Ij - IMF eLibrary

This is a working paper and the author would welcome anyIMF WORKING PAPER comments on the present text. Citations should refer to an

unpublished manuscript, mentioning the author and the dateof issuance by the International Monetary Fund. The viewsexpressed are those of the author and do not necessarilyrepresent those of the Fund.

WP/86/1 INTERNATIONAL MONETARY FUND

Fiscal Affairs Department

Indirect Taxation in Developing Countries:A General Equilibrium Approach

Prepared by A. Lans Bovenberg Ij

Authorized for Distribution by Ved P. Gandhi

September 17, 1986

Abstract

Indirect taxes are an important element in stabilization tax packagesthat aim at raising revenue in the short run. This paper evaluates, byusing a general equilibrium model, alternative instruments of indirecttaxation in middle-income developing countries. It uses data forThailand as an illustration and examines the effects on revenue,efficiency, equity, and international competitiveness. The paper showsthat the interaction between taxes and distortions caused by variouspolicies can be important for revenue and efficiency. It also revealssignificant backward shifting and a link between outward-looking supply-side tax policies and trade policies in industrial countries.

_!/ The author would like to thank Shanta Devarajan and HecktorSierra for providing data, Wouter Keller for providing computationalassistance, and Vito Tanzi, Ved Gandhi, Sheetal Chand, Partho Shome,Thanos Catsambas, Somchai Richupan, Wouter Keller, Shanta Devarajan,Sweder van Wijnbergen, Phil Young, and Sara Chernick for their commentsand suggestions.

©International Monetary Fund. Not for Redistribution

- ii -

Contents Page

Summary iii

I. Introduction 1

II. The Computable General Equilibrium Model 2

1. The computable general equilibrium framework 22. A particular general equilibrium model 43. Data requirements 6

HI. Simulation Results 14

1. Introduction 14a. Tax instruments 14b. The tables 14

2. Income-based VAT 203. Broad-based consumption VAT 214. Zero rating agriculture 225. Exempting agriculture 246. Exempting services 257. Existing domestic taxes on goods and services 268. Import tariff 289. Export tax on agricultural products 28

IV. Policy Conclusions 30

Text Tables

1. Thailand (1973): Expenditures and Receipts 82. Effects on Revenue and Relative Prices 153. Effects on Production Structure 164. Welfare Effects 175. Welfare Effects and Export Elasticities 29

Figures

1. Utility Structures of Domestic Household Sectors 14a2. Production Structures of Domestic Production Sectors 14b

Appendix. Algebraic Presentation of the Model 33

References 37


- iii -

Summary

This paper evaluates alternative -instruments of indirect taxationin middle-income developing countries. Indirect taxes are an importantelement in stabilization tax packages that aim at raising revenue inthe short run. In addition, revenue mobilization through indirecttaxation plays an important role in medium-run development plans thataim to raise domestic investment and public savings while reducingdebt-creating capital inflows. In evaluating indirect taxes the paperexamines not only their revenue effects but also their structuraleffects on the objectives of efficiency, equity, and internationalcompetitiveness.

This study uses, as illustration, data for Thailand. That countryprovides an interesting case for the study of indirect taxation becausethe characteristics of its indirect tax structure—cascading, ratedifferentiation, and the importance of trade taxes, for example—applyto many other middle-income developing countries as well.

In order to assess various indirect tax instruments, this paperadopts a general equilibrium approach. In contrast to a partial equilib-rium approach, this approach accounts for the interaction betweenvarious markets and taxes. These interactions can be important from apolicy point of view.

Regarding the efficiency objective, this study shows that taxinstruments interact significantly with other distortions. Therefore,"efficient" tax packages should take these distortions into account.In the presence of trade distortions, for example, efficient taxpackages should avoid implicitly taxing exports.

Regarding the revenue objective, the paper shows that the interac-tion between taxes, distortions, and various other economic policiescan be important. Each instrument of economic policy affects the basesof all tax instruments. To illustrate, this study shows that an economythat relies heavily on trade taxation can be on the downward slopingsegment of the export-tax Laffer curve, not only because of the erodingbase for the export tax itself, but also because of the additionaleffect of a depreciating exchange rate on the base of import tariffs.

Regarding the equity objective, this application reveals thatbackward shifting of indirect taxes may be as important, if not moreimportant, as forward shifting. Therefore, depending on the elasticityestimates, rate differentiation can be relatively ineffective in redis-tributing income from the demand side.

For developing countries that are improving the competitivenessof their export sectors, the study reveals an important link betweenoutward-looking supply-side tax policies and trade policies in indus-


- iv -

trial countries. If the industrial countries remove trade barriers,developing countries benefit from domestic outward-looking tax policiesthat simultaneously improve their competitiveness and world welfare,thereby benefiting the world economy as a whole. If the industrialcountries pursue protectionist policies, however, the developingcountries may suffer short-run welfare losses from domestic outward-looking tax policies that improve their competitiveness as well asworld welfare. Thus, protectionist policies in the industrial countriesprovide significant disincentives for developing countries to formulateoutward-looking tax policies that benefit themselves as well as theworld as a whole.


I. Introduction

Tax policy plays an important role in the efforts of many middle-income developing countries to improve their fiscal and economicperformance. In the short run, policymakers formulate tax packagesto complement expenditure restraint aimed at containing macroeconomicimbalances. Revenue mobilization through taxation is also an importantelement in medium-term development plans that aim to raise domesticinvestment and government savings while reducing reliance on debt-creating capital inflows. However, the need for increased fiscalrevenue has to be reconciled with the other objectives of economicpolicy, such as an efficient resource allocation, an equitable incomedistribution, and a competitive trade sector.

This paper assesses alternative indirect tax instruments indeveloping countries in the light of these objectives. In order to doso, this paper adopts a general equilibrium approach. In contrast tothe partial equilibrium approach, which is typically adopted inassessing taxes, a general equilibrium approach accounts for the inter-actions between various markets and taxes. These interactions arepotentially important from a policy point of view. Regarding therevenue and efficiency objectives of policy, for example, a generalequilibrium approach can explore how various taxes and other distortionsinteract with each other in determining tax revenue and distortingbehavior. An important theme in this paper is the effect of domestictaxes on goods and services on both revenue collected from, and distor-tions induced by, other tax instruments, especially trade taxes. Thus,despite its limitations, the general equilibrium approach representsan advance over other methods by offering a unifying framework thatcan highlight channels of interdependence that a partial analysis wouldnot uncover.

