+ All Categories
Home > Documents > Cyclical properties of fiscal policy in new member states of the EU

Cyclical properties of fiscal policy in new member states of the EU

Date post: 09-Dec-2016
Category:
Upload: martina
View: 212 times
Download: 0 times
Share this document with a friend
21
This article was downloaded by: [Moskow State Univ Bibliote] On: 27 September 2013, At: 01:03 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Post-Communist Economies Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/cpce20 Cyclical properties of fiscal policy in new member states of the EU Martina Dalić a a Zagreb School of Economics and Management , Zagreb , Croatia Published online: 16 Aug 2013. To cite this article: Martina Dalić (2013) Cyclical properties of fiscal policy in new member states of the EU, Post-Communist Economies, 25:3, 289-308, DOI: 10.1080/14631377.2013.813144 To link to this article: http://dx.doi.org/10.1080/14631377.2013.813144 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms- and-conditions
Transcript

This article was downloaded by: [Moskow State Univ Bibliote]On: 27 September 2013, At: 01:03Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Post-Communist EconomiesPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/cpce20

Cyclical properties of fiscal policy innew member states of the EUMartina Dalić aa Zagreb School of Economics and Management , Zagreb , CroatiaPublished online: 16 Aug 2013.

To cite this article: Martina Dalić (2013) Cyclical properties of fiscal policy in new member statesof the EU, Post-Communist Economies, 25:3, 289-308, DOI: 10.1080/14631377.2013.813144

To link to this article: http://dx.doi.org/10.1080/14631377.2013.813144

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Cyclical properties of fiscal policy in new member states of the EU

Martina Dalic*

Zagreb School of Economics and Management, Zagreb, Croatia

(Received 10 January 2013; final version received 8 March 2013)

The analysis of cyclical properties of government expenditure is performed for newmember states and Croatia using a panel data model. The sample covers 1999–2011.The analysis is performed for the main expenditure aggregates as well as theirsubcomponents. The disaggregated approach is useful in revealing potentialdifferences in the cyclical properties of individual expenditure components sincethey could move in different and possibly offsetting directions. Procyclical behaviour isfound for total general government expenditure as well as for its main components, i.e.current and capital expenditure. The exclusion of interest payments does not alter theresult that output expansion is strongly associated with growth in major expenditureaggregates. Furthermore, the same proportional reaction is found for capitalexpenditure, indicating the presence of the voracity effect. The subcomponents ofcurrent expenditure reveal a slightly different pattern. While government wage andnon-wage consumption behave in a procyclical manner, social transfers arecountercyclical. However, their countercyclical behaviour is not strong enough toovercome the procyclical influence of the wage bill and non-wage consumption on theoverall cyclical stance of current expenditure. The evidence of asymmetric behaviourof government expenditure over the cycle is weak.

Fiscal policy behaviour over the cycle has received increasing attention from researchers

in recent years. The stabilisation of cyclical fluctuations is a permanent task of fiscal policy

but the current crisis has again put the policy’s effectiveness and capacity to perform that

task under the spotlight of researchers and policy makers. The conventional wisdom is that

fiscal policy should be countercyclical and should smooth out the fluctuations of the

business cycle. However, a number of studies (Gavin and Perotti 1997, Talvi and Vegh

2000, Alesina et al. 2008, Ilzetzki and Vegh 2008) have revealed evidence of procyclical

behaviour in a number of countries, mostly developing ones.

This article presents the results of a study of the cyclical properties of fiscal policy in a

sample of new member states (NMS) of the EU and Croatia using a panel data model for

1999–2011. NMS, the countries that joined the EU in 2004 and later, have only

occasionally been included in previous research work owing to either non-existence or low

quality of data. The main reason for low data quality is related to intertemporal

comparability caused by structural breaks and rapid changes in these countries. However,

over time these problems have lost their relevance. On the other hand, since NMS have

been functioning within the same institutional framework as developed economies for

several years, it is interesting to shed some light on the issue of the (in)ability of NMS to

conduct their fiscal policies in a way similar to that used by developed economies, where

countercyclical properties are well documented (Lane 2003).

q 2013 Taylor & Francis

*Email: [email protected]

Post-Communist Economies, 2013

Vol. 25, No. 3, 289–308, http://dx.doi.org/10.1080/14631377.2013.813144

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

The focus of interest is directed on the cyclical properties of government expenditure. The

reason for this is twofold. As pointed out by Kaminsky et al. (2004), the investigation of the

cyclical properties of fiscal policy is sensible only in relation to its instruments, i.e. tax rates

and government expenditure. In that respect, deficit is an outcome produced by fiscal policy

instruments. On the other hand, apart from inherent difficulty in collecting tax rates, the search

for fiscal activism should be directed towards the expenditure side. For example, Alesina et al.

(2008) suggest that procyclicality is determined by expenditure, not taxes. Therefore, the

recognition of the cyclical properties of government expenditure represents the first step

towards understanding the reasons and causes of discretion embedded in expenditure.

The cyclical behaviour of the main expenditure aggregates as well as their

subcomponents is estimated in this study. An analysis of the main expenditure components

is potentially useful in revealing differences in the cyclical properties of individual

expenditure components, since examination of the main aggregates can be misleading if

subcomponents move in different and potentially offsetting directions. Furthermore, the

cyclical properties of expenditure subcomponents are useful for effective design of fiscal

adjustment, since the composition of spending is critical for the success of fiscal

adjustment (Ardagna 2004, Alesina and Ardagna 2009).

Following the view of Kaminsky et al. (2004) that macroeconomic policies in

developing countries are highly procyclical during times of crisis, we also examine

possible asymmetries in cyclical properties of expenditure aggregates as well as individual

subcomponents.

The structure of the rest of this article is as follows. It begins with a brief summary of

the theoretical issues related to the effectiveness of fiscal policy, and government

expenditure in particular, to achieve the stabilisation of cyclical fluctuations. This is

followed by a review of empirical evidence on cyclical properties of government

expenditure and a discussion on the causes of fiscal procyclicality. We then present the

data and estimation strategy and the panel regression results and give an interpretation of

the empirical evidence. The final section sets out the conclusions.

Theoretical aspects of fiscal cyclicality

In Keynesian theory,1 based on the assumptions of price and/or wage rigidities and

adaptive expectations, changes in aggregate demand can produce real effects on output

and employment. During the downturn, fiscal policy should smooth out the cycle by

lowering tax rates or increasing expenditure in order to increase aggregate demand and

help the economy to find the way out of the recession. During the expansionary phase

expenditure should be reduced and government saving increased. From the Keynesian

perspective there exists a clear view about the effectiveness of fiscal policy instruments, in

particular government expenditure, in achieving countercyclical effects. In other words,

the Keynesian multiplier is strictly greater than one.

