THEWORLD BANK POLICY PLANNING AND RESEARCH STAFF Infrastructure and Urban Development Department Report INU 2 Transport in the Input-Output System by Esra Bennathan and Mark Johnson August 1987 Discuission Paper This is a doc,jment published informally by the Woild Bank. The views and interpretations herein are those of the author and should not be attributed to the World Bank, to its affiliatrcd organizations, or to any individual acting on their behalf. Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized
Transcript
THEWORLD BANK
POLICY PLANNING AND RESEARCH STAFF
Infrastructure and Urban Development Department
Report INU 2
Transport in the Input-Output System
by Esra Bennathanand Mark Johnson
August 1987
Discuission Paper
This is a doc,jment published informally by the Woild Bank. The views and interpretations herein are those of the author and
should not be attributed to the World Bank, to its affiliatrcd organizations, or to any individual acting on their behalf.
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
INFRASTRUCTURE AND URBAN DEVELOPMENT PAPERS, Report INU 2
TRANSPORT IN THE INPUT-OUTPUT SYSTEM
by Esra Bennathanand Mark Johnson
DISCUSSION PAPER
Copyright @ 1987The World Bank1818 H Street, N,,W.Washington, D.C. 20433, U.S.A.
All Rights ReservedFirst Printing August 1987
This is a document published informally by the World Bank. In
order that the information contained in it can be presented with the least
possible delay, the typescript has not been prepared in accordance with the
procedures appropriate to formal printed texts, and the World Bank accepts
no responsibility for errors.
The World Bank does not accept responsibility for the views
expressed herein, which are those of the author(s) and should not be
attributed to the World Bank or to its affiliated organizations. The
findings, interpretations, and conclusions are the results of research
supported by the Bank; they do not necessarily represent official policy of
the Bank. The designations employed, the presentation of material, and any
maps used in this document are solely for the convenience of the reader and
do not imply the expression of any opinion whatsoever on the part of the
World Bank or its affiliates concerning the legal status of any country,
territory, city, area, or of its authorities, or concerning the
delimitations of its boundaries or national affiliation.
INTRODUCTORY NOTE
Structural analysis of industries in terms of their Input-Outputrelations may seem rather an oldfashioned pursuit but, to our knowledge, ithas not so far been done for the transport industries of developingcountries. This note is a modest first attempt in that direction. It ismodest in that it uses Input-Output tables essentially for descriptivepurposes, and also in the small number of countries covered. The resourcesavailable for it were limited. Most of the spadework was done by Mr. MarkJohnson, then a graduate student of economics at Princeton University, duringthe 3 months of 1985 that he spent as summer intern at the World Bank.
We are grateful to colleagues at the World Bank who helped us togain access to Input-Output materials. Mr. Alan A. Armstrong of theUniversity of Bristol, and Mr. Claiborne M. Ball of the U.S. Department ofCommerce, were kind enough to help us with information and advice.
ABSTRACT
The object.
1. This note describes the links between the transport industry andother economic sectors of production or demand, in COte d'Ivoire, India,Mexico, Philippines and the United States. The account is based on recentInput-Output tables of those countries. Its object is:
first, to indicate the extent to which the cost of transportservices is affected by the prices of other goods and services,required as inputs in the production of transport, and by theprices and productivity of the labor and capital employeddirectly by the industry; the effect of different tax systems onthe cost of transport services; the dependence of Transport onindustrial (intermediate) demand and on demand generated byprivate and government consumption, investment and exports('Final Demand'); and the extent to which the demand fortransport reacts to changes in aggregate output;
further, to demonstrate with realistic examples, ways in which I-O tables can be used to predict input requirements of thetransport industry, and of the demand for its output; theprecautions necessary in such exercises and in internationalcomparisons, and the meaning of different concepts of therelative 'total size' of a country's transport industry.
Input-Output accounting and issues relevant to Transport.
2. The note reviews in non-technical terms, the main features of I-0accounting, and its relation to National Income accounting. It is usual fornational accounts to be constructed with the help of relations discovered inI-0 accounts, so that the same qualifications attach to inferences drawn from
either. Input-Output accounts rarely attempt to credit Transport with theoutput of own-account transport operations. Value Added in Transport, andTotal Output of Transport are therefore typically understatements of the truevalue of a country's total transport activity. Further, the Transport costcomponent of a transaction between two industries could be debited to eitherthe buyer or the seller, as a purchase from Transport. Different conventionsare followed in this matter and the difference affects the apparent transportrequirements of different industries. Such differences in methodology orpractice need to be borne in mind when international comparisons are to bemade, whether of Input-Output data of different countries or of the Transportcontribution to their GDP.
Transport as a user of resources.
3. Considering first the share of intermediate inputs in the totalvalue of Transport output, differences between the 5 countries are partlyexplained by differences in prices, including the cost of labor per unit ofoutput.
4. As a broad generalization from a narrow base, the gross output ofthe transport industry is about 1.8 times Value Added in Transport.
5. The share of Value Added (the value of labor and capital iervicesemployed directly by the industry) in the cost of total output is typicallygreater for the transport industry than for the average of all industry. Arelatively light share of intermediate inputs in the cost of producingtransport services tends to insulate Transport from movements in the pricesof traded goods in the economy, and thus from the effect of competitiveforces acting on those prices. When added to the relative absence of directexternal competition with a country's transport industry, this relativeinsensitivity to movements in prices and productivity of other industriescauses the cost of transport to be dominated by the efficiency with which theindustry uses its own resources. This 'structural' status of Transport isrelevant of the effect of the competitive regime under which it operates.
6. By the same token, Transport is less sensitive than other industryto the 'cascading' effect of turnover taxes. The cost of transport to users,and the organization of the transport industry, should therefore be lessaffected, and benefit less, than other industry from the trend towards ValueAdded taxation. The change from one system to the other may indeed raise therelative price of transport services in the economy.
Demand for transport.
7. In Mexico and the USA, Transport sells the larger part of its outputto the Final Demand sectors, and the lesser part, to other industries forintermediate use. The converse is true for the 3 countries with relativelylower incomes per head: C6te d'Ivoire, India and Philippines.
S. The average of the shares of transport cost in the total outputvalue of industries is found in all 5 countries to be quite close to 2percent. While the average direct transport input coefficients of industriesis thus quite similar, this is not true of the 'total' coefficient whichrepresents the shares of different inputs required per unit (Dollar) increase
in each industry's sales to the Final Demand sector. In the case of therequired transport inputs, the total demand coefficients are significantly
higher than the direct coefficients which measure input requirements forprevailing output levels. If future demand for transport is predicted fromthe direct coefficients the error will be significant whenever the country'stotal output is expected to change significantly.