This study focuses on indirect taxation—taxes on goods andservices, and on international trade—for three reasons. First, middle-income developing countries rely heavily on these taxes. Second, theindirect tax structure is a potential area of reform in many of thesecountries. The efforts of some of these countries to liberalize tradepolicy require a shift away from trade taxation toward taxation ondomestic consumption. Existing systems of taxation on domestic consump-tion in these countries often incorporate some undesirable featuresas well. Third, in the interest of raising revenues in the short run,stabilization tax packages often contain measures pertaining to indirecttaxes. _!/

T 7For the tax content of stand-by arrangements, see Beveridge andKelly (I960).


This paper uses, as an illustration, data for Thailand. Thatcountry provides an interesting case for the study of indirect taxationbecause certain characteristics of its indirect tax structure, such ascascading and the reliance on trade taxes, apply to many other devel-oping countries as well.

This paper contains three main sections. Section II introducesthe applied general equilibrium methodology and describes the mainfeatures of the particular model that is applied in this paper.Section III contains the results tnat are derived from the model simula-tions. The reader who is mainly interested in the policy implicationscan skip it. The concluding Section IV draws together the major lessonsand policy implications from this study.

II. The Computable General Equilibrium Model

This section first introduces the computable general equilibriumapproach and discusses its weaknesses and strengths in analyzing taxpolicy. The section then describes the particular model applied inthis paper and elaborates on some of its limitations.

1. The computable general equilibrium framework

Computable general equilibrium analysis explicitly specifies theinteraction between rational economic agents, markets, and public policywithin a consistent macroeconomic framework. Generally, the modelsemployed in this analysis incorporate several industries, households,goods, and factors as well as international trade flows. Thus, theyaccount for both macroeconomic constraints and the linkages betweenvarious industries and domestic and international markets. Generalequilibrium modeling is firmly rooted in the clearly specified theo-retical framework of microeconomic theory; the decisions of the decen-tralized agents are based on optimizing behavior and basic parametersof taste and tecnnology. In most models, relative prices are endogenousand play a crucial role; they equilibrate demand and supply in allmarkets. Shoven and Whalley (1984) survey the recent applications ofthe general equilibrium methodology. Bovenberg (1985) evaluates thesignificance of general equilibrium modeling for public policy.

General equilibrium modeling has many limitations. The economicrichness of the models does not allow for the simultaneous estimationof all parameters. Therefore, most modelers "guestimate" the parametersbased on econometric estimates from the empirical literature. More-over, the models adopt many simplifying assumptions, such as perfectcompetition and flexible prices. Taking into account the weaknesses ofthe models, the policymaker should interpret the quantitative resultsonly as indicative of the rough order of magnitude of policy effects,

- 2 -


given particular assumptions about economic behavior as weil as aboutpolitical and institutional responses. While it is assumed that- themodel solutions exert a certain pull on the economy, the simulationsshould always be supplemented with what one believes and knows aboutthe economic, institutional, and political world, and how that complexreal world relates to the stylized model. Thus, the simulation resultsshould induce policymakers to start, rather than stop, thinking.

Despite the limitations, the general equilibrium approach offersa useful unifying framework relative to other methods of studying therelationship between tax policy and the objectives of revenue,efficiency, equity, and international competitiveness. Regarding therevenue objective, the general equilibrium framework forces policymakersto consider the constraints imposed by rational microeconomic behavior.The framework models the erosion of tht; tax base owing to substitutionaway from taxed commodities by rational agents. It also estimates theconsequences of the behavioral response to a particular tax for thebases of other taxes.

As regards efficiency, general equilibrium modeling is particularlyuseful for welfare analysis in a distorted (second-best) economy becauseit simultaneously specifies the various tax distortions and is based onoptimizing behavior. Thus, the framework can be used to explore howtaxes interact with existing distortions in affecting rational behavior.An important theme in this paper, for example, is the effect of domestictaxes on goods and services on both revenue collected from, and distor-tions induced by, other tax instruments, particularly trade taxes.

Turning to the equity objective, because general equilibrium modelsincorporate several household sectors they can study not only theeffects on economy-wide efficiency but also on welfare enjoyed by eachdisaggregated household. This study focuses on the distribution betweenrural and urban sectors and between the national economy and the restof the world.

In the study of the incidence of taxation, partial equilibriumapproaches commonly assume that both the intermediate and final compo-nents of the sales and excise taxes are fully shifted forward towardfinal demand. The general equilibrium approach—in contrast—makes noprior assumptions about the shifting of a tax but endogenously deter-mines the tax shifting. The next section shows that, depending on thestructure of the economy, taxes on goods and services may be (partly)shifted backwards.

With respect to structural adjustment and the objective of inter-national competitiveness, the general equilibrium framework models theeffects on exports, competing imports, and complementary imports ineach sector of an economy. How taxation affects the trade orientation

- 3 -


of an economy, that is, the mix of export-oriented, import-dependent,nontradable, and import-competing industries, can be studied by dis-aggregating the industries accordingly.

2. A particular general equilibrium model

This paper uses a general equilibrium tax model developed inKeller (1980). I/ This subsection gives a broad overview of the modelstructure including the limitations and underlying assumptions. TheAppendix contains an algebraic presentation of the model.