Neoclassical theory, following the tax-smoothing hypothesis (Barro 1979),

recommends that tax rates should be held constant over the cycle while the budget

balance and government debt should move in a countercyclical direction. With tax rates

unchanged over the cycle, the budget balance should move into surplus during expansion

while deficits and debt growth should occur during the downturn because of a temporary

increase in expenditure and a decrease in tax revenue. The stabilisation task should be

mainly taken care of by automatic stabilisers.

Neoclassical theory is focused on the structure of government expenditure and

minimisation of distortions caused by frequent changes in tax rates (Aschauer 1988). In

290 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

that respect it is less precise about cyclical movements in government expenditure.

Government expenditure is assumed to be exogenously determined. Woodford (2010)

points out that an increase in government expenditure is not an effective way to increase

aggregate demand, while a more effective strategy may be to change the structure of

government expenditure in a way that favours expenditure with a marginal utility

approximately equal to that of private expenditure.

Lane (2003) and Ilzetzki and Vegh (2008) emphasise that an optimal path of

government consumption depends on whether there exists complementarity or

substitutability with private consumption. If government consumption is a substitute

(complement) for private consumption the expected movement of government

consumption should be countercyclical (procyclical). They also argue that similar

reasoning could be applied to the movements of transfers and capital expenditure. The

neoclassical multiplier is generally lower than one and depends on monetary policy

reactions (Hall 2009, Woodford 2010). However, Christiano et al. (2009) have shown that

in the case of zero interest rates, i.e. deep and long-lasting recession, there is a possibility

of an expenditure multiplier greater than two.

The hypothesis of Ricardian equivalence (Barro 1974), based on rational expectations

and forward-looking economic agents that recognise the government’s intertemporal

budget constraint, has led to theoretical and empirical considerations of the expansionary

effect of fiscal contraction, i.e. non-Keynesian effects of fiscal policy. Feldstein (1980)

was the first to suggest that fiscal contraction can have expansionary effects on output if it

creates expectations of future reduction in tax rates. Blanchard (1990) has developed a

model in which even an increase in taxes can have expansionary effects if it is

implemented in the context of high indebtedness. Therefore, theories of non-Keynesian

effects of fiscal policy allow the existence of a negative fiscal multiplier, particularly if an

expenditure increase is perceived as permanent and implemented in an environment of

high indebtedness or already high government expenditure (Bertola and Drazen 1993).

Giavazzi and Pagano (1990) were the first to present evidence about the possible

existence of expansionary effects of fiscal contractions. This research triggered numerous

analyses about the hypothesis of expansionary contractions (Alesina and Ardagna 1998,

2009, Giavazzi et al. 2000, Alesina et al. 2002, Ardagna 2004, Hogan 2004, Afonso 2006).2

In summary, sharp differences exist among economic theories with regard to the

effectiveness of fiscal policy instruments to smooth out the business cycle. The Keynesian

perspective implies a negative (positive) correlation between government spending (tax

rates) and output over the cycle, while in the neoclassical perspective these correlations

should be around zero. However, all of them are focused on the need to stabilise the

cyclical fluctuations in output, i.e. countercyclical behaviour would then be the optimal

recommendation.

Empirical evidence on fiscal cyclicality

The empirical evidence has shown that the ability of policy makers to use fiscal policy

instruments in a countercyclical manner differs across regions (Gavin and Perotti 1997),

levels of development (Talvi and Vegh 2000, Kaminsky et al. 2004, Calderon and

Schmidt-Hebbel 2008) and time (Frankel et al. 2011). Developed economies, on average,

are capable of pursuing countercyclical policy, while in developing countries as well as in

many emerging markets the tendencies toward procyclical fiscal policy prevail.

Lane (2003) studied the cyclical behaviour of fiscal policy in a sample of 22 developed

OECD economies. He found that current expenditure exhibited countercyclical behaviour

Post-Communist Economies 291

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

mainly because of the countercyclical behaviour of government transfers. On the other

hand, government investment is highly procyclical, which produces acyclicality of total

expenditure. Alesina et al. (2008) confirm countercyclical behaviour of fiscal policy in

developed economies while Egert (2010) points to different behaviour of expenditure

subcomponents over the cycle, indicating that, for OECD countries, the majority of them

appear unaffected by the cycle.

Gavin and Perotti (1997) were the first to provide evidence of differences in fiscal

cyclicality between industrialised and less developed economies. They showed that fiscal

policy was procyclical in Latin America mainly because of procyclical behaviour of

government expenditure. On the other hand, Talvi and Vegh (2000)3 examined a sample of

20 industrialised and 36 developing economies and pointed out that procyclicality was not

a phenomenon related exclusively to Latin America but characterised a wider group of

developing economies. Akitoby et al. (2004) also found widespread procyclicality of

government expenditure among developing countries. Kaminsky et al. (2004), Ilzetzki and

Vegh (2008) and Calderon and Schmidt-Hebbel (2008) also confirmed the counter-

cyclicality or acyclicality of government expenditure in developed economies and

procyclical behaviour of fiscal policy instruments in developing economies.

All the above mentioned studies focused mainly on government expenditure because

data on tax rates are hardly available. Notable exceptions are Talvi and Vegh (2000), who

tried to overcome this problem by using the inflation tax as a proxy for all taxes.4 Again,

countercyclical behaviour of inflation was found in developed economies while

developing countries displayed procyclical behaviour as the inflation tax was declining

during expansions and increasing during recessions. Frankel et al. (2011) showed that the

ability of countries to pursue countercyclical policies has been changing over time. Their

research shows that during the last decade some countries have lost the ability to pursue a

countercyclical policy (i.e. Greece) while some countries, like Chile or Brazil, have gained

that feature. With the very well documented tendency of developing countries to pursue

procyclical policy, researchers turned their focus towards the causes of such suboptimal

behaviour in developing economies.

Causes of fiscal cyclicality

Two complementary sets of explanations have been suggested to explain the suboptimal

behaviour of fiscal policy and, in particular, government expenditure in developing

countries. The first set is related to financial conditions faced by developing countries.

Gavin and Perotti (1997) and Aizeman et al. (2000) suggest the loss of credit market

confidence and hence an intensified borrowing constraint as a limiting factor on pursuing

countercyclical policy. Owing to credit market constraints developing economies are not

able to finance higher deficits during downturns. This kind of argument is consistent with

the consequences of sudden stops in international capital flows observed by Kaminsky

et al. (2004).

Riascos and Vegh (2003) point to the incompleteness of domestic financial markets,

while Caballero and Krishnamurty (2004) recognise low financial depth as a major reason

for procyclicality. However, Alesina et al. (2008) note that the financial market

explanation is not complete since it cannot explain why governments do not prepare

themselves for bad times by creating savings and surpluses during good times if they know

that their access to capital markets may be constrained during a crisis.

