoy


Table A.13Buying-seliing spreads in percentage terms for foreign banknotes

BFRDKRDMDRPTAFFIRLLITHFLESCUKL

B

X4,14,6

25,05,35,34,45,04,9

22,85,0

DK'

8,6X

1,915,36,94,74,5

11,12,4

14,53,3

D

5,07,3

X48,2

8,16,46,67,72,6

30,06,3

GRJ

4,14,14,1

X4,14,14,14,14,14,14,1

E

3,83,83,85,6X

3,83,83,83,85,73,8

F

6,78,36,4

19,710,7

X—11,46,5

19,28,3

IRL

4,85,55,56,95,55,4X

6,15,56,83,0

i3

1,71,91,82,12,01,91,9X

1,91,91,7

NL

5,811,03,6

23,115,49,5

10,714,0

X21,710,0

p

2,81,61,3

—2,41,61,66,21,4X

1,6

UK

6,75,66,29,96,96,46,76,46,56,7

X

1 A fixed commission fee of about DKR 20 is charged.2 Regulated market; banks are free to chsirge additional fees.1 Banknote transactions are subject to a lax of 0,9%.Columns denote the place of currency conversion. Figures concern exchange operations involving (he local currency on one side of the deal.Source: Data collected from newspapers and banks.

the local currency.12 For example, the figure in the upperleft-hand corner signifies that exchanging Belgian francs inBelgium for Danish kroner banknotes and back would entaila loss of 4,1 %.

The cost of obtaining foreign banknotes fluctuates widely,depending on the country and the currency, but the bid-askspread will seldom fall short of 3 to 5 %. The banks in thepoorer Mediterranean Member States display the smallestspread.l3 The biggest losses are incurred when buying andselling in the countries of the 'centre' banknotes of reputedlyweak currencies like the drachma and escudo or, to a much

The data in Table A.13 should be seen as indicative because nationalmarkets for foreign banknotes are far from perfectly competitive. Asthe cost to foreign cash customers of gathering information rapidlysurpasses the benefits, total expenses charged by different banks withrespect to one and the same currency can vary by as much as 5 to 10percentage points, in spite of the fact that banknotes form a classicalexample of a homogeneous service.In Greece (where pricing in the foreign banknote market is regulated)and, to a lesser degree, Portugal (where the lion's share of retail bankingis still in public hands) this small spread stems from policy, inspired,inter alia, by a wish to secure a fair deal for the numerous banknote-purchasing foreign tourists. Two factors may go a long way towardsexplaining the low spread in Spain and Italy where there is no cleargovernment involvement. First, holding strong-currency banknotes in aweak-currency country is a less unattractive proposition for a bankthan the converse situation. Secondly, the possible existence of scaleeconomies in the management of foreign banknotes may lower trans-action costs in countries that still rely predominantly on cash as themeans of payment.

lesser extent, the peseta and lira. The worst case reported isthat of moving in and out of the drachma in Germany,which would wipe out nearly half of the original amount.

2.3. The costs of cross-border payments

Individuals and enterprises making cross-border bank trans-fers in the Community not only have to put up with currencyconversion costs, they are also subject to high paymentexecution fees and long delays compared to what it takes interms of cost and time to carry out a domestic bank transfer.

This is borne out by a BEUC report, based on a sample ofmore than 100 international bank transfers or eurochequesexecuted criss-cross throughout the Community, each timefor an equivalent of ECU 100 in the beneficiary's money. It-observed that total costs associated with bank transfers wereon average 12,1 % and that it took generally 5 working daysfor the recipient account to be credited. However, in anumber of cases charges rose to twice the average and thetime needed was much longer, sometimes lasting months.

Total costs of an international bank transfer from a Belgianfranc to a sterling account, for an equivalent of ECU 80,amount to about 18,5 %. The foreign exchange margin andthe explicit conversion fee equal 2,5 % or less than one-fifthof the total costs borne by the Belgian franc account holder.

261

Annexes

This anecdotal evidence is consistent with informationobtained on the costs for the German banking sector oftreating bank transfers. Whereas the execution and process-ing of a domestic transfer costs German banks aboutECU 0,5, the corresponding expenses for an internationaltransfer amount to ECU 10, with approximately ECU 2being directly attributable to currency conversion.

This contrasts with the minor costs charged in the USA,despite the fact that the American financial services marketis still fragmented owing to the MacFadden Act. In theUSA, a coast-to-coast cheque takes two working days anda fee ranging from 20 to 50 cents.

The relatively high expenses and long delays cross-borderpayments in the Community suffer from are in the first placedue to the existence of several technical barriers that needto be removed in the internal market framework.14 Buta single currency would significantly facilitate further thetreatment of cross-border payments since it would stronglysimplify banks' treasury management and accounting.Banks' treasury management would become simpler as longand short positions could be summarized in one currency.So would bank accounting as the number of "vostri' and'nostri' accounts of correspondent banks will be halved. Allsuch accounts vis-a-vis banks from other Member Statescould eventually disappear when every EC bank would holdan account with the European Central Bank.

2.4. The financial transaction cost savings froma single currency

The data on the volume of foreign exchange transactionsbetween EC currencies laid out in Section 2.1. and the evi-den&e reported in Section 2.2 on the cost of converting onecurrency into another can be combined to obtain someestimates of the exchange transaction gains a single currencywould bring, be it for the Community as a whole, individualMember States, or specific payment instruments.

Financial transaction costs for the Community as a whole

Exchange transaction costs for the Community as a wholecan be gauged from the turnover on the Community's for-eign exchange markets when it is supposed that EC residents,

Handling costs would be diminished by measures that would allowtransmitting electronically information on the payment of a eurochequein lieu of physically; so would the elimination of licensing constraintson electronic value-added networks and the development of commontechnical standards. These issues are dealt with in somewhat greaterdepth in HM Treasury (1989).

and only EC residents, purchase their foreign exchange ser-vices on EC markets (and those markets only). Two con-siderations suggest their correct measurement should bebased on the foreign exchange transactions arising frombusiness with non-bank customers only, thus leaving asideall interbank operations. First, the resources absorbed in thebank sector by interbank operations triggered by customerbusiness are eventually remunerated by the non-bank sectorthrough the foreign exchange margins and fees it pays tobanks. As a corollary, the inclusion of interbank operationswould give rise to double counting. Second, the bank re-sources currently employed in purely speculative or arbitrageoperations involving two EC currencies are unlikely to beput to another use following the creation of a single currency,but may instead be shifted to ecu/dollar, ecu/yen, dollar/yen, etc. operations.

It was estimated in Section 2.1. on the basis of Tables A.3and A.4 that between 34 and 43 % of foreign exchangeturnover on the Community markets concerns direct orindirect transactions-between two EC currencies. Relatingthis percentage range to the net turnover arising from busi-ness with non-bank customers, shown in Table A.I, an ex-change transaction volume of between ECU 4 100 and 5 200billion is arrived at for 1989.

The banking cost percentage to be applied to this aggregatevolume can be derived fairly accurately from the data onthe size distribution of current and capital settlements(Table A.l l) and the confidential information on bankers'charges for spot and forward transactions.

For want of specific information, a working hypothesisneeds to be adopted regarding the relative importance of theuse of forward contracts for current account payments. Thebasic assumption made in this respect is that the portion offoreign currency transactions that is hedged goes up withthe size of the amount. Thus, only 20 % of payments insidethe ECU 1 000 to 5 000 bracket are supposed to be boughtor sold forward, 30 % of the ECU 5 000 to 10 000 bracket,40% of ECU 10000 to 50000, 50% of ECU 50 000 to100 000. All transactions larger than ECU 100 000 are as-sumed to be covered. Hedging instruments other than ordi-nary forward contracts have been ignored since their use forcommercial purposes is still very limited.

On the basis of this set of assumptions, the data inTable A. l l and information collected on bankers' charges,the average currency conversion cost associated with currentaccount settlements can be assessed at 0,3 to 0,35%. Forcapital account transactions by the non-bank sector, costsare clearly lower as the average size of transactions is muchlarger and as the latter have been supposed not to involve

262


forward contracts. Costs have therefore been set at 0,1 to0,15%. The data on the only country (BLEU) for whichprecise information is available to this effect, suggest thatgross capital flows by the non-bank sector are at least fivetimes as large as current account flows. A more conservative4:1 ratio has been taken for the whole of the Communityon account of the capital controls still operated in 1989by several Member States, leading to an overall averagetransaction cost estimate of 0,15 to 0,2 %.

Applying this cost percentage to the ECU 4 100 to 5 200billion transactions' volume identified earlier, the Com-munity's financial transaction cost savings from a singlecurrency as derived from data on the EC's foreign exchangemarket turnover can be estimated to have lain in 1989between ECU 6,2 and 10,4 billion.

Obviously, if one were to include interbank operations,'financial' savings from a single currency would grow muchlarger. Assuming that 40 % of such foreign exchange oper-ations concern two EC currencies and that a quarter of thereported total take place with banks located outside theCommunity, the BIS survey data suggest that interbankoperations involving two EC currencies equal aboutECU 21 500 billion on an annual basis. Applying an averagecost of 0,075 % in accordance with the findings ofTable A. 12, exchange transaction costs connected purelywith interbank operations would amount to slightly morethan ECU 16 billion.

The ECU 6,2 to 10,4 billion range just arrived at doesnot include the exchange transaction costs associated withsmaller payment instruments like banknotes, traveller'scheques or eurocheques.

The gain from no longer having to obtain foreign banknotesupon the creation of a single currency can be gauged fromthe estimate of total foreign EC banknote sales and pur-chases based on Table A.7, and the bid-ask spreads onthe banknote market reported in Table A.13. The weightedaverage of costs related to banknote transactions in theBLEU, Germany, Italy and France amounts respectively to2,5 %, 2,5 %, 1,3 % and 3,8 %. The overall weighted averagefor these Member States together equals 2,3%. Applyingthe latter percentage with the banknote turnover estimate ofECU 57 to 85 billion put forward in Section 2.1, banknote-related transaction cost savings can be thought to havevaried in 1988 between ECU 1,3 and 2 billion. As regardseurocheques, potential savings can be set at around ECU 125million (40 million cheques with an average value ofECU 125 and costs usually within the 2 to 3 % range). Onthe assumption that exchange transaction costs associatedwith traveller's cheques amount to 3 % on average, single

currency savings related to the latter payments instrumentcan be put at about ECU 150 million. The economies to beobtained with regard to the cross-border use of credit cardsare likely to amount to between ECU 150 to 200 million,with average transaction costs somewhat smaller than 2 %and a total foreign EC currency payments volume of roughlyECU 10 billion.

Assembling the various cost components and expressingthem in prices of 1990, the financial transaction costs in-curred by the Community as a whole due to the absence ofa single EC currency can be estimated to lie between ECU 8and 13 billion or 0,17 and 0,27 % of the Community's GDP.

The validity of this cost range is confirmed by confidentialevidence on the revenue the Community's banking sectorderives from its foreign exchange related services.

This suggests that almost 5 % of the EC banking sector'sincome arises from intra-EC foreign exchange activities. Theshare of the banking sector in GDP standing at close to 6 %,it follows that this income is equivalent to about 0,25 % ofCommunity GDP. It is noteworthy that almost half of theEC banks' intra-EC foreign exchange revenue turns out toaccrue in the UK. This is largely due to the fact that asthe UK hosts the world's biggest foreign exchange market,claiming 25 % of global net foreign exchange turnover, alarge part of spot and hedging operations arising in otherEC Member States also involve a financial institution locatedin the UK. Thereby, many non-financial EC companiesoutside Britain 'import' indirectly foreign exchange servicesfrom the UK.

Individual Member States' transaction costs

It was shown in Section 2.1 that the degree of exposure toforeign EC currencies varies strongly from country to coun-try. The replies to the Commission's questionnaire providedevidence that there still exist strong differences betweenMember States regarding the prices banks charge for foreignexchange services. As a corollary, the transaction costsavings a single currency will generate will not be spreadevenly across the Community. The small open economiesand those characterized by a relatively inefficient bankingsector will be the chief beneficiaries.

Combining the data of Table A.8, A. 11 and the informationcontained in bankers' replies, it can be calculated that ex-change transaction costs borne by the BLEU equalled closeto ECU 1.2 billion in 1988 or 0,9% of the BLEU's GDP.In contrast, even if one assumes, in conformity with theBLEU data, that Germany's capital imports and exports bynon-bank enterprises and households to other Member

263

Annexes

States in foreign EC currency are five times as large as itsEC currency payment flows connected with current accountoperations, Germany's transaction costs would have am-ounted in 1988 to only ECU 2.23 billion or 0,11 % of itsGDP. Calculations relying on the capital flow figure re-ported in Table A.8 put the exchange transaction costs forFrance in 1988 at ECU 920 million or 0,11 % of GDP.15

Making the same relative assumption as for Germany withrespect to the size of capital flows would raise the cost to0,15 % of GDP. The same percentage would be obtainedfor Italy, where foreign exchange services appear cheaperthan in the Community on average.

The available information thus suggests that in relative termstransaction costs can be 8 times more important for smallopen economies than for the largest Member State whosecurrency is a generally accepted means of payment insidethe Community.

Although no precise figures can be put forward to this effect,all less developed Member States are also likely to reaphigher than average transaction cost savings from the cre-ation of a single currency. Ireland is strongly exposed toforeign EC currencies given the marginal international roleof the Irish punt and the country's high degree of openness,particularly vis-a-vis the UK economy. The strong volatilityof the pound has probably prompted systematic hedging onthe part of Irish traders, aggravating the exchange trans-action burden. The three most recent Member States standto gain relatively much given the poor efficiency and degreeof sophistication of the local banks' foreign exchange ser-vices. This holds especially for Portugal and Greece, whoseintra-EC trade relative to GDP exceeds the Communityaverage and whose currencies are virtually not accepted asmeans of international payment.

Cost reductions of cross-border payments

It can be estimated on the basis of BIS statistics that in 1988around 4,4 billion so-called paper-based credit transfers werecarried out in the 12 Member States.16 Assuming that 5 %of these transactions concerned a cross-border transfer insidethe Community and that, in conformity with the findingson the German banking sector, the difference in processingcosts between a domestic and an international bank transferequals ECU 6, the establishment of a single currency, alongwith internal market measures facilitating cross-border pay-ments, would generate an additional transaction cost savingof the order of ECU 1,3 billion.

15 For want of specific information on bank charges in Germany andFrance, average banking costs as derived in Section 2.2 were applied.

16 BIS (1989).

3. In-house costs

As regards the costs borne inside non-financial companiesfrom having to work in the Community with a variety ofcurrencies, a distinction can be made between direct andindirect costs. Some of the direct effects to be identifiedpresently are in principle quantifiable by means of a tho-rough internal audit of the company. The indirect effectsrelate primarily to the impact a single currency may haveon a transnational corporation's business organization andstrategy by removing a major factor of managerial com-plexity. They are likely to take more time to materialize andare impossible to quantify with any degree of precision.None the less, they may turn out to have a more pervasiveimpact on the business sector in the Community than thedirect effects.

3.1. Direct and indirect effects

Four sources of direct costs — or of direct in-house gainsfrom a single currency — can be identified.

First and most obvious, there are the people and equipmentemployed in treasury and accounting responsible for nettingthe company's foreign exchange exposure and overseeing themultitude of spot, forward, currency swap, etc. transactions.Yet, as these are primarily overhead expenses and businesswith non-EC countries will continue to require foreign ex-change management, the scope for administrative savingsfrom the establishment of a single currency is bound to belimited.

Second, multiple currencies lead to company cash beingpoorly remunerated or, conversely, to high interest costs ondebit positions. This is due, on the one hand, to the disper-sion of balances over an extensive number of accounts invarious currencies held by the parent company or the foreignsubsidiaries and, on the other, to the fact that banks offercash pooling services only for accounts in one and the samecurrency. With a single currency it becomes much moreconvenient to finance the various units of a corporate groupfrom a central source.

Third, the delay between the point in time of debiting anaccount and that of crediting the correspondent recipientaccount, the so-called 'float' is, as a general rule, much largerfor transfers involving different currencies than for paymentsin one and the same money. Part of this difference in delayswill be eliminated as a result of the measures in the frame-work of the 1992 programme towards the creation of a singleEC financial market, but executing and clearing cross-border

264

sduquenoy


payments in the Community will remain inherently morecomplicated, hence costly, in the absence of a single currency.

Fourth, multiple currencies may cause firms to incur anopportunity cost arising from the search for natural hedges.Rather than managing it, many enterprises try to avoid asmuch as possible the foreign exchange problem by matchingreceipts and payments per currency or by pursuing a deliber-ate policy of diversifying foreign exchange risk.17 To theextent that this strategy leads to business forgone, lessfavourable conditions at the purchase of inputs, or smallerprofit margins, the company undergoes an implicit cost.

A single currency in the Community will induce indirecteffects inside firms closely tied to the role of money as aunit of account. By virtue of the increased cost and pricetransparency it induces, a single currency will facilitate themonitoring of firms' internal transfer pricing practices aswell as the reporting and analysis of foreign subsidiaries'performances. The resulting simplification of central man-agement's task of control and evaluation should improvethe decision-making process in transnational companies andrender international business strategy in its various aspectsmore sound and easily adjustable.

Furthermore, a single currency is likely to exert a non-trivial influence on the cohesion between the various units oftransnational firms. Along with the establishment of thefour freedoms and the internationalization of share owner-ship and management, the use by business of a single cur-rency will nurture the emergence of enterprises with tho-roughly European business attitudes and strategies.

3.2. Size of in-house costs

Quantifying these in-house costs with a high degree of accu-racy is very difficult. No relevant statistical material is avail-able and none of the enterprises interviewed to this effectwas able to come up with a precise comprehensive measure.This is not surprising, since obtaining a rigorous estimatewould require a significant analytical effort on their side asthe cost from not working in a single currency is widely

diffused over various central departments and foreign sub-sidiaries of a transnational group. Even multinational com-panies having imposed or thought of imposing a singlecurrency for internal invoicing, payments and reporting pur-poses — often the currency of the headquarters, sometimesthe ecu or dollar — based their analyses on primarily quali-tative indications.

A small set of case-studies conducted by Ernst & Young onbehalf of the Commission services (see Box) suggests thaton average the cost of treasury and accounting personneland equipment employed due to the existence of multiplecurrencies in the Community could amount to about 0,1 %of intra-EC exports. As those expenses are largely fixed, thisfigure is likely to be much higher for small and medium-sized enterprises and lower for multinational enterprises. Forexample, in a recent survey of Belgian small and medium-sized enterprises, internal foreign exchange management wasestimated to occupy 1,2 % of personnel and cost around0,3 % of total turnover.l8 The other quantifiable direct costs,primarily those related to the dispersion of balances over anextensive number of accounts in various currencies and tocross-border payment delays can be reckoned to be around0,1 % of turnover in other Member States as well.

Given that a company's value-added amounts on average toabout 55 % of its turnover, these direct sources of in-housecosts would equal 0,36 % (i.e., 0,2/0,55) of the value-addedgenerated through intra-EC exports. As intra-EC exports ofgoods and services represent close to 20 % of CommunityGDP, one can conclude that the costs borne inside com-panies from having to operate in the EC with 12 moneysare equivalent to 0,07 to 0,08% (one fifth of 0,36%) ofCommunity GDP.

This figure is a lower bound estimate as it does not take intoaccount the non-quantifiable direct and indirect sources ofin-house costs related to avoidance of exchange risk exposureand managerial complexity. Evidence in support of this claimis that according to a large-scale business survey conductedby Ernst & Young (1990)19 one internationally active ECfirm out of three thinks a single currency will permit savingsunrelated to banking costs exceeding 0,5% of turnover inforeign EC markets.

17 Foreign exchange risk is said to be diversified if the depreciation riskrelative to one or more currencies is offset by an appreciation riskrelative to other currencies.

18 DePecunia, (1990), p. 31.19 Ernst & Young, (1990), p. 54.

265

sduquenoy

Annexes

Box A.I: Evidence on in-house costs

A consultancy firm (Ernst and Young) was asked by the Com-mission services to undertake a small number of case studieswith a view to obtaining an estimate of how large is the cost —as a percentage of turnover in other EC countries — to firmsof having to deal with multiple currencies in the Community.

Two companies were examined in Germany, France, Spain andthe United Kingdom. In order to enhance the representativecharacter of the sample, the selection of enterprises was madeso that one firm in each of the four countries produced onlydomestically, but with significant imported inputs and salesabroad, whereas the other company had a manufacturing basein more than one EC country.

The firms were requested to provide an estimate of total in-cremental costs due to the absence of a single currency and toisolate the administrative component of these expenses. Theirreplies are reported in the table below, the French enterpriseshaving been omitted as they felt unable to advance specific costfigures.

Country Company Incrementalcosts -

Administrative

Total costs(in-house + financial)

Germany

United Kingdom

Spain

0,1%0,05%0,4%0,1%0,5%0,0%

0,1%0,05%1,0%1,0%2,1%0,3%

German firm No 1, active in shoe manufacturing and otherleatherware, with a total turnover of more than ECU 300million, thought that dealing in various EC currencies absorbedlittle administrative resources, since currency management iskept simple (all cash flows in foreign currency being coveredforward), and invoicing and accounting largely automated. Thecurrency denomination of bank transfers was not perceived asa source of payment delays. With netting of short and longpositions in the same currency systematically pursued and anoverwhelming share of international settlements either in De-utschmark or in dollars, financial exchange transaction costs inEC currencies were seen as insignificant relative to EC exports.

German firm No 2, a very large multinational company inelectronics, with EC imports and exports amounting respectivelyto ECU 400 and 1 800 million, deems it incurs trivial exchange

transaction costs as all exports are payable in Deutschmarksand only a small fraction of imports is in foreign EC currency.Areas where the absence of a single currency leads to very smallincremental expenses are in marketing and the processing offoreign currency invoices.

UK firm No 1, a large company in garments and home fur-nishings, with a global turnover of nearly ECU 400 million, hasproduction facilities and an extensive retail distribution networkon the continent. It estimates the absence of a single currencycauses significant administrative costs in the domains of sales,marketing, invoicing and accounting. Despite the complex inter-national finance operations, it reckons the scope for savings intreasury are very limited because of the remaining need tomanage foreign exchange with respect to non-EC currencies.Major sources of other than administrative costs are the delaysin receiving funds as well as the maintenance of overdrafts inthe UK, whilst EC subsidiaries have surplus cash balances thatare uneconomic to remit to the parent company.

UK firm No 2, a small precision engineering firm with a turnoverof around ECU 10 million has too little market power andtherefore has to accept the importer's money as the currency ofsettlement. It only covers its foreign exchange exposure oncontracts denominated in US dollars. Exchange risk in ECcurrencies is not hedged; instead, the normal contract price isincreased by 0,5 % for sales in other EC markets. This uncoveredposition relative to EC currencies cost the firm dearly in 1988-89 when the company faced an exchange loss equal to 6 % ofits exports to other Member States.

Spanish firm No 1, an important frozen seafood manufacturer,with exports to other EC markets of about ECU 20 million,thinks administrative costs could be reduced significantly by asingle EC currency, although the figure mentioned in the tablerelates to the situation when all its foreign exchange operations,including those in non-EC currencies, would have been elimin-ated. The other costs would essentially be the payment of mar-gins and fees to the banking sector. The 2,1 % reported in thetable refers again to exchange transaction costs associated withall foreign currency operations.

Finally, Spanish firm No 2, an important manufacturer ofpetrochemicals, with sales in other EC countries of ECU 300million, saw little scope for administrative cost reductions be-cause its currency management is limited to simple forwardcontracts. Furthermore, in the petrochemical sector, the dollaroccupies a predominant position. Financial costs were thoughtto amount to 0,3 % of EC turnover, although in this caseas well the company was unable to distinguish between dealsinvolving EC and non-EC currencies.

266


4. Once-and-for-all adjustment cost ofintroducing a single currency

Against the durable benefits arising from the elimination oftransaction costs must be set the once-and-for-all adjustmentcosts in the transition to a single currency. National bank-notes and coins currently in circulation will have to be calledin and reissued. Firms and banks will undergo some 'menucosts' as they have to convert prices, balances and wagesinto the new base. They will also need to modify theiraccounting and computing systems. Some equipment, inparticular vending and teller machines, will have to be al-tered. Last but not least, the European public at large willhave to adapt to the change, much in the same way asFrench economic agents had to do upon the introduction ofthe 'new' French franc at the end of the 1950s and the Britishfollowing the decimalization of their currency some 20 yearsago.

5. Cost summary

By way of conclusion, it is appropriate to sum up the variouscost savings identified in the foregoing analysis. Table A. 14provides this summary picture.

Table A.14Cost savings on intra-EC settlements by single EC currency

(billion ECU. 1990)

Estimated range

Financial transaction costs:Bank transfersBanknotes, eurocheques, traveller'scheques, credit cards

Subtotal

In-house costsReduction of cross-border payments cost

Total

6,4

1,8

8,23,61,3

13,1

10,6

2,5

13,14,81,3

19,2

Note: Exchange transaction costs associated with several sources of in-house costs are notincluded in this table.

267

sduquenoy

Annexes

References

Bank for International Settlements (1989), Statistics on pay-ment systems in 11 developed countries, Basle, December.

Bank for International Settlements (1990), Survey of foreignexchange market activity, February.

Black, S. (1989), Transaction costs and vehicle currencies',IMF Working Paper No 89/96, November.

Demarsin, P. (1990), 'Volume et structure des marches deschanges en Belgique et a 1'etranger', Cahiers de la BanqueNationale de Belgique, No 6.

De Pecunia (1990), PME, risque de change et ecu, February.

Ernst and Young (1990), A strategy for the ecu, London.

HM Treasury (1989), An evolutionary approach to economicand monetary union, London, November.

Price Waterhouse (1988), 'The cost of non-Europe in finan-cial services', Research on the cost of non-Europe, Basicfindings. Vol. 9, Commission of the EC, Luxembourg.

268

sduquenoy

Annex B — Germany and the Netherlands: the case of a de facto monetary union

Annex B

Germany and the Netherlands:the case of a de facto monetary union

The Dutch-German quasi monetary union in the past decade,does not seem to have created major problems for the conductof economic policy in the Netherlands. Monetary policy hasfollowed closely the policy of the Bundesbank. The interestrate differential, both with respect to short and long-termrates, has narrowed gradually. This differential was signifi-cantly larger at the beginning of the decade, the narrowingprocess was interrupted after the 1983 depreciation but thedifferential has almost disappeared in the last two years. Nospeculation against the guilder seems to have occurred duringthe period under consideration, with short-term interest ratesnot experiencing greater short-term volatility in the Nether-lands.

Fiscal policy, without triggering credibility problems for theexchange rate commitment, was more flexible in the Nether-lands than in the Federal Republic, particularly in 1986-87when it responded to the sharp fall of gas revenues. In thefollowing years, despite a great need for fiscal consolidation,a certain degree of inflexibility frustrated the desire to reducethe deficit. As a result, the Netherlands experienced a differentpattern of the public debt to GDP as compared with Germany,implying a huge consolidation effort for the future. However,monetary policy has never been damaged by fiscal develop-ments.

Wage developments seem increasingly to reflect the externalcompetitive position as an instrument of improving domesticemployment. Such a flexible approach greatly facilitates thestabilization process.

As regards economic targets, it is highly doubtful if the Dutcheconomy would have performed better in a flexible exchangerate regime. The inflation performance would have been nega-tively affected by a depreciation of the guilder while thebeneficial economic effects of a depreciation would have beenoffset by negative effects on confidence, i.e. higher long-terminterest rates.

1. Introduction

In the EMS the exchange rate between the currencies of theFederal Republic of Germany and the Netherlands has been

most stable. Only two realignments in 1983 and 1979 respect-ively have seen the Deutschmark appreciate by two percent-age points more than the Dutch guilder on each occasion.Since 1984, the bilateral nominal exchange rate of theDeutschmark and the guilder has actually been stable(Graph B.I). Therefore, the experience of the Netherlandsprovides an obvious example of a de facto monetary unionbetween a small country and a large country. Indeed, thecountry which has the longest experience with fixed exchangerates in the EMS is also the country which has experiencedthe least variability and the smallest differentials in interestrates vis-a-vis the anchor country.

Several advantages to the Netherlands from the policy of astable HFL/DM rate can been stressed:(i) the Federal Republic of Germany is the most important

trading partner of the Netherlands;(ii) a stable relation between HFL/DM facilitates a high

degree of price stability (low inflation) providing anominal anchor for the smaller economy of the Nether-lands;

(iii) the credibility incentive of this policy is reflected in thegradual decline in the interest rate differential vis-a-visthe Federal Republic; this might be advantageous tothe process of budgetary consolidation;

(iv) the link of HFL/DM without conditions has given thefinancial markets confidence in the policy of the Dutchauthorities.

As financial markets are at present very open arid relatedbetween the two countries, any doubt about the exchangerate would now be translated in capital movements andpressures on interest rates. Integration of the financial mar-kets makes it impossible for a small economy to control thegrowth of money supply, as only internal sources can beregulated by an independent monetary policy. This arguesin favour of a small country pegging its exchange rate onthe stable currency of a larger important economic partner.

On the other hand, the Dutch and German example facili-tates the analysis of the potential costs of abandoning aneconomic policy instrument, i.e. the exchange rate. A largerburden will be placed on other instruments, particularly onfiscal policy. However, fiscal consolidation — which is oneof the main objectives of economic policy in the Netherlands— might conflict with the greater stabilization role fiscalpolicy has to play.