9. To sustain a unit increase in each industry's sales to Final Demand,Transport has to raise its output by more than the average industry. Demandfor transport is thus more sensitive to a general increase (or decrease) ingeneral production than the average industry.
Indicator of the relative size of Transport activity.
10. The note concludes with a discussion of the meaning of differentmeasures of the relative 'size' or importance of Transport in the nationaleconomy: Value Added; Gross Output; Final Demand, and the index ofsensitivity to general output changes. Each of these provides an answer to adifferent question.
I. INPUT-OUTPUT
1. This note describes the links between Transport and other sectorsof the economy, as they appear from the input-output tables of 5 countries:C6te d'Ivoire, India, Philippines, Mexico and the United States. To seewhat meaning can be given to such data, some readers may be helped by abrief, non-technical rehearsal of the main features and uses of input-output accounting and the chief points to be borne in mind whlen evaluatingI-0 information on Transport. A full statement of theory and practice willbe found in U.N. (1973).
The framework in outline.
2. To fix the mind, Fig.l presents the under.lying accountingframework of I-0 analysis, in rudimentary form. The essentials of themethod become clearest if one starts by thinking of a closed economy,without foreign trade, and moreover one in which the output of an industrycan be unambiguously identified with one well-defined commodity or service.
Fig.l: The Framework
Production aectors Final demand Totals
Production A F Qsectors
Primary Value Addedinputs (+ Non-Comparable Imports)*
Totals Q
* The I-0 tables of some countries record Non-Comparable Imports as partof Primary Tnputs. See Para.7 below.
-2-
3. The matrix (A) at the upper left corner of the scheme records theflows of intermediate goods and services, traded between differentproducing units, called industries. It can be a commodity x commoditymatrix (commodity i as an input -into commodity j); a commodity x industrymatrix (commodity i into the production of industry j), or an industry xcommodity matrix, or an industry x industry matrix. We concentrate on thelast type in which each industry has a row and a column, arrayed in eitherdirection in the same sequence. The crucial convention on dimension isthen that each industry distributes its output along its row: to otherproducers within matrix A, and to the Final Demand sectors (F). In theabsence of foreign trade, Final Demand consists essentially of Private andGovernment Consumption, and Investment.
4. With industry sales in the rows, industry columns record theinputs into each industry. They consist of intermediate goods and services(recorded in A), and of Primary Inputs. Primary inputs consist of theinduistry's own resources which contribute its Value Added: wages and grossprofits, and of indirect taxes that make up the difference between value tothe producer and value to the buyer.
5. In the absence of imports, row tot.als must equal column totals,for each industry (in Q): the value of an industry's sales (rows) equal thetotal, value of its output (columns).
6. The simplicity of this rudimentary scheme is destroyed by theexistence of foreign trade, and by the heterogeneity of most industries'output . Each of these complications matters for the empirical part ofthis note. Different I-0 tables deal with imports in different ways andthis affects the comparability of different tables, across countries. Theproblem of heterogeneous output is relevant because many industries besidesTransport produce transport services for their own account. We turn firstto the consequences of foreign trading, leaving the problem of cwn-accounttransport operations to later discussion (para. 14 below).
7. No problem arises from the existence of exports: they are usually,and simply, treated as a Final Demand sector, with a separate column intowhich each industry (in an industry x industry scheme) contributes thevaLue of its exports. Imports, however, infiltrate both the inter-industryexchanges in matrix A (as intermediate goods and services) and the FinalDemand sectors. They are subject to very different treatment in different1-0 tables. A widely but by no means universally adopted procedure is tosplit imports between (a) Comparable (or Competitive) Imports -- goods orservices similar in nature to domestic production, and (b) Non-Comparable(or Non-Competitive) Imports. Non-Comparable Imports are then treated as aspecial 'industry' with its own row (distributing to intermediate uses andFinal Demand), but no corresponding column. Comparable Imports, on theother hand, are merged with the outputs of the similar (comparable)domestic industries which therefore distribute their product, domestic orforeign imported, along their rows. What matters for the purpose of thisnote is that this method is not generally implemented. In some tables,Non-Comparable Imports are classified as a Primary Input (see note toFig.l); in others, they are treated as an 'industry' (i.e., left in matrixA of Fig.l). In yet others, the distinction between the two classes of
-3-
imports is not drawn, or drawn on principles unlike those followedelsewhere. This limits the comparability of tables, the more so because itis sometimes not clear what method was followed.
8. One further aspect of the treatment of imports in I-0 tables isrelevant to the connection between 1-0 and national accounting (para.9belo,w). In the method described in the previous paragraph, imports aretreated as a sector -- i.e., a column -- of Final Demand, with negativeentries. Imports are thus deducted from each industry's sales of 'its'product: row totals are clear of imports. In the columns of the inter-industry exchanges matrix (A), however, no deduction is made for importedinputs; column totals represent the total value of industry output,including the value of all inputs, imported or otherwise. Not all tablesfollow this scheme which is shown graphically in Fig.2; some have noImports column in Final Demand. But whether or not imports are deductedfrom total sales value, the presence of imports breaks the equality betweenrow and column totals: some industries buy more imports than they sellalong with their own 'comparable' product; others distribute more importsthan they absorb.
Relation to national accounting.
9. In most national accounts one finds an analysis of GDP accordingto its industrial origin. This is therefore a readily available statisticand the simplest measure of the relative contribution of a sector tonational product (though not, of course, a measure of 'total resources'devoted to the production of that sector's output -- see para. 42 ffbelow). In terms of accuracy or completeness, however, this measure issubject to the same qualifications that attach to I-0 totals of 'industry'output or Value Added: industrial origin analysis in the national accountsis typically based on the results of I-0 accounts.
10. In broadest terms, the relationship between I-O accounts andnational accounts is:
GDP = VALUE ADDED BY INDUSTRIES = FINAL DEMAND - IMPORTS
The GDP of the national accounts is thus identical (in principle) with thetotal of Value Added in t,ie I-0 account. The industrial origin analysis isthe same in each. Fig.2, which reflects U.S. methodology, demonstrates therelationship. (Department of Commerce, 1984.)
Three basic tables.