Political economy theories contribute to the second set of explanations for fiscal

procyclicality, which point to inadequate quality of institutions and political distortions as

292 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

the main causes of suboptimal fiscal outcomes. Institutional factors and political

distortions caused by weak institutions are suggested by Tornell and Lane (1999) and

Talvi and Vegh (2000). Battaglini and Coate (2008) suggest that a procyclical policy

results from political distortions caused by a struggle among regionally elected politicians

to direct government transfers to their regions, while Ilzetzki (2011) points to different

preferences of governments (alternating in power) about transfers to their electoral base.

Political corruption and rents (Alesina et al. 2008) and inadequate dispersion of

political power (Akitoby et al. 2004, Calderon and Schmidt-Hebbel 2008) lead to

procyclical fiscal results because of the government’s inability to resist spending pressures

during the good times, i.e. during times of abundant revenue. Suboptimal institutional and

political arrangements as well as inadequate credibility of the government prevent the

implementation of countercyclical fiscal policy during good times (and consequently

generation of budget surpluses) because of which the fiscal space for countercyclical

reaction (and generation of large deficits) in bad times is constrained or non-existent. Talvi

and Vegh (2000) therefore conclude that it is not the capital market that causes a

procyclical fiscal policy but the inability of governments to save surpluses generated

during good times because of political distortions and related weak institutional

arrangements under which economic policy decisions are made.

Data and methodology

Owing to the lack of data and strong structural breaks, new member states (NMS) were

only occasionally included in the previous research. Apart from an attempt to fill this gap,

the NMS sample is additionally interesting because of its economic and institutional

proximity to developed economies. During their accession to the EU these countries went

through institutional and legal alignment and all of them shared the same legal and

institutional framework with developed countries for half of the sample period. The results

will thus indicate whether the cyclical properties of their fiscal policy are closer to those in

developed economies or are more similar to those in developing countries. On the other

hand, the majority of the countries included in the sample are former communist/transition

economies, which means the results could also be viewed as indirect evidence about the

completeness of their transition processes. The list of countries included in the sample is

given in Appendix 1, which also presents a discussion of the implications that inclusion of

two small economies without a communist past (Malta and Cyprus) may have for the

results.

A balanced panel has been compiled for the sample of NMS and Croatia in 1999–

2011. The year 1999 is chosen as the first year of the sample period because during that

year the majority of the countries included in the sample began their EU accession talks. It

is also reasonable to assume that the quality of data improved afterwards. Croatia is

included in the sample since it completed its alignment with EU standards during the

period observed. The sources of fiscal data are the Eurostat data base, WDI data base (for

some control variables) and national sources for Croatia. The scope of fiscal variables is

related to general government. A detailed description of variables used in our regressions

is given in Appendix 2.

Our estimation strategy follows the standard approach outlined in the literature. As in

Fatas and Mihov (2003), Lane (2003) or Ilzetzki and Vegh (2008) the starting point is an

equation of the following type:

Dðlog f i;tÞ ¼ a1 þ a2Dðlog Yi;tÞ þ 1i;t; i ¼ 1 . . .N; t ¼ 1 . . . T; ð1Þ

Post-Communist Economies 293

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

where i (i ¼ 1 . . .N) denotes a country and t (t ¼ 1 . . .T) denotes a year. Equation (1)

captures the reaction of a particular component of government spending fi,t to

contemporaneous real output Yi,t. The coefficient a2 captures the fiscal stance of the

indicator under consideration; a positive value of a2 signals procyclical behaviour and a

negative value means countercyclical behaviour while a value equal to zero (statistically

insignificant) represents movement that is disconnected from the cycle or acyclical.

The baseline Equation (1) is extended by inclusion of additional control variables.

Lagged government size, openness and government indebtedness were included as

additional controls in order to estimate an equation of the form

Dðlog f i;tÞ ¼ a1 þ a2Dðlog Yi;tÞ þ a3ðlogFi;t21Þ þ a4DðlogOPENi;t21Þþ a5DðlogDEBTi;t21Þ þ 1i;t ð2Þ

where fi,t denotes the real value of fiscal indicator for country i during year t. Equation (2)

is estimated for the following expenditure components:

. total expenditure,

. total non-interest expenditure as total expenditure minus interest payments,

. current expenditure defined as total expenditure minus capital expenditure,

. current non-interest expenditure as current expenditure minus interest payments,

. government wage bill (wages),

. government non-wage consumption defined as current expenditure minus interest

payments minus wage bill minus social transfers,

. social transfers and

. capital expenditure defined as the sum of capital transfers and gross fixed capital

formation.

Interest payments are excluded from total and current expenditure because they cannot be

influenced by discretionary decisions. It is therefore interesting to observe the behaviour of

the respective aggregates without the influence of interest payments.

Yi,t represents the level of real GDP of country i during year t and its log difference

proxies growth in real output. Coefficient a2 is the elasticity of expenditure with respect to

output growth and represents the measure of cyclicality in the behaviour of the expenditure

category under consideration. As already noted, a negative value indicates a

countercyclical stance while a positive value of a2 indicates a procyclical stance.

F(i,t21) is the overall size of government of country i during year t 2 1, measured as the

ratio of total expenditure to GDP. Lagged government size is included to control for the

stabilising role of fiscal policy as argued by Akitoby et al. (2004). Therefore, the parameter

a3 indicates the persistence of the fiscal indicator under consideration, i.e. the speed

of its mean reversion. It is reasonable to assume that the overall size of the government

affects the way discretionary decisions on individual expenditure items are formulated.

Parameter a3 is therefore expected to have a negative value since the bigger the size

of the government the slower the expected changes in the size of individual fiscal

indicators.

OPENi,t21 represents openness measured as the ratio of exports and imports to GDP.

Lagged openness is included following Rodik (1996) and his evidence about the

government as a sector that insulates the economy from external risks and shocks.

Following Rodik (1996), more open economies, economies that accept the risk of being

294 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

open to external shocks, require bigger governments as an instrument of income

stabilisation. Therefore the sign of a4 is expected to be positive.

DEBTi,t21 is government indebtedness measured as the share of gross public debt in

GDP. Lagged government indebtedness controls for possible procyclicality driven by high

debt. The expected sign of a5 is negative, since an increase in indebtedness may lead to

more concerns about fiscal sustainability and induce government to implement more

cautious expenditure policies.

Equations (1) and (2) were estimated via a panel data model using the fixed effects

OLS estimator with the correction for first-order serial correlation in the error term. The

likelihood ratio test for the significance of fixed effects is employed and it confirms the

significance of cross-section fixed effects.

A problem related to the estimation strategy described concerns the potential reverse

causality between output growth and the dependent variable (i.e. growth in the fiscal

indicator). Estimation results provided by OLS would accurately estimate Equation (1)

and/or (2) if and only if output growth is exogenous relative to the fiscal policy indicator.