Following Johansen (19bO), the model adopts log-linear approxima-tions of all—possibly nonlinear—relationships around the initialsituation and is solved in percentage changes of the variables. Thus,the solutions describe the marginal effects of changes in taxes.They can also be used, however, as linear approximations for largerchanges. 2j

This study adopts a comparative static approach: it comparesmedium-run equilibria before and after a change in tax policy. Generalequilibrium models are not designed to simulate short-run changesbecause behavior and prices may adapt slowly. Thus, the referenceperiod, defined as the period for which our results become valid, ismore than one or two years. This is a period sufficiently long forprices and behavior to adjust. On the other hand, the reference periodshould also be sufficiently short so that changes in the stocks duringthe period are small compared to the level of the stocks; the staticmodel ignores induced changes in stocks, such as the accumulation ofcapital stocks. _3/ The reference period in this model is about threeyears, kj

T7This study used the computer software "The Keller Model, FreeUniversity Amsterdam, for IBM-PC," which was kindly provided by W.J.Keller.

2j In order to compute the effects of large changes in policy,global nonlinear methods require global information on the variousstructural relations. The method applied in this paper only requireslocal information. Linear models, however, can account for non-linearities by adopting a procedure of iterative linearization. SeeBovenberg and Keller (1964).

3/ Although the static model does not compute changes in stocks, thesimulated changes in prices provide some information on the dynamiceffects of taxation. To illustrate, profitability in each productionsector provides an indication for the effects on investment decisionsin each sector and the interindustry allocation of capital (seeSection III).

_4/ For a similar model of the Australian economy—the Orani model—the reference period has been estimated to be about one and a half totwo years. See Cooper and McLaren (1983).

- 4 -


In order to simulate the changes in the structure of productionthat are induced by taxation, the domestic production side distinguishesamong agriculture, services, and four industrial sectors. The indus-trial sectors are consumer goods, capital goods, intermediate goods,and infrastructure. The classification in the input-output tabledefines the six production sectors (see Appendix C of Devarajan andSierra (1985)). Each sector supplies its own (homogeneous) commodity.Thus, the model includes six domestically produced goods. _!/

The paper opts for this small-scale disaggregation because it seemssufficient to begin to explore some of the broad issues of indirect taxpolicy studied here. Moreover, in order to illustrate the generalequilibrium methodology, a small-scale aggregation is preferable becauseit is easier to interpret the results. For the study of more specifictax policy questions, such as the effects of specific excises, however,a large-scale multi-sector model is called for.

Turning to the behavior of producers, this study adopts the usualassumption of perfect competition: firms take prices as given. Theymaximize profits subject to a production function, which may allow forsubstitution between the inputs. The inputs consist of imports, inter-mediate deliveries from other domestic producers, labor, and capital.Labor is further disaggregated into four categories: rural, urban,blue collar, and white collar. Capital is sector specific and immobileduring the period of reference.

Transaction taxes are returned to households as lump-sum transfers.The transfer to the public household is the public budget. This paperassumes that the transfers to all private households are exogenouslygiven in terms of the foreign good. 2J Thus, a change in the revenuefrom transaction taxes accrues to the public household only. The publichousehold demands goods and factors in order to satisfy public wants.The utility function of the public household endogenously determinespublic consumption. _3/

In order to simulate the consequences of tax policy for the equityobjective, the private domestic households are classified into six cate-gories. The rural households are disaggregated into two households:the own-account rural household and the informal rural household. Boththese households supply rural labor. The informal rural household doesnot save and does not own any capital. The urban household sector

_!/This paper defines services as a good.2_l All foreign goods can be aggregated into a single foreign good

because Thailand takes the prices of foreign goods as exogenously given.3/ Under certain conditions (see Keller (1980)), this utility func-

tion can be derived from the preferences for public goods of privatehouseholds.

- 5 -


+consists of the own-account urban, the informal urban, the blue-collar,and the white-collar households. The latter two households are charac-terized by the type of labor they supply (blue collar and white collar).The other two urban households supply urban labor.

Each household maximizes a regular, neoclassical utility functionsubject to a budget constraint. Private savings are treated aspurchases of capital goods by households. Utility depends on demandfor goods and the supply of factors. The after-tax prices and thelump-sum transfer determine the budget constraint. Households takeprices and transfers as given.

The foreign household represents the rest of the world. Its budgetconstraint is the balance of payments constraint. The real exchangerate is free to reconcile Thailand's expenditures with its income andthe exogenously given foreign savings.

Following Arraington (1969), the model specifies product differen-tiation by country of origin: foreign goods are distinct from domesticgoods. Final demands for imports enter the model as follows. Domesticconsumption and investment demand for each tradable good is a compositeof imported final goods and domestically produced final goods. Thecomposite good is called the Armington good. The value of the substi-tution elasticity between foreign and domestic goods can be variedbetween the extreme assumptions of infinite (perfect substitution) andzero (perfect complementarity).

Exports correspond to the demand of the foreign household. Theassumption of product differentiation by country of origin yields lessthan infinitely elastic demand functions for a country's exports.Thus, Thailand is not small in the markets for its exports and has some"monopoly power." When domestic and foreign goods are imperfect sub-stitutes in the Armington aggregation and when export demands are notinfinitely elastic, the prices of domestic goods can move independentlyfrom the price of the foreign good and the terms of trade are not fixed.

The model adopts the neoclassical closure rule. Thus, it does notspecify independent investment functions. Savings behavior determinesdomestic investment. This paper models savings as direct demand for acomposite investment good, which is produced by a so-called investmentsector. The input demands of the investment sector correspond to theinvestment demands for Armington goods.

3. Data requirements

The basic data requirements for the model are as follows. First,the model needs a benchmark data set that describes for each householdand f irm sector the after-tax expenditures and receipts as well as

- 6 -


+transaction taxes paid on all goods and services. Second, the modelrequires information on the elasticities that describe the marginalbehavior of the sectors.

The initial expenditures and receipts are derived from the dataset for 1973 used in Devarajan and Sierra (1985). Both the datasources and the procedure adopted to arrive at a consistent data setare described in Appendix B of Devarajan and Sierra (1985). Table 1presents the accounts. Each column corresponds to the after-tax expen-ditures by a particular sector. These accounts provide information onthe structure of the Thai economy (in 1973). That structure is animportant determinant of the tax effects. Therefore, it is useful tocomment briefly on these accounts starting with the accounts of theproduction sectors.