11. I-0 accounting subjects its data to essentially three differenttreatments, each giving rise to a basic table.
12. (i) The transactions matrix of inter-industry exchanges recordsthe absolute values of goods and services flowing between industries in theaccounting period (one year). It is, of course, bordered by the blocks forFinal Demand, Primary Inputs and the two Totals (Fig.l). The sources ofthese data are censuses of production, the less common censuses ofdistribution, special studies, fiscal records and special estimatingefforts.
4-
Fig.2: The relation of GNP to the I/O account.
Gs National Product
in tf Nae on inci and Pro-duct Accounio In an h*u-Ouw Fotmat
13. Several major decisions have to be taken when making this table.One of these concerns aggregation. The way in which industries,commodities or producing units are aggregated affects the results andmeaning of any analysis based on. the table. In an industry x industryscheme, changing the method of aggregation will change the gross output ofindustry, just as would vertical integration of industries (e.g., sugar
refineries integrating with sugar plantations). It is quite common forTransport to be aggregated (in I-0 as well as national accounting) with
warehousing, and frequently also with communications. The result is a lossof information on Transport, throughout the (output) rows and the (input)
cclumns of the aggregate industry. A nuisance of special significancearises at the crossing of row and column, in the cell (the diagonal
element) that records intra-industry transactions. The size of the entry
in this cell is particularly sensitive to purely organizational features ofan industry: whether the railway contracts with independent road hauliersor merges with them, payments arising in the former case and vanishing inthe latter. Since such features of an industry, and changes in them, do
not have any necessary technological or economic meaning, one may wish to
remove these entries at the row-and-column crossing from the table forpurposes of analysis, and this was done for the empirical part of thisnote. If warehousing and communications are thrown in with Transport,however, the contents of the diagonal cell refer p4rtly to the results of
otherwise irrelevant organizational peculiarities of the Transportindustry, and partly to relevant exchanges between essentially different
activities (e.g., road haulage working on contract for Post Office).
14. A decision has further to be taken on the treatment of an
industry's 'secondary products'. Industries are normally defined by their
products, but many firms produce a variety of ;hose, including some that
are the principal products of other industries. Coke from gasworks is the
usual textbook example. But an equally important one is the production of
transport services, by firms outside the transport industry, for their own
account. The general treatment of 'secondary production' varies between I-O accounts of different countries. The recommended method is to define
principal products for an 'industry', split off 'secondary products' and
'redefine' (i.e., transfer) them together with the corresponding inputs, to
the industry that has them for its principal product. But the principlefollowed differs between countries, and the practice is likely to differ
even more. Comparability between different tables is especially affected
in the case of transport because the principle adopted in a country for thetreatment of secondary production is not necessarily extended to own-
account transport. The United Kingdom statisticians seek to transfer own-account transport operations to the Transport industry: the United States
Commerce Department does not. In the empirical part of this note we have
to assume that the U.S. method prevails in our various tables. An
immediate consequence of this omission is that the Value Added in
Transport, and its Gross Output as recorded in the tables underestimate the
total value of transport activities in the natural (not the organizational)sense of the word. Furthermore, the use of transport service by differentindustries will be underestimated whenever they operate transport on own-account.
-6-
15. A further major problem concerns the prices at which transactionsare valued: net of indirect taxes ('basic prices'), or at producer prices(including tax) or at purchaser's prices (including taxes, trade andtransport margins)? Most of the- tables that we examine opt for producerprices.
16. The treatment of transport cost in the prices at sale or purchase,requires yet another decision. The principle mcast widely adopted and,followed so far as feasible, is to debit the cost of transport to the buyerof the product, as a notionally separate purchase made from Transport (thef.o.b. method). As explained in para.14 above, we have to assume that thecosts thus transferred are solely those of transport (purchased from thetransport industry as defined in the I-0 classification).
17. (ii) The direct coefficient matrix, derived from the transactionsmetrix, records in its columns the inputs that the industry receives fromother industries, each expressed as a proportion of the receivingindustry's gross outeut. These are the industry's direct inputcoefficients: its direct intermediate requirements per units of its owngross production; in practice, per Dollar of its gross output.
18. (iii) The total coefficient matrix is obtained from the directcoefficient matrix, technically by matrix inversion, that is, by solvingthe simultaneous equations underlying the 1-0 account. The coefficientsthat emerge, in any one industry column, express the inputs from eachsector per Dollar of that industryy's gross output, for given final demands;account being taken of direct as well as indirect inputs: the fuel absorbedby Transport to carry the fuel, etc.
Uses and basic assmptions.
19. Input-output analysis has developed massively, from Leontief(1951) on. It quickly became a central tool in planning and forecastingand many types of policy analysis. Its use in international comparisons ofthe structure of production and the study of development was largelypioneered by Chenery (e.g. with Watanabe, 1958; with Clark, 1959).
20. The use of I-0 accounting for any of these purposes, and even forthe modest descriptive purpose of this note, is conditional on certainassumptions. Without them, the coefficients in the tables cannot bethought of as 'requirements' and 'requirements of specific inputs'. Thefirst basic assumption ishomogeneity of industry output: this ha; to beaccepted if the entries in I-0 tables are to be treated as technicalcoefficients rather than as coefficients thrown up by mere commercialarrangements or industrial organization. The second basic assumption isthat of proportionality of inputs to output. The postulated system lslinear, each input into a particular sector being assumed to vary in directproportion with that sector's output. It is, of course, known that suchstrict proportionality characterizes only a limited class of productionprocesses. The more usual assumption in economics is that inputs aresubstitutable (so that the unit cost of an output is a strictly concavefunction of the price of any one input). Nor can one accept constantreturns as an empirically valid condition of production in general. It
-7-
follows that the expected error in projections of input requirements for aspecified level of output, based on a given set of I-0 coefficients, willbe greater, the larger is the specified change in final demands, or thegreater the possible change in relative prices. Major abuses of I-0analysis become possible when this caveat is ignored or forgotten. As oneinstance of many, a study in the 1970s, working with fixed I-0coefficients, came up with truly alarming consequences to U.S. product andemployment from a 15 percent reduction of the output of the U.S. portindustry.
II. TRANSPORT
Relevant questions.
21. In order to throw light on the rcle of transport in the economy weshall use in'put-output data in the most obvious ways:
(i) To establish the input requirements of Transport, the sourcesof inputs and the balance between intermediate inputs andprimary inputs. The answers should point to some of the majordeterminants of the cost and efficiency of the activity andthe impact of different tax systems on transport.
(ii) To establish the destinations of transport services, thesources of demand for transport and the balance betweenintermediate and final demand. The answers are relevant tothe relative degree of sensitivity of the demand for transportto variations in national output.