The results of this analysis are tentative, however, becausethey are to a large extent judgmental. The difficulty is that acomparison cannot be drawn with possible economic policydevelopment and performance in the alternative case where

269

sduquenoy

Annexes

GRAPH B.I: The bilateral Deutsctunark/guilder exchangerate: 1979-90 (HFL per DM, annual average)

1,0778 79 80 81 82 83 84 85 86 87 88 89 90

GRAPH B.2: Growth of production potential in the manufac-turing sector in the Federal Republic of Ger-many and the Netherlands

(annual % change)

2,5

1.5

0,5

_l——————I——————I——————U

80 81 82 83 84 85 86 87

•NL

a flexible exchange rate would have been applied. The resultsshould not be directly applied to other countries and particu-larly not to large countries. Thus the present paper is primar-ily devoted to analysing the adjustment mechanisms used inthe Netherlands to cope with shocks which have asymmetri-cally affected both economies.

2. Comparison of economic performance

2.1. Macroeconomic performance

In terms of the supply side, the performance of the twoeconomies shows a general similarity in the evolution. GraphB.2 shows that the growth in production potential of themanufacturing sector of both countries decreased to below1 % during the first half of the 1980s but has acceleratedsince 1983-84. This acceleration has been more pronouncedin the Netherlands. Until 1984 the growth of potential outputin the Netherlands has been below growth in the FederalRepublic and superior afterwards.

Graph B.3 compares the growth performance of the twocountries. In the Netherlands real GDP growth was lower(between 0,5 and 1 % per year) in the period 1979-82. Afterthe 1982 recession both countries recovered at a similarspeed. After the oil price decline in 1986 real growth tendedto be lower in the Netherlands than in the Federal Republic.Prospects for 1989 and 1990 indicate a renewed narrowingin the growth gap.

While output in the energy sector has followed a particularpattern of development and the public sector has expandedmuch less than the private economy, growth in privateenterprises (excluding energy) has generally been (except in1986) more favourable in the Netherlands than in the FederalRepublic since 1984. During that period, following a deeperrecession than in the rest of the Community at the beginningof the 1980s, the Netherlands experienced a rather sustainedupturn in growth.

The comparison of the respective employment performancesreflects the different economic growth patterns (see GraphB.4a). Graph B.4b shows that the unemployment rate (Euro-

270


GRAPH B.3: Real growth in the Federal Republic of Germany and the Netherlands (1979-90) — Gross domestic product at constantmarket prices (annual % change)

4 -

-2 I—————I—————I—————I—————I—————I—————I——————1—————T1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990

NL

Table B.I

Growth in the value-added of enterprises in the Federal Republic of Germany and the Netherlands (excluding energy) (1980 prices)

Germany

Netherlands

1981

0,2

0,1

1982

-0,5

-0,8

1983

1,6

1,1

1984

3,4

4,5

1985

2,6

3,0

I9S6

2,6

2,2

1987

1,3

1,4

1988

4,3

4,5

1989

4,4

4,8

1990

4,0

3,8

271

sduquenoy

Annexes

GRAPH B.4a: Employment in the Federal Republic of Ger-many and the Netherlands (total economy)(changes in %)

-278 79 80 81 82 83 84 85 86 87 88 89 90

•ML ——D

GRAPH B.4b: Unemployment rate in the Federal Republicof Germany and the Netherlands (percentageof civilian labour force)

14

12

10

78 79 80 81 82 83 84 85 86 87 88 89 90

•ML

stat) is currently about 2,5 percentage points higher in theNetherlands than in the Federal Republic.

Between 1970 and 1979 inflation was very high in the Nether-lands. The monetary authorities tried, by linking the guilderto the Deutschmark, to achieve an appreciation of the guilderin order to break the wage-price spiral and to obtain alower inflation rate. After 1979, both countries experienced aperiod of gradual disinflation, during the first half of the1980s, while having experienced inflation rates above 5%

-immediately after the second oil price shock (Graph B.5).

The process of disinflation started in 1981 and in 1986 zeroinflation was achieved in the Federal Republic. Since 1987,the inflation rate in the Netherlands has been lower thanthat of the Federal Republic. Most likely, underlying in-flation was very similar during the period and, to a consider-able extent, the relatively large gap evidenced in 1989 wasthe result of diverging tax policies. While VAT has beenlowered in the Netherlands, excise duties were increased inthe Federal Republic. In addition, the price dampeningeffect of the oil price drop in 1986 has materialized in theNetherlands with a certain time lag as natural gas consump-

tion is more important in the Netherlands than in Germanyand the level of its prices is to a certain extent a discretionaldecision.

The overall increase in the price deflator of private consump-tion has actually been the same during the last decade,indicating that both countries pursue similar stability-oriented policies. This, indeed, is an important beneficialconsequence of the strong currency option chosen by theNetherlands.

Graph B.6 indicates significant differences in the develop-ment of the current balance of payments. These differencescan mostly be explained by different structural charac-teristics, since the dependence on energy imports differs.Whereas the Federal Republic is a net importer of energy,the Netherlands has become self-sufficient. Because of thetime lag with which prices of natural gas follow oil prices, thecurrent balance of the Netherlands improved significantlyrelative to that of Germany in 1981. On average, the currentaccount surplus of the Netherlands was 2,5 % of GDP largerthan that of the Federal Republic. After the fall in energyprices the relative position changed. In terms of GDP the

272

sduquenoy


GRAPH B.5: Inflation in the Federal Republic of Germanyand the Netherlands — Price deflator of privateconsumption (changes in %)

78 79 80 81 82 83 84 85 86 87 89 90

NL

GRAPH B.6: Current balance in the Federal RepubUc of Ger-many and the Netherlands (% of GDP)

78 79 80 81 82 83 84 85 86 87 89 90

•NL

German current account has exceeded the Netherlands cur-rent account by about 2,5 % since 1987.

2.2. Structural and sectoral characteristics

At the sectoral level, the two economies show a differentstructure (see Table B.2). In both countries, the service sector

is the most important (42,5 to 44%); the weight of themanufacturing sector is very high, although declining, andit shows a significant difference in the two countries (33 %in the Netherlands and 41 % in Germany); investment goodsare more important in Germany, while for consumer goodsthe difference is less marked.

Both economies are strongly export oriented (see TableB.3). But the Netherlands, as a small open economy, has a

\

Table B.2Production according to sector in the Federal RepubUc of Germany and the Netherlands (% of total)

Federal Republic of Germany

AgricultureEnergy, water, miningManufacturing sector

basic materialsinvestment goodsconsumer goods

ConstructionServices

1970

3,34,6

48,513,417,817,47,5

35,4

1980

2,46,6

41,611,016,514,06,8

42,6

1986

2,06,2

41,49,9

17,913,05,6

44,0

1970

6,26,5

34,36,49,8

18,19,6

43,4

Netherlands

1980

4,510,630,7

7,18,5

15,19,0

45,1

1986

5,313,232,7

9,18,7

14,96,3

42,5

Source: BDS.

273

sduquenoy

Annexes

Table B.3Export shares in the Federal Republic of Germany and the Netherlands (in % of total production)




1970

4,46,2

18,819,629,47,5

1980

5,96,9

27,027,837,514,0

19B6

8,74,2

33,033,219,119,1

1970

30,232,946,374,849,034,6

Netherlands

1980

42,049,153,782,458,037,8

1986

39,149,460,476,268,945,7

.Source/BOS.

significantly higher export ratio (almost 60 % of GDP) thanthe Federal Republic (33 %). This makes the Netherlandseconomy more dependent on developments in world trade.Export dependency is in general much higher in the Nether-lands than in the Federal Republic in all branches.

In addition to the export share of total production, thedegree of self-sufficiency is an important factor determininghow susceptible a country is to fluctuations in world demand.From Table B.4 it is clear that the economy of the FederalRepublic is more specialized than the Dutch economy. Theself-sufficiency of the Dutch economy in energy productionand consumption is also evident.

Table B.4Export-import ratio in the Federal Republic of Germany and the Netherlands (in %)




1970

1237

13311827759

1980

1417

11213423169

1986

1919

13814322086

1970

8686898767

109

Netherlands

1980

9687

1001298296

1986

10714010213476

112

Source: BDS.

3. Comparison of economic policies

3.1. Monetary policy and inflation

Graphs B.7 and B.8 show the development of interest ratesin both countries. Short-term interest rates moved almostin parallel, although in the period 1985-87 rates in theNetherlands were above prevailing German rates. The nar-rowing of the differential was not primarily linked to theround of monetary tightening but can be considered as a

274

return to a 'normal' situation after the reverse oil shock of1986-87.

The movement of long-term interest rates shows a continu-ous narrowing of the interest rate differential between the >two countries. In 1988, the interest rate gap actually disap-peared. The disappearance of the long-term interest ratedifferential in 1988 might however be due to the introductionof the withholding tax which had an effect on German long-term interest rates in particular. The development of long-term interest rates indicates a progressive diminution of


GRAPH B.7: Short-term interest rates in the Federal Repub-lic of Germany and the Netherlands — 1978-90 (quarterly averages)

78 79 80 81 82 83 84 85 86 87 88 89 90

NL

GRAPH B.8: Long-term interest rates in the Federal Republicof Germany and the Netherlands — 1978-90(quarterly averages)

78 79 80 81 82 83 84 85 86 87 88 89 90

NL ——D

medium-term expectations of a guilder depreciation againstthe Deutschmark.

Graph B.7 shows that short-term interest rates have not beenmore volatile than would have been suggested by importantchanges in exchange rate expectations. Indeed, fluctuationshave become much smaller than in the 1970s.

The fact that markets have actually attributed a large degreeof credibility to the exchange rate commitment is illustratedby the fairly close movement of short-term interest rates andthe progressive narrowing in the long-term interest ratedifferential.

The difference between long- and short-term interest ratescan serve as an indicator of the stance of monetary policy.Developments in the two countries show that monetarypolicy was tighter in the Federal Republic in the early 1980s.Although since 1984 no major differences are detectable,monetary policy in the Netherlands seems to have becomemarginally more restrictive compared to the Federal Repub-lic, since the yield curve is somewhat steeper.

3.2. Fiscal policy

In the period under review, net borrowing of the generalgovernment in the Netherlands was larger than in the Fed-eral Republic. While net borrowing declined during a periodof consolidation between 1982 and 1985 in the FederalRepublic, the fiscal balance deteriorated significantly in theNetherlands (7,1 % of GDP) in 1982. The Government thenintroduced a stabilization programme and up to 1985 thedeficit declined to about 5 % of GDP. The slight increase inthe deficit in the Federal Republic after 1986 was a conse-quence of a fairly low growth of nominal domestic demandand reductions in income taxes. On the other hand, despitefurther attempts to cut back the public deficit in the Nether-lands, it increased in 1986 and 1987 to more than 6% ofGDP, because of oil price-induced losses in revenues. In1986-87 these losses represented some 4,1 % of net nationalincome. The deficit in the Federal Republic has, over thewhole period, been reduced from 3% of GDP to 1%.However, in the Netherlands the deficit, corrected for thegas revenue factor, passed from 6 to 5 % of GDP. Therevenue from natural gas, which was 16% of the totalbudget income, fell to 4 % in 1989.

275

sduquenoy

Annexes

GRAPH B.9: The difference between long-term and short-term interest rates in the Federal Republic ofGermany and the Netherlands (percentagepoint)

78 79 80 81 82 S3 84 85 86 87 88 89

GRAPH B. 10. Net lending (+) or net borrowing (-) in theFederal Republic of Germany and the Nether-lands — (i*neral government (% of GDP)

78 79 80 81 82 83 84 85 86 87 88 89 90

•NL

GRAPH B. 11: Compensation per employee in the FederalRepublic of Germany and the Netherlands (an-nual % change)

078 79 80 SI 82 83 84 85 86 87

•NL — — D

89 90

GRAPH B.12: Real compensation per employee in the FederalRepublic of Germany and the Netherlands (an-nual % change)

78 79 ,80 81 82 83-84 85' 86 37 88 89 90

•NL

276


3.3. Wage developments

Between 1979 and 1982, wages in the Netherlands movedindependently of wages in the Federal Republic. However,in the period 1983-89, wages in the Netherlands have increas-ingly responded to the need to preserve external competitive-ness and to ensure a better employment performance. Thisdevelopment was merely a consequence of the agreementbetween the social partners in November 1982 to accept aperiod of wage moderation.

Since 1984, the increase in both nominal and real wages hasbeen lower in the Netherlands than in the Federal Republic.The significant change in wage behaviour seems to haveoccurred after the last relative depreciation of the guilder on21 March 1983. Nominal wages actually stagnated in 1984and have risen by about half the German rate since then.Wages in the industry sector (in the Netherlands wages in thepublic sector were frozen) show a slightly different pattern;because of similar developments in prices, real wage moder-ation has been more significant in the Netherlands than inthe Federal Republic (Graph B.12).

The exchange rate policy in the 1970s led to a sharp rise ofrelative prices in the Netherlands, an important factor of thedeterioration of the competitive position in those years. Tobreak the wage-price spiral with this policy a high price hasbeen paid; the bad competitive position resulting from itcould be reversed by the moderating effects of high unem-ployment figures on wages.

4. The asymmetric effects of shocks

4.1. Energy prices

Graph B.I3 shows the development of the terms of trade inthe Netherlands and the Federal Republic of Germany.Clearly the two oil price shocks have influenced the twoeconomies asymmetrically. The increase in oil prices at thebeginning of the 1980s led to a significant fall in the termsof trade of the Federal Republic relative to those of theNetherlands while the fall in oil prices in 1986 had the reverseeffect. All in all, movements in the terms of trade havebeen more volatile in the Federal Republic than in theNetherlands, where the terms of trade have improved gradu-ally over time.

The changes in the terms of trade are primarily reflected inthe external current balances. In the Netherlands, the changein the terms of trade led to a significant increase in thecurrent account surplus in 1981, while after the decline inoil prices the German current account rose significantly bothin absolute terms and also relative to that of the Netherlands.In spite of the similar performance of almost all other econ-omic target variables (with the exception of the public deficit)the current account moved independently. This, however, isconsistent with the operation of a monetary union in whichcurrent accounts play a smaller role in adjustment.

The response of economic policy in the Netherlands to thefall in the terms of trade relative to Germany has been:

(i) monetary policy has become marginally tighter relativeto Germany. Given the large asymmetry of the shock,the necessity for monetary policy response has beensurprisingly small. This means that the exchange ratecommitment has been allowed to become sufficientlycredible;

(ii) the budget deficit became larger in 1986-87 when oilprices went down and the government receipts fell. Thetarget of reducing the budget deficit at the end of theperiod has not been given up. However, as the budgetdeficit remained high (5 % of GDP in 1989) the publicdebt/GDP rate was still increasing.

(iii) in particular wage moderation and thus improved com-petitiveness protected the Dutch economy from employ-ment losses in the aftermath of the fall in oil prices.

In order to assess the appropriateness of the economic policyresponse in the Netherlands, we should consider the alterna-tive response, namely an exchange rate depreciation relativeto the Deutschmark and/or an appropriate fiscal and struc-tural policy. In the first case, because of an implied highergrowth in nominal income, the effect on the fiscal deficitwould have been reduced. However, it is by no means evidentthat interest rates would not have increased considerablybecause of the potentially significant shock to financial mar-kets. In particular, long-term interest rates would probablyhave increased, triggering a detrimental effect on investment.The extent to which the strong exchange rate option was aprecondition for wage moderation remains an open question.Furthermore, a decision to depreciate would imply competi-tive gains of much shorter duration, not to mention signifi-cant inflationary consequences.

In short, the response of the Dutch economic policy to theoil price fall has been a combination of wage moderationand a considerable short-term flexibility of fiscal policy,reducing the need for adjustment of monetary policy.

277

Annexes

GRAPH B.13: Terms of trade in the Federal Republic ofGermany and the Netherlands (goods and ser-vices) (1980=100)

112

110

108

106

104

102

too

98

%

9478 79 80 81 182 83 84 85 86 87 88 89

•NL —— D

GRAPH B.14: Competitiveness of the Federal Republic ofGermany and the Netherlands — Unit labourcosts (annual % change)

78 79 80 81 82 83 84 85 86 87 88 89 90

•NL

4.2. External trade cycle and the Dutch policyresponse

The Netherlands was hit harder than the Federal Republicof Germany by the slowing down of world economic activityin the early 1980s. This was due to less rapid growth in theexport markets and even some loss in market shares. Thisled to a decline in the Dutch share in overall world trade inmanufactures during the period 1980-82 while the Germanshare increased by almost 1 percentage point during thatperiod. Even the 1983 depreciation of the guilder relative tothe Deutschmark improved the Dutch position in 1983 only.Therefore, the relatively weak growth performance of theDutch economy at the beginning of the decade is mainlyattributable to divergent external conditions.

As to the policy response, it seems that fiscal adjustment hasbeen the main instrument to counterbalance the detrimentaleffects on economic targets and in particular on unemploy-ment. It is interesting to note that the relative and absoluterelaxation of fiscal policy in the Netherlands has not haddetrimental effects on exchange rate expectations.

Wages were not supporting the adjustment needs of theDutch economy during this period. Nominal wages grew

relatively fast in 1982. The implied deterioration in competi-tiveness contributed to the observed losses of export marketshare.

During the recovery of world trade in the second half of thedecade, exports in the Netherlands grew faster than in theFederal Republic, explaining at least partly the change inoverall growth performance. The faster increase in Dutchexports can be explained by more moderate wage increasesand thus improved competitiveness. Although a quantitativeassessment is difficult, it would appear that, without theexport led growth, the deterioration in the fiscal deficit dueto the fall in the terms of trade would have been moresignificant. But export performance would perhaps havebeen better with less deterioration of the fiscal deficit (mak-ing room for tax reductions).

4.3. Deutschmark weakness in 1988; Germanmonetary union

In 1988, the introduction of the withholding tax in theFederal Republic led to an increase in long-term interest

278


rates and to capital outflows, triggering a temporary weak-ness of the Deutschmark on foreign exchange markets.

The effect of this asymmetric shock on Dutch economicpolicies was not important. It led to a temporary narrowingof the short- and long-term interest rate differential. Theposition of the guilder in the EMS was hardly changed.

The German economic, monetary and social union will bea crucial test not only for the quasi monetary union betweenthe Netherlands and the Federal Republic of Germany,but also with respect to the overall working of the EMS.Monetary policy in Germany will be confronted by greateruncertainties both as regards the appropriate stance andthe interpretation of monetary indicators. The appropriatestance will crucially depend upon the extent to which fiscal

policy in Germany will become more expansionary thanbefore.

On the level of the interpretation of monetary policy, monet-ary targets will become less reliable indicators of monetarypolicy as money demand can become fairly volatile in Ger-many; other market-oriented indicators, including notablythe exchange rates will have a greater role to play. Therefore,exchange rate policy in the EMS will become more an issueof common interest and the definition of the appropriatemonetary policy stance will also depend upon economicpolicies in the EMS partner countries.

The crucial question will be whether the more expansionarystance of fiscal policy in Germany can be sustained withoutsignificantly changing the policy mix. A tighter monetarypolicy could prove necessary with implications for the stanceof monetary policy in the whole EMS area.

279

Annexes

Annex C

European and German economic andmonetary union: a comparisonThe purpose of this annex is to compare the economic effectsthat can be expected from the German economic, monetaryand social union (henceforth GEMSU) with those of EMU(for the Community) that are analysed more thoroughly inthis volume. This comparison has to take into account theobvious differences in terms of size, starting positions andspeed of transition. It also has to take into account the act ofpolitical union, which took place on 3 October 1990 and whichconstitutes another important difference. In spite of theseobvious differences it still appears reasonable to apply someof the same basic categories of economic effects that wereused in the analysis of EMU in this volume to the analysis ofGEMSU. The effects that seem most important in the case ofGEMSU are (i) efficiency gains, (ii) benefits in terms ofprice stability, (Hi) fiscal policy effects and, (iv) the loss ofthe exchange rate as an adjustment instrument during thetransition.' These four effects are discussed below after a briefintroduction which describes the way GEMSU was im-plemented and discusses the extent to which it has more ofa Community than a purely German dimension. The annexconcludes with some discussion of different speeds in the tran-sition.

1. Introduction

Preparations for an economic, monetary and social unionstarted after the first democratic elections in the GermanDemocratic Republic on 18 March 1990 and the details werefixed in a treaty (Staatsvertrag) between the two Germangovernments. That treaty took effect on 1 July 1990 andfrom that date the Deutschmark became the sole legal tenderin the territory of the GDR. All current payments (wages,rents, etc.) formerly expressed in East German marks wereconverted at the rate of 1:1, whereas assets and liabilitieswere converted in general at the rate of two to one (excepta lump sum of 4 000 per adult, 2 000 per child and 6 000 forpersons over 59 years). At the same time the GDR alsoadopted laws and other legal dispositions, mostly identicalto those in the FRG, that freed economic activity from theconstraints imposed by the previous regime, thus creatingthe legal framework for a market economy. Starting on

2 July the free movement of goods, services and productionfactors was established, thus totally liberalizing trade be-tween the former West and East Germanys with the excep-tion of some quotas on agricultural products. On 3 October1990 the former GDR legally became part of the FederalRepublic of Germany, in accordance with Article 23 of theGrundgesetz.

An important aspect of this annex is that it is somewhatmisleading to speak of an economic and monetary unionbetween the two parts of Germany. The West Germaneconomy, like that of other member countries, is alreadyintegrated into the Community to such a considerable extentthat the conditions inside the Community already comevery close to the definition of an economic union (the freemovement of goods and services and factors of production).It might therefore be more appropriate to speak of a partici-pation of the GDR in the economic union that has alreadyformed to a large extent in the Community.2 The monetaryunion aspect, is, of course, more specifically German, buteven here it is important to recognize that through theEMS the GDR participates immediately in a quasi-monetaryunion that already exists and incorporates part of the Com-munity. 3

For clarity of exposition and also to make the presentationcomparable to the analysis of EMU this note distinguishesbetween the monetary union aspect (the introduction of theDeutschmark in the GDR)4 and the economic union aspect(the freeing of economic activity and trade). The economicunion aspect can be considered to consist of two elements:internal liberalization, i.e. the creation of a market economy(through the elimination of price controls, recognition ofprivate property in capital, etc.) and external liberalizationi.e. the suppression of trade barriers. The jump to a marketeconomy is interesting in its own right, and might constitutea useful reference point for the more gradual transitionprocess that is currently taking place in other central andeastern European countries.5 However, given that all mem-ber countries already have market economies the second

1 Because of the obvious difference in size GEMSU should not haveimportant global implications.

If another country, e.g. Hungary, had taken the same measures as theGDR in the economic field, one would have to speak of accession to theEC.It is clear that the social, and hence fiscal, aspects do not have thisEuropean component.The monetary union aspect can also be said to comprise the fact that theWest German banking system was extended to the GDR and that alldispositions of the Bundesbank now apply to the territory of the GDRas well.This aspect is also relevant for those member countries that have Statecontrol over considerable parts of their productive sector.

280

Annex C — European and German economic and monetary union: a comparison

aspect (external liberalization) is more relevant for the com-parison with EMU.

2. Price stabilityBy adopting the Deutschmark, the GDR immediately gainedall the advantages of a stable currency and of the credibleanti-inflationary policy of the Bundesbank. Although overtinflation had not been high under the previous regime thiscan be considered an important economic benefit for thepeople in the GDR since experience has shown that periodsof liberalization often bring high inflation. Moreover, itwould have been impossible for the authorities in the GDRto establish quickly the reputation and hence credibility forprice stability the Bundesbank has acquired over the lastfour decades. This gain must therefore be considered as evenmore important than that which some member countries canexpect from EMU since most member countries have alreadyreduced inflation considerably and can to some extent ben-efit from the reputation of the Bundesbank by participatingin the EMS.

By simultaneously freeing trade, not only with the FRG,but in effect with the Community and the world, the GDRwas also able to make sure that the elimination of pricecontrols would not lead to inflation since prices, at least fortradable goods, stabilized quickly at the West German orworld level. Thus the monetary reform means for the GDRalso an instantaneous and radical reform of the price mech-anism, with an enormous, indeed revolutionary, impact onthe 'real' economy, which goes beyond its 'nominal' aspects.This price reform aspect of GEMSU represents a dimensionthat is only weakly present in the EMU case.

3. Efficiency gainsNotwithstanding the enormous initial adjustment lossesbeing experienced in the GDR economy, it is appropriate tofocus on the ultimate economic gains.

The most important long-term effect of GEMSU will be thegains from the complete integration of markets for goods,services and capital that was made possible by the introduc-tion of the Deutschmark in the GDR,6 The GDR did not

By adopting the Deutschmark the GDR eliminated all transaction costsin intra-Gennan trade and reduced the costs that arise in trade with therest of the world to trY low level enjoyed by exporters and importersfrom the GDR. For example, West German banking sources estimatethat a transfer from East to West Germany^cost about DM 100 beforeGEMSU. This would fall to the DM I a transfer is estimated to costinside the German banking system. These direct savings far exceed whatcan be expected from EMU because the cost of international transactionsinside the Community is estimated,to be about DM 20, only about onefifth of the cost of an intra-German one before GEMSU.

have a capital market prior to GEMSU; through this move,however, savers and investors in the GDR are put immedi-ately into contact via an efficient capital market and financialsystem. The introduction of the Deutschmark has thus re-duced the cost and increased the availability of credit andrisk capital, which will allow surviving or new enterprises tofinance the huge investment needed to modernize the capitalstock of the East German economy and facilitate the creationof new enterprises, especially small to medium-sized ones.Without access to this capital market, investment and em-ployment creation through new enterprises would be muchmore difficult and the catching-up process slower. Moreover,and more importantly, access to the world capital marketvia the Deutschmark will allow the private sector in theGDR to use foreign savings to finance the huge initial needfor investment; i.e. the common currency has eliminated forthe GDR the current account constraint. As discussed inChapter 6 of this study the importance of the current accountconstraint may already be declining in the Community be-cause of capital market liberalization. However, the GDRwould certainly have been forced to maintain capital controlshad a separate GDR currency been maintained. This benefitof a common currency should therefore be much moreimportant for the GDR than in the context of EMU.

The economic benefits the GDR economy can derive fromthe creation of an efficient capital market and the eliminationof the current account constraint depend on the extent towhich an investment boom in the GDR does materialize.An efficient capital market and access to foreign savings areimportant only if investment is strong. The extent to whichthis will be the case is difficult to judge. On the one hand ithas been argued that conversion of wages at 1 to 1 has ledto wages that are above productivity and that very littleinvestment might therefore take place in the GDR. On theother hand, it has also been argued that the GDR representsa market with a strong pent-up demand and productivitymight increase rapidly with modern capital equipment, theright incentives for labour and given that the general levelof education of the GDR work force is not too far fromWestern standards.

It is apparent that the liberalization of trade (the externaleconomic aspect of GEMSU) also benefits the GDR econ-omy immensely in the long run. In the case of EMU thisaspect is not very important since the internal market pro-gramme will already eliminate all barriers to trade inside theCommunity. In the case of Germany, however, the startingpoint is totally different in that, before GEMSU, the GDReconomy traded very little with market economies and thetrade with the Comecon partners was conducted at artificialprices and distorted by political considerations. By openingthe economy to international trade the GDR economy would

281

Annexes

reap all the well-known gains from trade through increasedspecialization and economies of scale.

The sudden introduction of free trade leads, however, tolarge adjustment costs for the GDR economy. As with alltrade liberalization programmes these adjustment costs arisein the beginning. They are of a once and for all nature andultimately much less important than the efficiency gains thatwill be reaped continuously once the GDR economy hasadjusted and integrated itself in the global division of labour.None the less, the extremely large losses of industrial pro-duction in the first months of GEMSU show clearly thatthe future large gains cannot be rapidly secured without acorresponding rapid initial adjustment shock.

Because of the different starting points the efficiency gains,as well as initial adjustment costs, from both the economicand the monetary union aspects should therefore, here also,be several orders of magnitude higher in the case of GEMSU,than in EMU.

4. Fiscal effects

Here the qualitative differences between GEMSU and EMUare also very large. In the fiscal domain formal politicalunification, which took place on 3 October 1990, broughtabout important changes with respect to the situation pre-vailing immediately after 1 July 1990. For example, as longas the GDR State existed with its own government it wasbound by the dispositions in the Staatsvertrag that stipulatedthat it could not finance any deficit without the approval ofthe FRG. Once the GDR ceased to exist this issue no longerarose. To respect the federal structure of the FRG, Ldnderhave been recreated on the territory of the GDR, and theseLdnder will, after a transition period of 5 years, enjoy adegree of fiscal independence similar to the existing Ldnderin the FRG. The appropriate comparison in EMU is there-fore the situation existing after political unification.7

The Ldnder on the territory of the GDR will receive consider-able transfers directly from the federal government.8 Theeconomic rationale for the different size of these transfers

This applies mainly to fiscal issues. Political unification does not alterthe free trade and free movement of labour already established in July.These Ldnder start with a much lower level of debt than their WestGerman counterparts. They should therefore be able, at least initially,to run much larger deficits. The Finanzausgleich, which provides forhorizontal transfers from richer to poorer Lander in the FRG, is notdue to be revised until 1994, but it will probably have to be adjustedconsiderably after that date.