The composition of the outputs over the industries is .charac-teristic of a semi-industrial country in which agriculture and servicesaccount for more than 40 percent of gross domestic production. Withinthe manufacturing sector, consumption goods, capital goods, and inter-mediate goods each provide about 15 percent of gross production. Theimportance of various inputs varies considerably among industries.Agriculture has by far the largest value-added share in total costs.Consumption goods and intermediate goods, on the other hand, dependmore on intermediate inputs.

The accounts also provide information on the trade orientation ofthe industries. Within the traded goods sector, industries can begrouped into three main categories: export related, exportable, andimport competing. Agriculture is export related because it sells amajor part of its output to the consumption goods industry, which isthe most export-oriented sector. Capital goods is an import-competingindustry. The level of imported capital goods for final use whichcombines with domestically produced goods in the Armington aggregation,indicates that the capital goods sector sells in a market with a highlevel of import penetration. The intermediate goods sector can beclassified as import dependent because of its large share of importedintermediate goods. Because of its moderate export share this sectoris also a tradable sector. The only "pure" nontradable sector isinfrastructure, yet a large part of its intermediate inputs is imported.Because it includes trade-related services, the services sector has amoderately high export share.

The accounts reveal that the intermediate goods sector depends onintermediate demands from other domestic industries to sell its output.The capital goods sector and the service sector, however, dependprimarily on final domestic demand.

- 7 -


Table 1. Thailand (1973): Expenditures and Receipts

(Positive entries correspond to expenditures, negative entries to receipts)

(In millions of bant)

Expenditures by/ onConsumption

Agriculture GoodsCapitalGoods

Intermediate Infra-Goods structure Services

TaxRate J_/

A. Domestic Production Sectors

AgricultureConsumption goodsCapital goodsIntermediate goodsInfrastructureServices

-96,652.42,183.8666.4

7,345.7154.4

1,476.9

3U.347.9-55,564.9

395.44,699.52,493.8724.4

289.4194.3

-51,392.413,631.3

991.91,090.6

9,204.03,669.01,677.4

-54,014.94,317.92,797.6

87.827.5

1,485.49,068.3

-27,101.4994.3

1,646.77,379.13,109.58,556.64,472.7

-54,772.7Investment goodAgriculture ArmingtonConsumption ArmingtonCapital ArmingtonIntermediate ArmingtonInfrastructure ArmingtonServices ArmingtonRural laborUrban laborBlue-collar laborWhite-collar laborCapitalImports

50,629.4

32,633.21,562.6

3,621.34,462.4

109.05,492.23,218.9

8,529.09,574.1

251.69,792.67,047.6

2,675.42,396.91,204.89,637.8

16,434.1

2,939.23,128.6

140.6b,143.53,086.2

1,113.55,508.35,261.92,888.2

12,786.52,049.3

0.0710.0190.0560.0210.0520.020

Output/income 96,652.4 55,564.9 51,392.4 54,014.9 27,101.4 54,772.7


Table 1 (continued). Thailand (1973): Expenditures and Receipts


(In millions oi bant)

Expenditures by/ on InvestmentAgricultureArmington

ConsumptionArmington

CapitalArmington

IntermediateGoods

Armington

Infra-structureArmington

ServicesArmington

B. Composite Goods Sectors

AgricultureConsumption goodsCapital goodsIntermediate goodsInfrastructureServicesInvestment goodAgriculture ArmingtonConsumption ArmingtonCapital ArmingtonIntermediate ArmingtonInfrastructure ArmingtonServices ArmingtonRural laborUrban laborBlue-collar laborWhite-collar laborCapitalImports

Out put/ income

57,030.726,607.2

46,088.5

-51,711.01,312.8 -57,195.12,207.7 -28,452.430,708.3 -57,840.53,997.18,722.94,7b2.2

164.4 1,845.2 11,752.0

51,711.0 57,195.1 28,452.4 57,840.5

4,131.816,184.5

37,705.0

-8,219.8-16,184.5

-39,294.7

4,088.0 — 1,589.7

8,219.8 16,184.5 39,294.7


Table 1 (concluded). Thailand (1973): Expenditures and Receipts


(In millions of baht)

Expenditures by/on

AgricultureConsumption goodsCapital goodsIntermediate goodsInfrastructureServicesInvestment goodAgriculture ArmingtonConsumption ArmingtonCapital ArmingtonIntermediate ArmingtonInfrastructure ArraingtonServices ArmingtonRural laborUrban laborBlue-collar laborWhite-collar laborCapitalImports

Output/ income

Public

6,967.019.2

213. U352.1634.5126.4

6,996.8

4,465.38,431.7

28,206.0

Own AccountRural

23,076.429,474.310,357.011,575.41,491.62,668.07,457.9

-45,513.6

-40,498.0

86,100.6

Own AccountUrban

C.

10,074.69,453.05,701.85,581.7900.3

1,890.66,482.0

-20,123.6

"

-19,901.0

40,084.0

InformalRural

InformalUrban

BlueCollar

WhiteCollar

Rest ofWorld

Households Sectors

—2,701.71,074.8961.9166.3219.7813.5

-5,932.5

—

5,937.9

2,902.21,078.4429.0384.066.487.7324.7

-2,405.2

-2,867.2

5,272.4

5,121.09,610. 15,959.55,023.4686.4

1,404. 18,518.1

-30,663.9

-5,574.3

36,322.6

2,996.83,545.62,509.63,253.7277.2

1,065. 13,939.5

-12,743.4-4,802.2

17,587.5

5,741.316,350. 11,186.58,236.2

11,013.9573.0

1,919.3479.7

-46,069.0

45,500.0

Source: Devarajan and Sierra (1985).

\_l Tax rate corresponds to domestic tax revenue (revenue from business tax and excises) per good relative todomestic expenditures (exclusive tax) on the good.


- 11 -

The household accounts show that the rural households spend arelatively large proportion of their budget on the agricultural good.Services are an important expenditure item for the blue-collar andwhite-collar households. Savings rates show considerable variationamong households and are related to the shares of capital income intotal income. The contrast in savings behavior between the rural andurban informal sectors is particularly noticeable.