We seek the answers to these simple questions with the help of relativelysimple tools of industrial structure analysis, formulated mainly by Chenerysome thirty years ago. More sophisticated and complex experiments would beneeded to explore the consequences for transport of technological change orgeneral economic growth. One would also need more data than are availableto us, in the tables for five very different countries. They areheterogeneous in more than one respect and while the mind strives naturallyto find reasons for differences between the data of the differentcountries, such rationalizations have to be viewed with caution.
The data.
22. The following is a summary description of the five sets of input-output tables that were reviewed for this note. For each, we note the yearof reference, the definition of the sector, the share of Transport in GDP(GNP) fiom the national accounts, the number of sectors distinguished inthe inter-industry matrix, the definition of the prices used for valuingthe flows of inputs or outputs, and special. features of the tables.
(i ) USAYear: 1977Sector definition: Transpurt and WarehousingGDP share of Transpott & Communications: 6.4Z, market pricesNumber of sectors distinguished: 85Valuation: at market prices (including tax on inputs)Value Added: includes taxes on sector output.
(ii ) PHILIPPINESYear: 1974, updated to 1981Sector definition: distinguishes 5 transport sub-sectors:busline, other public passenger transport, land freight,water, airGDP share of Transport and Communications: 5.4Z, market pricesNumber of sectors distinguished: 31Valuation: at market prices including tax on inputs. ValueAdded includes taxes on sector outputSpecial feature: the object of the table was the analysis ofenergy requirements. The energy (and fuels) and Transportsectors were therefore anatyzed in considerable detail.
iii ) IndiaYear: 1968/69, updated to 1979/80 pricesSector definition: distinguishes 2 sub-sectors:. rail, otherGDP share of Transport & Communications: 5.8Z, market pricesNumber of sectors distinguished: 89Valuation: at Factor Cost; indirect taxes included in PrimaryInputs (with Value Added)Special feature: high level of disaggregation of agriculturalsector.
(iv) COte d'IvoireYear: 1978Sector definition: Transport & CommunicationsGDP share of Transport & Communications: 9.6Z at Factor CostNumber of sectors distinguished: 33Valuation: market prices, including taxes on inputs. Value
Added includes taxes on output.
(v) Mexico
Year: 1975, transaction matrix updated to 1978 pricesSector definition: TransportGDP share of Transport & Communications: 6.3Z, market pricesNumber of sectors distinguished: 72Valuation: Basic prices (exclude taxes); indirect taxesincluded in Primary Inputs.
-9-
This account of the data used for this note points to their heterogeneity
- in the definition of the sector that includes Transport,- in the relation of the time to which the underlying production
data refer and the date of the prices applied to them,
- in the extent of disaggregation of the economy,- in the rule for valuing goods and services.
We ought to add at once the highly probable differences in the treatment ofimports in the different tables (see para. 23 below).
Uniform treatment of data
23. To improve comparability somewhat, we subjected our data to
uniform treatment in two respects:
(i) Intra-sectoral transactions -- in the diagonal cells of theinter-industry exchange table, at the crossing of the rows andcolumns for Transport -- were eliminated from the tables. Thereason was given in para. 13 above.
(ii) Where tables identified Non-Comparable Imports they wereincluded in Primary Inputs, with Value Added by the industryreceiving these imports; the technical reason being that twoof the tables merged this class of imports with Value Added(see para. 7 above). In the case of C6te d'Ivoire thisprocedure could not be followed because imports are notseparated into Comparable and Nlon-Comparable. Total imports,by industry, are given separately and when added into PrimaryInputs, they are found to form an improbably large proportionof the sum. The alternative is to assign Imports wholly toIntermediate Inputs and this was the course adopted.
Transport as user of resources
24. We employ two measures to indicate the dependence of Transport on
intermediate inputs (and thus, by implication, on Primary Inputs, measuredby the complement of the percentage share of intermediates). Forconvenience we code-name them U and D:
(i) U - The proportion of direct intermediate inputs in the grossvalue of output by Transport. The measure is computed fromthe direct input coefficients and was formulated by Cheneryand Watanabe (1958) for purposes of comparative structuralanalysis of industry. It is known also as the backwardlinkage coefficient because it reflects the extent to whichone activity is a source of demand for the outputs of otheractivities (Hirschman, 1958).
(ii) D - The index of the power of dispersion. It is computed from
the total coefficient matrix, as thie ratio of the average ofTotal Input Coefficients in the Transport column (the
- 10 -
'production function' of Transport) to the average of allTotal Input Coefficients in the matrix. It therefore measuresthe relative input-dependence of Transport, allowing for allthe circles of intex-industry exchanges. If D exceeds 1,Transport is a relatively intensive user of intermediateinputs; if D = .5, a Dollar increase in the country's finaldemand generates only half as much additional demand forintermediate inputs for Transport as for the country' averageindustry.
25. Table 1 shows the U and D indexes for the Transport sectors orsub-sectors identified in our 5 tables. It also shows the U index for eachcountry's total industry (col.2), and the weights (col.1) signifying therelative importance of Transport sub-sectors in the two countries(Philippines and India) for which no sector aggregate are available.
Table 1: Intermediate input req~uirements of all Industryand of Transpo.rt
Weights: U - Share of intermediate inputs D - RelativeZ of Value in gross output total depen-Added by dence onspecified ALL INDUSTRY TRANSPORT intermediatesector inputs
Sector Subsector(1) (2) (3) (4) (5)
USA, 1977 .4896transport &warehousing 100 .3229 .8438
Note: * Imports, not separated into Comparable and Non-Comparable, left wholly with intermediate inputs.
See para. 23 above.
The level of intermediate input shares.
26. Considering first the average share of intermediate inputs in grossoutput value, across all industries, one finds a notable degree ofsimilarity between countries, Mexico being the exception. The intermediateinput share in the first 4 countries lies between 44 and 51 percent, aboutthe level that one tends to find in most industrial or industrializingeconomies.
27. There is much less agreement between the shares of intermediateinput in the Transport sectors: the U values for U.S.A. and Mexico lyingwell below those for Philippines, India and COte d'Ivoire.
The role of relative prices.
28. Some part of the contrast noted in the last paragraph can beexplained by differences in relative prices. We consider specifically therelative prices of fuels and of labor. Table 2 compares the fuel inputcoefficients of Transport as recorded in our 5 tables with the prevailingprices for the main fuels. The two petroleum-rich countries stand out in
terms of both low fuel input coefficients (per unit of Transport output) andlow domestic fuel prices, kept at the time of the I-0 table below worldprices. If thet fuel input-coefficient for Mexico (.06063) is doubled,bringing it closer to the corresponding coefficients for the Philippines,India or the COte d'Ivoire, the share of intermediate inputs for theTransport sector would rise from the 0.2773 in Table 2, to 0.3373. Mexicothus underestimates gross output from Transport.