(as compared to what is planned inside the Community) canbe found in the much higher degree of labour mobility thatexists between these parts of Germany.9 The mobility oflabour between member countries of the EC is much lowerso that much larger differences in income can persist withoutleading to unacceptable migration flows. Although commoncitizenship, i.e. political unification, is certainly also import-ant, the fact that regional transfers between the two partsof Germany have a different dimension from those at theCommunity level can also be understood as a reaction to adifferent economic environment, i.e. a different degree oflabour mobility.

With the disappearance of a 'central' government in theGDR the size of the federal budget relative to those of theLander will be close to the present situation in the FRG; i.e.about 50 % of total expenditure, whereas the Communitybudget accounts only for 2 % of public expenditure in theCommunity. This implies that the issue of fiscal policy coor-dination has a different dimension than inside the Com-munity where the central level accounts for a very small partof total public expenditure.

5. Adjusting without the exchange rate

The present study has argued that the main cost of a monet-ary union is the loss of the exchange rate as an adjustmentinstrument in the face of asymmetric regional shocks. Thisidea was also the main argument against the introductionof the Deutschmark in the GDR since the economy of theGDR will go through a period of profound adjustmentwhose success and speed cannot be predicted a priori. Thechanges in the product mix in the GDR and the rapid growthof income that can be expected during the catching-up pro-cess imply that the real exchange rate between the two partsof Germany has to change significantly in the near future.During a transition period the need for real exchange rateadjustments will therefore be much greater between the twoparts of Germany than inside the Community. One couldtherefore argue that in this respect the potential cost of amonetary union could be higher in the German case.

As emphasized in Chapter 6 of this study the nominal ex-change rate is a useful adjustment instrument only if nominalwages are somewhat rigid. Had the GDR maintained aseparate currency, to be able to adjust the exchange rateduring the transition period, exchange rate changes wouldhave been effective only if wages could be relied upon not

For a further discussion of the issue of fiscal federalism see the contri-bution by Van Rompuy et al. in European Economy (1990).

282

Annex C — European and German economic and monetary union: a comparison

to react immediately. Given the high degree of integrationof the German labour market (and of most other marketsas well) it is, however, likely that workers would have beenable to 'see through' the effects of exchange rate adjustments.It is therefore doubtful that the exchange rate would havebeen a more useful policy tool for the GDR than for otherCommunity countries which have realized that exchangerate changes on their own cannot correct major external orinternal imbalances.10

Another issue that arises more strongly in the case ofGEMSU concerns the initial level of the exchange rate. Inthe case of EMU market exchange rates provide a reliableguide to the appropriate level at which to fix rates for themonetary union. Only for countries with large imbalanceswould it be appropriate not to fix exchange rates at the levelat which they trade in the markets. However, in the case ofGEMSU no such indicator existed and the conversion ofwages at one to one, i.e. an exchange rate of one, hadbeen chosen only on the basis of some rough calculationsregarding productivity. At present there is still considerableuncertainty whether, or at what future point in time, thislevel might reflect the competitiveness of the GDR economy.

6. Transition

The German economic and monetary union went into effectliterally almost overnight. In contrast, in the case of theCommunity it took over a decade to eliminate tariffs andthe internal market programme will have taken a numberof years to be completed since it was launched in 1985. Inthe monetary field it took several years of operation of theEMS to reduce exchange rate variability and achieve someconvergence in inflation. The introduction of a commoncurrency for the Community is also expected to take anumber of years. Political factors certainly were the mainreason for this difference in the speed of transition. However,there are also economic aspects to this choice.

In the economic field the standard argument for a slowliberalization of trade is that this reduces the social costs ofadjustment. In the German case the financial costs of thissocial aspect are borne by the FRG and this makes animmediate trade liberalization possible.

10 Chapter 6 shows that devaluations of the nominal exchange rate leadto a gain in competitiveness only in the very short run.

In the monetary field the main economic reason for notgoing immediately to full EMU is that the institution thatwill manage the common monetary policy needs to establishits credibility and needs to learn how to operate on a Com-munity-wide financial market. This consideration does notapply to GEMSU since in this case the Bundesbank, whichis an experienced institution with a well-established repu-tation for price stability, will be in charge of the jointmonetary policy.

Another reason for a gradual transition in the case of EMUis that convergence of inflation usually takes some time.However, as discussed in Chapter 8 this argument appliesonly if the introduction of a single currency is preceded bya period of fixed exchange rates. In the German case thejump to the single currency eliminated this consideration aswell.

For the so-called 'coronation theory' monetary union canonly be the crowning act of a long process of convergencein economic policy and performance. According to thistheory the only way to verify that all the conditions for astable monetary union are met is through this process ofprior convergence. In the case of GEMSU, however, theGDR had signalled its willingness to fulfil all the conditionsfor a stable monetary union through the Staatsvertrag, whichstipulates that all the dispositions of the Bundesbank applyto the territory of the GDR as well and establishes tightcontrols over the fiscal policy of the GDR.

In summary it is evident that the main arguments for agradual transition towards EMU that can be made at thelevel of the Community did not apply in the case of Ger-many.

7. Concluding remarks

The economic, monetary and social union started betweenthe FRG and the GDR in July 1990 and the political unionwhich followed in October 1990 represent a dramatic exam-ple of a 'jump' to a different regime. This example will beof most relevance not so much for the Community's EMUprocess but rather for countries like Poland, Hungary andCzechoslovakia that plan to institute a market economy andhave to decide how long the transition should be.

This study has argued that EMU can be considered a 'regimechange'. In a certain sense GEMSU provides a special andparticularly stark example of the profound effects a 'regimechange' can have.

283

sduquenoy

Annexes

Annex D

Shocks and adjustment in EMU:simulations with the Quest model

1. Introduction and summary

Any assessment of the loss of the exchange rate instrumentin EMU has to take account of the size of asymmetric shocksand the extent to which the exchange rate can be replacedby other instruments. The purpose of this annex is to identifythe different shocks individual countries have faced in therecent past (Section 2), as well as to evaluate the costs of theloss of the exchange rate in terms of output and inflation inrelation to alternative adjustment instruments (Section 3).

Both analyses are performed using the Quest model of theCommission. * Naturally, the results are influenced by theparticular structure and properties of this model. Therefore,as far as possible, the arguments are supplemented by theor-etical considerations justifying the results. The followingconclusions emerge.

Concerning the different types of shocks:(i) macroeconomic fluctuations in one quarter are only

partially determined by shocks occurring in that samequarter, and more by the effects of previous shocks, i.e.permanent shocks are more important than temporaryshocks;

(ii) of the temporary shocks to GDP (as measured by theirimpact in the first quarter), about one half are due toshocks in economic behaviour of private agents, onequarter are due to shocks in budgetary policy and aboutone-tenth are due to monetary policy shocks (the smallsize of the latter is due to the fact that monetary policyis only effective with some lags, in the model); govern-ments may therefore be considered to be a non-negli-gible source of temporary shocks;

(Hi) the importance of shocks in the behaviour of privateagents is smaller in European countries than in theUnited States for the real side of the economy, but notas regards inflation, and therefore wages.

Concerning the loss of the exchange rate instrument:

(i) the exchange rate instrument serves to frontload realexchange rate changes needed to adjust to a shock; but

given that unemployment affects wages, this has thedrawback that real adjustment to the new equilibriumlevel is delayed in time;

(ii) the cost of losing the exchange rate instrument in termsof output diminishes the higher the degree of indexationof wages to prices and the stronger the reaction of wagesto unemployment;

(iii) alternatives to the exchange rate may be fiscal policy,wage policy, or a supply measure (e.g. social securitycontributions of employers); but relative to a devalu-ation, each of these instruments implies either a realwage loss or an increase in the government deficit;

(iv) the loss of the exchange rate instrument does not necess-arily imply a cost in terms of inflation.

2. Sources of macroeconomic fluctuations in theCommunity

In the context of EMU, the loss of domestic monetarypolicy and the exchange rate instrument implies that nationalgovernments may lose some control over their economyin the face of country-specific shocks. This could increasemacroeconomic fluctuations and therefore reduce welfare.Accordingly, it is particularly important to identify as far aspossible the different categories of shocks that may influencean economy and the way in which they will evolve in EMU.The purpose of this section is therefore to describe differentcategories of shocks and give some empirical account ofthem. The latter is done using the Quest models for Ger-many, France, Italy, the United Kingdom, and the UnitedStates for comparison.

Emphasis is given to the short-term effects of shocks andthe extent to which they are due to government policies.The former are captured by looking at the first-quarterinstantaneous impact of different shocks. Even thoughshocks in reality often have an impact which lasts for alonger period, this makes it possible to focus on sources ofshort-term fluctuations, and therefore of uncertainty. Asregards the influence of government policies, it is assumedthat the stabilization function of the government impliesthat the monetary and fiscal authorities attempt to influencevariables such as employment, income, price levels, rates ofinflation, rates of investment and saving, and economicgrowth.2 In doing so, the authorities have to take accountof several factors influencing these variables, including theconsequences of their own policies. Observed macro-economic fluctuations are therefore the outcome of a combi-

1 See Bekx e( a/. (1989).

284

See Alien (1976).

sduquenoy

Annex D — Shocks and adjustment in EMU: simulations with the Quest model

nation of the working of the economy and policy inter-ventions. To the extent that the latter are unexpected, theyalso act as shocks and therefore as a source of macro-economic fluctuations.

2.1. Categories of shocks

A shock cannot be defined without reference to a giveneconomic system. In the context of EMU, the referenceeconomic system is taken to be a Member State of EMU.Secondly, a set of variables has to be defined which areconsidered to be determined endogenously and simul-taneously inside the system. These can be taken to be theusual macroeconomic variables such as output, inflation andemployment. A shock may then be defined as any eventwhich has a direct or indirect impact on the endogenousvariables of the reference system without, however, beingdetermined by it. Given this definition, a country-specific orlocal shock may be defined as a shock having a direct impacton only one country. Similarly, a common shock may bedefined as a shock having a direct impact on all membercountries of the union. A second distinction arises from theduration of the shock. A temporary shock is an event thatdisappears after a short period of time, while a permanentshock is an event that remains present over the whole timeperiod analysed (this does not preclude that it eventuallydisappears).

There were two common shocks relevant to the EMU con-text which had a large impact over the last two decades: theoil price hikes of 1973-74 and 1979 (a typical sector-specificshock) and the increase in interest rates in the United Stateswhich provoked the 1981-82 recession (i.e. a common exter-nal shock for the Community). In addition, the reducedimport demand in developing countries due to the debt crisis,wildly fluctuating commodity prices and the current-accountdeficit of the United States may also be considered to be inthis category. This also holds for the dollar exchange ratemovements of the 1980s, although after the Plaza/Louvreagreements these movements have become more or less thejoint responsibility of the United States and its G7 partners.This is important as it implies for the Community the partialinternalization of a foreign variable as a domestic policyinstrument.

Country-specific or local shocks may be subdivided in anumber of categories.

(i) An important source of local shocks arises from domesticpolicy instruments, which may be broadly divided betweenmonetary and budgetary policy. It might be argued that theseinstruments in fact react to changes in domestic economic

performance, are therefore endogenously determined andnot to be considered as shocks. This is certainly true forsome elements of budgetary policy, mainly those parts whichact as automatic stabilizers (social benefits, progressivetaxes) or which behave pro-cyclically (interest payments ondebt). Other elements, however, are of a more discretionarynature. Unless it is assumed that fiscal fine-tuning is stillpossible, changes in these elements will therefore act asrandom shocks to the national economy. Moreover, it mightbe argued that domestic policies are exogenous for the pri-vate sector as a whole.

(ii) A second important set of country-specific shocks relatesto changes in domestic natural, human or capital resources.Examples of the first are the oil extraction which started inthe second half of the 1970s in the United Kingdom or theclosing-down of coal mines in several Community countries.Changes in human resources arise as a consequence of demo-graphic movements or modifications in the composition ofthe labour market (participation rates, workforce compo-sition). Since resource shocks, notably concerning humanresources, usually evolve gradually over time, they areclassed as permanent. The same can be said of capital re-sources to the extent that technical progress is concerned.

(iii) A third set of country-specific shocks consists ofchanges in behaviour of economic agents (households,firms). These changes may arise from changes in taste, busi-ness climate or any other form of news that has a randompattern in influencing economic behaviour. In econometricterms, these shocks are contained in the residuals of thebehavioural equations.

(iv) Finally, if the concept of a shock is defined ratherbroadly, a fourth category of shocks is formed by inertia.This can be explained as follows. A broad definition ofa shock would define it as all events without which theendogenous variables of the economic system would notchange. Under this definition, past values of endogenousvariables of the system, or inertia, which influence the pre-sent endogenous variables are also to be considered as ashock. They are the consequence of the existence of perma-nent shocks. Empirically, not much is known about thecauses of lags in economic behaviour. Usually they areassumed to be related to information costs, adjustment costsor the formation of expectations.

There is no unique measurement system available for shocks,be they common or country-specific. The main issue involvedhere is that shocks have different dimensions. How may anoil-price increase be compared to an earthquake, for in-stance? To circumvent this problem, the impact of a shockon one of the endogenous variables may be taken as a

285

sduquenoy

Annexes

yardstick. Since the endogenous variables of the system areall determined simultaneously, a representation of the systemwhich determines them is needed in order to be able to assessthe final impact of a shock on any particular endogenousvariable. A natural candidate for such a representation is aneconometric model. Despite the fact that each econometricmodel has its unique features, any model that is a represen-tation of the neo-Keynesian/classical synthesis will probablybe a reliable guide.

2.2. Methodology

The methodology adopted may best be described by as-suming that the model which is used is linear or linearized.In this form, a dynamic simultaneous model may then bewritten asyt = Ayt B Ly t_ Dx u,,

with y an nxl vector of endogenous variables, x a kxl vectorof exogenous variables and u an nxl vector of disturbances.The reduced form of this model isyt = MB + MLy t_, + MDxt + Mu,,

with M = (I - A) ~ ' . In first differences, this may be written as

yt ~ vt-i = yu + y* + y*>with the three components on the right-hand side represent-ing the contribution of the changes in lagged endogenousvariables, exogenous variables and disturbances, respect-ively, to the change in the endogenous variables:

yu t _! - y t_2] = MB MDx

yx[ = MD[xt - x t _, ] = MB + MLy t_2

- y t - i>yut = M[ut - u t _ i ] = MB + MLy t_2 M«t

This breakdown of contributions may be divided into moregroups than the three presented here, down to the level ofthe individual variables. Furthermore, the results dependcrucially on which variables in the model are endogenousand which variables are exogenous. For the results presentedhere, the existing division between exogenous/endogenousvariables has been taken.

Technically, in order to calculate the contribution of eachindividual component to the change in period t, a static

simulation is performed for period t- 1 in which the vari-ables of the component concerned are set equal to theirvalues in period t.3 As may be seen from the above equations,the difference between the values of the endogenous variablescalculated this way and the observed values in period t — 1gives the contribution of the component considered. Byrunning T static simulations for T periods t= 1, ... T, thechanges in all endogenous variables due to one particularcomponent may be calculated for a whole period. This maybe repeated for each component. The relative importance ofany component as a source of shocks may then be obtainedby calculating the observed variance of each component anddividing it by the sum of the variances of all components. Thelatter has been taken as numerator instead of the variance ofthe endogenous variable in order to have the sum equal to100%, even though this has the disadvantage that the co-variances between the different contributing factors (whichare zero theoretically) are not taken into account. In practice,the variances are calculated after a stationary transformationof the variables in order to remove predictable elements:growth rates for GDP, first differences for unemploymentrates, first differences relative to previous-period GDP forbudget balances, etc. This implies, for instance, that thecontribution to the variance of the growth rate of GDP ofan exogenous variable which grows at a constant rate iszero.

Example:Let y l x t be the change in GDP (endogenous variable ylt) inperiod t with respect to period t - 1 due to the changes in allexogenous variables. Furthermore, let Vlx be the calculatedvariance over the sample period of this component after thestationary transformation [ylxt — yu- i l /y t t - i - Similarly,V1L and Vlu may be calculated, Then the relative varianceof GDP due to changes in all exogenous variables is definedas:

V,»/[V1L + V lx + V,J

The method described calculates the contributions of differ-ent components to the variance of the change in all endogen-ous variables between two consecutive periods. This impliesthat the effects of temporary shocks will be given relativelymore importance than those of permanent shocks. In prin-ciple, this could be overcome by calculating contributionsto changes over a longer time horizon, for instance eight

3 Indicated by italics in the equations. In practice, the contribution of thelagged endogenous variables is calculated by taking the difference be-tween the observed change in the endogenous variables and the sum ofthe other contributions.

286


quarters.4 Because of the technical complexity of such simu-lations, this was not done. Nevertheless, the extent to whichpermanent shocks play a role can be read off from therelative importance of the lagged variables, even though theorigin of the permanent shocks is unknown.

2.3. Results

The method outlined above was applied to the Commission'sQuest models for Germany, France, Italy, the UnitedKingdom and the United States.5 The period covered was1980.11 to 1987.Ill, which is a period containing many ofthe shocks described above,

Each row in Table D.I represents, for an endogenous vari-able (i.e. one element of the vector y), the variance of oneexplanatory component as a percentage of the sum of thevariances of all explanatory components, as described above.The explanatory components have the following meaning:Inertia: all lagged variables (calculated as residual on thebasis of all other components);Foreign: all foreign variables (foreign demand, foreignprices, autonomous components of the balance of pay-ments);Monetary: interest rates and dollar exchange rates;Government expenditure: expenditure components ongoods and services in constant prices, subsidies, autonomoustransfer components, but not automatic stabilizers;Government revenue: tax rates, rates of social security con-tributions and autonomous revenue components;Resource: demographic factors, exogenous shocks to thelabour force or productive capacity;Behaviour: all disturbances of behavioural equations.

The first result is the importance of inertia, or permanentshocks, in explaining the variance of almost all the variablesin the table. Only for inflation (Italy and the UK) and thegovernment deficit (all except the UK) is the contributionof permanent shocks lower than 60%. The exceptions forinflation are caused by the relative importance of behav-ioural shocks, and that of government revenue (for Italy).The latter could be due to shocks in social security contri-bution rates. Permanent shocks are less important for the

4 This was done in the study by Fair (1988), who used stochastic simulationsto calculate the respective contributions of different equation residualsafter one and eight quarters.

5 See Bekx el at. (1989). Updated and re-estimated versions were used forall models, as well as a new model for Italy.

government deficit (except in the UK) due to the relativeimportance of shocks in government tax rates (revenueshocks) and, to a minor extent, in expenditure. Permanentshocks are especially important for the unemployment rate(due to demographic factors) and the current account (prob-ably due to J-curve effects). This corresponds to a prioriexpectations, since these phenomena are known to be slowin practice and to take effect only with substantial lags.

Of the variance due to temporary shocks, the largest part isgenerally due to behavioural shocks. For GDP, for instance,this accounts on average for almost half of the variance dueto temporary shocks. Of the remaining part, almost 25 % isdue to government budgetary policy and some 10 % due tomonetary policy. On the whole, the government thereforehas a non-negligible role as a source of temporary shocks.The importance of the government also holds for variablesother than GDP, and most notably for the governmentdeficit for obvious reasons. Furthermore, the effect of taxrates (government revenue) on inflation is considerable, no-tably in the European countries.

Temporary shocks in monetary policy have a weak impacton inflation. This does not mean that monetary policy hasno effect on inflation, but only that its short-term (firstquarter) effects are rather weak. The strong impact of monet-ary policy on the Italian government deficit corresponds tothe large Italian government debt.

For all behavioural shocks except for those to inflation, theEuropean countries are more symmetric relative to eachother than with respect to the United States, for which theshare of behavioural shocks in the sum of variations isalways the largest. For inflation, the performance is stillrather asymmetric, which is mainly due to Italy and theUnited Kingdom.

2.4. Conclusion

The evidence presented in this section seems to indicate thatpermanent shocks are in general more predominant thantemporary shocks. Speaking generally, this is among otherthings a sign that shocks, even after their origin has disap-peared, may affect economic behaviour persistently throughthe lags that operate in the economy. While nothing specificcan be said about the nature of the permanent shocks, itappeared that of the temporary shocks to GDP (as measuredby their impact in the first quarter), about half is due toshocks in economic behaviour of private agents, a quarteris due to shocks in budgetary policy and about one-tenth isdue to monetary policy shocks. Governments may therefore

287

Annexes

be considered to be a non-negligible source of temporaryshocks. Furthermore, it appeared that shocks in the behav-iour of private agents are of similar importance among

European countries when compared to the United States forthe real side of the economy, but not as regards inflation,and therefore wages.

Table D.I

Relative variances of shocks determining endogenous variables, 1980.11 to 1987.HI1% of sum of variances)

Shocks

Inertia Country-specific

Foreign Monetary Govern men I

Expenditure

Real GDP

Inflation1

Employment1

Unemployment rate2

Government deficit3

Current account3

DFIUKUSDFIUKUSDFIUKUS

DFIUKUS

DFIUKUSDFIUKUS

66,669,173,869,361,761,171,114,648,370,369,972,276,869,666,9

79,181,680,488,079,4

24,245,818,976,545,0

81,281,280,581,777,7

3,73,02,33,02,4

1,81,10,31,20,73,61,82,92,81,83,43,13,12,02,7

1,01,50,63,11,4

2,43,13,13,22,9

3,85,43,23,62,70,41,10,30,90,74,22,02,73,11,8

3,93,13,22,12,7

10,32,5

18,93,33,0

1,60,93,21,63,0

4,72,73,52,83,61,91,10,21,20,85,46,43,33,22,3

3,63,53,01,82,5

19,13,47,56,1

17,7

3,33,13,33,13,6

4,24,34,63,23,18,5

10,140,3

7,03,4

3,71,92,92,92,2

3,13,13,22,02,9

40,240,250,9

5,822,5

3,03,13,33,12,9

2,72,82,93,02,61.81,10,31,20,7

2,51,73,02,81,8

2,73,13,12,02,7

0,91,50,53,11,5

3,23,13,33,22,9

14,312,89,7

15,123,924,514,544,040,323,4

10,614,18,4

15,623,1

4,12,53,92,17,04,45,02,62,18,9

5,25,43,44,17,1

1 Growth rate.1 • First d ifTerence.3 First difference as % of last period nominal GDP.Source: Simulations with the Quest model of the Commission.

288


3. The loss of the nominal exchange rate as anadjustment instrument: an evaluation

As developed in Chapters 2 and 6, the main cost of EMUis that nominal exchange rates can no longer be used tobring about the adjustment to a new equilibrium after theeconomy has suffered an external or internal shock. The aimof this section is to propose an assessment of the associatedcost with the help of Quest simulations. Alternative adjust-ment mechanisms are also discussed, and it is shown howtheir performance can be compared to that of exchange ratechanges.

For the sake of simplicity, only two types of country-specificshocks are considered: a demand shock, consisting of a 5 %drop in the volume of demand for French exports, and adomestic supply shock, originating in a rise in employers'social security contributions by one percentage point ofGDP in Germany. Most of the comments concentrate onthe demand shock since this is the one for which an exchangerate policy can be most easily justified.

Although the quantitative results of these simulations arespecific both because of the nature of the shock and becauseof the properties of the country model, fairly general con-clusions can be drawn from this exercise. Indeed, resultsfrom the first simulation are representative of the effects ofan external or internal demand shock for a generic MemberState, while the results of the second simulation are charac-teristic of a wide class of supply shocks of domestic origin.

3.1. Nature of the cost

As discussed in Chapter 6, nominal exchange rate changesdo not result in real exchange rate adjustments in the longrun. This is also a property of most empirical macro-economic models. Hence, the loss of the exchange rate instru-ment has no strong long-term implications. But short- andmedium-term costs are incurred when prices are sticky sincethe exchange rate can no longer be used to smooth adjust-ments. The problem is to evaluate these costs.6

Suppose a given country experiences an unexpected drop inthe foreign demand for its products. Let the working of

Throughout this section, the exchange rate is considered as an instrumentin the baseline regime. This corresponds to the EMS situation, wherecentral parities rates can be modified in a discretionary way when ashock occurs. Shocks are assumed to be unanticipated, but when theyoccur, they are supposed to be fully observed by the authorities, whocan then choose to modify the exchange rate. The exchange rate peg isalways supposed credible, and therefore monetary policy is entirelydetermined by the foreign interest rate.

the model be represented in a simplified fashion within anaggregate supply/demand framework (Graph D.I). Aggre-gate demand depends positively on the real exchange rate,whereas aggregate supply is downward-sloping.7 In the longrun, the shift in the demand curve from AD to AD' impliesa depreciation of the real exchange rate (equilibrium movesfrom A to A1). But in the short run, prices are sticky andthe real exchange rate depends on the nominal exchange rate.Without nominal exchange rate adjustment, the economymoves to short-term equilibrium at B. Since B is below fullemployment, prices tend to fall and the economy movestowards A' along the demand schedule. Adjustment costsdepend on the length of the move from B to A', which inturn depend on the degree of wage and price flexibility.

GRAPH D. 1: Effects of a demand shock with and withoutexchange rate change

Output

Nominal exchange rate depreciation can be used to reduceadjustment costs by 'frontloading' changes in the real ex-change rate instead of having to wait for the completion

This may come, for example, from the labour supply curve: by reducingreal wage income, a real depreciation reduces the supply of labour.

289

sduquenoy

Annexes

of wage/price adjustments. Immediate depreciation limitsunderemployment in the short run and therefore the cost ofadjusting to the new equilibrium real exchange rate.

Without nominal exchange rate changes, the adjustment costcan only be reduced by speeding up the move from B in thedirection of A'. Candidate instruments are: (a) governmentborrowing (demand policy), (b) wage moderation (supply/demand mix), and (c) cost reduction (supply policy). Thefirst would by, say, an increase in public spending reducethe effect of the export fall on GDP. The second wouldbring forward the shift from domestic to foreign demand,just as a nominal depreciation of the exchange rate woulddo. Ultimately, increased public expenditure would lead toreal exchange rate appreciation, and thereby make adjust-ment more difficult, while nominal wage cuts would inducetoo much real depreciation. Such effects would have to beremedied, perhaps by a combination of both instruments.In the mean time, the adjustment in terms of real GDP couldbe favourably affected.

3.2. Evaluation with the Quest model

For simulation purposes, one has to choose the size of theinitial devaluation. Empirically, it can be chosen so that itcorresponds to the eventual real exchange rate adjustment.Alternatively, one could aim at minimizing the deviationfrom baseline of some target variable (i.e. nominal GDP)over the simulation period. In practice, both methods yieldcomparable results as long as the exchange rate change is aonce and for all adjustment, similar to an EMS realignment.8

Graph D.2 presents the output and real exchange rate effectsof the 5 % drop in the demand for exports with and withoutaccompanying devaluation policy.

Without devaluation, price and wage rigidity precludes thenecessary real exchange rate adjustment, and the initial effectof the shock is to dampen output. Real GDP drops by 1,3 %in the first year. By frontloading this adjustment, a 7 %devaluation significantly reduces the output cost of the shockduring an initial period of three years.

However, in the medium run wages and prices react to thedisequilibrium in goods and labour markets, and withoutdevaluation the output loss is progressively reduced as the

The practice of realignments within the ERM can best be approximatedby such a once and for all change. At a more abstract level, the possibilityof an optimal exchange rate management could be taken into account;instead of 'frontloading' the depreciation, optimal control simulationscould be used to determine the exchange rate path that would minimizethe present value of a given macroeconomic welfare function.

real exchange rate depreciates. Devaluation is an obstacleto adjustment in the medium term, since the initial boost inoutput reduces real wage (and real exchange rate) adjust-ments. In this respect, devaluation leads to trading off aspeedier adjustment in the short run against delays in themedium-term adjustment. As demonstrated in Box D.I, thisis a fairly general result which holds in so far as wages arenot completely insensitive to labour market developments.

GRAPH D.2: Output and real exchange rate effects of anexport demand shock with and without ac-companying devaluation policy

Real exchange rate

Output

-1.4 4-- -

Source Quest simulators wuh French submodel b)1 the Commission services.

290

sduquenoy


Box D.I: Short-term and long-term effects of a devaluation policy

It is apparent in Graph D.2 that although devaluing the currencyhas the advantage of reducing the immediate output cost of theexport demand shock, it also implies that further adjustment tothe new equilibrium is delayed. Hence, devaluation seems toimply a trade-off between short-term gains and long-term costs.

A small reduced model of the wage-price block of a standardmacroeconometric model helps to understand why. The modelconsists of four equations:

(1 )w = 5pc + (1-6) pc_, - 0v

(2) p = w

(3)pc = ae + ( l -a)p

(4 )v= - t [ E _ , / P _ , -1]

Upper case letters refer to variables in levels while lower-caseletters refer to rates of change (x = Log X - Log X _ , ) .Equation (1) is a standard Phillips curve equation, with 8 and0 respectively characterizing the speed of indexation and thedegree of wage responsiveness (flexibility); equation (2) givesGDP prices as a function of wage costs, equation (3) linksconsumption prices pe to GDP prices and the exchange rate,with a being the imports to [GDP + imports] ratio, and finallyequation (4) is a reduced form equation of the demand andemployment blocks which gives the unemployment rate as afunction of the lagged real exchange rate.