However, there is a possibility of a causal relation going from fiscal indicator to output

growth, i.e. reverse causality. On the other hand, one should keep in mind, as underlined

by Agnello and Cimadomo (2009), that reverse causality could be weak in the case of

budgetary subcomponents since their influence on economic activity could be feeble.

The issue of reverse causality is tackled by estimating the panel regression using the

instrumental variables (IV/2SLS) method. The usual problem with the IV method is the

choice and construction of an appropriate instrument for real GDP growth. Since our

sample is composed of small open economies, we follow Lane (2003) and construct the

instrument for real GDP growth as a foreign trade-weighted average of the trading partners

GDP growth rates.5

Empirical results

Table 1 presents a pair-wise correlation between fiscal indicators and their determinants. A

simple correlation analysis suggests a positive correlation between output growth and

major expenditure aggregates, indicating their procyclical stance. A positive correlation,

i.e. procyclical stance, is also indicated for government wage and non-wage consumption

while a pairwise correlation between output growth and social benefits is not statistically

significant.

Least squares

The OLS estimates for our baseline specification (1) are reported in Table 2 and indicate

the procyclical stance of all expenditure components except social benefits. The

coefficients for output growth are positive and highly significant while the coefficient in

the regression for capital expenditure is even higher than one, signalling the existence of

the voracity effect6 (Tornell and Lane 1999). Only the coefficient in the regression for

social benefits is negative but statistically insignificant.

The results reported in Table 2 serve as a benchmark for the results of our main

regression Equation (2) where additional controls are included and which are reported in

Table 3. The results for the main expenditure aggregates are shown in columns 2–6 while

the results for current expenditure subcomponents are presented in the last three columns

of Table 3.

Post-Communist Economies 295

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Table

1.

Sim

ple

correlationanalysis.Correlationsbetweendifferentfiscal

indicators

anddeterminants.Annual

data(1999–2011).

Total

expenditure

Totalnon-

interestexp.

Current

expenditure

Currentnon-

interestexp.

Capital

expenditure

Wages

Non-w

age

consumption

Social

benefits

Realoutput

(inlogdifferences)

0.255

0.419

0.19

0.218

0.279

0.237

0.316

20.049

3.41

5.96

2.50

2.89

3.77

3.16

4.31

20.65

Indebtedness

(%ofGDPin

logdifferences)

20.231

20.461

20.261

20.304

20.063

20.332

20.211

20.177

23.08

26.72

23.51

24.13

20.83

24.55

22.79

22.33

Openness

(%ofGDPin

logdifferences)

0.000

20.411

0.059

0.056

20.104

0.041

0.013

0.105

0.00

20.53

0.77

0.73

21.34

0.05

0.18

1.37

Governmentsize

(%ofGDPin

logs)

20.385

20.479

20.339

20.342

20.303

20.307

20.23

20.317

5.39

27.07

24.67

24.7

24.11

24.17

23.06

24.33

Note:t-statistics

initalics.

296 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Table

2.

Cyclicalityofgovernmentexpenditure,baselineregression.

Dependentvariable:fiscal

indicator,real

valuein

logdifferences.Estim

ation:OLS(annual

data,1999–2011).

Dependentvariable

Total

expenditure

Totalnon-

interestexp.

Current

expenditure

Currentnon-

interestexp.

Capital

expenditure

Wages

Non-w

age

consumption

Social

benefits

12

34

56

78

9

Realoutput

(inlogdifferences)

0.519***

0.600***

0.364***

0.426***

1.49***

0.566***

0.636***

20.083

(0.130)

(0.143)

(0.118)

(0.120)

(0.325)

(0.172)

(0.185)

(0.156)

� R,

20.351

0.169

0.357

0.356

0.092

0.284

0.13

0.352

Number

of

observations

169

169

169

169

169

169

169

169

Note:***,**and*indicatethat

thevariable

issignificantat

1%,5%

and10%

respectively.

Autocorrelationandheteroskedasticity

consistentstandarderrors

areshownin

parentheses.

Post-Communist Economies 297

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Total expenditure and its two major subcomponents, current and capital expenditure,

exhibit procyclical behaviour. Coefficients for output growth are positive and significant at

the 1% level in the regressions for total and capital expenditure, while the coefficient in the

regression for current expenditure is significant at the 5% level. The exclusion of interest

payments from total and current expenditure does not alter the result that output expansion

is strongly associated with expenditure growth. For example, a 1% increase in real output

would lead to a 0.38% increase in real total expenditure while real total non-interest

expenditure would grow by 0.40%.

At the same time, the coefficient for output growth in the regression for capital

expenditure is again higher than one. A 1% increase in real output causes an increase in

real capital expenditure by 1.4%. The existence of the voracity effect, signalled by more

than proportional procyclical reaction of capital expenditure to output growth, indicates

also that capital expenditure could be considered the most important tool of aggressive

discretionary fiscal policy in our sample of countries.

While the procyclical behaviour of total expenditure emerges from the procyclical

behaviour of both current and capital expenditure, the major subcomponents of current

expenditure exhibit a slightly different pattern. As shown in the last three columns of

Table 3, the wage bill (wages) and government non-wage consumption are procyclical,

while the negative coefficient on output growth in the regression for social benefits gains

significance, showing their countercyclical behaviour. The countercyclical behaviour of

social benefits is consistent with the existence of well-developed automatic stabilisers built

into this expenditure category. However, the countercyclical behaviour of social benefits is

not strong enough to overcome the procyclical influence of wage and non-wage

consumption on the cyclicality of current expenditure.

Regarding our control variables, we find evidence of a stabilising effect produced by

lagged government size, since the coefficient of lagged government size is statistically

significant and negative in all regressions. This effect is particularly pronounced in the

case of capital expenditure, confirming the presence of strong discretion associated with

changes in capital expenditure. The coefficient for growth in lagged indebtedness is

significant and has the expected negative sign for total and current non-interest

expenditure, current expenditure and for some of its subcomponents. However, in the

regression for capital expenditure the coefficient for growth in lagged indebtedness is

statistically significant but positive. It appears that an increase in government indebtedness

generally constrains the growth of non-interest expenditure and its subcomponents but in

the case of capital expenditure it seems that an increase in public indebtedness feeds

directly into its growth.

Instrumental variables

The results discussed above indicate procyclical behaviour of government expenditure

with the exception of social transfers, which exhibit countercyclical properties. However,

they do not take account of possible endogeneity of real output growth, that is, shocks to

fiscal policy that may affect real output growth. In order to take account of possible reverse

causality we instrument real output growth with weighted growth rates of the most

important trading partners. In that respect the OLS results reported in Table 3 serve as

useful benchmarks.