29. The relative price of labor is more of a structural fact than therelative price of diesel fuel. The low share of intermediate inputs in thegross output of U.S. Transport (Table 1, col.3) may be explained in part byrelatively high labor cost in the U.S.A. A low share of intermediate inputsimplies a high share of Primary Inputs within which labor accounted in theU.S.A. for 67 percent. The average share of labor in Primary Inputs, acrossall industry, was only 59 percent. The labor cost element in the cost of
U.S. Transport is thus higher than in U.S. industry in general and thisshould raise the share of Primary Inputs and depress that of intermediateinputs in gross Transport output. Relatively high labor cost and low fuelprices should thus explain some (unknown) part of the comparatively lowshare of intermediate inputs in U.S. Transport product. It is noteworthythat the 1979 input-output accounts of the United Kingdom -- also a countrywith relatively high labor cost, but with much higher relative fuel prices -- show a U-index for Transport of .453, close to values found forPhilippines, India and C6te d'Ivoire (Table 1, cols 3 and 4). The UnitedKingdom's U-index for all industry, on the other hand, is practically thesame as that of the United States: .48 against .49.
A numerical conclusion: the typical value of U.
30. On the basis of the preceding discussion we hazard a mere numericalconclusion: that the average share of intermediate inputs in the grossoutput of Transport in developing countries lies close to 45 percent. Itwould then follow that Gross Output of Transport is about 1.8 times the
12 -
Tab: 2: Fuel input coefficients of transport and fuel prices
Fuel input as Price per gallonproportion of in USc
Fuels sector gross Diesel Premiumoutput value gasoline
Philippine 1981land fre ht diesel, gasoline .19915 120.4 225.8public t nsport 1f n I .2591
U.S.A., 19transpor & diesel, petroleumwarehous g refining,
electricity .09014 54.3 71.9
Mexico, 19 ;transpor .tion petroleum
refining .06063 11.0 67.0
Source of Lel prices: Hu.ghes (1986).
- 13 -
value of Primary Inputs into Transport. If the Non-Comparable Importscomponent of Primary Inputs can be ignored (see Table 3, below), TransportGross Output would be some 1.8 times Value Added in Transport, asascertainable from national accounts. The meaning that can be attached toGross Output is discussed in para;. 44 below.
Intermediate Inputs vs. Primary Inputs into Transport.
31. In the U.S.A., the main Transport sub-sectors of the Philippines,in India and Mexico, the share of intermediate inputs in the gross outputvalue of Transport is below the national average of this share (Table 2,cols 2,3,4). In those countries, Transport is intensive in Primary Inputs.The exception is Cote d'Ivoire where the U-index of Transport lies close tothe average for all industries.
32. Transport is largely to be classed as a non-traded good, (orservice). In most countries, transport services are, neither actually norpotentially, in competition with imported services of the same kind. Exceptin countries that serve as important transit routes to others, internationaltrade in transport services is confined to air and some shipping services.In general, therefore, wor]d levels of costs (or efficiency in production oftransport services) exercis;e little constraint on the costs and efficiencyof national transport industries. While direct competitive pressure fromthe outside is thus inegligible there could still be indirect externaleffects on domestic transport cost levels, transmitted through the prices oftraded goods required in the production of transport service. Therelatively low share of such intermediate inputs (including importedintermediate goods) in the gross output value of Transport reinforces theeffect of an absence of importable substitutes for transport: it helps toshelter the cost of domestic transport from international costs, prices andthe changes in them that follow from technical progress and otherproductivity improvements. Even in open economies, therefore, the lower thevalue of the U-index, the more will the relative cost of transport begoverned by the levels of competition and efficiency within the industry,and not least by wage costs per unit of output (Table 3).
Effects of the tax system.
33. Two great classes of taxes on production of goods and services arethe cascading type, exemplified by turnover taxes, and Value Added Tax. Thedistorting effect of the former (inducing the flour miller to integrate withthe baker) is a major argument in favor of the latter. Transport, however,tends to have a relatively low share of intermediate inputs and a high oneof Value-Added in gross output. Other things being equal, therefore,Transport may be relatively favored by a cascading system and should in anycase benefit less than other industry from the institution of a VAT system(Kay and Davis, 1987).
14 °
Table 3: The composition of Primary Inputs*
Country and sector Payment to labor Property-type Non-comparableZ income Z imports Z **
U.S.A., 1977. transport and 67 23 4warehousingall industry 59
Philippines, 1981. transport and 40communications
. all industry 33
COte d'Ivoire, 1978. transport & 57 41
communications. all industry 33
Mexico, 1978* transport 37 60 2
Notes: * India's I-0 accounts do not distinguish labor and property-typeincome in gross value added by industry.
** Non-comparable imports cannot be distinguished for Coted'Ivoire. Total imports were assigned as intermediate inputs.See para. 23 above.
Total,input requirement of Transport.
34. Total input' coefficients represent the input requirements inresponse to a unit change in final demand, calculated through the sequenceof rounds of inputs into inputs. The coefficients in any one column thusdepend on all other coefficients in the table. (See para. 18 above).Averaging these total coefficients in the Transport column and comparing theaverage with that of all total coefficients in the matrix one obtains the D-index described in para. 24 above. Its values (Table 1, col.5) confirm whatappeared from comparisons between the direct input coefficients forTransport and all industry, except in the case of India. In COte d'Ivoire,India (non-rail transport) and two lesser transport sub-sectors of thePhilippines, the demand response of transport to a unit increase in generaloutput is more intensive in intermediate inputs than is true of industry ingeneral. In all other cases, the demand generated by transport for theoutput of other activities is less than for industry as a whole. In India,therefore, the direct input coefficients are a less reliable base forpredicting requirements than in the remaining 4 countries.
- 15 -
Transport as supplier of services: transport demand.
35. Once again, we employ two summary measures to indicate the sectoraldestination of Transport output, or sectoral sources of demand for transportservices, and the sensitivity of transport demand to changes in generaloutput. These counterparts of the U and D indexes in the precedingdiscussion are code-named W and S.
(i) W The proportion of gross Transport output going to otherindustries as intermediate services. The ratio is computedfrom the direct input coefficients and, like the U index, wasdefined by Chenery and Watanabe (1958). It is also known asforward linkage coefficient because it expresses the supportthat Transport (or any given sector) supplies to otherindustries. A relatively high W denotes relative predominanceof intermediate uses of Transport services.