The solution of this model yields a first difference equationinP:

(5) [ l -8( l -a ) ]p = ( l -S) ( l -a )p_! + 8 a e + (1-S)a e_ .

Equation (5) can be either solved or simulated for differentvalues of 8 and 0. Graph D.3 shows the time profile of the realexchange rate after a 10% devaluation for 8 = 0,5, a = 0,2and T = 0,07, with 0 = 0 (rigid wages) and 0 = 0,3 (flexiblewages). After the initial depreciation, the real exchange raleappreciates as domestic prices rise, however overshooting thenew long-term equilibrium as is the case in the Quest simulations.

GRAPH O.3: Real exchange rate effects of a devaluationunder alternative wage rigidity hypotheses

I 2 3 4 5 6 7 8 9 10 11 12 13

The reason behind this phenomenon can be understood byintegrating (5):

(6) [I -8(1-a)] P, = (1-5) ( ! -<*)? ,_ , + 8 a E t + (1-8) aE t _, + 0 T [ S U E U _ , / P U _ I - I]

The log-run equilibrium is characterized by E, = P, = E ,_ , =Pt_ ]. However, this only holds if:0tZuE^,/Pu M - 1 = 0

As long as the Phillips curve coefficient 0 is not zero, long-runequilibrium implies that the integral of the real exchange ratehas to be zero. This is because depreciation leads to output gainsand therefore to tensions on the labour market which imply thatthe real wage is above its equilibrium level.

This shows that the output and employment gains initiallyprocured by a devaluation not only disappear in the long run,but also have to be compensated by further output and employ-ment losses.

291

sduquenoy

Annexes

3.3. Methodology of the simulations

The purpose of the simulations is to measure the cost oflosing the nominal exchange rate and to study the propertiesof each alternative instrument individually. The results arepresented as deviations from a baseline, which is assumedto be optimal. That is, it is assumed that the instruments ofeconomic policy have been set so that the targets for GDPgrowth, inflation, unemployment and the balance of pay-ments are tracked as closely (and with as little year-to-yearvariation) as possible in the baseline, given the existing trade-offs and the budget constraint of the government.

After the setting of the instruments, the economy is subjectto a shock in an uncontrollable variable (here the demandfor exports). Substantial shocks would have an impact onthe target variables and, since some of them will inevitablybe adversely affected, call for a reoptimization in the direc-tion of the baseline values. As the exchange rate instrumentis lost, the question is whether it should be replaced byanother instrument. In a pure target-instrument frameworka la Tinbergen, it could be argued that, since balance ofpayments equilibrium disappears as a target for each individ-ual country in EMU, there remain just enough instrumentsto reach the other targets. But if the use of the instruments,and the frequency with which they can be changed, is restric-ted, it is interesting to see what the properties of alternativecandidates would be, compared to those of exchange rateadjustments. Such restrictions are realistic as long as a pre-cise Tine-tuning' is ruled out.

The design of a comparative exercise in the case of an exportshock is therefore as follows:

(i) First run a simulation of a sustained drop in foreigndemand with exchange rates and interest rates fixed innominal terms.9 This moves the economy away fromthe baseline, which is assumed to represent the optimalequilibrium path. The period of simulation should belong enough to reach the new equilibrium characterizedby an x % real depreciation, although in practice itwould be difficult to reach a steady state. (Let T be therequired length of the period.)

(ii) Combine (i) with a once-and-for-all depreciation by x %which represents the policy that would be chosen inorder to 'frontload' the real exchange rate adjustment.

(iii) Use an alternative instrument in order to achieve thesame output loss as in simulation (ii) (measured by thepresent value of real GDP over T years).

The adjustment paths can be judged on different criteria:(a) the result at the end of the period;(b) the variability in growth rates along the way;(c) the present value accumulated over the period (i.e. the

equivalent change in period 1 of the total change overthe [1,T] period).

All three criteria are employed in the presentation of theresults. The variability is measured as the root-mean-square-error (RMS) of the variable itself, or of its growth rates, indeviation from the baseline. The present value is presentedas a percentage difference from the corresponding startingvalue in the baseline. The discount rate is the actual long-term interest rate of the country in the 1980s, that is about10 % for Germany and 15 % for France. Real variables arediscounted by the appropriate real interest rates of that time,4 % for Germany and 6 % for France.

3.4. The exchange rate versus alternativeinstruments

To illustrate the cost of losing the exchange rate instrument,a sudden permanent drop of 5 % in the demand for Frenchexports has been simulated.10

This shock is combined with either:(a) fixed nominal exchange rates (EXP),(b) a depreciation of 7 % (DEV),(c) an increase in government investment (GVT),(d) an ex-ante reduction of the wage rate in the first quarter

of the simulation (WMOD),(e) a reduction of social security premiums paid by em-

ployers (SCC).

Without changes in nominal exchange rates (simulation (a)),the 5 % export shock would be deflationary and ultimatelylead to a real depreciation of some 7 % in the 7th year ofthe simulation (Table D.2, panel A).11 Real GDP would

Nominal interest rates are kept constant at baseline values throughoutthe exercise. This is consistent with the following assumptions: (a) neitherthe shock, nor the realignment are anticipated; (b) the country is small,and therefore developments in the home economy do not affect theinterest rate of the monetary union.

The nature and the size of the shock are only illustrative. However, anexport shock of that size is not unrealistic. In fact, for 1986 as a whole,the residual of a conventional export equation for France was about5%,Examination of the simulation results shows that real depreciationapproximately reaches a plateau after seven years. However, this is stillnot a steady state.

292


have almost returned to the baseline by then, but in themean time the present value of the loss would amount to 6percentage points, and the present value of real losses inincome per wage earner would also be about 6 % (TableD.2, panel C).

According to the methodology detailed above, in simulation(b) the nominal exchange rate is depreciated by 7 % —roughly the same value as the real exchange rate change inyear 7 of simulation (a) — at the beginning of the period.This depreciation reduces the present value of the outputloss by 1,3 percentage points (Table D.2, panel C). Thedisinflationary impact of the demand shock is more than

wiped out and the price level is 1,6 % above the baseline inyear 7. In spite of the output gains, the present value of thereal wage (-6 %) is lower than for the pure export shock,because of the deterioration in the terms of trade.

The next three simulations are calibrated in order to achievethe same reduction in the present value of output loss asthrough the devaluation. Therefore, present values of realGDP deviations (Table D.2, panel C, column 2) all amountto 4,5 percentage points. For that purpose, an increaseof government investment by 0,15% of GDP would besufficient. This would reduce the loss in real income to 4,6 %and leave GDP prices below the baseline. The deficit on the

Table D.2Summary results for export demand shock: France

Standard Phillips curve

A — Results in year 7: differences from baseline

EXPDEVGVTWMODSSC

RealGDP

-0,27-0,60

0,160,03

-0,03

GDPdeflator

-6,741,58

-5,63-8,60-7,03

Real wagerate

-1,65-1,41-1,35-1,88-1,37

Currentbalance1

-0,53-0,43-0,65-0,46-0,57

Governmentdeficit1

0,500,370,680,450,57

Unemploymentrale

0,230,300,180,090,12

Realexchange rale

7,215,326,009,407,55

B — Root mean square deviations in growth rates

EXPDEVGVTWMODSSC

GDPgrowth

0,530,290,450,540,51

Inflationrate (PCP)

0,961,030,791,180,96

Real wagerale

0,240,350,200,330,21

C— Present value in percentage of baseline

EXPDEVGVTWMODSSC

NominalGDP

-24,08,8

-20,1-34,5-27,2

RealGDP

-5,8-4,5-4,5-4,5-4,5

Real wagerate

-5,6-6,0-4,6-7,6-4,9

Currentbalance2

-4,0-3,3-4,6-3,8-4,2

Governmentdeficit2

3,02,53,92,73,4

1 % of nominal baseline GDP5 % of nominal GDP.EXP = export demand shock (-5%); DEV = EXP -t- exchange rale devaluation (7%); GVT - EXP + government expenditure (0,15% of GDP); WMOD - EXP + wage moderation

=- --

(0,45 %); SSC = EXP + social security contributions (0,3 % of wage bill)Source: Quest simulation by the Commission services

293

Annexes

government's financial balance would, however, substan-tially increase: measured by its present value as a percentageof the present value of nominal GDP in the baseline, it wouldincrease by 50 % compared to the case of depreciation.

An alternative adjustment mechanism is that of wage moder-ation exerted through income policy. Direct wage control isrepresented here by a downward shock in wage growth by0,45 % in the first quarter of the simulation. The governmentmay not have direct control over wage negotiations, but itcould always use the instrument of changing the socialsecurity premiums to be paid by employers. The procedureset out above would imply a reduction in social securitycontributions by 0,3 % of the wage bill. That would be moreof a supply-side shock, since wage costs are directly affectedbut the net wage income of employees is not. In the case ofwage moderation, the downward adjustment of the wagerate per employee would leave the present value of realincome below that of the nominal exchange rate adjustmentand not improve the variability of GDP growth and inflationcompared to the baseline. The supply-side shock would leadto more favourable results, at the cost of an increase in thegovernment deficit.

Since the current balance loses its relevance as a targetvariable under an EMU system and since real GDP wasselected as a benchmark, the costs and benefits of usingthe different instruments can be discussed by focusing oninflation on the one hand, and the trade-off between a largergovernment deficit and a smaller decline in real wage incomeper employee on the other hand.12

Regarding inflation, all simulations except devaluation exhi-bit a fall in the price level. However, since the baseline pathis supposed to be optimal, there is no reason to consider thisas a superior result. A better yardstick is provided by thevariability of the inflation rate (Table D.2, panel B). Thegovernment expenditure instrument turns out to outperformthe other instruments by this criterion, wage moderationbeing the worst because it increases the deflationary impactof the demand shock.

Comparing the present value effects on the governmentdeficit and the real wage rate (Graph D.4) shows that:(i) it is possible to improve upon the effect of the de-

preciation on real income, most by the governmentinvestment scenario and somewhat less by reducingsocial security contributions, both at the expense of ahigher government deficit;

Another important variable is unemployment. The effects on unemploy-ment go hand in hand with those on real GDP, except for substitutioneffects in the wage moderation and social security contribution simula-tions.

(ii) direct wage control would lead to approximately thesame government deficit result as depreciation, at thecost of losses in real wages;

(iii) income policy as a whole (combination of wage moder-ation and social security contributions reduction) couldapproximate the efficiency of the exchange rate instru-ment rather well since for the same real income loss,the present value of the increase in government deficitwould be about 0,6 percentage points of GDP.

The cost of losing the exchange rate instrument when otherinstruments are available can be assessed on Graph D.4. Itis the implicit utility loss arising from the move from point'devaluation' on to the wage moderation/social security/government investment schedule.13

GRAPH D.4: Real wage loss/government deficit trade-offswith alternative instruments

I 5

4 -

Wage moderation

a Devaluation

^Social securityi Governmentinvestment

2 2,4 2,8 3,2 3,6 * 4,4 4,8

Government deficit (present value)

Sourer: Quest simulation! by the Commission services.

As the loss of the exchange rate instrument implies either a highergovernment deficit or a real income loss, the associated cost can onlybe measured using a utility function.

294

sduquenoy


Three conclusions emerge from this exercise:(i) The loss of the exchange rate instrument is unambigu-

ously a cost for output, but not for inflation in the caseof a demand shock.

(ii) If no other instrument is available to replace the ex-change rate, the Quest model leads to the inference thatthis cost is equivalent to an immediate, once and for all1,3 % loss in real GDP in the case of a negative 5 %shock to the demand for French exports.

(iii) If other instruments are available, the loss in real GDPcan be eliminated at the cost of an increase in thegovernment deficit. In the case of a negative 5 % shockto the demand for French exports, the budgetary costis an increase in the deficit by 0,6 % of GDP lasting forone year.

3.5. Demand shocks under alternative wagebehaviour

It is well known that the effectiveness of exchange ratepolicy strongly depends on the behaviour of wages. A higherflexibility, i.e. an increase in the responsiveness of wages tolabour market conditions, and a speedier indexation bothaffect the reaction of the economy to a demand shock andalso the impact of an exchange rate change. As shown inChapter 6, Appendix 6.3, the aggregate supply curve of theFrench Quest model can be considered steep in comparisonto other multinational models. This is because of both rela-tively rapid indexations and a strong responsiveness of wagesto unemployment. Therefore, it is important to examine thesensitivity of the results to a change in wage behaviour.

Graph D.5 depicts the output and income effects of a de-mand shock with exchange rate 'frontloading' policy (lower

case letters) and without changes in the exchange rate (uppercase letters). In each case, three versions of the model arecompared: the standard one, a model with a 100 % increasein wage responsiveness to the labour market conditions (highflexibility) and a model with slower indexation, doubling themean lag (see Box D.2).

GRAPH D.5: Effects of a demand shock upon real GDP andreal wages with and without offsetting exchangerate policy

(upper case letters refer to the fixed exchange rate case; lower case lettersrefer to devaluation)

•5 7 )

| 6,5 •_eL"a 6 ;

™ f

£ 5

4,5

43

High flexibrlii) HIGH FLEXIBILITYo c

Slow indexation

Standard caê

STANDARD CASt

SLOW INDEXATION

4 5 6

Real GDP loss (present value)

aurt, Quest simulations b> the Commission

295

Annexes

Box D.2: Wage-price equations in the Quest model

The wage-price block in the country models for Germany,France, Italy, the UK and the USA have the common generalspecification:14

(1) w - gj(L) - pc + c, . g2(L).(l -a), upro + C2 . LUR + c,. (LUR - LUR ,)

(2)pc = a . pm + (1-a) . p(3) p = g3 (L) . w - c, . g2(L) . upro + c4 . ucap

All variables are in rates of change, except for the unemploymentrate, LUR, which is in levels. So, the Phillips curve equation (1)contains both the effect of the level of unemployment on therate of change in the wage rate, and the effect of a change inthat level (hysteresis). The speed of adjustment to a change inthe price level is represented by the distributed lag functiongi(L). In the long run there is therefore perfect indexation.Equation (2) defines consumption prices as a weighted averageof import and value-added prices, where the weight a reflectsthe degree of openness of the country. A labour productivity(upro) term is present in both the wage equation and the value-added price equation. Its coefficient is constrained so thatchanges in labour productivity have no effect on nominal wagesin the reduced forms. In addition, equation (3) contains a ca-pacity utilization (ucap) effect.

The estimated coefficients are given in Table D.3. The distrib-uted iag functions are characterized by their average lag in TableD.4.

14 Some differentiations which only have a short-term effect have been left out of thispresentation, as is the error-correction term which ensures a constant labour-incomeshare in the long run.

Table D.3Coefficients of the wage-price equationsin the Quest country models

<r, c, c, c4

D 0,90 -0,09 -0,84 0,15F 0,30 -0,13 -0.41 0,14I 0,37 -0,22 -0,63 0,23UK 0,31 -0,11 -1.36 0,09USA 0,50 -0,22 -0,13 0,04

Table D.4Average lag (longest lag)of the distributed lag functions in quarters

f i (L) E2<L) gj(L)

D 0,67(2) 0,67(2) 1,33(4)F 0,67 (2) 0,67 (2) 0,67 (2)I 1,00(3) 1,00(3) 1,33(4)UK 0,33(1) 1,33(4) 1,33(4)USA 1,00(3) 1,00(3) 5,5(11)

High flexibility of wages to changes in labour market conditions(Graph D.5) is represented by doubling the coefficients c2 andc3 for France.

Slow indexation is imposed by replacing g[(L) in the Frenchwage equation by a distributed lag over 4 quarters which doublesthe average lag to 1,33 quarters.

It is apparent that at fixed exchange rates as well as withexchange rate policy, a higher wage flexibility reduces theimpact of a demand shock on output because the shocktranslates faster into a real wage decline. Regarding index-ation, a slower indexation increases the output loss bothwith fixed exchange rates and a frontloading exchange ratepolicy. This is because low indexation reduces the speed ofadjustment in both cases. This is not a peculiarity of the

model but rather a fairly general result, as demonstrated inBox D.3.

The output cost of losing the exchange rate can in each casebe represented by the distance between upper case and lowercase points. This cost is reduced by both a higher wageflexibility and a speedier indexation. Once again, this resultappears to be fairly general, as demonstrated in Box D.3.

296

sduquenoy


Box D.3: Exchange rate policy and wage behaviour

The purpose of this box is to evaluate how the cost of losingthe exchange rate instrument is affected when wage behaviourchanges. For that purpose, analytical results are derived from asimple illustrative model.

The model is a simplified linearized representation of the work-ing of large-scale neo-Keynesian macro models like Quest. Itconsists of five equations:( l ) y = d + ax

(3) x = a (e - p) - m y + x0

(4) p = w(5) w = 5 (ae + ( l -a)p) + 0y

where y is real GDP, d is real domestic demand, x is net exports,a the imports/[GDP+ imports] ratio (equal to the exports/[GDP+imports] ratio), e the exchange rate, p the GDP deflator,w the wage rate. All variables are expressed as percentagedeviations from the baseline, hence the model is linear.

Equation (1) is a linearized expression of goods market equilib-rium, equation (2) links domestic demand to real domesticincome, equation (3) gives net exports as a function of thereal exchange rate, domestic demand and an exogenous termrepresenting the demand shock, equation (4) gives GDP pricesas a function of wage costs, and equation (5) is a Phillips-typeequation with 5 representing the indexation parameter and 0representing the responsiveness of wages to labour market con-ditions (flexibility). It is assumed that wages are fully indexedin the long run, so 5LR = 1. However, indexation lags (e.g. dueto overlapping contracts) lead to incomplete indexation in theshort run. Only two periods are considered: the short run (firstperiod result), and the long run.

Partial resolution of the model gives two equations for theaggregate demand and aggregate supply curves:

(AD)y = A(e-p) + B XQ

(AS) p = [0 (1 - 6 (1 -a))] y + [5 a / ( I - 8 (1 -a))] e

where A and B depend on the parameters in equations (1) to(4). Complete solution yields the following expression for y:(6) y = [B(l - 5 ( l -a ) ) / ( l - 5( l -a) + A0)] x0

+ [A(l - 8 ) / ( l - 5 (I-a) + A0)]e

The above expression can be used to assess the effect on outputof different shocks.

Demand shock with fixed exchange rates

This case corresponds to the effects of a demand shock in EMU.With e = 0, equation (6) yields:

(7) y = L (©, 8) x0 L'0 < 0 L'g < 0

Simple calculations show that partial derivatives of L withrespect to 0 and 5 are both negative. This means that for

an economy facing an adverse demand shock, a higher wageflexibility or a speedier indexation of wages on prices reduce theoutput loss. Referring to Graph D. 1, both an increase in 0 and6 increase the slope of the aggregate supply curve, therebyreducing the output loss.

Exchange rate shock

Equation (6) can be used to assess the effects of a devaluation.With x0 = 0, equation (6) yields:(7) y = M (0, 6) e M'e < 0 M'5 < 0

Thus, the output gain resulting from a devaluation is reducedby a higher flexibility or a higher indexation. A special case isthat of instantaneous indexation (5 = 1), where the output gainis reduced to zero. This is the familiar interdependence betweenwage behaviour and the effectiveness of exchange rate policy(see for example Marston, 1985).

Demand shock with offsetting exchange rate policy

The above model can be used to interpret the results of Questsimulations. Putting y equal to zero in equation (AD) gives thenecessary long-run real depreciation,15 i.e.:(8) e - p = - B X0 / A

Frontloading the long-run real exchange rate depreciation corre-sponds to an immediate nominal depreciation of the same mag-nitude. This yields:(9)y = [8 a / ( I - 8(1-a) + A0)] B x^, i.e.(9') y = N (0, 5) x,, N'e <Q N'8 > 0

The effectiveness of such an exchange rate policy for reducingthe output loss is therefore reduced if wages are flexible, butincreased by a speedier indexation.

The cost of losing the exchange rate instrument

The output cost of losing the exchange rate instrument is givenby the difference between the output effects of the demand shockwith and without exchange rate devaluation:(10) C = - [(1 - 5 ) / ( I - 8(1- a) + A0)]Bxo

Since we are interested in the effect of negative demand shocks(\Q < 0), (10) can be rewritten in terms of the elasticity of outputwith respect to the demand shock:(11)C = [(1 - 8)/(l - S(l-a) + A0)]B,i.e.(11') C = Q(0, 8) Q'e < 0 Q'6 < 0

Both partial derivatives are negative. Thus, the cost of losingthe exchange rate instrument in terms of output is a decreasingfunction of _ the degree of wage flexibility and the speed ofindexations. Equation (11) also shows that this cost is nil withinstantaneous indexation.

This assumes that the long-run aggregate supply curve is vertical.

297

Annexes

3.6. A supply shock

An increase in social security contributions paid by theemployers affects the real wage cost, the price level and thegovernment deficit, but leaves ex-ante aggregate demandunchanged. Hence, it can be considered as a pure supplyshock.

It is well known that the choice of an exchange rate reactionto a supply shock entails a trade-off between output andinflation. Since a supply shock is both inflationary andcontractionary, the response can be either to appreciate (inorder to counteract inflation) or to depreciate (in order tobring the real exchange rate back to the baseline value).Hence, the reaction depends on the preferences of the policy-maker.

In this context, the loss of the exchange rate instrument isnot a cost of the same nature as in the case of the demand

shock. However, the loss of this instrument reduces thepossibilities of trading off growth against inflation.

This is illustrated by Graph D.6 which, in the case of the1 % shock in German social security contributions, gives thepresent values of the GDP loss and the discounted changein inflation for different values of the exchange rate change.The resulting schedule depicts the trade-off between growthand inflation facing the policy-maker. Assuming the loss ofthe exchange rate would force him to stick at point Ainstead of choosing point B, the associated cost would be thecorresponding utility loss.

This cost cannot be assessed without an explicit utility func-tion. However, except for extremely strong preferences foreither growth or price stability, it can confidently be assessedto be much smaller than in the case of a pure demand shock.

GRAPH D.6: Exchange rate policy and output/inflation trade-off after a supply shock: Germany

GDP loss16

14

12

10

Appreciation

Dep reflation

-2

0,1 0.2 0,3 0.4 0,5 0,6

Discounted change in inflation

0,7 0.8 0,9

298

sduquenoy


Appendix: Detailed simulation results

Table D.A1Export demand shock : France, high wage flexibility


Real GDP GDP deflator

EXPDEVGVTWMODSSC

0,38-0,09

0,390,510,46

-11,12-2,55-9,62

-11,67- 10,57

Real wagerate

-2,33-2,08-1,97-2,32-1,97

Currentbalance1

-0,40-0,33-0,51-0,36-0,44

Governmentdeficit1

0,390,270,550,370,46

Unemployment Realrate exchange rate

-0,02 12,500,11 9,79

-0,03 10,64-0,10 13,21-0,07 11,81


EXPDEVGVTWMODSSC

GDP growth

0,580,320,500,580,56

Inflationrate (PCP)

1,631,111,401,701,53

Real wagerate

0,370,400,320,410,33

C — Present value in percentage of baseline

EXPDEVGVTWMODSSC

NominalGDP

-37,6-1,8

-32,6-44,1-38,2

Real GDP

-4,2-3,4-3,4-3,4-3,4

Real wagerate

-8,5-8,2-7,3-9,8-7,6

Currentbalance2

-3,8-3,1-4,3-3,6-4,0

Governmentdeficit2

2,62,13,42,32,9

1 % of nominal baseline GDP.; % of nominal GDP.EXP = export demand shock (-5%); DEV = EXP + exchange rale devaluation (7%);(0,45%); SSC = EXP + social security contributions (0.3% of wage bill).Source : Quest simulation by the Commission services.

GVT = EXP + government expenditure (0.15% of GDP): WMOD = EXP + wage moderation

299

sduquenoy

sduquenoy

Annexes

Table D.A2Export demand shock: France, slower indexation


EXPDEVGVTWMODSSC

Real GDP

-0,39-0,62-0,21

0,00-0,06

GDP deflator

-5,88-1,98-4,60-8,50-6,47

Real wagerate

-1,52-1,28-1,15-1,89-1,20

Currentbalance1

-0,56-0,43-0,70-0,47-0,61

Governmentdeficit1

0,520,380,760,450,61

Unemploymentrate

0,280,290,190,110,13

Realexchange rale

6,244,924,819,286,91


EXPDEVGVTWMODSSC

GDP growth

0,520,260,420,540,50

Inflationrale (PCP)

0,840,790,651,150,87

Real wagerate

0,230,390,170,330,18

C— Present value in percentage of baseline

EXPDEVGVTWMODSSC

NominalGDP

-21,46,4

-16,8-33,7-25,1

Real GDP

-6,1-4,6-4,6-4,6-4,6

Real wagerate

-5,0-6,3-3,9-7,5-4,0

Currentbalance2

-4,0-3,2-4,8-3,8-4,4

Governmentdeficit'

3,12,44,32,73,7

1 % of nominal baseline GDP.2 % of nominal GDP.EXP = export demand shock (- 5 %); DEV = EXP + exchange rale devaluation (7 %); GVT - EXP + government expenditure (0,2 % of GDP); WMOD = EXP + wage moderation (0,6 %);SSC = EXP + social security contributions (0,4% of wage bill).Source: Quest simulation by the Commission services.

300


Table D.A3Increase in social security contributions (1 % of nominal GDP) paid by employers : Germany


Exchange rate change

-3%|-2% > depreciation-1%J0%1% 12% } appreciation3% J

Real GDP

-3,2-3,0-2,9-2,6-2,4-2,3-2,1

GDP deflator

8,66,74,83,01,2

-0,5-2,2

Real wagerate

-2,5-2,7-2,9-3,1-3,3-3,5-3,7

Currentbalance1

0,10,20,20,20,20,10,1

Governmentdeficit1

0,40,40,40,40,50,50,6

Unemployment Realrate exchange rate

1,4 -5,11,3 -4,41,3 -3,61,2 -2,91,1 -2,21,0 -1,51,0 -0,8


Exchange rale change

-3%1-2% ? depreciation-1% 10%1% |2% | appreciation3% J

ODP growth

0,730,620,530,500,530,600,72

Inflationrate (PCP)

1,35

1,17

1,02

0,910,860,870,95

Real wagerate

0,670,630,600,580,570,590,61

C — Present value in percentage of baseline

Exchange rale change

-3%1-2% > depreciation- 1 % J0%1% 12% \ appreciation3% J

NominalGDP

29,623,517,611,76,00,3

-5,2

Real GDP

-6,8-8,1-9,4

-10,8-12,1-13,4-14,8

Real wagerate

-4,6-5,9-7,3-8,6-9,9

-11,3-12,6

Currentbalance7

-0,3-0,3-0,4-0,6-0,7-0,8-1,0

Governmentdeficit2

-2,1-1,4-0,8-0,1

0,6

1,21,8

1 % of nominal baseline GDP.1 % of nominal GDP.

301

sduquenoy

Annexes

References mission of the EC, Directorate-General for Economic andFinancial Affairs, March.

Alien, P. R. (1976), 'Organization and administration of a Fair, R. C. (1988), 'Sources of economic fluctuations in themonetary union', Princeton Studies in International Finance United States', Quarterly Journal of Economics, May, pp.No 38, Princeton University. 313-332.

Marston, R. (1985), 'Stabilization policies in open econo-Bekx, P., Bucher, A., Italianer, A. and Mors, M. (1989), mies', in Jones, R. W. and Kenen, P. B. (eds), Handbook of'The Quest model (version 1988)', Economic Papers 75, Com- International Economics 2, Amsterdam, North-Holland.

302

sduquenoy

sduquenoy

Annex E — Exchange rate regimes in the EC: simulations with the Multimod model

Annex E

Exchange rate regimes in the EC:simulations with the Multimod model

1. Introduction and non-technical summary1

Standard model simulations (like those presented in AnnexD) help to assess the effectiveness of alternative policy instru-ments when a given economy is hit by a specific shock, andmore precisely to assess the cost of losing the exchange rateinstrument in such circumstances. However, such simula-tions cannot provide a more thorough evaluation of theperformance of alternative monetary policy and exchangerate regimes for two reasons: first, the quality of a policyregime cannot be judged with respect to a single perturbationaffecting an identified variable, but rather when the economyfaces disturbances affecting all variables simultaneously;therefore, model simulations should aim at assessing theperformance of alternative regimes in an environment ofrepeated random shocks rather for a specific, single shock;secondly, the major consequences of the choice of an ex-change rate regime, and especially of that of a preciselydefined and highly disciplinary one like EMU, cannot besupposed to be simply ignored by agents, particularly witha single currency; for example, it cannot be assumed thatagents would not be aware of the fact that participation inEMU implies that inflation in the domestic economy shouldnot diverge from those of other EMU partners; in otherwords, if the so-called 'Lucas critique' of policy evaluationwith standard econometric models (which states that esti-mated model parameters cannot be assumed to be invariantwhen the policy regime changes) has some relevance, it iscertainly in the case of such a major and highly visiblecommitment resulting from an international treaty.2

The response to the first of these two limitations is touse the technique of stochastic simulations (see Box E.I).Basically, it consists in subjecting the model to repeated

The model simulations presented in this annex were performed under theresponsibility of the Commission services with the Multimod model ofthe IMF. The support from the IMF staff, and in particular Messrs PaulMasson and Steven Symansky, is gratefully acknowledged.See Lucas (1976). This is not the place to discuss the relevance of theLucas critique in the more general context of policy evaluation. However,it is worth mentioning that as EMU would be a regime change of anexceptional magnitude and clarity, a large part of the arguments putforward in the 'critique of the critique' (see for example Sims, 1982) donot hold in that context.

random shocks affecting a whole set of variables simul-taneously.3 With the policy regime represented by a rulelinking policy instruments (in the present case the short-terminterest rate) to observed deviations from target for keyeconomic variables (like inflation and output), the result ofsuch simulations represents for a given regime the behaviourof the economy in an environment of random disturbances.As for standard simulations, stochastic simulations areperformed around a reference path (the baseline) where allavailable instruments are supposed to be used in an optimalway. Therefore, the aim of the authorities as described bythe rule is to bring the economy back to the baseline. Theeconomic criterion for judging the relative performance ofdifferent rules is their ability to minimize on average thedeviation from baseline of key economic variables.