The IV/2SLS estimation results are reported in Table 4 and generally confirm the OLS

results. The coefficient for real output growth maintains its high significance while in the

case of current expenditure and wages the level of significance improves. The coefficient

298 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Table

3.

Cyclicalityofgovernmentexpenditure:mainregression,OLSestimates.

Dependentvariable:fiscal

indicator(realvaluein

logdifferences).Estim

ation:OLS(annual

data,

1999–2011).

Dependentvariable

Total

expenditure

Totalnon-

interestexp.

Current

expenditure

Currentnon-

interestexp.

Capital

expenditure

Wages

Non-w

age

consumption

Social

benefits

12

34

56

78

9

Realoutput

(inlogdifferences)

0.384***

0.397***

0.196**

0.258***

1.446***

0.326**

0.761***

20.212**

(0.088)

(0.058)

(0.076)

(0.074)

(0.267)

(0.155)

(0.094)

(0.090)

Governmentsize,lagged

(%ofGDPin

logs)

20.357***

20.426***

20.195***

20.224***

21.148***

20.122*

20.392*

20.185*

(0.071)

(0.77)

(0.066)

(0.070)

(0.362)

(0.064)

(0.143)

(0.100)

Openness,lagged

(%ofGDPin

logdifferences)

0.128***

20.009

20.007

0.19***

20.107

0.214***

0.032

0.179***

(0.039)

(0.044)

(0.063)

(0.051)

(0.227)

(0.045)

(0.099)

(0.068)

Indebtedness,lagged

(%ofGDPin

logdifferences)

20.049

20.067**

20.096***

20.125***

0.283**

20.219***

20.009

20.099*

(0.032)

(0.032)

(0.029)

(0.027)

(0.129)

(0.035)

(0.053)

(0.057)

� R,

20.454

0.481

0.441

0.491

0.166

0.409

0.234

0.434

Number

ofobservations

169

169

169

169

169

169

169

169

Note:***,**and*indicatethat

thevariable

issignificantat

1%,5%

and10%

respectively.

Autocorrelationandheteroskedasticity

consistentstandarderrors

areshownin

parentheses.

Post-Communist Economies 299

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Table

4.

Cyclicalityofgovernmentexpenditure:mainregression,instrumentalvariables.

Dependentvariable:fiscal

indicator(realvaluein

logdifferences).Estim

ation:instrumentalvariables(annual

data1999–2011).

Dependentvariable

Total

expenditure

Totalnon-

interestexp.

Current

expenditure

Currentnon-

interestexp.

Capital

expenditure

Wages

Non-w

age

consumption

Social

benefits

12

34

56

78

9

Realoutput(inlogdifferences)

0.355***

0.383***

0.168***

0.203***

1.551***

0.304***

0.783***

20.340***

(0.057)

(0.040)

(0.054)

(0.060)

(0.487)

(0.103)

(0.111)

(0.077)

Governmentsize,lagged

(%ofGDPin

logs)

20.357***

20.447***

20.195***

20.197***

21.148***

20.122**

20.392***

20.16*

(0.071)

(0.079)

(0.067)

(0.066)

(0.292)

(0.064)

(0.143)

(0.096)

Openness,lagged

(%ofGDPin

logdifferences)

0.127***

0.131***

20.007

0.001

20.102

0.213***

0.032

0.011

(0.039)

(0.029)

(0.064)

(0.065)

(0.247)

(0.043)

(0.098)

(0.060)

Indebtedness,lagged

(%ofGDPin

logdifferences)

20.052

20.073***

20.099***

20.116***

0.292**

20.221***

20.07

20.1*

(0.032)

(0.023)

(0.029)

(0.028)

(0.139)

(0.038)

(0.055)

(0.058)

� R,

20.454

0.502

0.441

0.463

0.165

0.409

0.234

0.409

Number

ofobservations

169

169

169

169

169

169

169

169

Note:***,**and*indicatethat

thevariable

issignificantat

1%,5%

and10%

respectively.

Autocorrelationandheteroskedasticity

consistentstandarderrors

areshownin

parentheses.

300 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

for output growth in the regression for social benefits gains in size and significance,

strengthening the conclusion that their behaviour is countercyclical. However, the

countercyclical position of social benefits is still too weak to offset the effect of the

procyclical stance of other current expenditure components. Therefore, current

expenditure maintains its procyclical stance. The existence of the voracity effect for

capital expenditure indicated by OLS is additionally confirmed.

In summary, with the notable exception of social benefits, our regression analysis

reveals the procyclical stance of total government expenditure and its main

subcomponents. The procyclical stance of total expenditure is influenced by procyclical

behaviour of both current and capital expenditure. Furthermore, the procyclicality of

capital expenditure is particularly strong. In that respect it seems that cyclical properties of

government expenditure in NMS are still much closer to those in developing than in

industrialised countries. However, the countercyclical behaviour of social benefits is

similar to that in developed economies and indicates the existence of well-developed

automatic stabilisers built into this expenditure category.

The role of business cycle asymmetries

The possible asymmetry in the fiscal response to output shocks was already outlined by

Gavin and Perotti (1997). Owing to the economic and political costs of recessions and

possible insensitivity of certain expenditure (such as unemployment compensation) to the

business cycle at high levels of economic activity it is possible that the fiscal reaction during

bad times is stronger: fiscal policy may behave differently during good and bad times.

Kaminsky et al. (2004) draw a distinction between good and bad times, defining good times

as years during which the growth rate is higher than or equal to the median growth rate. Bad

times are years during which the growth rate is lower than the median rate. Following this

definition, we define the variables for good times and bad times as follows:

GOOD_TIMESi;t ¼ 1 if Dlog ðYi;tÞ $ �Y ð3Þ

BAD_TIMESi;t ¼ 1 if Dlog ðYi;tÞ , �Y ð4Þ

where Dlog (Yi,t) is the growth rate of country i at time t and �Y is the sample median growth

rate.

Panel regression Equation (2) is extended as follows:

Dðlog f i;tÞ ¼ a1 þ aþ2 Dðlog Yi;tÞ *GOOD_TIMESi;t þ a2

2 Dðlog Yi;tÞ *BAD_TIMESi;t

þ a3ðlogFi;t21Þ þ a4DðlogOPENi;t21Þ þ a5DðlogDEBTi;t21Þ þ 1i;t ð5Þ

Panel regression Equation (5) allows us to test whether the cyclical properties of fiscal

policy during good times (a2þ) are different from those during bad times (a2

2) or whether

there is an indication that the behaviour of expenditure during bad times drives the overall

results. In order to formally test for asymmetric effects Wald tests are used. The first tests

the null hypothesis (a2þ) ¼ (a2

2), i.e. the equality of coefficients associated with good and

bad times. The second tests the linear restriction (a2þ) ¼ (a2) ¼ 0, i.e. coefficients that are

equal and jointly not statistically different from zero.