(ii) S The index of sensitivity indicates the relative importance of asector as a supplier of intermediate inputs to other sectors --the comparison being with the average of all industries. Itis computed from the total coefficient matrix, as the ratio ofthe average of coefficient in the Transport row (distributionof output) to the average of all coefficients in the matrix.It therefore measures the dependence of Transport on industrydemand for its output, relative to that kind of dependence forthe country's industry as a whole. It is important to notethat when this index exceeds unity, Transport has to increaseoutput relatively more than other sectors for a unit increasein final demand in every sector.
Transport requirements: demand for transport.
36. The presumed omission of own-account transport from the Transportaccount in our I-0 tables is likely to have more of a distorting effect onthe reported structure of demand for transport than on its inputcomposition. The totals for transport services sold (row totals) and forresource requirements for transport (column totals) will, of course, beaffected to an equal degree. But there is less reason to expect differencesin input structures between professional and own-account transport than inthe structure of demand (or destination of output) for the two modes ofoperation. Own-account transport tends to be used, to a greater extent thanprofessional transport, for industrial purposes, as intermediate service.The true-W-index is thus likely to be underestimated in Table 4 (cols 3 and4). The size of the necessary correction remains unknown.
37. The demand structure of transport seems less likely to be astructural fact of the industry than its input structure. Table 4 showsTransport in the U.S.A. and Mexico to depend more on final demand than doesthe industrial sector as a whole. In the Transport sector of the Coted'Ivoire and the main Transport sub-sectors of India and Philippines, on thecontrary, transport services are intermediate goods to a greater extent thanindustrial output in general. This does not hold for rail in India or for
16 °
public transport, water or air transport, water or air transport in thePhilippines which depend more than the average industry on sales to FinalDemand: as anyone acquainted with the countries and transport modes wouldexpect.
Table 4: Share and intensity7 of intermediate demand for theoutput of all industry, and of Transport
Weights: W - Proportion of gross output S - RelativeZ of Value for intermediate use total dependenceAdded by - on demand forspecified ALL INDUSTRY TRANSPORT intermediatesector use
sector subsector
(1) (2) (3) ()(5)
USA, 1977 .5619Transport & 100 .5268 2.3029Warehous ing
38. While the distribution of total Transport output betweenintermediate use and Final Demand differs a good deal between the 5countries there is striking similarity in the average direct transport inputcoefficients of their industries (Table 5). Sensitive though thesecoefficients are to the varying degrees of aggregation in the differenttables, the means of the direct coefficients cluster around 2 percent. In 4of the 5 countries, the dispersion of coefficients around the mean is alsoquite similar. The average impact effects of changes in transport prices onindustrial costs seems therefore to be broadly similar in these verydifferent countries. There is much less uniformity in the total transportcoefficients which are highly sensitive to aggregation differences.Predictions of future demand for transport based on direct inputcoefficients will obviously be biased downwards. Table 5 suggests that theerrors will be significant. When all repercussions have been taken intoaccount, the average transport input coefficient is doubled in 3 of the 5countries, and raised by one-third in the other two.
39. Table 5 also shows the average of direct transport inputcoefficients for the groups of most intensive users, those producers whoshould be most sensitive to changes in productivity and prices of thetransport industry. It may be difficult to identify this group without 1-0information. The group of main users -- in terms of their share of totaltransport services sold to producers -- is much more readily known: they arethe industries on whose activity and development transport operators dependmost heavily. In principle, those two groups need not overlap, neitherfully nor even partially. Main users may be industries with transport inputcoefficients well below the national average, but a high total output valuerelative to the total inter-industry sales of the transport industry. A'main user' who is not also a 'most intensive user' will have a larger totaloutput value than the least intensive among the 'most intensive' users. Weilli4strate by comparing the 5 most intensive users of Philippines landfrelght transport with the 5 main users:
A A' B A A' Bzz z z z A
Five most intensive users Top five main users:Basic metals 5.2 7.3 4.8 1. Food products 2.7 3.8 23Paper, publishing 3.6 5.1 2.0 2. Construction 2.5 4.5 12Chemicals 3.6 5.1 5.0 3. Textiles 3.2 4.4 5.3Forest, wood prod. 3.5 4.7 5.0 4. Forest,wood prod 3.6 4.7 5.0Textiles 3.2 4.4 5.3 5. Basic metals 5.2 7.3 4.8
- 6. (Trade) (2.4 2.9 17122.1 1-5 50.1
[1-4,63 [62.3]
Note: A - Direct transport input coefficient.A' - Total transport input coefficient.B - Share of total transport services delivered to industries.
Table 5: Average transport reQuirement per unit (Dollar) of output
Country, definition Direct Requirements Total RequirementTransport sector, andtotal number of sectors Mean S.D. CV Mean S.D. CV
Note: SD - Standard DeviationCV - Coefficient of Variation
- 19 -
If one concentrates on commodity producing sectors, 3 of 5 Philippinesindustries figure in either group. If Trade is included among the 5 mainusers -- an industry for which forecasts may be more difficult than forcommodity sectors -- the overlap shrinks to 2 in 5. Even excluding Trade,the two most important users of this major branch of Philippines transport,with a combined share of 40 percent of transport services provided tointermediate users, are not among the most intensive users. By comparisonwith all industry, however, all the top 5 main users (whether Trade isincluded or not) have transport cc-*fficients above the general average.This is the norm in the 4 I-0 tables that permit the comparison, as appearsfrom the following comparison of average direct transport coefficients forall industry and the group of main users. (For the practical purpose ofdetermining required volumes of transport service, the average of industrycoefficients is not usually relevant: one wants individual coefficients forindividual industries expected to grow more or less than the average. Herewe use averages merely for a quick characterization of the relativeintensity of transport use by different producer groups.) A and B have thesame meaning as in the preceding panel:
A B A B
PHILIPPINESUSA (Land freight)All industry 2.04 All industry 2.110 main users 4.01 40 5 main users 2.9 62.3- omitting Trade 3.96 32 - omitting Trade 3.4 50.1
INDIA(Not rail transport) COTE D'IVOIREAll industry 2.5 All industry 2.010 main users 4.0 47 5 main users 7.0 57- omitting Trade 3.0, 34 - omitting Trade 5.2 18
The result of the comparisons, however contingent on variation in the formand extent of aggregati6n in different tables, suggests that the 'mainusers' of transport are transport-intensive industries relative to theaverage of the producing sector. Other things being equal, changes in theefficiency and unit price of transport (that is, the cost of the service toits user) should affect their costs and profits more than the effect onindustry at large. Being main sources of demand for transport, theirreaction will in turn have a relatively strong effect on Transport. Onefinds also that Trade and Construction are in all cases among the main usersof Transport.