The second limitation raises deeper methodological issues.Basically, the response is to use a model whose represen-tation of agents' behaviour is structural enough to remaininvariant with respect to a change in the economic policyregime.

Unfortunately, there is no way to make sure that any particu-lar model fulfils this condition. However, it is generallyaccepted that structural models are less subject to changesin parameters when the policy regime changes than smallreduced form models whose equations represent a mix ofgenuine behaviour and policy reactions.4 Another necessarycharacteristic of the model, at least for the comparison ofexchange rate regimes, is that it includes forward-lookingexpectations. This is of paramount importance for the rep-resentation of financial and exchange markets, since theirrevocable fixing of the exchange rate, i.e. the expectationof no future exchange rate changes has to impact on currentexchange rates and bond rates. This is also important forlabour and goods markets, where expectations of futureinflation impact on current wage claims and price setting.

By using a structural model incorporating forward-lookingexpectations, it can be expected that a large part of theobjections to econometric policy evaluation with empiricalmodels is removed. It is still possible that EMU could inducesome genuine behavioural changes that would not be takeninto account in the simulations, e.g. regarding wage/price

Stochastic simulations can also refer to shocks affecting the parametersof the model. The aim of such simulations is not to assess the performanceof a given economy (supposedly accurately represented by the model) inan environment! of random shocks, but rather to assess whether the(forecasting or simulation) results are robust.A typical example of the latter is a two-equation reduced form determin-ing output and inflation.

303

Annexes

flexibility. However, in the absence of a methodology forevaluating possible changes of this kind, ad hoc changes inthe parameters of the model can only be used for illustrativeand normative purposes, but not for an ex ante evaluation.Conversely, it should be recalled that forward-looking mo-dels are frequently very reactive as expectations of future

changes trigger immediate adjustments, and that since expec-tations are by definition not observed, parameters of themodel which describe the formation of expectations are notdetermined with precision. Therefore, it could also happenthat the impact of EMU on for example wage and priceformation is overestimated.

Box E.1: A schematic representation of the simulationtechniques

Standard simulation techniques consist in subjecting the modelto a shock to a particular variable (normally exogenous, but itcan also apply to endogenous variables). The model then givesthe impact of the shock on the macroeconomic variables, asubset of which are policy objectives. Assuming that the authori-ties have full information on the nature and the magnitude ofthe shock, a policy action can be taken to offset some of itsundesirable effects. This setting is schematically represented inthe top panel of Graph E.I.

The method of stochastic simulation can be described withinthe same framework. The core of the methods remains identical,but three major changes are introduced as represented in themiddle panel of Graph E. 1:

(i) the model is now being subjected to repeated multipleshocks affecting at each period a whole range of variables;these shocks are not anticipated before they occur;

(ii) the policy action is no more supposed to be discretionary,but rather takes the form of systematic reactions to devi-ations from target in the policy objective variables; thesereactions are given by a set of policy rules which character-ize an economic policy regime.

(Hi) agents are supposed to form forward-looking expectationsof key economic variables; as expectations are based onthe solution of the model for future periods, informationregarding the policy regime is embedded in these futurevalues.

In addition, shocks themselves can depend on the policy regime;this happens obviously in the case of the intra-EC exchangerate shocks in EMU, but could also arise from regime-inducedbehavioural changes.

The bottom panel of Graph E.I explains the working of thesimulations as regards the particular case of inflation.

The simulations presented in this annex have been run usinga slightly modified version of the IMF Multimod model (seeBox 2.5 in Chapter 2 and Appendix E.I for an overview ofthe model, and Appendix E.II for the modifications). Thissmall model, which incorporates both Keynesian and classi-cal features, has a forward-looking financial block whichincludes a determination of the exchange rate through anuncovered interest rate arbitrage condition. Wage-price, con-sumption and investment equations also incorporate for-ward-looking behaviour, although backward-looking index-ation and liquidity constraints on consumers are also takeninto account. Since Multimod has individual models for fourEC Member States (D, F, I, UK), the impact of EMU hasonly been simulated for those four countries. However, theaim of the exercise is to provide an overall assessment of theconsequences of EMU, not.to discuss how specific countrieswould be affected.

A further reason to focus on general rather than individualcountry results is that due to the estimation technique usedby the IMF, country models are very similar to each other.This is because only statistically significant differences in the

parameters of the equations were kept, while a commonvalue was imposed in other cases. Therefore, country modelsexhibit fairly similar reactions in deterministic simulations.This does not mean, however, that results of stochasticsimulations are biased towards the optimistic side. First,while the models are similar they are not identical becauseof asymmetries in structure (as regards for example the shareand geographical pattern of foreign trade, the size of publicdebt, etc.) and in certain behavioural parameters (the degreeof backward-looking wage indexation differs from one coun-try to another). Second, as developed in Appendix E.I, tothe extent that behaviour is more asymmetric in reality thanrepresented in Multimod, this implies that residuals of thebehavioural equations differ and are not correlated acrosscountries. Since those residuals are used to generate theshocks in the simulations, any asymmetry which would behidden in the estimated equation would reappear in the formof an asymmetry in shocks.

The first step in the design of the simulations is to representadequately the different exchange rate and policy regimes.Four different regimes are simulated (see Table E.I):

304


GRAPH E . I : Standard and stochastic simulations compared

Single shock Mode!

Pobqaction

Standard simulations: overview

Macro \anablesIPniicy objectives)

Stochastic simulations: overview

Inflationexpectations

Stochastic simulations: example

(i) a clean float, in which all countries use monetary policyto target domestic variables;

(ii) the EMS as it could presumably function in Stage I,which is modelled in accordance with the assumptionsspelled out in Chapter 2; following the German anchorhypothesis, it is supposed to remain asymmetric; it isalso supposed that all Member States, including theUK, would participate in the narrow band of the ERM;

(iii) a hypothetical asymmetric EMU (AEMU), which is notconsidered a realistic option but serves as a technicalintermediate regime between the EMS and EMU, sinceit is characterized by both irrevocably fixed exchangerates and the maintenance of the German leadershiphypothesis;

(iv) EMU, in which monetary policy is supposed to keepthe same orientation as for Germany in AEMU, butwhose scope is extended in order to cover the economyof the Community (in practice, the four aforementionedcountries) instead of that of the FRG.

While the choice of these regimes is quite straightforward,their precise definition in terms of rules involves specificassumptions. First, the monetary policy rule followed bythe authorities reflects a certain weighting of the policyobjectives, notably output and inflation. It was chosen tomimic closely the practice of monetary targeting followedby central banks, and thus to give a high priority to thecontrol of inflation in the definition of the policy rule. Alsoall countries are supposed to follow the same rule althoughdifferences in structure or policy preferences could implythat the optimal rule differs across countries. Second, model-ling the EMS is especially difficult due to the existence ofbands and the possibility of realignments. It was decided totry to capture the essential features of the ERM of the late1980s, which has been much more disciplinary than the earlyEMS. However, the evolving nature of the system meansthat the simulated regime should be regarded as one amongdifferent possible representations of the system. Third, forthe sake of clarity the monetary rule is kept unchanged inthe EMU, although assuming no change in preferences it isno longer optimal. Fourth, no fiscal policy rules are intro-duced. The two latter assumptions imply the cost-benefitassessment of EMU could err on the pessimistic side.

The second major ingredient for stochastic simulation is thederivation of the shocks to the model. For non-policy shocks,the standard technique is to consider the residuals of theestimated equations as a sample of the shocks to endogenousvariables. This means that departures from the behaviourrepresented by the equation are interpreted by the policy-maker as unanticipated shocks. An important issue iswhether the size of and correlations between these shocks

305

sduquenoy

Annexes

change between regimes, for instance because shocks couldbecome more symmetric as argued in Chapter 6. For allvariables save exchange rates, the size of and correlationsbetween shocks have been kept unchanged among regimes.While this may introduce a bias against the EMU regime, ithas the advantage that the regimes can be compared moreeasily. For exchange rate shocks, however, some assump-tions have to be made since the correlation between exchangerate shocks is necessarily equal to one in the EMU regimes,close to one in the EMS regime but presumably lower in afree float. Here, the observed shocks for the three EMSmembers and the United Kingdom are used as being rep-resentative for the different regimes.

The adopted methodology implies that the results of thestochastic simulations are entirely driven by three factors.The first concerns the changes in exchange rate shocksbetween the regimes. As shown below, this implies that thedisappearance of intra-Community exchange rate shocks isconducive to inflation stability and also output stability.Secondly, there is the impact on the monetary and exchangerate policy of the Member States, which is completely inde-pendent in the float regime, strongly geared towards theGerman monetary policy in the EMS and asymmetric EMUregimes and which becomes a common monetary and ex-change rate policy in the symmetric EMU regime. The degreeof monetary policy independence strongly influences thepossibilities of accommodating real and inflationary shocks.Finally, there is the impact of the regimes on private behav-iour through forward-looking expectations, notably con-cerning wage and price discipline. The absence of devalu-ation possibilities relative to other Community countries inthe EMU regimes imposes more discipline on wage and pricebehaviour, as can also be demonstrated analytically. Thecombination of these three factors shows that the float,EMS, asymmetric EMU and symmetric EMU regimes, re-spectively, all improve upon each other in terms of reducinginflation variability. For the EMS, the reduction of intra-Community exchange rate shocks compared to the floatregime is a decisive factor. This factor also plays a role forthe asymmetric EMU regime relative to the EMS, helped bythe stronger wage and price discipline imposed by the fixedexchange rates. The reduction in inflation variability in thesymmetric EMU regime compared to the asymmetric EMUregime may largely be ascribed to the fact that monetarypolicy, while still as anti-inflationary, is now geared to theaverage Community inflation and output levels instead offocusing exclusively on the German situation. Taking thefour regimes in the same order, there is no such uniformreduction in output variability. The relatively minor effectof the diminished exchange rate shocks on output variabilityin the EMS compared to the float regime does not compen-sate for the virtual loss of the monetary policy and exchangerate instrument used in the float regime for output stabiliza-

tion. In the asymmetric EMU regime output variabilityreduces compared to the EMS because of the strong effectsof the absence of the devaluation instrument on wage andprice discipline. The symmetric EMS improves on this be-cause, while excluding the possibility of depreciation, thecommon monetary policy is concerned with average Com-munity output rather than with that of Germany alone.

These results naturally depend on the assumptions madeand methods used, which may all be subject to importantcaveats, notably concerning the impact of the EMU processon (forward-looking) expectations. Nevertheless there is atleast a qualitative indication that there are elements whichmay compensate for the loss implied by the absence of theexchange rate instrument, presumably even to an extentthat there is a net improvement in macroeconomic stability.Important conditions for this to become true are the monet-ary policy orientations of the future EuroFed, which shouldbe geared to price stability, and the behaviour of privateagents, which should take into account the constraints im-posed by EMU.

2. Monetary policy and exchange rate regimes

2.1. General description

The simulations with the Multimod model are used to ana-lyse the properties of four exchange rate regimes.

1. Free float: each G7 country pursues its own monetarypolicy to stabilize output (low priority) and inflation (highpriority) under a system of floating exchange rates. Identicalpolicy rules, which serve to mimic the practice of monetarytargeting, apply to all countries of the model.5

2. EMS: the exchange rates of the four individual Com-munity countries in the Multimod model are linked in asemi-crawling peg version of the European Monetary System(NB: it is therefore assumed that the United Kingdomtakes part in the system). The overall monetary policy isdetermined by Germany, which is in a free float with respectto non-EMS currencies. The monetary policies of the threeother EMS countries are such as to maintain the paritywith respect to the DM, subject to margins of fluctuationcorresponding to those observed in 1985-87 in the actualEMS. These margins are used to give a small weight to

In the Multimod model, the exchange rate of the block for smallerindustrial countries is pegged to the DM. This assumption is maintainedthroughout.

306


considerations of domestic monetary policy. There is a re-alignment rule based on the price differential with Germany:each time the differential amounts to 8 %, a devaluation of4 % takes place.

3. Asymmetric EMU: this is the same regime as the EMS,but with zero margins and no possibility for realignment.The asymmetry comes from the fact that monetary policyis determined exclusively by the German monetary policyobjectives.

4. Symmetric EMU: this is the same regime as the asymmet-ric EMU, with the same monetary rule as for Germany, butwith as monetary policy objectives the average performanceof output and inflation for the four EMU members insteadof that of Germany.

In each of these regimes, all countries are assumed to pursuemonetary policy by means of a policy reaction function forthe short-term interest rate. Due to the assumptions of freecapital mobility, perfect asset substitutability and perfectforesight, exchange rates follow uncovered interest rate par-ity with respect to the dollar. The expected exchange rate isdetermined by the model (i.e. model-consistent expec-tations), given an exogenous long-term nominal steady statevalue of the exchange rate.

2.2. Free float regime

For countries in a floating rate regime, monetary policy issupposed to target a nominal aggregate. This implies that amonetary policy rule has to be devised. Standard monetarypolicy practices are money supply targeting and nominalGDP targeting. Since explicit introduction of a money sup-ply target would introduce disturbances coming from moneydemand behaviour, the choice has been made for a ruletargeting macroeconomic flow variables.

A first possibility, as for example chosen by Frenkel etal. (1989) for their simulations, is nominal GDP targetingspecified in level terms:( l)( i- ib) = I00a(y-y b )

where i is the short-term nominal interest rate expressed inpercentage points, y is the logarithm of nominal GDP, thesuperscript b indicates baseline values, and a = 0,25 (1 %deviation of nominal GDP raises the interest rate by 0.25percentage points).

This kind of specification has some undesirable medium-term properties: it implies that after an inflationary shock,monetary policy not only aims at reducing inflation, but

Table E.I

Monetary policy rules in the four simulated regimes

Germany Other Member Slates<F, I, UK)

Float

EMS

AEMU

EMU

Policy aims at stability of domestic inflation (high priority) and real GDP {low priority)Same policy reaction function for all four Member States

Same as float

Same as float

High priority for DM exchange rate stabilityLimited domestic stabilization within narrow bandsRealignment rule: devalue by 4% when the price leveldifferential reaches 8 %

Irrevocably fixed DM exchange ratesZero marginsZero policy autonomy

Same as float, but with Community variables replacing domestic variables in the monetary policy reaction function

307

Annexes

also at eliminating the deviation of prices from their baselinelevel. This does not correspond to standard behaviour ofcentral banks, which are generally considered to be indiffer-ent to price levels. Thus, an alternative rule is:

(!') (i - ib) = 100 [ o, (Ji - 7ib) + a2 (q - qb) ]

where q is the logarithm of real GDP, and n is the inflationrate (p - p_i), with p the logarithm of the absorption pricedeflator.

It is easy to show that this kind of rule is consistent withthe standard monetary policy practice of setting the moneysupply target for the year ahead in accordance with desiredoutput and inflation. Moreover, values of at and a2 can bederived from the parameters of a money demand equationlike that of Multimod (see Appendix E.III). After adding apartial adjustment mechanism for the interest rate this finallyyields as monetary policy rule:

(1") (i - ib) = 0,5 { 100 [ 2(7t - 7tb) + 0,4 (q - qb)]}+ 0,5 {i - ib}_,

This implies that the interest rate reaction to a 1 % deviationof real GDP is 20 % of that to a one percentage pointdeviation of inflation. This distribution of weights for theobjectives puts a strong emphasis on inflation compared toother stochastic model exercises.6

2.3. The EMS regime

Any model representation of the EMS is bound to be ad hocdue to several factors. First, there is the exchange ratemanagement inside the bands in the form of intra-marginalinterventions. This problem has been solved by imposing astrong linearity in the interest rate reaction function whenthe bands are approached. Secondly, there is the debate onthe asymmetric character of the EMS due to monetaryleadership of Germany.7 Since most empirical research con-firms this hypothesis, it is retained for the simulations, imply-ing that Germany conducts its monetary policy indepen-dently and the other EMS members are firmly constrainedin their monetary policy because their parity with the DMhas to be respected. Thirdly, there is the issue of realign-ments, which in the past were not subject to specific rules

For instance, the exercises presented at the Brookings conference 'Empiri-cal evaluation of alternative policy regimes', 8-9 March 1990, whichincluded an output-inflation rule, all gave at least the same weight tooutput as to inflation; see Bryant el al. (1990).See De Grauwe (1988) or Von Hagen and Fratianni (1989).

but should be so in the model. Here a realignment rule basedon the experience of the 1985-87 period has been adopted.There are thus several elements which make this represen-tation of the EMS distinct from the past or from what canbe expected in Stage I. Indeed, ever since its conception, theEMS has been an evolving system difficult to capture by thestrict rules of econometric modelling. Consequently, theresults for this regime are especially uncertain and shouldbe regarded as emanating from a stylized EMS rather thana precise description of past, present or future forms of thesystem.

2.3.1. Monetary policy rule

In the EMS regime, the non-German ERM Member Statesare supposed to target their exchange-rate vis-a-vis the DM.Fluctuation bands, however, allow for short-term deviationfrom the central rates. These are a distinctive feature of theEMS as opposed to full monetary union. Exchange-ratefluctuations within the bands may act as absorbers for smallshocks, and fluctuation bands are useful as long as realign-ments are not ruled out.

Monetary policy reaction functions for non-German ERMmembers therefore combine targeting of domestic variablesand exchange-rate targeting. The rule has to be designed inorder to allow for some limited flexibility, but to ensure thatexchange rates are kept within the bands.

A straightforward solution is to specify a standard linearreaction function with inflation, output and exchange ratesas arguments. Starting with a quadratic utility function:

(2) U = (i - ib)2 + a j (n - ?cb)2 + a2 (q - qb)2

+ a3 (e - eb)2

where e is the logarithm of the nominal exchange rate (localcurrency per DM), and an interest rate term is added inorder to avoid instrument instability, a non-linear reactionfunction for the interest rate can be derived by differentiating(2).

(3) (i - ib) = 100 [(*! (n - *b) + a2 (q - qb) + a3 (e - eb}]

where c^ = - (a^lOO) [Sit/Si], a2 = - (a2/100) [5q/5i], a3 =- (a3/100) [Se/Si]

Interpretation of the parameters at of the reaction functionis thus straightforward. They are a combination of the par-ameters from the utility function and the elasticities of themodel.

308


Even with a high priority for exchange rate stability, how-ever, (3) does not ensure that the exchange rate is kept withinthe fluctuation bands. This may be obtained by adding astrongly non-linear term to (3), allowing to combine someflexibility with fixed fluctuation margins:

(3') (i - ib) = 100 [ttj (7t - Jib) + a2 (q - qb) +a3 { (e - eb) + B (e - eb)0}]

Recent exchange rate behaviour within the ERM shows thatobserved exchange rate flexibility within fluctuation bandshas been reduced to a large extent since 1987. The de factoband width is close to ± 1 %, or even below, for BFR, HFLand FF vis-a-vis the DM, instead of de jure ± 2,25 % fornarrow-band currencies. Accordingly, the choice has beenmade to calibrate the B and 0 parameters in order to exhibitstrong non-linearity for exchange-rate deviations from thecentral rate above I % (see Appendix E.IV, which gives B =2.1018and© = 11).

In order to choose relative weights of the domestic andexchange rate targets, the working of the system in a simula-tion context has to be analysed. Since perfect capital mobilityis assumed, exchange rate dynamics can be represented bya set of three equations:

e = -

Equation (4) follows from (3), with i° representing the dom-estic terms (inflation and output) and eb assumed to be equalto the ERM central rate eP. Equation (5) is the open interestrate parity condition, with ea being the (log of) expectedDM exchange rate and i* the foreign interest rate. Equation(6) gives the expected exchange rate, which is supposed tobe the central rate for the next period. This is obviously asimplification, but it allows to restrict the analysis to a single-period set of equations. However, (6) would be replacedby a similar terminal condition in a dynamic multi-periodsystem.

Without expected realignment, i.e. eP+1 = eP, (6) reducesto:(6')ea

+1 - eP

Equations (4) to (6) lead to the following equations for theinterest rate and the exchange rate:

(7) i = (1 - u) i* + u i° = i* + u (i° - i*)(8) e = eP + u (i* - i°)where u = [1/(1 + a3)], with 0 =£ u =£ 1

As is clear from (7), n measures the weight given to thedomestic targets relative to that given to the exchange ratetarget.

The value of u is of some importance for the properties ofthe system since this parameter determines the degree offreedom left to monetary policy for domestic targeting (aslong as small interest rate variations are concerned, i.e.variations which only imply exchange rate movements withinbands).

Since membership of the ERM implies that priority is givento exchange rate stability, it seems reasonable that the valueof u be below 0,5. For the simulations, it was assumed thatH = 0,33 (i.e. a3 = 2), which means that ceteris paribus, theinterest rate response to domestic disturbances for ERMcountries is one third of that of non-ERM countries.

Since ERM membership already dampens interest rate varia-bility, no adjustment process has been taken into accountlike in (T). Hence, the reaction function for ERM membersis:

(3") (i - ib) = 100[2(7C - Jib) + 0.4(q - qb) + 2{(e - eb) +2.1018(e-eb)11}]

2.3.2. Realignment rule

Realignments are a distinctive feature of the EMS. Thus, arealignment rule has to be specified. Obviously, no suchrule exists in practice, but for simulation purposes a primecandidate is a rule like: devalue the central rate by z%when the price differential reaches x %. This may include aconstant real exchange rate rule as a special case (z - x),but encompasses a wide range of practices.

Derivation

Such a rule has to be chosen subject to the following require-ments :

(a) the bands should be overlapping, since otherwise discreteexchange rate jumps offer a 'one way bet' for speculators;hence, z < 2L (where L is the band width, i.e. 2,25 %);

(b) incomplete offsetting of price differentials, since completeoffsetting (i.e z = x) both lowers discipline and raises thepossibility of self-fulfilling speculative attacks;8 hence,z < x;

See Obstfeld (1989).

309

Annexes

(c) consistency with the monetary policy rule; since agentsknow the rule, realignments are fully anticipated.9 Hence,one has to examine the consequences of future realign-ments for current exchange rates and interest rates. Thiscan be done on the basis of equations (4) to (6) above.Suppose a z % devaluation of the central rate is expectedfor the year ahead.10 Hence, ea = e? + z. Equation (5)leads to:

(5') eP - e = i - (i* + z)

Hence, a z % expected devaluation is equivalent to a rise inthe foreign interest rate of z percentage points. Thus,

(T)i = (l-u)(i* + z) + ui°

(8') e = eP + u (i* + z - i°)

A z % expected devaluation of the central rate for the yearahead triggers a uz % immediate depreciation and a(1 - u)z % rise in the domestic interest rate. Condition (c)is therefore tantamount to n =S L/z, since the immediatedepreciation should be inside the band width L.

Empirical implementation

Since its creation, the EMS has experienced 12 realignmentsinvolving 39 bilateral DM central rate changes. Table E.2gives some simple statistics based on this experience, whichshow that:

(i) the average size of the realignments, which was greaterthan the band width prior to 1985, has been reduced forall countries since 1985; during the 1985-87 period, theaverage size of the realignments has been 3,5 % for thenarrow-band ERM countries;

(ii) the degree of offsetting of price differentials, which wasclose to 100 % in the first years of the system, has alsobeen reduced for all members with the exception ofIreland; during the 1985-87 period, it has been close to50%.

The Delors report'' states that in Stage I realignments willstill be possible, but that efforts will have to be made tomake the functioning of other adjustment mechanisms moreeffective. Already, no realignment has taken place fromJanuary 1987 until December 1989. The realignment rule forthe simulations must therefore incorporate a strong disciplin-ing effect for high inflation countries. As the latter wasalready the case in the 1985-87 period, a conservative choiceis to derive the rule from this experience.

Thus, the simulations are run with z = 4 % and z/x = 50 %,i.e. x = 8: a 4 % devaluation occurs each time the pricedifferential with Germany reaches 8 %. With u = 0,33, theimmediate depreciation in the year before devaluation isuz = 1,33 %, which is consistent with the band width of2,25 %. The rise in interest rates is (1 - u)z = 2,66 percent-age points.

9 This is the case in the Multimod model.10 Strictly speaking, expectations also take into account the possibility of

fluctuation within the bands.See Committee for the Study of Economic and Monetary Union inEurope (1989).

Table E.2

Realignments in the EMS, 1979-87

B/L DK Narrow band ERM

1979-83

Number of realignmentsAverage sizeOffsetting of price differentials1 (%)

1985-87

54,9

296,6

74,4

139,8

47,1

105,4

45,2

43,5

56,3

64,4

22

203,2

225,1

101,2

All data refer to bilateral DM exchange rales.' Central rate variation over the period divided by the accumulated differential of consumer price indices relative lo Germany since the last realignment, monthly data.Source: Calculated using data from the Commission of the European Communities.

275,3

89,7

Number of realignmentsAverage sizeOffsetting of price differentials' (%)

21,5

28,5

22,5

36,5

24,5

54,5

34,7

77,2

3 —4,7 —

50,7 —

93,5

52,7

123,8

52,1

310


2.4. The EMU regimes

Asymmetric EMU

In the asymmetric EMU regime, Germany and the non-EM U countries follow the same monetary rule ( 1 ") as before.The non-German EMU countries follow German monetarypolicy completely:

(!'") (i - i») = (iD - ibD)>

with the subscript D indicating German variables.

Symmetric EMU

In the symmetric EMU regime, all EMU members followthe monetary rule (1"), but as it is assumed EuroFed wouldtarget Community variables, the output and inflation objec-tives are expressed as EMU averages:

(1") (i - ib) = 0,5 [100 ( 2(nEMU - rc*>EMU)

with the EMU averages defined as (given that only D, F, Iand UK are represented in the model):

with yj real GDP of country i.

This is obviously a simplification, since the size and thereforethe economic weighting of the four countries in Communityaverages are not exactly identical.

3. Results from deterministic simulations

This section illustrates the effects of the EMU regime on theeffectiveness of budgetary policies and on the behaviour ofeconomic agents by means of some deterministic simula-tions. Two types of simulations are discussed: a governmentexpenditure increase and a supply shock (autonomous priceshock). In all cases the shocks were given to the Frenchmodel since this allows to highlight the constraints imposedon monetary policy by the EMS or asymmetric EMU regime.

Generally speaking, the pattern of GDP reactions(Table E.3) illustrates both the properties of the model andthe monetary and exchange rate policy assumptions. As aconsequence of the Keynesian features of the model, anincrease in government expenditure is expansionary in theshort term, but as a consequence of the forward-lookingcharacter of the model the first-year multiplier is lower than

Table E.3Effects of 1 % government expenditure increase in France

(% deviation from baseline)

Variable

GDP

Short-term interest rate1

Real effective exchange rate

Inflation1

1 Difference with respect 10 baseline in percentage poiAEMU = Asymmetric EMU.Source: Mullimod simulations.

Regime

FloatEMSAEMUEMU

FloatEMSAEMUEMU

FloatEMSAEMUEMU

FloatEMSAEMUEMU

nts.

Year 1

0,590,800,870,80

0,14-0,05

0,050,07

0,870,430,220,32

0,020,250,320,30

Year 2

0,310,710,860,75

0,470,210,080,16

1,070,790,530,66

0,340,520,570,56

Year 3

0,080,510,700,57

0,540,270,070,17

1,081,040,820,98

0,300,480,480,50

Year 4

-0,070,290,480,33

0,480,260,050,18

1,011,181,041,21

0,230,340,300,36

YcarS

-0,140,100,250,09

0,370,200,040,15

0,911,241,161,35

0,170,170,090,17

311

Annexes

for most empirical models of the French economy. It is alsodecreasing rapidly with time.

The simulations also illustrate the familiar Mundell-Flemingproposition that with capital mobility fiscal policy is moreeffective in a system of fixed exchange rates than in a systemof flexible exchange rates. While the Mundell-Fleming resultis derived in a system with fixed prices through differencesin the crowding out of investment and foreign trade, thepresent simulations obtain the same effect through the im-pact of the government expenditure increase on output andinflation and the ensuing reaction of interest rates in accord-ance with the monetary policy reaction function. In the floatregime, interest rates increase to choke off the higher levelsof output and inflation, thus provoking a nominal and realappreciation which further reduces output. In the EMS andEMU regimes, the interest rate reacts less vehemently dueto the exchange rate constraints in the EMS and the common(asymmetric or symmetric) monetary policy in the EMUregimes. The interest rate increase in the asymmetric EMUregime is equal to that in Germany and therefore representsthe spill-over effect of the French expansion on Germaninterest rates in the form of higher French export prices toGermany which increase inflation and of higher Germanexports to France which increase German economic activity,leading to an interest rate increase to dampen this price andoutput surge.