Table 5 reports the estimates of the panel model (5) using the instrumental variable

estimation procedure with the test results shown at the bottom of the table.

Post-Communist Economies 301

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Table

5.

Cyclicalityofgovernmentexpenditure:identifyingasymmetricbehaviour.

Dependentvariable:fiscal

indicator(realvaluein

logdifferences).Estim

ation:instrumentalvariables(annual

dana1999–2011).

Dependentvariable

Total

expenditure

Totalnon-

interestexp.

Current

expenditure

Currentnon-

interestexp.

Capital

expenditure

Wages

Non-w

age

consumption

Social

benefits

12

34

56

78

9

Realoutput *Goodtimes

(inlogdifferences)

0.281

0.389**

0.192

0.21

1.358*

0.051

0.512*

20.04

(0.206)

(0.146)

(0.166)

(0.170)

(0.743)

(0.231)

(0.260)

(0.185)

Realoutput *Bad

times

(inlogdifferences)

0.384***

0.354***

0.197*

0.245

1.742***

0.475**

0.774***

20.457***

(0.133)

(0.104)

20.116

(0.126)

(0.528)

(0.189)

(0.256)

(0.132)

Governmentsize,lagged

(%ofGDPin

logs)

20.365***

20.444***

20.22***

20.226***

21.170***

20.173*

20.44***

20.146

(0.071)

(0.079)

(0.083)

(0.083)

(0.373)

(0.089)

(0.168)

(0.097)

Openness,lagged

(%ofGDPin

logdifferences)

0.118***

0.12***

0.172***

0.183***

20.141

0.208***

0.174***

0.18***

(0.042)

(0.032)

(0.045)

(0.047)

(0.258)

(0.049)

(0.065)

(0.059)

Indebtedness,lagged

(%ofGDPin

logdifferences)

20.053*

20.08***

20.109***

20.129***

0.284**

20.235***

20.046

20.97*

(0.031)

(0.026)

(0.028)

(0.025)

(0.138)

(0.038)

(0.044)

(0.055)

� R,

20.447

0.488

0.461

0.485

0.164

0.405

0.23

0.438

Number

ofobservations

169

169

169

169

169

169

169

169

H0:a

þ¼

a- ;Ftest

0.119

0.005

0.000

0.019

0.133

1.88

0.406

2.022

H0:a

þ¼

a-¼

0;Ftest

10.22***

12.29***

3.84***

5.22***

10.53***

4.99***

8.73***

13.169***

Note:***,**and*indicatethat

thevariable

issignificantat

1%,5%

and10%

respectively.

Autocorrelationandheteroskedasticity

consistentstandarderrors

areshownin

parentheses.

302 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

The coefficients for real output indicate acyclical behaviour of total expenditure during

good times and procyclical behaviour during bad times. However, there is no strong

statistical evidence that the two parameters are different from each other. Although the

coefficient for real output associated with good times is statistically insignificant, while the

one associated with bad times is positive and significant at 1%, the corresponding Wald

statistics cannot reject the null hypothesis of the equality of coefficients. Jointly they are

significant at the 1% level.

Similar behaviour is found for current expenditure as well as for its subcomponents.

Statistical significance of coefficients for real output in equations for current expenditure,

current non-interest expenditure, wages and social benefits indicates a possible difference

in behaviour between good and bad times but the formal statistical test does not confirm

that. Associated Wald statistics reject the hypothesis that the two coefficients are

statistically different.

However, there is evidence that total non-interest expenditure, capital expenditure and

non-wage government consumption are procyclical in good as well as in bad years. The

coefficient for real output is significant and positive for good as well as bad years andWald

statistics cannot reject the null hypothesis of their equality.

Conclusions

Total expenditure of general government and its main subcomponents are procyclical, i.e.

output expansions are associated with expenditure growth and vice versa; a deterioration

in growth rates is associated with a decline in expenditure. Furthermore, we found

evidence of the voracity effect for capital expenditure and its disproportionate procyclical

reaction to output changes. The only exception is social transfers, which behave in a

countercyclical way. There is no firm evidence of possible asymmetric behaviour over the

cycle.

These results show that the cyclical properties of government expenditure in NMS are

more similar to those in developing countries than in developed economies. It seems that

their institutional alignment with the EU and subsequent EU membership did not create

enough pressure to change the circumstances of fiscal decision making in a direction

which would favour countercyclical behaviour. However, countercyclical behaviour of

social benefits is similar to that in developed economies, signalling probably well-

developed automatic stabilisers built into this expenditure category. Additional research is

needed to uncover the causes of the procyclicality we found but, for the time being, further

efforts are constrained by the lack of data.

Notes

1. This summary of the Keynesian theory is based on Romer (2006).2. In the light of the current crisis, Perotti (2011), one of the most distinguished researchers in this

area, reviewed the relevance of the hypothesis on expansionary fiscal contractions andconcluded that his subsequent analysis cast some doubts on the conclusions of the hypothesisand its relevance for the current time.

3. The final version of this paper was published in the Journal of Development Economics in 2005.However, the version published in 2000 as an NBER Working Paper had a profound influenceon the empirical work during the first part of the last decade so that the version from 2000 isusually quoted in the literature.

4. Gavin and Perotti (1997) also investigated the cyclical properties of inflation with similarresults. In developing economies, the inflation tax increases with output growth (i.e.

Post-Communist Economies 303

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

countercyclical behaviour) while in Latin America the inflation tax accelerates when outputgrowth is low, indicating procyclical behaviour.

5. The GDP growth rates of trading partners were weighted by the share of an individual partner’sforeign trade in total foreign trade of the country in question, normalised to one. The number oftrading partners included in the calculation was determined so as to encompass at least 60% of acountry’s foreign trade.

6. Tornell and Lane (1999) pointed out that during times of abundant revenue (good times) in acountry with weak institutions, influential (and polarised) informal groups attempt toappropriate for their own goals and interest the highest possible share of increased revenue. As aresult, with output growing, certain expenditure categories grow more than proportionately andvice versa. During the downturn, a decline in output is more than proportionally reflected in adecline in these expenditure categories.

References

Afonso, A., 2006. Expansionary fiscal consolidations in Europe: new evidence. European CentralBank Working Paper, no. 675/2006.

Agnello, L. and Cimadomo, J., 2009. Discretionary fiscal policies over the cycle: new evidencebased on the ESCB disaggregated data. European Central Bank Working Paper, no. 1118.

Aizeman, J., Gavin, M. and Hausmann, R., 2000. Optimal tax and debt policy with endogenouslyimperfect creditworthiness. Journal of international trade and economic development, 9 (4),367–395.