- 20 -
Sensitivityz of Transp2ort demand to changes in ,production.
40. The 'index of sensitivity' (S), shown in the final column of Table4, is computed on the rows of the total requirements table. Entries in theTransport row of those tables show the production of Transport servicesrequired, directly and indirectly, per Dollar of output delivered to FinalDemand by the respective column-industries. The index then compares theaverage of these required Transport inputs (to other industries, per Dollarof additional final demand for their products), with the average of totalinput coefficients in the table.l It therefore measures the average effecton Transport of equal increases in sales to Final Demand by all sectors,relative to the corresponding average demand effect on all sectors. (Sincewe are dealing with total coefficients, all intermediate demands for inputsto support such a Final Demand expansion are, of course, brought into thereckoning.)
41. The sensitivity index for the Transport sectors of the U.S.A.,India, Cbte d'Ivoire and Mexico, and for the land-freight subsector of thePhilippines is always well above unity (Table 4). In terms of the index,therefore, Transport is distinctly more sensitive to 'general' outputexpansion than the average industry: it has to expand its output by morethan the average industry to support this type of output expansion. It is,of course, unlikcely that Final Demand will ever expand in this particularfashion, by equal amounts for each industry's output. But unless onespecifies the enItire pattern of incremental demands -- a standard step inplanning exercises -- the sensitivity planning exercises the sensitivityindex seems as good a measure as any other for determLning the relativeeffect of general, marginal output expansion on the demand for the output ofspecific sectors. The magnitudes of the Transport sensitivity indexreported in Table 4 are such that one may assume the conclusion ofrelatively high output-sensitivity of Transport demand to hold even formoderately uneven distributions of increments to final demand.
The 'size' of the Transport sector.
42. Different summary indicators of the total size of an industry areusually answers to different questions. In considering alternatives, it istherefore essential to be clear about the meaning of the indicator and theuse to which it is to be put. Most uses require that the indicator shouldpermit logically valid comparisons with other aggregates.
43. The Value Addesd by Transpert is usually the most availableindicator since it is. part of the standard presentation of nationalaccounts. It is the sector's contribution to GDP or GNP and can thereforebe expressed as a percentage of them. Like other indicators drawn fromnational statistics, it is safer to use for comparisons over time in thesame country than for international comparisons. It is difficult tointerpret the difference between the percentage of Value Added due toTransport in COte d'Ivoire (9.6Z in 1978) and the United States (6.4Z in
1/ I.e., it is formed as the ratio of the average of the Transport rowentries to the average of the average of all row entries.
- 21 -
1977): the former being Value Added by Transport and Communications, thelatter, Transport and warehousing; the former, at Factor Cost and thelatter, at Market Prices; the U.S. national accounts including value addedby Trade or Public Administration and Defense while those are omitted byCGte d'Ivoire (or so it appears), yielding a smaller base. A more generalobjection to the Value Added measure is that in most countries it excludesthe value added in own-account transport operations. The objection is noless valid for any other measure that does not include a deliberatecorrection for this omission. But while the computation of value added byTransport will be done with different degrees of perfection in differentcountries, the concept is clear: the value, at local prices, of the annualgross product (including capital consumption) of the labor and capitalemployed in what is defined ae the transport industry.
44. Since the industry's labor and capital do not produce transportservices unaided by supplies from other industries, Value Added is not ameasure of 'total resources devoted to the production of transportservices'. Gross Output may seem to be the right answer to this questionbecause it is the sum of Value Added and Intermediate Inputs in Transport.We have suggested that for given technology, industrial organization anddifferential productivity, Gross Output in Transport is not quite twiceValue Added (para. 30 above). Unfortunately, Gross Output is useless as ameasure of the relative 'size' of Transport because it cannot be related toother aggregates: obviously not to GDP which is net of intermediate inputs,nor to a grand total of Gross Outputs which involves double-counting (thegross output of petroleum refining being partly also in the Gross Output ofTransport, and so forth). The concept has, therefore, limited uses.Planners may use it as the base from which to project or plan changes in thesupply of transport to all uses. Since Gross Output measures Industryturnover, it can be employed in estimating the return to changes in turnovertaxes. Also, the comparative movement over time of Value Added and GrossOutput (the time path of our U index) can serve as an indicator ofdifferential productivity change. T'his is obviously safest done in thecourse of projecting; in ex post studies one faces the problem of having tocontrol for irrelevant changes in industrial organization or statisticalaggregation.2
2/ India's Seventh Five Year Plan expects Gross Value Added in Transport torise annually at 7.1 percent, and Gross Output Value, at 8 percent,between 1984/5 and 1989/90. The difference is the expected annualincrease in the productivity of the transport industry. (Government ofIndia, 1985, vol.1, p.62.)
- 22
45. If the meaning of 'total resources devoted to the production ofTransport' is to be: resources for the production of a commodity calledTransport, then the appropriate measure is, simply, Final Demand fortransport service. 3 The breakdown of Final Demand by industrial origin ofthe goods or services is not usually a part of the standard presentation ofnational accounts: one has to look for it in the country's I-0 tables.Schematically, and i&,noring the complication of imports and, specifically,of trade deficits,
Proportion of Gross OutputFinal Demand 2 sold to Final Demand : Value Added
Proportion of Value Addedin Gross Output
or, in terms of our code-names U and W, with obvious notation:
FD - [(1 - w) / (1 - U)] VA.
Comparing the values for U and W in Tables 1 and 4, it appears that in mostof our cases -- W being larger than T - Final Demand for Transport islikely to be less than Value Added in TransportA4
3/ To demonstrate, assume a two-industry economy, with industries A and M,and the following transactions table:
A M Final Demand Total
A / 3 7 10M 2 / 9 11
VA 8 8TOTAL 10 11
Resources devoted to production of A:-A's Value Added 8M's contribution to production of A 2Less A's contribution toproduction of M -3Final Demand 7
4/ The conclusion is tentative because we ignore the different ways in whichimports are treated in different I-0 tables. In the preceding footnoteit was thus assumed that row and column totals for each industry areequal. They are normally unequal (see para. 8 above).