A second important aspect of the regimes is their impact onthe behaviour of economic agents. More specifically, insetting wages and prices in the EMU regimes, householdsand firms have to take into account that unwarranted wageand price increases can no longer be compensated at intra-Community level in the form of currency depreciation rela-tive to Community trade partners. A regime change basicallyimpacts on wage and price discipline through an effect onprice expectations. The extent to which price expectationsaffect actual wages and prices is however not known andcan be modified by institutional changes in, for example,wage bargaining. Both types of effect can be illustrated bya supply shock, for instance by giving a one-period increasein the autonomous component of the price equation (thuskeeping the price equation endogenous) and comparing theensuing behaviour of inflation upon this price shock.

By affecting price expectations through the knowledge thata price shock will not be compensated through an intra-Community devaluation, the EMU regime should causeprices to return faster to equilibrium than the float or EMSregimes. This proposition is illustrated, using a small theor-etical model in Box E.2, which shows that the first-yearinflationary impact of a price shock under EMU is notnecessarily lower than in a floating regime, but that the

return to equilibrium is faster afterwards. In fact, inflation(relative to baseline) eventually has to become negative underthe EMU regimes due to the fact that the price level has toreturn to equilibrium, while this not the case in a float wherethe price level rise may be permanent.

This theoretical result is confirmed by the simulations pre-sented in Table E.4, which shows the results of a single-period 5 % shock to French inflation under the four regimesanalysed with Multimod. Under the float regime, the in-flationary shock induces a strong increase in interest rates,causing a strong first year appreciation of the French francrelative to the Deutschmark (and other currencies). Afterthis first year, however, the exchange rate starts to depreciatein order to compensate for the general increase in the pricelevel. Since agents are forward-looking they incorporate thisin their expectations so that inflation returns only graduallyto equilibrium (assumed to be the baseline). In the EMSregime, this picture is followed for the first two years withinthe margins allowed by the exchange rate bands of the EMS.In the second year the price level differential with Germanysurpasses 8% so that, according to the EMS rules in themodel, a 4 % devaluation is triggered off in year 3, whichprovides again some relief for the price level. Since thedevaluation compensates only for 50 % of the price leveldifference, there is a nominal anchor implying that inflationeventually has to become negative to compensate for earlierincreases (see also Graph E.2 and Box E.2). The EMS istherefore seen to have an impact on wage and price disciplinewhich is however weakened by the possibility of devaluation.The devaluation possibility and intra-EC parity fluctuationsbeing absent in the two EMU regimes, they show the fastestreturn to equilibrium for inflation, the asymmetric EMUregime being faster because it completely lacks an interestrate reaction.

While these simulations show the impact of the EMU regimeon wage and price discipline through changes in expec-tations, the degree of forward-looking behaviour in EMUis an open issue. While it is impossible to estimate the effectsof EMU on behaviour, some illustrative simulations mayshow the potential impact of behavioural changes. To thatend, Table E.5 compares the results of a 2 % price shock inFrance under the EMU regime for two different speeds ofindexation of prices to price expectations. Given the sameforward-looking price expectations, this simulation thereforeoffers the possibility to analyse changes in behaviour. Thesimulation illustrates that from the second year onwards,inflation returns faster to equilibrium with normal than withslow indexation. Output, on the other hand, deteriorates lessin the short run with slow indexation than with normalindexation, but in the medium run is slower to return toequilibrium, thus once more illustrating the trade-offs in-volved with wage and price behaviour.

312


Table E.4Effects of 5 % shock to inflation in France

1 Difference with respect to baseline in percentage points,AEMU = Asymmetric EMU.Source: Multimod simulations.


GDP

Inflation1

FF/DM exchange rate

Short-term interest rate1

Real effective exchange rate1

Regime

FloatEMSAEMUEMU

FloatEMSAEMUEMU

FloatEMSAEMUEMU

FloatEMSAEMUEMU

FloatEMSAEMUEMU

Year 1

-1,77-1,26-1,06-1,36

5,365,535,475,49

-4,06-1,56

00

5,001,490,081,30

5,243,852,933,15

Year 2

-3,48-3,01-2,71-2,90

3,863,272,542,75

0,30-0,30

00

5,684,760,091,45

5,255,224,544,55

Year 3

-4,08-3,95-4,07-4,02

2,261,36

-0,060,45

5,584,0400

4,311,270,061,12

3,943,944,864,68

Year 4

-3,81-4,09-4,75-4,48

1,20-0,44-1,82-1,10

9,825,2600

2,620,230,030,69

2,523,244,203,94

YearS

-3,09-3,83-4,68-4,31

0,56-1,37-2,65-1,85

12,515,4600

1,280,05

-0,000,34

1,472,492,942,73

Table E.5Effects of a 2 % price shock in France under EMU with normal and slow price indexation

1 Dividing the indexation coefficient by 2, i.e. coefficient k in Appedi* E 11.2 Difference with respect to baseline in percentage points.Source: Multimod simulations.


GDP

Inflation2


Speed

NormalSlow1

NormalSlow1

NormalSlow1

Year 1

-0,52-0,42

2,151,95

1,271,18

Year 2

-1,14-1,02

1,091,26

1,841,86

Year 3

-1,60-1,54

0,180,51

1,902,10

Year 4

-1,78-1,86

-0,46-0,11

1,612,00

Year 5

-1,70-1,93

-0,80-0,51

1,131,67

313

Annexes

GRAPH E.2: Effect on price level of a 5 % price shock in France% deviation from baseline

Float EMS EMU

Source: Mullimod simulations.

314

sduquenoy


Box E.2: EMU, expectations and wage and price discipline

An analytical example with a two-country, two-period modelmay illustrate the effect of fixed exchange rates on expectationsand therefore on wage and price discipline, compared to asituation of free float. The model used resembles that of Begg(1990), albeit without taking account of fiscal policy, and followsin broad lines that of the Multimod model in the way it wasused to compare different exchange rates regimes in this Annex.

The model is entirely written in deviations from a baseline, suchthat all exogenous variables disappear. Lower case letters denotelogarithms of variables except for the nominal interest rate, r,which is in levels.

There are two prices in the model, the output price p and theabsorption price [up + (l-u)e] which is a weighted average(with n the share of GDP in [GDP 4- imports]) of the outputprice and foreign prices converted into domestic currencythrough the nominal exchange rate e (since foreign prices areexogenous, they are not shown). The output price equation is areduced form of a wage-price block with overlapping contracts,cf. Masson et al. (1988), and gives the growth rate (p - p_ , )of output prices as a partial adjustment to expected future(absorption price) inflation [u(p+] - p) + (l-M)(e+ | - e)],together with an effect of the deviation of output from baseline,y (proxy for Phillips curve effect). Hence, there is perfect priceindexation with a lag due to overlapping contracts:

- p_!> = (1 - cp)(p_, - p_2) (p[u(p+, - p)ty + u

The last term in the output price equation, u, represents atemporary price shock.

Output y is determined by a number of exogenous variableswhich are not shown since they do not change relative to base-line, by a negative effect from the real interest rate (with forward-looking inflation), and by a positive effect from the real exchangerate (e - p):

(2) y = -B(r - [u(p+, - p) + (l-n)(e+ ] - e)]) +n(e - p)

The nominal exchange rate in terms of local currency per foreigncurrency is determined by uncovered interest parity:

(3)e = + r *

The monetary policy rule, which may be derived from monetarytargeting with an inverted money demand equation, specifiesthat nominal interest rates increase if inflation or output areabove their baseline levels:

(4)r = a,[u(p a2y

In equations (I) to (4), all parameters are assumed to be regime-independent, except a, and a2 in the monetary policy rule (4).

In EMU, they will typically be smaller than in a float. Thesecond element which changes between float and EMU is theforeign interest rate r*. When the country forms an EMU withthe foreign country, they will have a common monetary policy.Therefore, in EMU, the foreign and domestic interest rates areequal, whereas the foreign interest rate is assumed to be exogen-ous in float:

(5)r' = CT)r o = 1 (Float)o = 0 (EMU)

For the analysis of the effect of EMU on wage and pricediscipline, assume for the moment that the future output pricep+ ] and exchange rate e+ ] are given. After some tedious butstraightforward calculations, the system of equations (1) to (4)may then be solved for the price level p in terms of past andfuture price levels, past and future exchange rates and the shocku:

(6)p = ?i2p_2 + Tt3p+l + *4t| - Jt2 — Tt3 - Jt4)e+ 1 0u

Independent of the expressions for the parameters, the importantfeature of this equation is that the long-run price level is homo-geneous of degree one in the exchange rate, i.e. in the long runthe real exchange rate is constant (this is due to the fact thatthe shock, u, is temporary). In a two-period framework, it maytherefore be assumed for simplicity that p + ) = e+1, such thatthe equation may be rewritten (deleting lagged variables whichdo not differ from baseline) as:{7)p = - jc4)e4 + l

In a float, the short-run effect of a price shock u is determinedby the size of the parameters Tij and 0 (which all depend on a(,a2 and a), and by the expected depreciation e+l > 0. In EMU,the expected depreciation, and therefore expected price level, iszero: e+ , = p + , - 0. The difference in the effect of a temporaryprice shock between EMU and float is therefore:

(8) [dp/du]EMU - [dp/du]FLOAT = @EMU - ©FLOAT- (1 - it, - rc2 - njlde+i/du]

This expression implies that sufficient conditions for wagesand prices to decrease faster in EMU than in a float after aninflationary shock (du > 0) are:

(9a)0EMU ~ ®FLOATTti — Jt-» — n

< 0

0

The proof for (9b) is straightforward, and simply requires thesubstitution of the expressions for the Ttj's. Condition (9a) is notnecessarily valid, however, the size of the two 0's being depen-dent on the a's. In EMU, the first-period effect of a price shockon inflation may therefore be stronger than in a float. Since the0's represent a once-and-for-all shock to prices, their effect willdie out over time in a multi-period framework, whereas theeffect of the expected depreciation continues to have an upwardeffect on the price level in a float relative to EMU.

315

Annexes

In a float, the price level may remain high because the de-preciation compensates the effect on the real exchange rate,which is zero in the long run. In EMU, the price level will haveto return to the baseline level because this is the only way tohave the real exchange rate unchanged in the long run, giventhe fixed nominal exchange rate. In terms of inflation, thisimplies that in a float inflation will be positive after a priceshock and then return monotonically to zero. In EMU, inflationwill also be positive, then turn negative (due to the return of theprice level to the baseline) and eventually return to zero. Theaverage inflation level after a price shock will therefore be higherin a float than in EMU. Secondly, the return of inflation to thebaseline level (and below, in EMU) after the initial shock willbe faster in EMU due to the effect of expectations on futureexchange rates.

Exchange rate expectations therefore may be said to enhancethe wage and price discipline in EMU.

The previous arguments are illustrated by a simulation with theabove theoretical model in Graph E.3, giving a price shock du =0,05, setting long-run values e+T = p+T = 0,115 in the case offloat (steady state from an initial simulation with T large andlong-run values equal to zero), using the parameter values q> =0,5, u = 0,8, T = 0,5, B = 0,25, n = 0,2, a, = 2 (0,5 in EMU)anda2 = 0,4 (0,1 in EMU).

GRAPH E.3: Effects of a 5 % price shock in floatand EMU, theoretical model

0,05

-0.020 2 4 6 10 U 14 16 18 20 22 24 26

Time

-Float -EMU

Solution of the model

Equation (5) implies that r — r* = or. Using this result,substitute equation (3) in equations (1), (2) and (4):

!') (P - P-i) ~ P>+ (l-u)ar + iy + u

(2') y = -B(r - [u(P+1 - p) + (l-u)ar]) + Ti(e + ]- ar -P)

(4') r = a,[u(p - p.,) + ( l -Me+i - ar - e_,)] + a2y

Next, solve equation (4') for the interest rate r:

(4")r = [a,u(p - p.,) + a,(l-u)(e+ | - e_,) + a2y]/£2

with ft = 1 + d|{l - u)a.

Substituting (4") in (!') and (21), solving for p and rearrangingterms gives equation (6):

(6) p = i u , p _ , n3p+| i 4 _ ,

where the n's and & are defined as:

n, = [2.- <p + a,u{ir - (p(l

K2 = -(1 - cp)/0

K3 = |i[<p + SB]/<D

0 = i/d>,

with as auxiliary symbols:

T = S + a[n - (1 - u)0]

£ = [tfl + acp(l-u)]/[il + a2rj

Using these definitions, the proof of (9b) follows straightfor-wardly:

(9b) (1 - 7i, - 7i2 - 7t4) = 1 - [2 - ip

- u)[Sr

= [<pu +

> 0

316


4. Methodology of stochastic simulations

4.1. General methodology 12

To describe the methodology, it is assumed that the modelcan be written in linear form as

(9) y = A y ( - l ) + Bx + Cy6 + u

Current endogenous variables y depend on values for theprevious period, y(— 1), as well as on expected values forthe following period, y6 = E(y( + 1)/I), where 1 is the infor-mation available in the current period. The vector of errorsmay include some that are identically zero, in particular forequations that are identities. The stochastic simulations areperformed by repeatedly replacing the error vector u by adraw from its observed statistical distribution; therefore anestimate of the co-variance matrix of the u's is needed. Thisis complicated by the fact that ye is unobservable.

In Multimod, equations with expected variables were esti-mated using McCallum's (1976) instrumental variablesmethod. Therefore, errors from the first stage regressionshould capture the expectations errors. If the expectationserrors are called v = y(+l) - y6, the equation residualsmay be decomposed into two parts: the structural residualsu and the expectations errors v:

(10) y = A y(- 1) + B x + C y(+ 1) + u - C v

In practice, autoregressive time series were fitted for thevariables for which the model has expectations (except forthe exchange rate equations, see below); the forecasts fromthese equations — say yf i t (+I ) — were substituted intoequation (9) in place of y6. The structural residuals werecalculated residually as(11)u = y - [Ay(-l) + B x + Cyf i t (+ l ) ]

A second problem concerns the serial correlation of theresiduals u, which was not taken into account in the esti-mation of the model due to the estimation techniques (e.g.time series/cross-section estimation). Therefore, autoregres-sions of the following form were fitted to the u's calculatedby equation (11):

(12) u = ( p u _ | + a + b t + e

The e vector then constitutes the innovations or 'shocks' inthe model. In simulation, the model constituted byequations (9) and (12) was solved together, given a draw forthe vector e of shocks. Shocks therefore have persistenteffects both through the u's and through the dynamics ofthe model. Solution to the model in period t replaces y6 bythe value calculated for y(+ 1), given information availableat t (call it yhat(t,t+ 1)), which itself depends on x(+ 1) aswell as on y( + 2), (which in turn is replaced by its modelsolution, yhat(t,t + 2), etc.). Some terminal condition is im-posed on y( + T). The effects in future periods are thusassumed to be correctly anticipated; however, future shocksare not anticipated before they occur. This implies that theforward-looking simulations have to be redone for each timeperiod in which a new shock is applied, so that, for instance,yhat(t + 1 ,t + 2), yhat(t,t + 2) and y(t + 2) may each have dif-ferent values. This greatly increases the number of simula-tions needed.

In practice, residuals calculated for the period 1973-8413

were used to estimate the parameters of equation (12) andthe 98 x 98 co-variance matrix describing the E vector. Adistinction was made between exchange rate shocks andshocks to other behavioural equations. For the latter, a first-order autoregression plus a time trend was estimated foreach of the residuals u. The approach for the former isdescribed below. These residuals e were then correlated,giving a co-variance matrix V. Table E.6 presents the esti-mated standard errors of the e's, save for the exchange rate,which is treated below. The table also excludes shocks toshort-term interest rates, since they act as a predictablepolicy variable.

The stochastic simulations were run by making repeateddraws from a random number generator, given standardnormal variates w, where

E(w) = 0

E(ww') = I

A Cholesky decomposition14 was performed on the matrixV, yielding a lower triangular matrix D such that

DD1 = V

This section draws on Masson and Symansky (1990).

This period is determined by the fact that some of the equations containforward-looking variables up to the period t + 4. By assuming a chainrule of forecasting (as actually implied by the use of the fitted values ofthe AR(1) regression), this period could be extended to 1988. For someresiduals, however, this resulted in unusually large errors (e.g. the capitalstock equation in France).A special routine was developed to perform the Cholesky decompositionwithout needing the matrix V to be non-singular.

317

Annexes

Table E.6Estimated standard errors of non-exchange rate shocks,

Private consumptionOil consumptionCapital stockExports of manufactured goodsImports of manufactured goodsImports of commoditiesMoney demandLong-term interest rateGNP deflatorExport priceProduction capacity

Estimated standard error of E in the regression u = ip u _ , + a1 Private and public consumption, capital-importing developin1 1973-84.1 All developing countries.4 Exports of primary commodities.

D

0,63,60,62,43,06,64,11,11,01,51,2

+ b t + e.g countries.

1973̂ 88

F

0,82,60,82

1,81,75,1

10,11,31,21,61,2

i

1,33,21,33,52,58,66,31,81,01,82,1

UK

1,07,41,02,12,16,98,21,33,32,71,2

USA

0,72,30,72,92,16,53,21,41,52,40,9

Japan

0,93,00,93,22,9

10,31,91,21,63,31,5

Canada

1,03,31,04,58,37,75,10,92,53,21,3

Smallerind. coun.

0,62,90,61,93,85,28,21,52,73,01,5

l%)

Rest ofworld

1,5'——

6,93—4,7"—————

The generated shocks w were premultiplied by D, yielding 4.2. Treatment of exchange rate shocksshocks e* with the same properties as the e's, that is withestimated co-variance V:e* = D w

E(e* e*') = E(D w w'D') = D D' = V

The model was first calibrated to track a smooth baseline.For each year from 1990 to 1999, a draw was made for thee's; using inherited y(— 1), which depends on past E'S, themodel is solved forward to the terminal date, which in eachcase was taken to be 2020. No shocks were applied to theyears 2000-20, in order to leave a sufficient period at theend so that the simulations over the period of interest, 1990-99, would not be much affected by the terminal conditionson the expectations variables. This procedure was replicated43 times, giving a total of 430 draws for the e's.15

One of the issues discussed at the Brookings conference 'Empiricalevaluation of alternative policy regimes', 8-9 March 1990, WashingtonDC, was whether calculating root-mean squared deviations over thecomplete period 1990-99 would bias the results. Calculating them forthe period 1995-99 showed as only major difference for the Communityaverage that the improvement in inflation variability between the floatand EMS regime almost became zero.

The approach for the exchange rate shocks is crucial for theevaluation of EMU. It is also more complicated. Considerthe uncovered or open interest parity (UIP) condition usedin the Multimod model:

(13) e = Ee+1 - r + r* + u,

with e the (log of the) exchange rate expressed in domesticcurrency per dollar, r the domestic interest rate, r* thedollar interest rate, u a residual and E the model-consistentexpectations operator.

Let ef be the (1 year) forward exchange rate, and ea+1 be

the (unobservable) true expectation for e +1. Consider thefollowing definitions:16

fd = ef — e forward discount

cp = r — r* — fd country premium

16 These definitions are mainly adapted from Frankel (1989).

318


rp = ef — ea + j

ee = ea+1 - Ee

ft. = f . — *>a+ 1

+ 1

risk premium

expectational error

forecast error

Hence, the residual u which represents the shocks to theexchange rate equation can be expressed as:(14) u = e - ea

+| + ea+| - Ee+1 + r - r*

= (r - r* - fd) + (e^ - e*

= cp + rp + ee

+ (e*+l - Ec+1)

The first term in the above equation is the country premium,the second the risk premium and the third the expectationalerror. This therefore highlights the three factors contributingto the exchange rate shocks:

(i) incomplete capital market integration leading to non-zero country premium (i.e. deviation from coveredinterest rate parity);

(ii) non-zero exchange rate risk premium arising from aver-sion towards risk and more generally imperfect substitu-tability between comparable assets in different cur-rencies;

(iii) expectational errors arising from differences betweenthe agents' exchange-rate expectations and those rep-resented by the model-consistent expectations Ee+1; amajor cause for these differences may be changes inhypotheses regarding the long-term equilibrium realexchange rate.

Note that the forecast error e+ j - ea+ j does not affect the

residual u. Any estimation of u should therefore exclude theforecasting error.

Since Ee+ ( is not observed, estimation of u requires substi-tuting an estimator e^+ j for Ee + j in equation (13), yielding:(13') u = e - ea

+, + r - r*

Possible solutions are:(a) ea

+) = e+1 (observed shocks solution),(b)ea

+| = e ('random walk' hypothesis),(c) to use the model to forecast ea

+ 1

(d) to derive ea+ 1 from a partial model.

These four possibilities will each be examined in turn.

Possibility (a): observed shocks

A first possibility is to calculate the residual using the ob-served exchange rate values in period t+ 1 (i.e. substituting

e+1 for ea+, in (13'). The residual ua corresponding to this

assumption is simply:

ua = cp + rp + fe = e - e + 1 + r - r*.

The estimated residual therefore includes the forecast errorinstead of the expectational error. This is not a desirableapproach, since the forecast error should theoretically beexcluded from the shock because it also depends on newsarising in period t+ 1.17

Possibility (b): "random walk* hypothesis

A second possibility is that taken by Masson and Symansky(1990), and would correspond to the same treatment as forother equations with expectational variables. In this case, itwould consist of assuming a random walk to be the bestpredictor for the exchange rate in period t+ I (i.e. ea

+1 =e), which makes the estimated residual ub of the equationequal to:

ub = cp + fd = r - r*.

In other terms, expected depreciation is always zero, ex-change rates follow a random walk, and changes in theforward discount reflect changes in the risk premium. Othersources of error such as the expectational error are removedfrom equation (14). This is consistent with the empiricalevidence put forward by Meese and Rogoff (1983). Never-theless, it would seem difficult to combine it at the sametime with the uncovered interest parity condition in the wayMasson and Symansky (1990) do, since the exchange rateremains undetermined in (13) as it drops out of the equation.The correct random walk residuals would be e+ i — e, thevariance of which would probably be very close to thatof (a).

Possibility (c): model-consistent forecast

Methods (a) and (b) represent two extreme possibilities:either, as in (a), to assume that all the observed variance ofexchange rates is due to deviations from the UIP condition,or, as in (b), to attribute almost all of this variance to 'news'arising in period t + 1.

A third route would be to use the model itself in forecastingmode to calculate the expectation, at period t-1, of theexchange rate in period t+1. Inserting this expected ex-

This is the method originally used for the Multimod stochastic simula-tions by Frenkel el al. (1989).

319

Annexes

change rate in the uncovered interest parity condition givesa residual:u,, = e - e+ 1 + r

Given a simulation over T + 1 periods and a terminal valuee+T + 1 for the exchange rate, backward substitution usingthe simulated interest rate values r+ i and r* + j for periodst + 1 to t + T expresses the estimated residual as dependingon the terminal value and the cumulative simulated interestrate differential:18

T(15)uc = e-e+T+1 + r - r*+ S [r + j - r* + i].

Though conceptually attractive, this approach also has itsproblems.

Exchange rate expectations as computed by the model de-pend obviously on expectations regarding economic policy.In a retrospective simulation, one has to make sure thatsimulated policy expectations are not confused with observedshifts in policy. Ideally, representation of economic policyshould be based on equations representing standard behav-iour of the authorities. Monetary policy in Multimod obeysan interest rate reaction function. As a consequence, agentsexpect interest rates to behave in accordance with this policyrule. However, since the interest rate reaction function comesfrom an inverted money demand function, it incorporatesshifts in the money supply, implying that policy changeswere to a large extent anticipated. For fiscal policy, thesame kind of problem arises since the fiscal policy stance asmeasured by budget deficits is partially anticipated.

Possibility (d): long-run equilibrium consistent expectationswith a partial model

Given the problems in using a complete model simulation,an alternative method based on a partial model can bedesigned that would be consistent with long-run equilibriumlevels of interest rates and exchange rates. The idea was toretain the basic logic of equation (15), but to make theassumptions regarding economic policy much more trans-parent. Therefore, a reduced model has been built whichconsists of the following equations for each of the G7countries and the smaller industrial countries zone:

(17) R = nR_, + (l-u)R + V(18) V = (pV_, + at + B + e

If it is assumed that the long-term interest rate is related to the short-term interest rate through a perfect term-structure relationship, thisresult is equivalent to that used by Hooper and Mann (1989) or Ko-romzai et al. (1987), who explain the exchange rate as a function of anequilibrium rate and a cumulative long-term interest rate differential.

with: T = log of real exchange rate with respect to dollarR = real interest rateR* = US real interest rateR = long-run equilibrium value real interest rate~ (2,8 %)t = time

This model consists of (16), i.e. equation (13) transformedin terms of real exchange rates and real interest rates, andan interest rate reaction function (17), also in real terms,which ensures that all real interest rates converge to thesame long-run equilibrium level R.

The only purpose of equation (17) is to capture a very crudeproxy of monetary policy shifts. In order to take sustainedshifts in the policy stance into account, equation (18) describ-ing the autocorrelation process and trend movements of theresiduals of the interest rate equation is added. Interest rateequation (17) was estimated using pooled time-series/cross-section data for the eight countries over the period1973-88, and î has a common value equal to 0,68. Theautocorrelation coefficient <p in equation (18) was next esti-mated for the eight countries separately, having relativelysmall values (smaller than 0,33 in absolute value).

Equations (16) to (18) were then simulated (with a forwardsimulator) until the year 2015 ten times, starting from 1979onwards until 1988. That is, for each year from 1979 to 1988a separate forward-looking simulation has been run withu = 0, taking into account the available information as rep-resented by the initial values of the variables and of the errorterm e, to yield a simulated real exchange rate t.

As in equation (15), backward substitution in (16) leads to:T

(16') T = T+T+] - r + r * - I [R + i- R*+i]

where T+ T-H is the real long-term equilibrium exchange rateof the baseline simulation and R and R* are simulated realinterest rates (using equations (17) and (18) for periods t+ 1... t + T).

Hence, the estimated residual is:ud = T - T

Equation (16') clearly exhibits two major factors affectingexchange-rate expectations: expectations regarding the long-term real exchange rate, and expectations regarding futurepolicy.

320


Comparison of the results

Table E.7 below gives the measure of exchange-rate shocksderived from methods (a), (b) and (d) (results for method (c)were not judged satisfactory for the reasons spelled outabove). Since these methods do not share the same assump-tions regarding the underlying exchange rate model, the rawresidual results from methods (b) and (d) have been correctedfor autocorrelation, trends and for different degrees of inte-gration.

As expected, these methods give clearly different pictures ofthe stochastic component of exchange-rate unstability.Given the evident limitations of method (a), possible candi-dates to generate the shocks are methods (b) and (d).

The major problem with method (b) is its reliance on staticexpectations and the inconsistency of the random walk hy-pothesis with the UIP condition of the model. '9

The major problem with method (d) arises from a very cruderepresentation of expectations of future policies. Apart fromthis admittedly important limitation, however, method (d)

Table E.7Comparison of exchange rate shocks

Rate

USD/DMUSD/FFUSD/LITUSD/UKLUSD/CADUSD/YENUSD/Smaller ind. court.FF/DMLIT/DMUKL/DM

Observeddeviation

from UIP (a)

17,117,417,714,05,1

17,013,65,13,7

10,2

'Randomwalk' +UIP(b)

0,71,11,62,21,21,91,31,31,61,8

Modelexpectations

withpartial

model (d)

11,210,910,69,76,4

16,86,2

4,22,96,5

(a) Observed; the shock is defined as the standard deviation of the residual ua defined aboveover the period 1979-88.

(b) 'Random walk': the shock is defined as the standard error of the equation UL, = <p*ub( I — I ) + a*t + v, with ub the 'random walk' residual r-r*. t is lime and v the shock.Estimated over 1979-88.

(d) Model expectations: the shock is defined as the standard error of Ihe equation uc = <p*uc(t— l)+s'u (1-2)4- a't-t-v, with ac the residual from the partial model simulations asdefined in (16). Estimaded over 1981-88 with either s - 0 or a = 0.

Although early research by Meese and Rogoff (1983) brought somesupport to the random walk hypothesis, further work by the sameauthors (1988) brought evidence, contrary to the random walk hypoth-esis, of an effect of interest rate differentials on exchange rate move-ments, albeit non-significant. Other empirical evidence against the staticexpectations hypothesis is brought by Frankel and Froot (1987,1989)on the basis of survey data. Their findings lead to reject clearly the staticexpectations hypothesis, and to highlight the existence of systematicexpectations] errors.

is more satisfying since it takes expectational errors intoaccount in a way which is consistent with the model itself.Method (d) has therefore been chosen for the simulations.

4.3. Stochastic properties of exchange ratesunder alternative policy regimes

4.3.1. Principles

As outlined above, experiments with alternative policy re-gimes raise a number of difficult issues.

Strictly speaking, neither coefficients nor statistical charac-teristics of shocks are invariant with respect to changes inthe exchange rate regime. Since these issues warrant moredetailed study, a relatively conservative position is justifiedfor the Multimod stochastic simulations: apart from theireffect on expectations as represented in a model-consistentway, policy and exchange rate regimes are supposed to beneutral with respect to the internal structure and stochasticproperties of the model (leaving aside the exchange rateequation).