Akitoby, B. et al., 2004. The cyclical and long-term behaviour of government expenditures indeveloping countries. IMF Working Paper, no. 04/202.

Alesina, A. and Ardagna, S., 2009. Large changes in fiscal policy: taxes vs. spending. NBERWorking Paper, no. 15438.

Alesina, A., Campante, F. and Tabellini, G., 2008. Why is fiscal policy often procyclical? Journal ofEuropean Economic Association, 6 (5), September 1006–1036.

Alesina, A. et al., 2002. Fiscal policy, profits and investment. American economic review, 92 (3),June 571–589.

Alesina, A. and Ardagna, S., 1998. Tales of fiscal adjustment. Economic policy, 27, 489–545.Ardagna, S., 2004. Fiscal stabilizations: when do they work and why. European economic review,

48, 1047–1074.Aschauer, D.A., 1988. The equilibrium approach to fiscal policy. Journal of money, credit and

banking, 20 (1), 41–62.Barro, R.J., 1974. Are government bonds net wealth. Journal of political economy, 82, 1095–1117.Barro, R.J., 1979. On the determinants of the public debt. Journal of political economy, 87 (1),

940–971.Battaglini, M. and Coate, S., 2008. Fiscal policy over the real business cycle: a positive theory.

NBER Working Paper, no. 14047.Bertola, G. and Drazen, A., 1993. Trigger points and budget cuts: explaining the effects of fiscal

austerity. American economic review, (March), 11–26.Blanchard, O.J., 1990. A comment. In: O.J. Blanchard and S. Fisher, eds. NBER Macroeconomics

annual 1990. Vol. 5, 111–115.Caballero, R. and Krishnamurthy, A., 2004. Fiscal policy and financial depth. NBERWorking Paper,

no. 10532.Calderon, C. and Schmidt-Hebbel, K., 2008. Business cycles and fiscal policies: the role of

institutions and financial markets. Central Bank of Chile, Working Paper, no. 481.Christiano, L., Eichenbaum, M. and Rebelo, S., 2009. When is the government spending multiplier

large. NBER Working Paper, no. 15394.Egert, B., 2010. Fiscal policy reaction to the cycle in the OECD: pro- or counter- cyclical? OECD

economics department working papers, no. 753 [online], Available from: http://dx.doi.org/10.1787/5kmft7pthb27-en

Fatas, A. and Mihov, I., 2003. The case for restricting fiscal policy discretion. The quarterly journalof economics, 118 (4), 1419–1448.

Feldstein, M., 1980. Government deficit and aggregate demand. NBER Working Paper, no. 435.

304 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Frankel, J.A., Vegh, C.A. and Vuletin, G., 2011. On graduation from procyclicality. NBER WorkingPaper, no. 17619.

Gavin, M. and Perotti, R., 1997. Fiscal policy in Latin America. In: B.S. Bernanke and J. Rotemberg,eds. NBER macroeconomics annual 1997. Vol. 12, 11–72.

Giavazzi, F. and Pagano, M., 1995. Non-Keynesian effects of fiscal policy changes; internationalevidence and Swedish experience. NBER Working Paper, no. 5332.

Giavazzi, F., Japelli, T. and Pagano, M., 2000. Searching for non-linear effects of fiscal policy:evidence from industrial and developing countries. European economic review, 44 (7),1259–1289.

Hall, R.E., 2009. By how much does GDP rise if the government buys more output. BrookingsPapers on Economic Activity, (Fall), 183–231.

Hogan, V., 2004. Expansionary fiscal contractions evidence from panel data. Scandinavian journalof economics, 106 (4).

Ilzetzki, E. and Vegh, C.A., 2008. Procyclical fiscal policy in developing countries: truth or fiction?NBER Working Paper, no. 14191.

Ilzetzki, E., 2011. Rent seeking distortions and fiscal procyclicality. Journal of developmenteconomics, 96 (2), 30–46.

Kaminsky, G.L., Reinhart, C.M. and Vegh, C.A., 2004. When it rains it pours: procyclical capitalflows and macroeconomic policies. NBER Working Paper, no. 10780.

Lane, P.R., 2003. The cyclical behaviour of fiscal policy: evidence from the OECD. Journal ofpublic economics, 87 (12), 2661–2675.

Perotti, R., 2011. The ‘austerity myth’: gain without pain? NBER Working Paper, no. 17571.Riascos, A. and Vegh, C.A., 2003. Procyclical fiscal policy in developing countries: the role of

capital market imperfections [online], Available from: http://econweb.umd.edu/,vegh/papers/riascos-vegh.pdf

Rodik, D., 1996. Why do more open economies have bigger governments. NBER Working PaperSeries, no. 5537.

Romer, P.M., 2006. Advanced macroeconomics. 3rd ed. New York: McGraw-Hill/Irwin.Talvi, E. and Vegh, C., 2000. Tax base variability and procyclicality of fiscal policy. NBER Working

Paper, no. 7499. Published also as Talvi, E. and Vegh, C., 2005. Tax base variability andprocyclicality of fiscal policy. Journal of development economics, 78, 156–190.

Tornell, A. and Lane, P.R., 1999. The voracity effect. American economic review, 89 (1), 22–46.Woodford, M., 2010. Simple analytics of the government expenditure multiplier. NBER Working

Paper, no. 15714.

Appendix 1. The sample

The countries included in the sample are Bulgaria, Czech Republic, Croatia, Cyprus, Estonia,Hungary, Malta, Lithuania, Latvia, Poland, Romania, Slovakia and Slovenia.

The sample includes two small island economies, Malta and Cyprus, that joined the EU in 2004but do not share the same economic background (i.e. communist past) as the other countries includedin the sample. However, together with former communist/transition economies they also wentthrough the institutional and legal alignment with EU standards which represented an opportunity toimprove and strengthen the institutional arrangements of fiscal policy making. Therefore, they areincluded in the sample to capture the possible effects of the unique opportunity all of these countrieshad to improve their fiscal policy making. Furthermore, for the same reason Croatia is included in thesample too.

Malta and Cyprus are small economies which account for 0.07% of the combined sample GDP.Furthermore, regression Equation (2) was estimated for a sample from which Malta and Cyprus wereexcluded. As indicated in the Tables A1.1 and A1.2, when these two countries are excluded from thesample the significance and sign of the coefficients is generally unchanged (in particular thecoefficients for output growth and government size) and therefore we present the results for the fullsample of NMS.

Post-Communist Economies 305

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Table

A1.1.

Cyclicalityofgovernmentexpenditure:adjusted

sample,OLSestimates.

Dependentvariable:fiscal

indicator(realvaluein

logdifferences).Estim

ation:OLS(annual

data,

1999–2011).Sam

ple:NMSwithoutMalta

andCyprus.