- 23 -
46. Another aspect of relative 'size' is measured by the index ofsensitivity (paras 40, 41 above): the effect of marginal changes in outputon the demand for Transport relative to the corresponding effect on allindustry. The measure is comprehensive in that it counts all intermediatedemands provoked by Final Demand increases. The index has its limitations,as discussed earlier, but it takes one as far as a purely static use of I-Oaccounts allows, in answering a question that does not contain a specificstatement about the expected or planned behavior of the economy.
47. None of these indices of the total 'size' of the transport sectorcover the country's transport activity exhaustively. The gap consists ofown-account transport, industrial and personal. Most input-output accountsknown to us do not attempt to reclassify the outputs and the correspondinginputs of industrial own-account transport operation to the transportindustry. The overall importance of this omission (in terms of tlheproportion of the country's total freight transport cost excluded from theaccount of the transport industry) should be the larger, the smaller is theshare of rail, water and air in freight transport. A rough calculation forthe United States suggest that Value Added by Transport and Warehousingshould be raised by at least 12 percent to cover own-account transport offreight by road. There is no reason to believe that the size of theomission varies systematically between countries with the degree ofindustrialization or income per head. Its chief effect is to understate thelevel of intermediate anid primary inputs into transport and the level of itsoutput, and to misrepresent the transport I-0 coefficients of individualindustries or commodities. The other indicators derived from our I-0 tablesrefer to the response of Transport, relative to the response of otherindustries, to changes in general output, as a purchaser of intermediateinputs or a supplier of service. These indicators (D in Table 1, and S inTable 4) may still be valid for industrial transport activity as a whole,provided that the input coefficients for own-account transport in responseto a general increase in output behaves in the same way as demand for theoutput of the defined transport industry. On balance, the validity of theseindices may be less affected by the failure to reclassify own-accounttransport to the transport industry than the effects on input and outputlevels and relative industrial transport requirements.
48. Private own-account transport, predominantly of passengers, isrepresented in I-0 accounts by its purchased inputs, merged in the PersonalConsumption Expenditure (and government consumption) accounts of FinalDemand. In developed countries, this category of own-account transportoperations constitutes a sizeable component of transport output in the broa,.:sense. In the United States, personal consumption absorbs 26 percent of thetotal output of transport industry services: even without governmentconsumption (partly also personal transport) its shares is thus more thanhalf of that taken by intermediate uses of transport (Table 6). Butpersonal consumption expenditure on petroleum refinery products (used partlyfor other household purposes) and automobile repairs amounts to almost twiceas much as personal expenditure on services bought from the transportindustry. Personal expenditure on motor vehicles and equipment, in 1977,exceeded that of industry by almost one-fifth. In developing countries,ownership of transport means is less prevalent. Taking India as an example(Table 6) one finds that the share of personal consumption in the total
24 -
Table 6: Indicators of the share and componentsof personal transport
Delivered to:
Intermediate users*: Personal GovernmentFrom Industry all industry consumption consumption
a. USA (1977)
Transport & Warehousinge $ millions 58 805 33 210 7 194
e 100 42 6Motor Vehicles & Repair. Rs millions 8 611** 1 614 3 672
I 100 19 43Motorcycles, Bicycles & Repair
Rs millions 2 890** 1 706 334I 100 59 12
Notes : * Intra-industry transactions excluded.** Value of output delivered to industries plus delivery to Gross
Fixed Investment.
- 25 -
output of the transport industry (30 percent), and its share relative tothat of intermediate industrial use, are quite close to those in the UnitedStates. Personal consumption expenditure on petroleum products (needed alsofor many other household uses) is however, very much smaller relative toexpenditure on transport industry services. Personal expenditure on motorvehicles, motorcycles, bicycles and repairs was only 28 percent of suchpurchases by classified industries;5 if government purchases are added, thepercentage rises, but only to 64 percent of purchases by industries. Thesequantities, beside confirming what is well-known, indicate that transportindustry services to persons or passengers are a better representation oftotal personal transport output in developing than in developed countries;but in no sense a full representation.
5/ Motorcycles and bicycles for intermediate use went to the 'othertransport' industry, which includes services by rickshaws of all kinds.
- 26 -
REFERENCES
I. The Input-Output tables and eMplanatory literature.
COTE DFIVOIRER6publique de C6te d'Ivoire, Les comptes de la Nation, 1978. Minist6rede l'Economie, des Finances et du Plan. Abidjan, 1980.
INDIAGovernment of India, A technical note on the Sixth Plan of India.Perspective Planning Division, Planning Commission, New Delhi, 1991.
Salvja, M.R., Input-Output tables for India: Concepts, Construction andApplications.
MEXICOR6public of Mexico, Matriz de Insumo-Producto Ano 1978. Secretaria deProgramacion y Presupuesto.
PHILIPPINESRepublic of the Philippines, Philippine Statistical Yearbook, 1984.National Economic and Development Authority.
UNITED STATES
Department of Commerce, Bureau of Economic Analysis, The detailed Input-Output structure of the U.S. Economy, 1977. vol. 1.
--The Input-Outputstructure of the U.S. Economy, 1977. Survey of Current Business, May,1984.
-- ,Definitions andconventions of the 1972 Input-Output study. Bureau of Economic AnalysisStaff Paper, Philip Ritz.
---------------------------------- Gerald Silverstein, New structuresand equipment by using industries, 1977, Survey of Current Business,November 1985.
--.- , and T. Watanabe, 1958, International comparisons of thestructure of production. Econometrica, vol. 26, N04.
Hirschman, Albert 0., 1958, The strategy of economic development. NewHaven, Yale University Press.
Hughes, Gordon A., and T. Tavana, with computer implementation byM,Ahmad and S. Pal, 1986), Petroleum product prices, 1970-1980. WorldBank Transportation Department, mimeo.
India, Government of, 1985, The Seventh Five-Year Plan, vol. 1.Planning Commission, New Dehli.
Kay, J.A., and E.H. Davis, 1987, The VAT and services. World Bank,Development Research Department, Discussion Paper, Report DRD 245.Washington, D.C. mimeo.
Leontief, W.W., 1951, The structure of the American Economy. 2ndedition. Oxford University Press.
-------- , and others, 1953, Studies in the structure of theAmerican economff. New York, Oxford University Press.
United Nations, 1966, Problems of Input-Output tables and analysis.Statistical Office of the United Nations, Studies in Methods, series F,N014.
, -1973, Input-Ouptut tables and analysis. StatisticalOffice of the United Nations, Studies in Method, Series F, N014, rev.lBy Alan A. Armstrong.