This leaves open the choice of parameters for the stochasticproperties of exchange rate equations: variance of shocksand co-variance between exchange rate shocks affecting dif-ferent currencies. In this respect, experience is limited. Sincethe breakdown of the Bretton Woods regime, the DM andthe pound sterling have been floating with respect to the USdollar, together with the yen and the Canadian dollar; theFrench franc and the Lira have been tightly linked to theDM for most of the period, (even if there was dejure floatingin 1973-79). Hence, it is not clear from which experience theparameters for generalized floating, EMS or EMU shouldbe drawn.

Since experience is limited and samples are small, the choicehas been made to consider, when justified on a priori econ-omic grounds, that national experiences with exchange rateregimes were drawn from the same statistical distribution.An obvious example regards the EMS: both the French andItalian experiences (and also in some respect the Germanone) are characteristic of EMS performance. It seems there-fore valid to derive, for example, the possible stochasticbehaviour of the pound after participation in the ERM fromthe average of German, French and Italian experiences.

Currencies differ, however, by their international role: otherthings being equal, the DM seems to be a closer substitute tothe US dollar than other European currencies. Accordingly,shocks due to portfolio shifts away from the dollar assets

321

sduquenoy

Annexes

would probably in a pure floating rate regime impact moreon the DM than on other European currencies. Althoughhard evidence of clear-cut distinctions among Europeancurrencies does not exist, evidence from the floating-rateperiod tends to confirm this stronger link (Table E.8): thedollar exchange rate variability was higher in 1973-79 forthe DM than for other major European currencies, in spiteof de facto pegging to the DM for the latter.

Table E.8Observed variability of bilateral dollar exchange rates___________________________________<

DM FF LIT UKL

Standard deviation ofthree-month logarithmicchanges, July 1973-Febru-ary 1979 5,8 4,8 5,1 4,8


Table E.9

Summary of hypotheses for exchange-rate regimes

Intra-ERM regime Regime vis-a-vis ROW

Observed regime(1979-88)

DM, FF, LITUKL

Simulated regimes

Free float

DMFF, LIT, UKL

EMS

DM, FF, LIT, UKL

EMU

DM, FF, LIT, UKL

Peg (P) Joint floating (JF)Floating (Fl) Floating (F2)

FlFl

Fixed

JFF2

JF

JF

Table E.9 gives an overall qualitative presentation of themethodological choices derived from the above principles.Technical hypotheses are detailed below.

The observed baseline regime (1979-88) can be summarizedby four major features regarding intra-ERM exchange ratebehaviour and behaviour of the dollar exchange rate forboth the ERM members and the pound sterling (Table E.9):

Peg (P) for DM, FF and LIT cross-rates; Joint floating (JF)for DM, FF and LIT dollar rates; Floating with respect toboth the other European currencies (Fl) and the US dollar(F2) for the pound sterling.

The hypotheses for the simulated regimes are derived fromexperience with the observed regime in the following way:

(i) In a free float regime, the degree of exchange rate varia-bility within the EC (as measured by shocks) is supposedto be close to that of the pound sterling with respect tocurrencies of the ERM. With respect to the dollar, theDM would retain the present aggregate variability ofERM currencies and the other currencies that of thepound.

(ii) In. the EMS and in EMU, exchange rate variabilitywould be close to that experienced in the ERM, exceptfor fixity within the EC in EMU.

4.3.2. Empirical implementation

The triangular variance-co-variance matrix used to generatethe shocks summarizes all information regarding the sto-chastic features of the simulations. It can schematically bedecomposed into four parts:

Domestic shocks

Domestic shocks

Exchange rate shocks

" AB

Exchange rate shocks

-

DC D _

Part A

Part A represents the (own and cross-country) variances andco-variances between the shocks other than for the exchangerate. As a consequence of the neutrality hypothesis outlinedabove, this part is supposed be independent of the exchangerate regime.

Part B

Part B represents the co-variances between the exchange rateshocks and the 'domestic' shocks in different countries. Theobserved co-variances represent a mixture of the presentEMS regime for Germany, France and Italy and a free float

322

sduquenoy


for the dollar exchange rates of the other G7 countries. Theimposition of co-variances for the different regimes to besimulated which would simultaneously satisfy the restric-tions imposed by the regime (e.g. equal co-variances for thecurrencies participating in the EMU regimes) and economicreality is difficult to realize.20

Therefore, in order to avoid any bias through the introduc-tion of largely arbitrary assumptions, the co-variances be-tween the shocks to any dollar exchange rate and any 'dom-

Table E.10

Observed (1981-88) and simulated correlation coefficients forexchange rate shocks

Observed

USD/DMUSD/FFUSD/LITUSD/UKLUSD/CADUSD/YENUSD/Smaller ind. coun.

Simulated

Free float

USD/DMUSD/FFUSD/LITUSD/UKL

EMSUSD/DMUSD/FFUSD/LITUSD/UKL

Asymmetric!Symmetric EMU

USD/DMUSD/FFUSD/LITUSD/UKL

USD/DM

10,8250,9330,7610,0430,5290,761

10,7150,7150,715

10,8850,8850,885

11I1

USD/FF

10,8980,635

-0,1010,6780,820

10,7150,715

10,8850,885

111

USD/LIT

10,750

-0,1710,7450,856

10,715

10,885

11

USD/UKL

10,0260,6800,548

1

1

1

As an example, consider the estimated co-variance between shocks tothe USD/FF rate and French private consumption. Due to the presenceof the EMS, the shocks to the USD/FF rate are largely determined bythose to the USD/DM rate. If the causality of the shocks were entirelyfrom the USD/FF rate in the direction of French private consumption,this co-variance could be used for the free float regime. However, thiscausality is unlikely in the case of free float, so this approach would bedifficult to justify.

estic' variable have been put equal to zero. While this is notreally necessary for the co-variances involving shocks tonon-Community currencies, the zero co-variance hypothesishas also been introduced in those cases for want of consist-ency.

PartC

Part C represents the co-variances between the exchangerate shocks themselves. Appendix E.V describes more fullythe modifications to the observed correlations between theexchange rate shocks in order to obtain appropriate corre-lations for each of the regimes. Table E. 10 below shows thecorrelation coefficients between the different exchange rateshocks before and after modification. As expected, the orig-inal correlation between shocks affecting dollar exchangerates is high inside the ERM, and somewhat lower betweenthe pound and the ERM currencies. Shocks to the Canadiandollar/US dollar exchange rate are uncorrelated to othershocks, while shocks to the yen/ dollar exchange rate areindeed correlated with those affecting European currencies.Finally, the currencies of the (mainly European) smallerindustrialized countries appear to be linked to those of theERM.

Part D

Part D represents the variances of the exchange rate shocks.If V; is the calculated variance of the exchange rate shockto the dollar rate of currency i, i=l,..7, with the samecountry correspondences as before, the variances given inTable E.I 1 are assumed for the different regimes.

Table £.11

Simulated variances for exchange rate shocks

Free float

EMS + asymmetric/symmetric EMU

(v , -V4Vj

(V, H

vi

h V2 + V3)/3 i =i =i =

h V2 + V3)/3 i =

i =

12,3,45,6,7

1,2,3,4

5,6,7

V| is the observed variance of the shocks to the dollar exchange rate of country i.D = ] , F = 2, 1 = 3, UK = 4. Canada = 5. Japan = 6. smaller industrial countries = 7.

For the free float, the same dichotomy as before betweenthe DM as a major international currency and the otherthree EMU currencies is introduced. The DM shocks are

323

Annexes

given the average of the calculated variances for the threepresent EMS members in the model, whereas the shocks forthe FF, LIT and UKL receive the calculated variance of theUKL (see Table E. 12). The non-EMU variances are assumedunchanged.

For each of the three EMU regimes, the EMU currenciesget the (high) variance associated with either the DM as theanchor currency (EMS and asymmetric EMU) or the ecu(symmetric EMU regime).

Table E.12Standard deviations of exchange rate shocksin the simulated regimes (percentage points)

USD/DM USD/FF USD/LIT USD/UKL

AssumptionsFree floatEMSEMU

PM: Observed

10,910,910,9

11,2

9,610,910,910,9

9,610,910,910,6

9,610,910,99,6

5. Results from stochastic simulations

The differences, among regimes, in the variability of macro-economic variables in the stochastic simulations can be as-cribed to a limited number of factors since the only elementsthat change between regimes are the exchange rate shocksand the monetary policy rule for the short-term interest rate.

In the model, exchange rate shocks have short-term nominaland real effects. If exchange rate shocks become more sym-metric among Community countries, it may be expected thatnominal and real variability due to this factor is lower onaverage in the Community in the EMS regime compared tofloat, and again lower in EMU compared to EMS. Asillustrated by a simple example in Box E.3, this is due to thefact that the variance of intra-Community exchange ratesreduces to zero in EMU, therefore eliminating a source ofvariability.

The effect of the monetary policy rule in the different regimesis twofold. On the one hand, it may restrict national auton-omy in conducting monetary and exchange rate policy,

Box E.3: Exchange rate shocks and macroeconomic stability

A very simple analytical example may illustrate why the disap-pearance of intra-Community exchange rate shocks will reducethe variability of macroeconomic variables such as inflation andoutput.

Suppose the world consists of three countries, say 1 = Germany,2 = France and 3 = the USA. As an example, consider the effectof exchange rate shocks on prices. Write the semi-reduced formof the price equations for Germany and France as follows:

Pi = anwi + ai2(-e2i + ?2> + Mei + PS) + zi

P2 = a22w2 + a2l(e2l + Pi) + a23(£21 + el + ?3> + *2>

where (in logarithms and in deviation from the baseline): pt isthe shock to the domestic absorption price level in country i,expressed in local currency, Wj the shock to the domestic outputprice level (assumed to be equal to the wage level, which in turnis assumed to be fully indexed to the absorption price level, suchthat Wj = PJ), e21 is the shock to the log of the FF/DM exchangerate, C] is the shock to the log of the DM/USD exchange rate,Zj represents all other shocks in the model, a;i is the share ofdomestic output in domestic absorption of country i and a;: theshare of imports from country j in the domestic absorption ofcountry i (asi + aj2 + aj3 = 1). These equations explain thedomestic absorption price by a weighted average of itself (via

the linkage domestic output price-wage level-indexation to ab-sorption price) and the import prices from the other twocountries. For simplicity, it is assumed that the shocks p3 andZj are independently distributed, such that for the analysis athand they may be put equal to zero without influencing theresult.

Thus, imposing wf = pj, p3 = Zj = 0, the two equations maybe solved to give the following reduced form:

Pi = eiP2 = e2, + e.

This implies that the arithmetic average of the variance ofabsorption prices in the Community is equal to:

[var(p,) + var(p2)]/2 = var(e,) + [var(e2l) + cov(e,,e2l)]/2

In EMU, the last expression in this equation reduces to zero. Ifthe DM/USD exchange rate variance var(e,) remains unchangedin EMU (as assumed in the simulations), the average varianceof prices in the Community will therefore decline if [var(e21) +cov(e|,e2|)] > 0, which is the case for the exchange rate shocksused since the correlation between DM/USD shocks and intra-Community exchange rate shocks is small, even though it maybe negative due to past tensions in the EMS, when there waspressure on the US dollar.

324


which may increase macroeconomic variability, as discussedin Chapter 6 and Annex D. On the other hand, the restric-tions on exchange rate policy will be incorporated in thebehaviour of private agents, which may therefore enhancewage and price discipline, leading to faster readjustment toshocks and therefore to less variability of macroeconomicvariables.

The combined impact of these factors on the variabilityof macroeconomic variables is presented in index form inTable E.13, taking the free float regime as a reference.

The differences in the effects on the individual Communitycountries are marked by the asymmetric role of Germany intwo of the four regimes, where it retains full autonomy inits monetary policy with the other Community countrieslosing autonomy. The implication is that the float, EMS andasymmetric EMU regime for Germany only differ in termsof exchange rate shocks whose decrease reduces inflation

variability and, to a minor extent, output variability. The bigchange for Germany occurs with the switch from asymmetricEMU to symmetric EMU. In the latter regime, the Germanmonetary policy objectives for output and inflation onlycount for one fourth,21 so that a loss of monetary policyautonomy occurs. Since the common monetary policy reac-tion function attaches a high priority to low inflation varia-bility, the latter does not deteriorate in EMU compared tothe EMS, due to the compensating effect of diminished intra-EC exchange rate shocks. The counterpart of this positiveresult for inflation variability is an increase in output varia-bility in Germany in EMU compared to each of the otherthree regimes. This represents a 'worst case scenario', how-ever, since only the role of monetary policy is taken intoaccount, neglecting, for instance, the role of fiscal policy inoutput stabilization.

This weight is model-specific, since only four Community countries arerepresented.

Table E.13Stochastic simulations with Multimod: root-mean squared deviations from baseline, float = 100

EMS Asymmetric EMU Symmetric EMU

Germany

GDPPrivate consumptionGross investmentTotal exportsTotal importsAbsorption deflatorGNP deflatorCapacity utilization1

Short-term nominal interest rate1

Nominal effective exchange rateExchange rate wrt2 USDCurrent account balance3

Inflation1


100100100100100100100100100100100100100100

94949698869196959167

102988682

9693969983899595896499988380

108105103102848087

1028567

103968682

France



Nominal effective exchange rateExchange rate wrt USDCurrent account balance3

Inflation1


100100100100100100100100100100100100100100

13515419690

11083

104115126

1365

1189868

104889783675563

100580

64767241

9380868462516094560

66756042

325

Annexes

Table E.13 (continued)

Free float EMS Asymmetric EMU Symmetric EMU

Italy




Inflation1


100100100100100100100100100100100100100100

13114713199

127119142126132

1781

14012581

11411611595

10789

104116690

7911910471

10310510996998093

108670

821149070

United Kingdom

GDPPrivate consumptionGross investmentTotal exportsTotal importsAbsorption deflatorGNP deflatorCapacity utilization'Short-term nominal interest rate1


Inflation1


100100100100100100100100100100100100100100

9810512595

10897

11295

1941598838674

8197909497697779540

93917561

8096879495627379520

96896863

EUR 4 (average)



Exchange rate wrt USDCurrent account balance3

Inflation1


100100100100100100100100100100100100100

10912913996

11297

11610315183

1059675

9310010293877181926480928162

90969794826475906283917362

USA

GDPNominal effective exchange rateInflation1


100100100100

100100101102

9999

101100

99100101103

326


Table E.13 (continued)

Asymmelric EMU Symmetric EMU

Japan

GDPNominal effective exchange rateExchange rate wrt USDInflation1


100100100100100

10010199

10099

100101999998

100101999999

1 Difference wilh respecl lo baseline in percentage points.2 wrl = wilh respecl lo.3 Difference with respect to baseline, % baseline nominal GDP.

For the Community countries other than Germany, signifi-cant effects occur between each of the regimes. Comparingthe EMS as modelled here to the float regime, inflationvariability decreases in France and the United Kingdom,but increases in Italy. In the former two countries, thedecrease in intra-EC exchange rate shocks and wage/pricediscipline effects dominate the loss of monetary policy. InItaly this is not the case, which may be due to the nature ofthe historical shocks and more backward-looking wage andprice behaviour. The loss of monetary policy increases out-put variability in France and Italy due to the smaller realeffects of exchange rate variability reduction. Only in theUnited Kingdom is the wage and price discipline effectstrong enough to reduce output variability slightly. Themove from EMS to the asymmetric EMU regime shows asimilar pattern for the three countries, with the eliminationof intra-EC exchange rate shocks and effects on wage andprice discipline having a strong downward impact on thevariability of both output and inflation, more than compen-sating the small additional loss of monetary policy auton-omy. When the common monetary policy treats the policyobjectives of all EMU members symmetrically, these gainsare further strengthened since they regain part of monetarypolicy control while still enjoying the wage and price disci-pline effects of the abandoned intra-EC exchange rate instru-ment.

The picture for the Community average in terms of macro-economic variability is conditioned by the individual countryresults. With the EMS as baseline, EMU would reduceoutput variability on average by some 15-20%, while in-flation variability could be some 25 % lower. As seen above,this result is not uniform for all countries due to the asym-metric position of Germany in the EMS. Due to the highpriority of low inflation variability in the monetary policyrule, inflation stability in Germany is guaranteed, however.

To obtain an idea of the actual size of the reductions invariability of macroeconomic variables that can be expected,

the percentage reductions could have been applied to theabsolute values of the variations resulting from the simula-tions (given in Appendix E.VI). These measures of variabilityare model-dependent, however, and strongly influenced bythe choice of similarity of coefficients used for most equa-tions, which could overstate the results. Moreover, there isthe problem of choice of baseline. Rather, the reductionsshould be applied to the observed variability of the growthrate of GDP and the level of inflation (see Box 6.4 in Chapter6). On that basis, a reduction of output growth variabilityof 20 % would correspond to a reduction in Member States'output growth variability ranging between 0,3 and 0,7 per-centage points, and an inflation variability reduction of25 % would be equivalent to a reduction in Member States'inflation variability ranging from 0,6 to 1,5 percentagepoints. It should be noted that all these reductions refer tothe variability of output growth or inflation, and not theirlevels.

6. Caveats and general evaluationThe results of the stochastic simulations presented above aresubject to a number of important caveats. Taking thesecaveats into account is necessary in order to put the resultsinto their proper perspective.Generally speaking, the methodology used to evaluate theimpact of EMU on the variability of macroeconomic vari-ables is well adapted to the problem (for instance, since itmeets most of the Lucas critique of policy evaluation withempirical models) and may be considered as being up todate. While this has the advantage that the maximum ofempirical information which economic research can deliveron the stability properties of EMU is obtained, this alsoimplies that the results bear the uncertainty associated witha methodology which is still under development.The Multimod model must be considered a reasonablecompromise as regards size and basic assumptions in theclass of multicountry models, notably concerning its in-

327

Annexes

elusion of model-consistent expectations and endogenoustreatment of exchange rates. Desirable as these features mayseem, they also are a possible source of uncertainty since theformer could make the results err on the optimistic sideconcerning the adjustment of variables to expectations, whilethe latter introduces an element of uncertainty in the floatingexchange rate regimes induced by the endogenous modellingof exchange rates, which is known to be a highly fragileexercise. Furthermore, the imposed similarity of most coef-ficients among countries may underestimate the behaviouralasymmetries but overestimate the asymmetries of historicalshocks, so that the individual country results should ratherbe considered as draws from the same population than asprecise indications.

The main limitation of the policy regimes introduced in themodel is that they focus exclusively on monetary policy andlack a fiscal policy rule. While this has the advantage that theeffects of the individual policy regimes are more tractable, itimplies a departure from reality. For instance, in the case ofGermany, it may be imagined that part of the loss in outputvariability could be compensated through fiscal policy. An-other problem related to the realism of the regimes concernsthe way in which the EMS was modeled. Firstly, becausethere exists no such thing as a 'devaluation rule' in the EMS,and secondly because the EMS is very much an evolvingsystem of which it is difficult to imagine which State couldserve as a proper baseline alternative to EMU.

A more intricate problem is related to the monetary policyrule used. As shown in Section 2.3.1, the parameters ofthe interest rate reaction function should be regarded as acombination of the parameters of the policy-makers' objec-tive function and the multipliers of the model, and shouldtherefore change among regimes since the multipliers are notinvariant to regime changes. Nevertheless, the parameters ofthe interest rate reaction function were kept constant be-tween regimes. Since the parameters chosen correspond toa free float regime, the corresponding rule is not optimalin the other regimes (this is equivalent to saying that theelimination of non-cooperative monetary and exchange ratepolicies has not been taken into account in the simulations).Removing this obstacle, however, could therefore only in-crease the gains of EMU in terms of macroeconomic stab-ility.22

Some final remarks relate to the generation of the shocksintroduced for the stochastic .simulations. As discussedabove, there exists no standard methodology for the calcu-22 Related to this point is the fact that macroeconomic stability has not

been evaluated using an explicit macroeconomic welfare function, butrather on the basis of individual variables. This is only a problem,however, if the variability of output and inflation would move indifferent directions.

lation of exchange rate shocks mainly because of the diffi-culties in modelling exchange rate behaviour. The presentapproach was found to give exchange rate shocks whichseem a priori acceptable, but departs from the latest IMFmethodology (which itself is also evolving). A further diffi-culty relates to the absence of shocks to exogenous variables(both policy and non-policy), implying that changes to thesevariables were fully anticipated in the past. Doing so wouldhowever have introduced additional uncertainty concerningthe distinction between anticipated and non-anticipatedchanges in these variables. Finally, it was assumed that theobserved shocks in behaviour would be the same for allregimes, which again is more instrumental in providing trac-table results than necessarily adding to the realism of themodel.

Appendix E.I: Overview of the Multimod model

The Multimod model used for the simulations in this Annexwas developed at the International Monetary Fund by Mas-son et al. (1988, 1990). It may be considered as intermediatebetween purely theoretical and full-scale empirical macro-economic models. The model links submodels for the fourmajor Community countries (Germany, France, Italy andthe United Kingdom), the United States, Japan, Canada,and several 'rest of the world' zones. The country modelscontain some 10 econometric equations each, and may there-fore be considered to be small. An important specific featureof the model is that the labour market has been substitutedout, notably in the wage-price block. The nature of themodel is such that it hardly has any forecasting abilities, butcan usefully be employed for policy simulations around agiven baseline.

The theoretical structure of the model is fairly transparentand incorporates both neo-Keynesian and classical features.Forward-looking expectations are combined with wagestickiness, for instance. Furthermore the model is based onintertemporal utility maximization of households but alsoassumes the existence of liquidity constraints. Forward-look-ing expectations in the model are equivalent to model-con-sistent expectations of future variables. Consequently, whena shock occurs, economic agents are supposed to have fullknowledge of the state of the economy and notably of thepolicy rules (as represented by the model). This implies thatshocks are unanticipated when they occur, but that theirconsequences are fully seized. Forward-looking expectationsappear notably in the price equation (which is a reduced formequation of the wage-price block), the private consumptionequation, the investment equation and the financial block.The latter is not estimated except for a money demandequation (not used in the present simulations); since the

328

sduquenoy


short-term interest rate is determined by a regime-dependentreaction function, the long-term interest rate is determinedby a pure arbitrage condition with respect to short-terminterest rates and the exchange rate follows uncoveredinterest rate parity, i.e. the expected change in the exchangerate in period t+ 1 with respect to period t equals the interestrate differential in period t.

A specific feature of the model is that economic behaviouris imposed to be highly symmetrical across countries. Econ-omic structures (e.g. as regards the size and geographicalcomposition of foreign trade, the level of public debt etc.)differ among countries, but only statistically significant dif-ferences in parameters across countries were maintained,meaning in practice that except for the trade equations andwage rigidity parameter in the price equation most equationshave identical parameters. This may underestimate the be-havioural asymmetries among the G7 countries, causingthem to reappear in the equation residuals in the form of arelatively poor statistical fit and the appearance of moreasymmetric shocks than otherwise would be the case for thestochastic simulations through the co-variance matrix.

This point may be illustrated by a simple example. Supposethat there exist two equations estimated separately for twocountries, for instance y( = ct|X| + U j and y2 = a2x2 + u2,with Uj the residuals. When estimated jointly, i.e. imposing acommon coefficient a, the equations become y, = ax, + e,and y2 = ax2 + e2, with the e{ residuals different from theUj. For a numerical example, take a, = 1, a2 = 3 and a =2. Suppose that for a particular date the variables have thevalues y, = 2, y2 = 4 and x t = x2= 1, such that there is aperfect symmetric shock Uj = u2 = 1 for the equations withasymmetric behaviour. The shocks to the equation withsymmetric behaviour are now more asymmetric, i.e. C j = 0and e2 = 2. In terms of co-variances between shocks, thiswould imply a lower co-variance between the asymmetricshocks.

Appendix E.II: Modifications to Multimod model

In addition to the introduction of explicit currency perDeutschmark equations for France, Italy and the UnitedKingdom, the Multimod model has been modified in threerespects.

1. Tests with price shocks given to the model have pointedto instability problems, i.e. diverging oscillations in pricescaused by the asynchronous movement of prices, productioncapacity and output. The IMF has suggested that this behav-iour of the model, which occurs in particular for Italy, mightbe due to the weak price elasticities of the Italian import

and export equations, whereas it could also be related to thehigh level of public debt in the baseline and the ensuingcontractionary effect on taxes. The solution adopted to avoidthis problem of diverging oscillations has been to change thedynamic structure of the price equations for the G7 countriesand the zone of smaller industrial countries. In particular,the long-run effect of the degree of capacity utilization hasbeen made instantaneous, without having repercussions inlater periods. While this does not change the long-runproperties of the price equation, this ensures a bettersynchronous movement of prices and the degree of capacityutilization towards a long-run equilibium after the occur-rence of a price shock.

The original Multimod price equation is as follows:

ln(PY/PY_!) = a + b*f(CU) + k*ln(P+1/P)+ (l-k)*ln(PY_1/PY_2)

with PY =GDP deflator

P = Absorption deflator

f(CU) = function of degree of capacity utilization.

This equation implies a long-run effect of f(CU) on the rateof inflation of b/k. This equation can be interpreted as beingderived from a static equation with a Koyck lag on theexpression (b/k)*f(CU) + ln(P+|/P). The modification con-sists in assuming that the Koyck lag operates solely onthe rate of expected inflation m(P+ ,/P), and leads to thefollowing equation:

ln(PY/PY^) = a + b/k*[f(CU) - (l-k)*f(CU_!)]+ k*ln(P+1/P) + (l-k)*ln(PY_,/PY_2).

This formulation has the same long-run elasticity of priceswith respect to the expression in the degree of capacityutilization, but has it frontloaded instead of applying gradu-ally through time according to the geometrical Koyck lag.

2. The second modification to the model deals with thestrong effect of endogenous tax rates in response to changesin the debt/GDP ratio, as mentioned above. The responseof tax rates with respect to changes in this ratio, which waspreviously imposed to be equal to 0,3, has, in accordancewith ideas of the IMF in this respect, been put equal to alower value, namely 0,1.

3. The definition of the user cost of capital has a treatmentof tax credit which implies that investment is assumed to be100% debt-financed. In some circumstances, this couldimply that increases in interest rates could decrease the usercost of capital. To avoid this undesirable behaviour, 50%debt-financing was assumed.

329

Annexes

Appendix E.III: Derivation of the nominal GDPtargeting rule

A targeting rule involving inflation and output is derivedbelow under the assumption of standard money supply tar-geting by the authorities.

Let the money demand equation be:(I) Md - p = u(aq - bi) + (1 - u)(Md - p)_, + c

where Md is nominal money demand, q is real output, p isthe absorption price level, and i is the nominal interest rate.All variables except the interest rate are in log form.

Suppose the monetary authorities have inflation and outputlevel as ultimate objectives, money stock as intermediatetarget and the (short-term) interest rate as instrument. Letrt*, q* and i* be the target values for inflation, output andthe nominal interest rate.23 Assuming the authorities knowthe money demand function (I), the money target is set at alevel consistent with the ultimate targets, i.e.:(II) M* = (p_! + TC*) + u(aq*-bi*) +

( l -H)(M d -p )_! + c.

Note that output enters the equation in level form, but thatonly variations in prices (inflation) are taken into account.Monetary authorities target the output level, but there is nomemory of past prices. Hence, only inflation is taken intoaccount, instead of price levels.

Putting money demand equal to money supply yields:

n - TC* -f u a (q - q*) - u b (i - i*) = 0

where n = p - p_, is the inflation rate.

Differentiating with respect to baseline leads to:

n - Jtb + u a (q - qb) - u b (i - ib) = 0.

Hence, the associated interest rate reaction function is:

(III) (i - ib) = [l/(ub>] (TC - 7Cb) + [a/b] (q - qb).

Values of a, b and u can be derived from the Multimodmoney demand equation (pooled estimate for the G7countries, see Masson et al. (1990)): u = 0,21, a = 0,94,b = 0,024. Hence,(IV) (i - ib) = 100 [ 2 (n - icb) + 0,4 (q - qb) ].

This equation implies a strong short-term reaction to in-flationary shocks since a 1 % rise in inflation implies a 2 %rise in the nominal interest rate. In order to avoid excessiveinstrument instability in the case of temporary shocks, apartial adjustment mechanism is added to (IV), yielding:

(V) (i - ib) = 0.5 [100 ( 2(7i - Tib) + 0.4 (q - qb)) ] +

Appendix E.IV: Calibration of the parameters ofthe interest rate reaction function in accordancewith recent EMS behaviour

The effective margins of fluctuation vis-a-vis the Deutsch-mark have already been reduced to 1 % or even less for mostparticipants in the narrow-band ERM (Table E.AI).

Table E.AI

Indicators of exchange rate variability within bands

BFRDKRFFIRLHFL

1984-87

0,660,930,680,920,25

1987-89

0,330,950,630,330,17

Average root-mean square deviation from trend; trends are computed by using exponentialsmoothing technique.Source /Calculated using data from the Commission of the European Communities.

Calibration of parameters B and 0 in (3') allows to combinequasi-linearity for small deviations and strong non-linearityin the neighbourhood of band limits. From the followingequation

(3') (i - ib) = 100 [Q! (IE - rcb) + a2 (q - qb) + a3 { (e - eb)+ B (e - eb)®}]

it appears that:

[8i/5e] - 100 a3 [ 1 + 0 B (e - ].

2J Alternatively, the target value for the nominal interest rate could bederived from inflation and real interest rate targets.

Table E.A2 below gives for a-, = 2, 0 = 11 and B = 2.1018

the values of both interest rate deviations (from (3')) andpartial derivative [6i/5e) for different values of exchange ratedeviation.