Dependentvariable

Total

expenditure

Totalnon-

interestexp.

Current

expenditure

Currentnon-

interestexp.

Capital

expenditure

Wages

Non-w

age

consumption

Social

benefits

12

34

56

78

9

Realoutput

(inlogdifferences)

0.429***

0.449***

0.210***

0.260***

1.601***

0.351**

0.763***

20.223**

(0.081)

(0.083)

(0.068)

(0.067)

(0.270)

(0.138)

(0.068)

(0.093)

Governmentsize,lagged

(%ofGDPin

logs)

20.413***

20.408***

20.219***

20.224***

21.428***

20.104

20.423*

20.188*

(0.061)

(0.058)

(0.071)

(0.070)

(0.447)

(0.072)

(0.155)

(0.102)

Openness,lagged

(%ofGDPin

logdifferences)

0.143***

0.054

0.000

0.007

0.384

20.013

0.033

0.003

(0.043)

(0.079)

(0.070)

(0.072)

(0.220)

(0.115)

(0.111)

(0.070)

Indebtedness,lagged

(%ofGDPin

logdifferences)

20.028

20.019

20.089***

20.102***

0.461**

20.197***

20.003

20.084

(0.036)

(0.043)

(0.032)

(0.032)

(0.176)

(0.039)

(0.067)

(0.059)

� R,

20.481

0.488

0.460

0.486

0.243

0.396

0.267

0.424

Number

ofobservations

143

143

143

143

143

143

143

169

Note:***,**and*indicatethat

thevariable

issignificantat

1%,5%

and10%

respectively.

Autocorrelationandheteroskedasticity

consistentstandarderrors

areshownin

parentheses.

306 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Table

A1.2.

Cyclicalityofgovernmentexpenditure:adjusted

sample,instrumentalvariables.

Dependentvariable:fiscalIndicator(realvaluein

logdifferences).Estim

ation:instrumentalvariables(annualdana1999–2011).Sam

ple:NMSwithoutMaltaand

Cyprus.

Dependentvariable

Total

expenditure

Totalnon-

interestexp.

Current

expenditure

Currentnon-

interestexp.

Capital

expenditure

Wages

Non-w

age

consumption

Social

benefits

12

34

56

78

9

Realoutput(inlogdifferences)

0.412***

0.461***

0.176***

0.220***

1.868***

0.349***

0.774***

20.326***

(0.058)

(0.057)

(0.046)

(0.040)

(0.223)

(0.072)

(0.100)

(0.063)

Governmentsize,lagged

(%ofGDPin

logs)

20.413***

20.408***

20.219***

20.223***

21.143***

20.104

20.423*

20.188*

(0.060)

(0.058)

(0.071)

(0.069)

(0.452)

(0.071)

(0.155)

(0.098)

Openness,lagged

(%ofGDPin

logdifferences)

0.142***

0.053

0.002

0.008

0.373

20.013

0.032

0.007

(0.041)

(0.079)

(0.070)

(0.072)

(0.214)

(0.119)

(0.108)

(0.069)

Indebtedness,lagged

(%ofGDPin

logdifferences)

20.03

20.018

20.092***

20.106***

0.487**

20.197***

20.004

20.094*

(0.036)

(0.044)

(0.032)

(0.032)

(0.192)

(0.039)

(0.073)

(0.057)

� R,

20.481

0.488

0.460

0.485

0.241

0.395

0.267

0.421

Number

ofobservations

143

143

143

143

143

143

143

143

Note:***,**and*indicatethat

thevariable

issignificantat

1%,5%

and10%

respectively.

Autocorrelationandheteroskedasticity

consistentstandarderrors

areshownin

parentheses.

Post-Communist Economies 307

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013

Appendix 2. Description of variables

All fiscal variables, except those for Croatia, are from the Eurostat data base available from http://epp.eurostat.ec.europa.eu/portal/page/portal/statistics/search_database. National sources andauthor’s calculations are used for Croatia. All fiscal variables are related to general governmentand correspond to the ESA95 statistical standard.

. Growth rate of real total expenditure. Real total expenditure is calculated as total expenditureof general government in national currency deflated by consumer prices (2005¼1). Growthrates are calculated as the difference in the logarithm of real total expenditure in nationalcurrency.

. Growth rate of real current expenditure. Total current expenditure of general government innational currency is calculated as the difference between total expenditure and capitalexpenditure in national currency and then deflated by consumer prices (2005 ¼ 1). Growthrates are calculated as the difference in the logarithm of real current expenditure in nationalcurrency.

. Growth rate of real total non-interest expenditure. Total non-interest expenditure of generalgovernment is calculated as the difference between total expenditure of general governmentand interest payments in national currency and deflated by consumer prices (2005 ¼ 1).Growth rates are calculated as the difference in the logarithm of real total non-interestexpenditure in national currency.

. Growth rate of real current non-interest expenditure. Current non-interest expenditure iscalculated as the difference between current expenditure of general government (calculatedas explained above) and interest payments in national currency. Current non-interestexpenditure is then deflated by consumer prices (2005 ¼ 1) to compute real current non-interest expenditure. Growth rates are calculated as the difference in the logarithm of realcurrent non –interest expenditure in national currency.

. Growth rate of real government wages. Government wage bill in national currency is deflatedby consumer prices (2005 ¼ 1) to compute real government wages. Growth rates arecalculated as the difference in the logarithm of real government wages in national currency.

. Growth rate of real government non-wage consumption. Government non-wage consumptionis calculated as current expenditure of general government in national currency minusgovernment wage bill minus interest payments minus social transfers. The real values arecomputed using consumer prices (2005 ¼ 1) as deflator. Growth rates are calculated as thedifference in the logarithm of real non-wage consumption in national currency.

. Growth rate of real social benefits. Real social benefits are computed as social benefits innational currency deflated by consumer prices (2005 ¼ 1). Growth rates are calculated as thedifference in the logarithm of real social benefits in national currency.

. Growth rate of real capital expenditure. Capital expenditure is computed as the sum of capitaltransfers and gross fixed capital formation. The resulting figures are then deflated byconsumer prices (2005 ¼ 1) to compute real capital expenditure in national currency. Growthrates are calculated as the difference in the logarithm of real capital expenditure in nationalcurrency.

. Indebtedness. Calculated as a share of gross public debt in GDP. The source of data is theEurostat data base.

. Growth rate of real output. Calculated as the difference in the logarithm of real GDP inconstant local currency units (with 2005 as the base year). The source of data is the Eurostatdata base.

. Government size. Calculated as the share of total government expenditure in GDP.

. Openness. Calculated as the share of the sum of imports and exports in GDP. The source ofdata is WDI available from http://databank.worldbank.org/data/home.aspx.

308 M. Dalic

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

01:

03 2

7 Se

ptem

ber

2013


Recommended