This section now turns to the basic features of the system ofindirect taxation in Thailand many of which are common to the systemsof indirect taxation in other middle-income developing countries. _!/Several of these features, including substantial rate differentiation,cascading, and discrimination against international trade, are inconsis-tent with the policy objectives of efficiency, equity, and internationalcompetitiveness.

The business tax and excise taxes are modeled as ad valorem taxes.The tax rates differ among goods according to the observed tax revenuefrom each good. Table 1, Part A, indicates that the system of domesticindirect taxation is characterized by substantial rate differentiation,which creates a number of problems for efficiency, equity, and competi-tiveness.

Regarding the efficiency objective, effective tax rates shouldsystematically vary among goods. If the effective tax rates are notsystematically related to elasticities of demand and supply, however,economic decisions could well be more severely distorted than under auniform rate structure. International competitiveness may also behurt; when a large part of the nontradable service sector is exempt,production factors leave the tradable sectors. Regarding the equityobjective, differentiated rates are likely to be less effective inserving tnis objective than commonly presumed because the differentiatedrates may be shifted backwards.

Both the excise taxes and the business tax are basically cascade-type manufacturers' taxes and apply to both domestic intermediate demandand domestic final demand. Thus, the tax rate pyramids when goods movethrough the channels of production. This process of cascading distortsintermediate demand and supply decisions. Moreover, the rate differen-tiation that results from cascading may distort final demand andsupply decisions more than necessary and lead to uncertain—and oftenundesirable—distributional effects. Cascading also impedes exportsbecause the tax component of input costs is typically not refunded whengoods are exported.

!_/ For a more detailed discussion of the systems of indirect taxa-tion in developing countries, see Gandhi (1977) and Tanzi (1983).


- 12 -

As regards trade taxes, the export tax on agricultural productsis an ad valorem tax on foreign demand for the agricultural product.Ad valorem taxes on demands for the foreign good by domestic industriesana by the Armington sectors represent the import tariffs on interme-diate and final imports, respectively. Trade taxes produce distortionsin consumption and production decisions as well as inequities; theybenefit sectors that produce import substitutes at the expense ofsectors that depend on exports. In addition, trade taxes hurt theobjective of export-led growth. During the usual process of develop-ment, trade taxes gradually lose their significance for revenue andprotection. Domestic economic activities take over as an importantsource of revenue. Domestic industries mature and are able to faceinternational competition, thereby reducing the need for protectionthrough trade taxation. At the same time, the development of strongexport-oriented industries demands lower taxes on international trade.

Corporate taxes and income taxes are modeled as follows. Corporatetaxes appear as taxes on capital employed in the production sectors,while income taxes appear as taxes on labor income received by house-holds.

The paper now turns to the specification of the elasticities thatdescribe marginal behavior of the sectors. _i_/ For each household,income and price elasticities completely determine marginal behavior.Lacking information on tne income elasticities, the simulations assumeunitary income elasticities. Labor supply is fixed.

The following formula determines the price elasticities for eachhousenold:

Here e^-, is the uncompensated price elasticity of good i with respectto the after-tax price of good j. The first term at the right-handside stands for the compensated price elasticity while the second termrepresents the income effect. c.: and n^- are the income share and theincome elasticity of good j, respectively. The elasticities of substi-tution, Oji, are determined by nested Constant Elasticity of Substitu-tion (CES) utility functions. Nested utility functions allow for themodeling of differentiated substitution possibilities by definingvarious levels of aggregates and specifying distinct elasticities ofsubstitution between these aggregates. 2J

77Tne elasticities refer to the reference period, which is abouttnree years. See subsection 2 above (p. 4j.

2J See Keller (1980) for a derivation of (1) as well as for a moredetailed discussion on nested CES utility functions.


- 13 -

figure 1 reports the nested utility structures along with theelasticities of substitution adopted for the domestic private house-holds. The unitary elasticity between total consumption and capitalgoods (i.e., savings) implies a rixed propensity to save independent ofthe rate of return. I/ Most studies on consumer demand systems findmoderate substitution between consumption goods; the (partial) elas-ticity of substitution of 0.5 between consumer goods reflects thisfinding. Public demand is price inelastic, which is represented by azero substitution elasticity.

The trade elasticities reflect the ease of substitution betweendomestic and foreign goods. The utility structure of tne foreign house-holds corresponds to export-demand elasticities of 6. At the importside, the elasticities of substitution between domestic and foreigngoods in the Armington sectors are fixed at 2. Because of the lack, ofempirical estimates for Thailand itself, these values for the tradeelasticities are based on the best judgment, given empirical evidenceon other countries in Goldstein and Khan (1978) and Dervis, de Melo,and Robinson (1982).

For each production sector, price elasticities describe marginalbehavior. The price elasticity of good i with respect to after-taxprice j, E J J , is defined at a constant output level ana is given by

Here, GJ is the cost share of good j and O'-M is the elasticity ofsubstitution. For each domestic production sector, the nested CESstructures and the substitution elasticities that determine the priceelasticities are presented in Figure 2. The unitary elasticity betweentotal product acid (distribution) services implies fixed costs shares.Devarajan and Sierra (1985) provided the substitution elasticitiesbetween the factors of production, which reflect moderate substitution.The other substitution elasticities indicate more limited substitutionbetween aggregate intermediate input ana value added, between domesticand imported intermediates, and between domestically produced inter-mediate inputs.

if Empirical studies on savings behavior in developing countrieshave not resolved whether an increase in interest rates will raise thesavings rate. See, for example, Ebrill (1984).


- 14 -

III. Simulation Results I/

1. Introduction

a. Tax instruments

This section reports the simulation results induced by raisingseparately eight taxes—six domestic taxes on goods and services andtwo trade taxes. The tax measures are normalized such ttiat real publicconsumption can increase by 1 percent.

The first domestic tax is an income-type value-added tax (.VAT)with uniform rates and a tax base that comprises both domestic consump-tion demand and domestic investment demand but excludes intermediatedemand. Given inelastic labor supply, the effects of this tax areequivalent to the effects of a lump-sum tax. This tax is the referencetax; it does not induce any new incentive effects or distortions becauseagents do not affect the tax paid by changing their behavior and becauseit distributes over the domestic households. Comparing the effects oftax instruments with those of the reference tax isolates the incentiveand distributional effects of these tax instruments.