330


Table E.A2

Interest rate reactions to exchange rate deviations

Exchange ratedeviation (%> = I00(e-eb)

0,000,250,500,751,001,251,501,752,002,25

[6i/Se]'

2,002,002,002,022,446,10

27,37120,53452,56

1465,11

Interest ratedeviation (%) = (i-ib)

0,000,501,001,502,042,976,46

22,3685,92

303,77

1 In percentage points.

Appendix E.V: Derivation of simulatedcorrelations among exchange rate shocks

Let R" (i not equal to j) be the calculated correlation coef-ficient between the dollar exchange rate shocks of country iand country j, with ij=l,..,7 (D=l , F = 2, 1 = 3, UK =4, Canada = 5, Japan = 6, smaller industrial countries = 7).Table E.A3 shows the modifications made to the observedcorrelation coefficients in accordance with the principlessummarized in Table E.9. The modifications to the corre-sponding co-variances take place via the correlation coef-ficients that are associated with the variance-co-variancematrix.24

For the free float, the correlations between the exchangerate shocks of the four EMS members are assumed to beequal to an average of the calculated correlations betweenthe exchange rate shocks of Germany, France and Italy vis-a-vis the UK. For the correlations between the shocks for thefour EMS members and the other currencies, a dichotomy isassumed between the position of the DM and the other threecurrencies. The DM being more of an international currencythan the other three currencies, it is assumed that the corre-lation of shocks between the DM and the non-EMS currenc-ies can be approximated by the average of the calculatedcorrelations of the DM, FF and LIT with any third-country

currency. The three other EMS countries, France, Italy andthe UK, are assumed to be in a similar position with respectto third-country currencies, and therefore all receive thecorrelation of the exchange rate shock for the UK withrespect to each non-EMS currency shock. The correlationsamong the third-country currency shocks are assumed to beunchanged.

Table E.A3Simulated correlations for exchange rate shocks

Free float

EMS

Asymmetric EMU/symmetric EMU

(R14 + R24 + R34}/3 ij

(Rij + *2j + R3j)/3 *jR4: ij

ti ^j

(R12 + RI3 + R23)/3 ij

(Ri j + R2j + R3j>/3 ij

RJJ ij

1 ij

(R,j + R2j + R3j)/3 ij

11 '

j

= 1,2,3,4= 1,2,3,4= 1= 5,6,7= 2,3,4= 5,6,7- 5,6,7= 5,6,7

= 1,2,3= 1,2,3,4= 1,2,3,4= 5,6,7= 5,6,7= 5,6,7

= 1,2,3,4= 1,2,3,4= 1,2,3,4= 5,6,7= 5,6,7= 5,6,7

Using the formula cov(x,y) = R *ox*oy, with cov(.) the co-variance,R the correlation coefficient and a the standard deviation.

RJJ is the correlation between calculated exchange rate shocks relative to the dollar betweencountries i and j.D = I, F = 2, I = 3, UK = 4, Canada = 4, Japan = 6, smaller industrial countries = 7.

For the EMS regime, the correlations between the exchangerate shocks for Germany, France and Italy are taken tobe equal to the average calculated correlations. For thecorrelations between these three countries and the UK, theaverage of the correlations between the three EMS membersis also taken. The correlation between any EMS memberand any third currency is assumed to be equal to the averageof the correlations between the DM, FF and LIT and thatcurrency. Again, the correlations among the third currenciesare left unchanged.

For the asymmetric EMU and symmetric EMU regime, thesame correlations as for the EMS regime are used, exceptthat now the correlations for exchange rate shocks betweenEMU currencies are equal to one.

331

Annexes

Appendix E.VI: Stochastic simulations with Multimod:root-mean squared deviations from baseline

Germany



Nominal effective exchange rateExchange rate wrt2 USDCurrent account balance3

Inflation1


France



Exchange rate wrt DMExchange rate wrt USDCurrent account balance3

Inflation1


Italy




Inflation1


United Kingdom

GDPPrivate consumptionGross investment

Free (loai

3,63,27,49,57,35,75,14,13,58,6

18,63,42,36,5

3,46,79,1

10,212,912,110,83,55,4

21,328,83,83,2

10,0

3,56,9

13,49,9

10,39,59,13,74,5

17,523,32,43,09,4

6,79,3

13,5

EMS

3,43,07,29,36,35,25,03,93,25,8

19,03,32,05,4

4,610,417,89,2

. 14,210,111,24,16,82,9

18,94,53,16,8

4,610,117,69,8

13,111,312,94,76,03,1

19,03,43,87,6

6,69,9

16,9

Asymmetric EMU

3,43,07,19,56,15,14,94,03,15,6

18,63,31,95,2

3,66,08,98,58,76,86,83,63,10,0

18,62,92,34,2

4,08,0

15,69,5

11,18,59,54,43,10,0

18,62,93,16,7

5,49,1

12,2

Symmetric EMU

3,93,37,79,76,14,64,54,23,05,8

19,23,32,05,3

3,25,47,88,78,16,36,63,33,00,0

19,22,91,94,2

3,67,3

14,79,5

10,27,78,54,03,00,0

19,22,82,76,6

5,49,0

11,7

332


Free float Asym metric EM U Symmetric EMU

Total exportsTotal importsAbsorption deflatorGNP deflatorCapacity utilization1



Inflation1


7,712,512,86,85,7

20,719,94,35,3

11,9

8,48,4

12,214,46,5

11,13,2

19,73,64,6

8,37,58,79,95,43,10,0

18,63.94.07,3

8,37,37,89,45,43,00,0

19,23,83,67,5

EUR 4 (average)



Exchange rate wrt USDCurrent account balance3

Inflation1


4,56,9

11,29,69,8

10,39,84,74,93,03.53.69,6

4,98,9

15,59,2

11,010,111,44,97,3

19,23,73,57,3

4,26,9

11,49,08,57,48,04,43,1

18,63,32,96,0

4,16,6

10,99,18,16,77,54,33,0

19,23,22,66,0

USA

GDPNominal effective exchange rateInflation1


2,716,32,3

12,2

2,716,32,4

12,5

2,716,12,4

12,3

2,716,42,4

12,6

JapanGDPNominal effective exchange rateExchange rate wrt USDInflation1


4,128,633,74,0

15,8

4,129,133,64,0

15,6

4,129,033,64,0

15,5

4,129,033,64,0

15,61 Difference with respect to baseline in percentage points.2 wrt = with respect to.3 Difference with respect lo baseline, % baseline nominal GDP.Note: To obtain convergence, the shocks lo the equation residuals were scaled down by a factor of 10. Consequently, the results have been multiplied by 10. Tests have shown ihat this operation isrelatively inconsequential for the results.

333

Annexes

References

Begg, D. (1990), 'Alternative exchange rate regimes: the roleof the exchange rate and the implications for wage-priceadjustment', in European Economy (1990).

Bryant, R.C., Hooper, P. and Mann, C. L. (1990), 'The corestochastic simulations: a pre-conference overview', paperpresented at the Brookings conference 'Empirical evaluationof alternative policy regimes', 8-9 March, Washington, DC.

Committee for the Study of Economic and Monetary Union(1989), Report on economic and monetary union in the Euro-pean Community, Commission of the European Communi-ties.

De Grauwe, P. (1988), 'Is the European Monetary Systema DM-zone?', CEPS Working Document No 39, October.

European Economy (1990), 'The economics of EMU', specialissue, forthcoming.

Frankel, J. A. (1989), 'Quantifying international capital mo-bility in the 1980s', Working Paper No 2856, NBER.

Frankel, J. A. and Froot, K. A. (1987), 'Using survey datato test standard propositions regarding exchange rate expec-tations', American Economic Review 77; pp. 133-153.

Frankel, J. A. and Froot, K. A. (1989), 'Forward discountbias: is it an exchange risk premium?', Quarterly Journal ofEconomics, pp. 139-161.

Frenkel, J. A., Goldstein, M. and Masson, P. (1989), 'Simu-lating the effects of some simple coordinated policy rules',in Bryant, R. C., Currie, D. A., Frenkel, J. A., Masson,P. R. and Fortes R. (eds), Macroeconomic policies in aninterdependent world. The Brookings Institution, CEPR andIMF, pp. 203-239.

Hooper, P. and Mann C. L. (1989), 'The emergence andpersistence of the US external imbalance, 1980-87', PrincetonStudies in International Finance No 65, Princeton University.

Koromzai, V., Llewellyn, J. and Potter, S. (1987), 'The riseand fall of the dollar: some explanations, consequences andlessons', Economic Journal 97, pp. 23-43.

Lucas, R. E. Jr. (1976), 'Econometric policy evaluation: acritique1, Carnegie-Rochester Conference Series on PublicPolicy. No 2, pp. 19-46.

Masson, P., Symansky, S., Haas, R. and Dooley, M. (1988),'Multimod — A multi-region econometric model', StaffStudies, IMF.

Masson, P., Symansky, S. and Meredith G. (1990), 'Multi-mod Mark II: A revised and extended model', IMF Oc-casional Paper No 71, July.

Masson, P. and Symansky, S. (1990), 'Stochastic simulationresults for Multimod', paper presented at the BrookingsConference 'Empirical evaluation of alternative policy re-gimes', Washington, DC, 8-9 March.

McCallum, B. T. (1976), 'Rational expectations and theestimation of econometric models: an alternative procedure',International Economic Review 17, pp. 484-490.

Meese, R. and Rogoff, K. (1983), 'Empirical exchange ratemodels of the seventies: do they fit out of sample?', Journalof International Economics 14, pp. 3-24.

Meese, R. and Rogoff, K. (1988), 'Was it real? The exchangerate-interest differential relation over the modern floatingperiod', Journal of Finance 43, pp. 933-948.

Obstfeld, M. (1988), 'Competitiveness, realignment andspeculation: the role of financial markets', in Giavazzi, F.,Micossi, S. and Miller, M. The European Monetary System,Cambridge University Press.

Sims, C. (1982), 'Policy analysis with econometric models',Brookings Papers on Economic Activity, Vol. 1, pp. 107-152.

Van der Ploeg, F. R. (1990), 'Macroeconomic policy coordi-nation during the various phases of economic and monetaryintegration in Europe', in European Economy (1990).

Von Hagen, J. and Fratianni, M. (1989), 'Credibility andasymmetries in the EMS', paper presented at the CEPS/IMFconference 'Exchange rate policies in selected industrializedcountries', 12-14 October, Brussels.

334

Detailed contents

Preface 5

Executive summary 9

Part A — Synthesis and economic principles 15

Chapter 1 — Synthesis 17

1.1. What alternative regimes are to be compared? 17

1.2. The conceptual framework 18

1.3. The main benefits and costs 20

1.3.1. Concerning efficiency and growth 201.3.2. Concerning price stability 221.3.3. Concerning public finance 231.3.4. Concerning adjustment without exchange rate changes 231.3.5. Concerning the international system 24

1.4. The impact through time and space 25

1.4.1. Transitional issues 251.4.2. Regional impact 271.4.3. Benefits and costs by country 27

1.5. The Community as an advantageous monetary area 28

Chapter 2 — The economics of EMU 31

2.1. EMU and alternatives 31

2.1.1. The economic content of EMU 312.1.2. Alternatives to EMU 40

2.2. The economic impact of EMU 44

2.2.1. Building blocks for the analysis of EMU 452.2.2. Costs and benefits 48

Part B — The main benefits and costs 61

Chapter 3 — Efficiency gains 63

3.1. Exit exchange rate transaction costs 64

3.1.1. External or financial costs 64

335

3.1.2. In-house costs 673.1.3. Summary evaluation of transaction cost savings 68

3.2. Exit exchange rate uncertainty 68

3.2.1. Exchange rate variability in the Community 703.2.2. Exchange rate variability and trade 713.2.3. Exchange rate variability and capital movements 743.2.4. Efficiency gains from the elimination of exchange rate un-

certainty 74

3.3. Building on 1992 75

3.4. Indirect and dynamic gains from economic and monetary union 77

3.4.1. Integration and dynamic gains 773.4.2. Quantifiable dynamic gains from economic and monetary

union 80

3.5. Business expectations and growth 83

Chapter 4 — Benefits of stable prices 87

4.1. The cost of inflation 87

4.1.1. The effects of anticipated inflation 874.1.2. The effects of unanticipated inflation 91

4.2. The cost of disinflation 93

4.3. Requirements of a stable and credible monetary regime 95

4.3.1. A statutory mandate for price stability 964.3.2. Central bank independence and price stability 97

Chapter 5 — Implications for public finance 100

5.1. Budgetary policy in a monetary union 101

5.1.1. The need for autonomy 1025.1.2. The need for discipline 1065.1.3. The need for coordination 113

5.2. Impact on income and expenditure of governments 120

5.2.1. Seigniorage revenue losses 1205.2.2. Interest rates and public debt service 1235.2.3. Taxation and the provision of public goods in EMU 128

Appendices 131

Chapter 6 — Adjusting without the nominal exchange rate 136

6.1. Do nominal exchange rate changes have real effects? 138

336

6.2. Will asymmetric shocks diminish in EMU? 140

6.2.1. Country-specific shocks 1406.2.2. When are common shocks asymmetric? 1426.2.3. Trend real exchange rates 1456.2.4. Conclusion on shocks 146

6.3. Factor adjustment 147

6.3.1. The impact of EMU on real wage flexibility 1496.3.2. Regional mobility 151

6.4. Shocks, adjustment and macroeconomic stability 152

6.5. Adjustment through external financing 158

6.5.1. The external constraint 1596.5.2. Effects of EMU on the external constraint 161

6.6. Budgetary policy as an alternative to the exchange rate instrument 162

6.6.1. The need for budgetary policy 1626.6.2. The effectiveness of budgetary policy 163

6.7. The role of central public finance in assisting regional adjustment 166

6.7.1. Central public finance in existing federations 1666.7.2. Central public finance in EMU 168

Appendices 169

Chapter 7 — External dimensions 178

7.1. Recasting of international currencies 178

7.1.1. The ecu as a vehicle currency 1797.1.2. EM U and the official monetary sector 1837.1.3. Seigniorage effects 1857.1.4. Portfolio adjustments and exchange rate effects 186

7.2. The Community's part in international cooperation 189

7.2.1. EMU and international cooperation 1907.2.2. Global coordination gains from EMU 190

7.3. Towards a better international monetary regime 194

7.3.1. The limits of the present regime 1947.3.2. Is a multi-polar system desirable? 195

Appendix 196

337

sduquenoy

Part C — The impact through time and space 201

Chapter 8 — Transitional issues 203

8.1. Benefits and costs by stage 203

8.1.1. Price stability 2038.1.2. Efficiency gains 2048.1.3. Public finance effects 2058.1.4. Adjusting without the nominal exchange rate 2058.1.5. External effects 205

8.2. Stability in the transition 2058.2.1. Speculative attacks 2068.2.2. Currency substitution 207

8.3. How much convergence remains to be achieved? 208

8.3.1. How to achieve convergence 209

8.4. Speed of transition 210

Chapter 9 — Spatial aspects 212

9.1. Principles 212

9.1.1. Concepts: regions and regional problems 2129.1.2. Geographical effects of economic integration 213

9.2. Basic facts and trends 2159.2.1. Empirical evidence 2159.2.2. The response in EC policies 216

9.3. Businesses' perceptions of regional handicaps and competitiveness 217

9.4. Wider strategic considerations 220

9.4.1. The less favoured regions and economic union 2209.4.2. The less favoured regions and monetary union 2239.4.3. Regime change and regime structure 225

Chapter 10 — National perspectives on the costs and benefits of EMU 235

10.1. Currencies in the narrow band of the exchange-rate mechanism 23510.1.1. Germany 23510.1.2. France 23610.1.3. Italy 23710.1.4. Belgium 23810.1.5. Netherlands 23910.1.6. Luxembourg 23910.1.7. Denmark 24010.1.8. Ireland 240

338

sduquenoy

10.2. Currencies in the broad band of the exchange-rate mechanism 241

10.2.1. Spain 24110.2.2. United Kingdom 242

10.3. Countries outside the exchange-rate mechanism 243

10.3.1. Greece 24310.3.2. Portugal 244

Annexes

Annex A — Exchange transaction costs 251

1. Introduction 251

2. Financial transaction costs 2512.1. The volume of foreign exchange transactions in the Community 2512.2. The costs of foreign exchange 2582.3. The costs of cross-border payments 2612.4. The financial transaction cost savings from a single currency 262

3. In-house costs 264

3.1. Direct and indirect effects 2643.2. Size of in-house costs 265

4. Once-and-for-all adjustment cost of introducing a single currency 267

5. Cost summary 267

Annex B — Germany and the Netherlands: the case of a de factomonetary union 269

1. Introduction 269

2. Comparison of economic performance 270

2.1. Macroeconomic performance 2702.2. Structural and sectoral characteristics 273

3. Comparison of economic policies 2743.1. Monetary policy and inflation 2743.2. Fiscal policy 2763.3. Wage developments 277

339

sduquenoy

4. The asymmetric effects of shocks 2774.1. Energy prices 2774.2. External trade cycle and the Dutch policy response 2784.3. Deutschmark weakness in 1988; German monetary union 278

Annex C — European and German economic and monetary union:a comparison 280

1. Introduction 280

2. Price stability 281

3. Efficiency gains 281

4. Fiscal effects 282

5. Adjusting without the exchange rate 282

6. Transition 283

7. Concluding remarks 283

Annex D — Shocks and adjustment in EMU: simulations withthe Quest model 284

1. Introduction and summary 284

2. Sources of macroeconomic fluctuations in the Community 284

2.1. Categories of shocks 2852.2. Methodology 2862.3. Results 2872.4. Conclusion 287

3. The loss of the nominal exchange rate as an adjustment instrument:an evaluation 2893.1. Nature of the cost 2893.2. Evaluation with the Quest model 2903.3. Methodology of the simulations 2923.4. The exchange rate versus alternative instruments 2923.5. Demand shocks under alternative wage behaviour 2953.6. A supply shock 298

Appendix: Detailed simulation results 299

340

Annex E — Exchange rate regimes in the EC: simulations withthe Multimod model 303

1. Introduction and non-technical summary 303

2. Monetary policy and exchange rate regimes 3062.1. General description 3062.2. Free float regime 3072.3. The EMS regime 308

2.3.1. Monetary policy rule 3082.3.2. Realignment rule 309

2.4. The EMU regimes 311

3. Results from deterministic simulations 311

4. Methodology of stochastic simulations 3174.1. General methodology 3174.2. Treatment of exchange rate shocks 3184.3. Stochastic properties of exchange rates under alternative policy

regimes 3214.3.1. Principles 3214.3.2. Empirical implementation 322

5. Results from stochastic simulations 324

6. Caveats and general evaluation 327

Appendices 328

341

sduquenoy

List of tables and graphs

Tables

1.1. Economic mechanisms generating benefits and costs, by stages ofEMU, as in the Delors Committee Report 26

2.1. Major features of 1992 and EMU 322.2. Indicators of city and regional inflation divergence within Canada 372.3. Major features of the three EMU stages 402.4. Realignments in the EMS, 1979-89 422.5. Major categories of effects of EMU 502.6. Assumptions and methods of quantitative evaluations of EMU 532.7. A schematic presentation of costs and benefits of alternative

exchange rate regimes 56

3.1. Currency transaction losses in a (hypothetical) round-trip through10 countries 66

3.2. Cost savings on intra-EC settlements by single EC currency 683.3. Bilateral exchange rates 693.4. Bilateral nominal exchange rates against 20 industrialized

countries 713.5. Bilateral nominal exchange rates against EUR 12 currencies 713.6. Bilateral nominal exchange rates against ERM currencies 723.7. State aids in the European Community, by main objective 773.8. The influence of the internal market on the expectations of firms 793.9. The medium-run growth bonus as a percentage of static efficiency

gains 795.1. Summary presentation of the fiscal policy and public finance

effects of EMU 1015.2. Medium-term effects of the saving decline 1055.3. Debt sustainability conditions in the Community 1095.4. Short-term spill-over effects of a bond-financed fiscal expansion

under alternative exchange rate regimes 1155.5. Spill-over effects of a rise in government expenditures by 2% of

GDP, Multimod simulation 1165.6. Spill-over effects of a rise in German government expenditures

by 2 % of GDP under an EMU regime, Quest simulation 1175.7. Gross seigniorage revenue effects of monetary union 122

6.1. Effects of economic structure on domestic prices after an increaseof 10% in oil prices, using 1980 input-output tables 144

6.2. Regional net migration in the EC, the USA and Sweden 1516.3. Central public finance in five major federations 1666.4. Fiscal autonomy and Fiscal equalization in existing federations 1686.5. Short-term price elasticity of money wages 1716.6. Elasticity of money wages with regard to unemployment 171

343

7.1. Functions of international currencies 1797.2. Trade invoicing currencies of the six major industrialized

countries 1807.3. Shares of different types of exchange rate regimes in 1975, 1981

and 1989 1847.4. Currency composition of official reserves 1857.5. Size and currency composition of world financial wealth, 1988

and 1981 1877.6. Construction of the world reference portfolio 198

'9.1. Disparities in per capita GDP in EUR 12, 1950-87 (Theil coef-ficient) 215

9.2. The structural Funds and the EC budget, 1970-90 2179.3. Evolution of aspects of regime structure with the degree of inte-

gration 226

10.1 Selected economic indicators for EC Member States, 1980-89 24510.2 Eurobarometer survey, autumn 1989 245A.I. Total foreign exchange market activity 252A.2. Foreign exchange market activity, by type of transaction: gross

turnover 252A.3. Currency composition of foreign exchange market activity:

gross turnover 253A.4. Measurement of the share of transactions between EC currencies

in total foreign exchange turnover 254A.5. Volume of EUR 12 goods trade, 1988 254A.6. Member States' current and capital account transactions 255A.7. Currency breakdown of banknote transactions, 1988 256A.8. Balance of payments transactions with other EC countries in EC

currencies (exclusive of the domestic currency) in 1988 257A.9. Currency distribution (%) of payments by (or to) domestic resi-

dents to (or from) EC Member States (current account trans-actions and foreign direct investment) 258

A.10. Sum of intra-EC imports and exports of goods, 1988 258A.I I. Size distribution of trade account settlements through means of

payments other than banknotes 259A. 12. Interbank bid-ask spreads expressed in percentage terms (data

collected in November 1989) ' 259A.13. Buying-selling spreads in percentage terms for foreign banknotes 261A.14. Cost savings on intra-EC settlements by single EC currency 267

B.I. Growth in the value-added of enterprises in the Federal Republicof Germany and the Netherlands (excluding energy) (1980 prices) 271

B.2. Production according to sector in the Federal Republic of Ger-many and the Netherlands (% of total) 273

B.3. Export shares in the Federal Republic of Germany and theNetherlands (in % of total production) 274

B.4. Export-import ratio in the Federal Republic of Germany and theNetherlands (in %) 274

D.I. Relative variances of shocks determining endogenous variables,I980.IIto 1987.III 288

D.2. Summary results for export demand shock: France 293D.3. Coefficients of the wage-price equations in the Quest country

models 296

344

sduquenoy

D.4. Average lag (longest lag) of the distributed lag functions in quar-ters 296

D.A1. Export demand shock: France, high wage flexibility 299D.A2. Export demand shock: France, slower indexation 300D.A3. Increase in social security contributions (1 % of nominal GDP)

paid by employers: Germany 301

E.I. Monetary policy rules in the four simulated regimes 307E.2. Realignments in the EMS, 1979-87 310E.3. Effects of 1 % government expenditure increase in France 311E.4. Effects of 5 % shock to inflation in France 313E.5. Effects of a 2 % price shock in France under EMU with normal

and slow price indexation 313E.6. Estimated standard errors of non-exchange rate shocks, 1973-88 318E.7. Comparison of exchange rate shocks 321E.8. Observed variability of bilateral dollar exchange rates 322E.9. Summary of hypotheses for exchange-rate regimes 322E.10. Observed (1981-88) and simulated correlation coefficients for

exchange rate shocks 323E. 11. Simulated variances for exchange rate shocks 323E.12. Standard deviations of exchange rate shocks in the simulated

regimes (percentage points) 324E.13. Stochastic simulations with Multimod: root-mean squared devi-

ations from baseline, float = 100 325E.A1. Indicators of exchange rate variability within bands 330E.A2. Interest rate reactions to exchange rate deviations 331E.A3. Simulated correlations for exchange rate shocks 331

Graphs

1.1. A business perception of the microeconomic impact of EMU 101.2. An economist's perception of the macroeconomic impact of EMU 101.3. Simplified schema of the effects of EMU 121.4. Detailed schema of the effects of EMU 19

2.1. Evolution of relative consumer prices in three monetary unions,1950-88 37

2.2. Evolution of relative wage levels in three monetary unions, 1950-88 382.3. A graphical representation of alternative exchange rate regimes 43

4.1. Unemployment and inflation in OECD countries, 1970-85 904.2. Real GDP per capita and inflation in OECD countries, 1975-85 914.3. Inflation and growth of real GDP per capita 914.4. Inflation and its variability in OECD countries 924.5. Inflation and variability of output growth for OECD countries,

1955-85 934.6. Central bank independence and inflation 985.1. Effects of a fall in UK household saving under three exchange rate

regimes 1045.2. Debt ratios and sustainability gaps in the Community, 1989 109

345

5.3. Evolution of public debt and primary deficit in selected MemberStates 110

5.4. Budget and current account balances, 1989 1135.5. Seigniorage revenues in the Community, 1988 1215.6. Long-term nominal interest rates and inflation, average 1985-89 1245.7. Ex-post real long-term interest rates and inflation, average 1985-89 1245.8. Long-term interest rate and inflation differentials with respect to

Germany 127

6.1. Schematic overview of the chapter 1376.2. Adjustment with and without devaluation 1396.3. The symmetry and asymmetry of sector-specific shocks in the EC:

1979-88 1446.4. The symmetry of shocks and trade barriers in the EC 1446.5. Real wage rigidity 1456.6. Real depreciation and growth, 1973-88 1476.7. Relative unit labour cost and relative unemployment rates inside

Germany and the Community 1486.8. The disciplinary effect of EMU on inflation after a price shock 1506.9. The dispersion of regional unemployment rates 1516.10 Macroeconomic stability of EMU 1546.11. The effectiveness of budgetary policy in EMU 1646.12. The trade-off between budgetary policy and wage adjustment 165

7.1. Currency breakdown of transactions in foreign exchange markets,1989 181

7.2. EMU and global coordination: an illustration 193

9.1. Enterprise assessment of factors of regional competitiveness 2199.2. Spain and Portugal — trade and investment flows with EUR 10 22210.1. Variation of EC Member State currencies with respect to the

Deutschmark (1980 = 100) 246B. 1. The bilateral Deutschmark/guilder exchange rate, 1979-90 (HFL per

DM, annual average) 270B.2. Growth of production potential in the manufacturing sector in the

Federal Republic of Germany and the Netherlands 270B.3. Real growth in the Federal Republic of Germany and the Nether-

lands, 1979-90 — Gross domestic product at constant market prices(annual % change) 271

B.4a. Employment in the Federal Republic of Germany and the Nether-lands (total economy) (changes in %) 272

B.4b. Unemployment rate in the Federal Republic of Germany and theNetherlands (percentage of civilian labour force) 272

B.5. Inflation in the Federal Republic of Germany and the Netherlands— Price deflator of private consumption (changes in %) 273

B.6. Current balance in the Federal Republic of Germany and the Nether-lands (% of GDP) 273

B.7. Short-term interest rates in the Federal Republic of Germany andthe Netherlands — 1978-90 (quarterly averages) 275

B.8. Long-term interest rates in the Federal Republic of Germany andthe Netherlands — 1978-90 (quarterly averages) 275

B.9. The difference between long-term and short-term interest rates inthe Federal Republic of Germany and the Netherlands (percentagepoint) 276

346

B.IO. Net lending ( + ) or net borrowing ( — ) in the Federal Republic ofGermany and the Netherlands — General government (% of GDP) 276

B.ll. Compensation per employee in the Federal Republic of Germanyand the Netherlands (annual % change) 276

B.I2. Real compensation per employee in the Federal Republic of Ger-many and the Netherlands (annual % change) 276

B.13. Terms of trade in the Federal Republic of Germany and the Nether-lands (goods and services) (1980 = 100) 278

B.I4. Competitiveness of the Federal Republic of Germany and theNetherlands. Unit labour costs, annual % change 278

D. 1. Effects of a demand shock with and without exchange rate change 289D.2. Output and real exchange rate effects of an export demand shock

with and without accompanying devaluation policy 290D.3. Real exchange rate effects of a devaluation under alternative wage

rigidity hypotheses 291D.4. Real wage loss/government deficit trade-offs with alternative instru-

ments 294D.5. Effects of a demand shock upon real GDP and real wages with and

without offsetting exchange rate policy 295D.6. Exchange rate policy and output/inflation trade-off after a supply

shock: Germany 298

E. 1. Standard and stochastic simulations compared 305E.2. Effect on price level of a 5% price shock in France 314E.3. Effects of a 5% price shock in float and EMU, theoretical model 316

347

sduquenoy

Date post:	11-Jan-2023
Category:	Documents
Upload:	khangminh22
View:	0 times
Download:	0 times

CD1. European Economy. Basic editions. 44/1990. One ...

Documents