The other four types of VAT that are simulated are consumptionbased; they apply zero rates to investment demand and intermediatedemand. The first is an ideal-consumption VAT (or expenditure tax),which applies the same tax rate to all consumption goods. The sectionthen examines a differentiated rate structure for consumption goodsas well as the difference between a zero rate and an exemption, bysimulating the effects of both zero rating and exempting agriculture.There is an important difference between zero rating and exemption.In contrast to a zero-rated sector, an exempt sector pays tax on itsintermediate inputs and must either pass the tax forward or absorb it.The fourth VAT exempts the service sector, which for administrativereasons is difficult to tax. The last domestic tax is the existing taxon domestic transactions. The simulated trade taxes are a uniformimport tariff and an export tax on the agricultural good.

b. The tables

Tables 2, 3, and 4 contain the simulation results for the eighttax instruments. The simulation results in the tables use the initialequilibrium (before any tax changes) as the benchmark.; the changescontained in the tables are measured with respect to the initial

JV Given the technical nature of this section, the reader who ismainly interested in the policy implications of the simulation resultscan skip this section and continue with Section IV, which summarizesthe major lessons and policy implications from this study.


- L4a -

FIGURE 1

UTILITY STRUCTURES OF DOMESTICHOUSEHOLD SECTORS

1 Giveninelasticsupply


- 14b -

FIGURE 2

PRODUCTION STRUCTURES OF DOMESTICPRODUCTION SECTORS

1 For the agricultural sector these elasticities are 0.9.


Table 2. Effects on Revenue and Relative Prices

A. Change in tax wedge

Incomebased

Consump-tion based

relative to the initialfor 1 percent increase in public

B. Relative changes in

AgricultureConsumption goodsCapital goodsIntermediate goodsInfrastructureServices

C. Relative changes in

InvestmentAgricultureConsumption goodsCapital goodsIntermediate goodsInfrastructureServices

0. Relative changes in

Average capital rentalRural laborUrban laborBlue-collar laborWhite-collar labor

0.270

consumption

0.353

VATAgriculturezero rate

Agriculture Servicesexempt exempt

ExistingDomestic

TaxStructure

UniformImportTariff

ExportTax

Agriculture

after-tax price (in percent)

0.559 0.490 0.358 31.92 I/ 1.108 102.4

producer prices (in percent x 100)

-12.6-3.94.34.34.515.2

final demand

30.814.423.430.429.231.541.6

factor prices

-8.8-15.1-4.421.3259.5

-17.4-5.67.94.77.9

15.6

prices (in

6.018.030.141.637.743.250.2

-8.9-3.74.64.24.012.6

percent x 100)

3.5-8.952.459.158.059.968.0

-5.9-2.83.13.52.812.0

2.9-5.946.451.550.851.860.6

-19.8-8.43.52.12.827.0

3.716.127.938.536.838.525.9

-23.71.14.59.64.8

16.8

18.8-2.67.120.511.420.622.3

-3.711.541.439.934.531.7

48.5-3.417.955.575.134.534.9

-3,442.0-1,536.8-255.0-609.9-242.5-363.9

-365.9-3,432.1-1,437.2-203.2-306.6-242.5-349.1

(in percent x 100)

-9.2-20.70.126.0259.5

-8.8-10.9-6.918.8254.0

-11.8-15.2-8.717.0253.0

-15.0-23.4-7.418.8249.5

-24.6-31.5-18.47.3

246.2

-7.6-11.14.030.7249.2

-1,462.8-3,864.9

296.4377.9568.3

Source: Author's simulations.

I/ refined as the change in the tax wedge relative to its initial level.©International Monetary Fund. Not for Redistribution

Table 3. Effects on Production Structure

A. Changes in capital

Average rentalAgricultureConsumption goodsCapital goodsIntermediate goodsInfrastructureServices

B. Changes in exports

AgricultureConsumption goodsCapital goodsIntermediate goodsServices

C. Changes in imports

Imports

Incomebased

earnings in

-8.8-15.4

2.5-9.0-6.3-4.4-0.8

relative to

3.96.9-0.6-3.8-18.2

relative to

-3.7

Consump-tion based

each sector

-9.2-21.03.20.6-3.24.5-3.4

final demand

5.49.8-1.1-4.3-18.7

its initial

0.7

VATAgriculture Agriculturezero rate exempt

Servicesexempt

ExistingDomestic

TaxStructure

UniformImportTariff

ExportTax

Agriculture

(in percent x 100)

-8.8-10.9-11.1-4.9-6.9-3.7-9.1

in each

2.86.5-0.6-3.8-15.2

level (in

-3.7

-11.8-15.3-14.5-5.7-12.6-5.5-9.1

sector (in percent

1.84.9-0.4-3.1-14.5

percent x 100)

-5.9

-15.0-23.52.0

-5.6-9.7-7.5-15.3

x 100)

6.114.8-0.5-1.9-32.4

-3.8

-24.6-31.9-15.6-20.6-32.6-21.4-8.5

7.3-1.9-0.6-8.6-20.2

-13.1

-7.6-11.2-18.035.0-38.1-0.5™*6» 8

1.2-19.9-5.7

-35.9-37.9

-90. 2

-1,462.8-3,935.33,655.8-934.5465.4-340.3251.5

-2,102.82,711.7

35.4552.2439.0

-477.3



Table 4. Welfare Ef fec t s

VATIncome Consump- Agriculture Agriculture Servicesbased tion based zero rate exempt exempt

ExistingDomestic

TaxStructure

Uniform ExportImport TaxTariff Agriculture


\J National income includes income of the public sector, but excludes income of the foreign sector.2J Aggregate income comprises the incomes of all household sectors including the public sector and the foreign sector.

A. Changes

PublicOwn-accountOwn-account

in real income relative

ruralurban

Informal ruralInformal urbanBlue collarWhite collarRest of the

B. Change

C. Change

D. Change

world

in real

in real

in real

+100.0-37.3-34.2-38.5-34.3-11.2155.6-5.6

national income

83.8

aggregate income

-31.8

aggregate income

-2.5

to its initial level

100.0-38.8-31.2-50.6-21.6-9.9153.4-4.5

relative

84.0

relative

-10.2

relative

-0.8

100.-31.-37.-39.-22.-19.143.-5.

(in percent x 100)

00214870

to its initial level

65.

to initial

-38.

0

national

2

to change in public

-3.0

100.0-32.7-36.5-40. .9-23.4-18.2145.7-5.3

(in percent

43.1

income (in

-66.7

consumption

-5.2

100.0-38.7-31.9-47.9-24.6-9.4151.9-4.9

x 10

71.2

100.-38.-34.-39.-35.-10.156.-6.

,000) I/

44.

08517055

6

100.0-36.0-32.7-29.9-37.7-2.6146.0-25.5

242.2

100.0-1,215.8

610.6-1,941.3

386.61,406.71,093.8-117.7

-6,864.9

percent x 10,000) 2/

-31.3

(in

-2.5

-89.

percent)

-7.

1

0

-286.8

-22.4

-9,305.1

-727. 0


equilibrium. Except for the discussion of the income-based VAT insubsection 2 below, the discussions of the tax instruments will use anequilibrium after an alternative tax instrument has been raised toincrease real public consumption by 1 percent as the benchmark.. Sucna comparison isolates the differential incentive and distributionaleffects of taxes and amounts to comparing two columns in the tables.

Table 2 contains the results for revenue and relative prices.Part A indicates the revenue effects by presenting the change in thetax wedge (the difference between the before-tax price and the after-taxprice) as a percentage of the after-tax price tnat is required to raisepublic consumption by 1 percent. Because the existing tax on domestictransactions is characterized by a multiplicity of rates, however, theabsolute change in the wedge depends on the size of the initial taxrate. Therefore, the change in this tax instrument is defined as thechange in the tax wedge as a percentage of the initial tax wedge.Part b contains the effects on the relative after-tax supply prices ofdomestic goods, which the domestic industries receive. The consequencesfor the demand prices of the investment good and the Armington goods,which the domestic households pay, are presented in Part C. Relativechanges in these demand prices indicate that taxes are shifted forward.Part D contains the effects on wages and average capital earnings.Given the equality between costs—including profits—and revenues,factor prices of production factors employed in a particular industrytend to be positively related to after-tax supply prices of tne goodproduced in that industry and negatively related to the demand prices ofintermediate inputs. Relative changes in these factor prices indicatethat taxes are shifted backward. All price changes are measured withrespect to the fixed world price of foreign goods.

Part A of Table 3 reports the effects on the production structure.It presents the changes in the capital earnings Cor profitability) ineach domestic production sector, which provide an indication for theeffect of tax policy on the interindustry allocation of capital; overtime, capital moves toward those sectors in which the capital rentalrises relative to average capital earnings. Part & of Table 3 containsthe consequences for international trade—exports of each industry andtotal imports.

Table 4 contains tne welfare effects. This paper uses negative"compensating variations" to measure the income effects for the indivi-dual households. The compensating variation corresponds to the transfer,following the tax change, that is required to maintain the household'sutility from private goods at its initial level. Thus, a positivecompensating variation reflects a negative income effect. The incomeeffect for private households does not include the welfare effects fromthe 1 percent change in public consumption. This effect, however, is

- la -


- 19 -

reflected in the income effect for the public household. The aggregateincome effect is computed as the sum of all the income effects. Tnisaggregate measure—the negative "excess burden"—reflects the efficiencyeffects of a tax change; a positive efficiency effect (or negativeexcess burden) indicates that a Pareto improvement might be realized sothe gainers could compensate the losers.

The distribution of the income effects over the households reflectsdistributional consequences. By assigning "social" weights to thesedistributional effects, social welfare can be computed. The nationalincome effect assigns a zero weight to the foreign households; it isthe sum of the domestic income effects. The aggregate income effectdefined above attaches a uniform weight to all households—includingthe foreign household. Therefore, this measure reflects the conse-quences for world efficiency. It is expressed both as a percentage ofnational income and as a percentage of the.increase in public consump-tion. This latter percentage corresponds to the "welfare" cost ofdifferent tax instruments per unit of additional real tax revenueraised.

In order to interpret the aggregate efficiency effect, equation (3)writes the aggregate income effect relative to national income I withrespect to its origin

Here t1 is the vector of shares of transaction taxes paid by (householdor firm) sector i in national income and reflects the existing distor-tions in the economy. The symbol q1 stands for the vector of relativechanges in real demand or supply by sector i. The formula shows thatthe efficiency effect is positive (negative) when the volume of taxedtransactions increases (decreases) or when resources are relocatedtoward a more (less) heavily taxed sector. The intuition is thatbecause the marginal benefit of taxed items exceeds the marginal costs,larger transactions raise real income. In the presence of severalexisting distortions, the magnitude and sign of the aggregate efficiencyeffect is determined in a complicated way and often is not intuitivelypredictable. The methodology presented in this paper allows for theexplicit examination of the complex interaction between existing distor-tions and tax policy.

Expression (3) not only illustrates the interaction among taxdistortions in a second-best world, but it also provides a link withthe revenue effects. The elements in the expression for the efficiencyeffect correspond to the changes in tax revenue that are induced bythe changes in the tax base. Thus, the efficiency effect reflects thechange in tax revenue beyond tne change in revenue assuming unchangedtax bases.


- 20 -

2. Income-based VAT

This subsection discusses the changes in the initial equilibriumthat are induced by the reference tax: an income-based VAT. I/ Thechanges in the initial equilibrium are heavily influenced by the composi-tion of public expenditure because the tax increase involves a transferof purchasing power to the public sector.

An examination of the supply prices of domestic goods (Table 2,Part B) shows that the producer prices of agricultural and consumptiongoods decline, while the supply price of services increases most. Thedeclining price for agricultural goods is explained by the transfer ofpurchasing power from the private households, which demand most of theagricultural good, to the public household, which demands almost noneof the agricultural good (see Table 1). The producer price of theconsumption goods sector falls because intermediate purchases of thecheaper agricultural product account for more than half of the inputcosts in this sector. The costs of the service sector, on the otherhand, rise because increased public demand for white-collar labor raiseswhite-collar wages. This cost effect combined with rising aggregatedemand for services owing to relatively large public purchases causesthe producer price of services to increase.

The relative profitability of capital in each sector (Table 3,Part A) indicates the relevance of the interindustry demand structurefor intermediate goods. Although the public household demands very fewconsumption goods, the consumption goods sector benefits most from theexpansion of public demand because of the declining costs of agricul-tural intermediate inputs.

The exports by sector (Table 3, Part B; reflect the changes inproducer prices, which apart from the export tax on the agriculturalgood are the prices that foreigners face. The sectors with decliningsupply prices—agriculture and consumption goods—increase their exports.The exports in the other sectors decline. Domestic-demand pressure andrising costs hit exported services particularly hard. Imports declinebecause public demand is not import intensive.

World efficiency, which is represented by the aggregate incomeeffect (Table 4, Parts C and D), tends to be positively related to thevolume of imports, because imports are relatively heavily taxed (seeequation (3)). Thus, declining imports reduce efficiency. This effi-ciency effect illustrates that, although a lump-sum tax has no conse-quences for efficiency in a nondistorted economy, it certainly doeshave consequences for efficiency in a distorted economy.

ITThese changes are presented in the first columns of Tables 2-4.


3. .Broad—based consumption VAT

This subsection compares a consumption-based VAT, which zero ratesinvestment goods, with an income-based VAT, discussed in the previoussubsection. The discussion in this subsection measures ali effects ofthe consumption-based VAT with respect to the effects of the income-based VAT. JY In doing so, the discussion focuses on the differentialincentive and distributional effects of the consumption-based andincome-based VATs. By comparing two taxes that yield the same changein real public expenditure, the discussion is not affected by the compo-sition of public expenditure.

Zero rating the investment good increases the producer prices ofthe industries that supply investment goods (capital goods and infra-structure) or that supply intermediate goods to these industries (inter-mediate goods) (Table 2, Part B). The producer price of the agriculturalgood, on the other hand, decreases because domestic demand shifts awayfrom the agricultural good toward the investment good.

Not only these relative producer prices, but also the factorprices, indicate that the zero rate is partly shifted backwards(Table 2, Part D; and Table 3, Part A). The production factors thatare mainly employed in the sectors depending on investment demand—urban and blue-collar labor, and capital employed in the capital goodsand infrastructure sectors—gain. Rural labor and agricultural capital,on the other hand, lose. Capital employed in the consumption goodssector gains because of the reduced cost of intermediate agriculturalinputs.

The backward shifting can be quantified by subtracting the firstcolumn from the second column in Tables 2 and 3. Introducing a1 percent consumption VAT instead of an income VAT to raise publicconsumption, increases the producer price of capital goods relative toagricultural goods by about 0.23 percent (Table 2, Part B) and capitalearnings of the capital goods sector relative to the agriculturalsector by about 0.43 percent (Table 3, Part A).

Zero rating investment stimulates both imports and exports(Table 3, Parts B and C). Imports increase because the investment goodconsists primarily of import substitutes and goods produced by import-dependent sectors. Exports increase because investment requires rela-tively few exportables and goods produced by export-dependent sectors.Consequently, zero rating investment reduces domestic demand for thesegoods, thereby reducing the prices that foreigners face.

T/This comparison amounts to the differencebetween the second andfirst columns of Tables 2-4.

- 21 -


- 22 -

World efficiency benefits from increased international tradebecause both the exports of agricultural products and imports are taxedheavily. The efficiency effect, however, is rather small—about1.7 percent of additional real tax revenue (Table 4, Part D). Whilethe foreign sector benefits from the increases in export supply andimport demand, real national income is practically unaffected.

Domestic distributional effects indicate that both forward andbackward shifting is relevant (Table 4, Part A). Households that haverelatively high savings rates and supply labor that is employed ininvestment goods industries—the urban sectors—gain. The informalrural sector, which does not save and depends on agricultural labor,suffers more severely than the own-account rural sector, which benefitsfrom its relatively high savings rate but suffers from the decliningearnings in the agricultural sector. Relative to the efficiencyeffects, the equity effects are substantial. To illustrate, substi-tuting a consumption-type VAT for an income-type VAT to finance1 percent of public consumption, decreases real income of the informalrural sector by 0.12 percent while raising real income of the informalurban sector by the same percentage.

4. Zero rating agriculture

This subsection analyzes the consequences of narrowing the tax baseof the consumption-based VAT by zero rating the agricultural good. Itcompares the effects of this narrow-based VAT with the broad-based con-sumption VAT, which was discussed in the previous subsection. _!/ Zerorating the agricultural good is often advocated because of distribu-tional reasons.

The increases in the relative producer price of the agriculturalgood, the rural wage, and the profitability of the agricultural sector,indicate that the zero rate is partly shifted backwards. The producerprices of the nonagricuitural domestic goods decrease except for thesupply price of the consumer good, which increases owing to the highercost of the agricultural intermediate input (Table 2, Part B).

Capital and labor employed in the nonagricuitural sectors sufferfrom the shift of domestic demand to the agricultural good (Table 2,Part D; and Table 3, Fart A). The differential rate structure hitsthe consumption goods sector particularly hard. This sector not onlysuffers from decreased domestic demand but also from increased inputcosts because of the higher price of agricultural inputs.

The effect of the zero rate for the agricultural good on interna-tional transactions and world efficiency is just the reverse of theeffects of the zero rate for the investment good. International trade

JY This comparison amounts to the difference between the third andsecond columns of Tables 2-4.


declines because the zero-rated good is now produced by an export-related sector and not by import-intensive sectors (Table 3, Parts Band C).

Increased domestic final demand for tne agricultural good crowdsout direct foreign demand as well as intermediate demand from theexportable consumption goods sector. Imports decline because domesticdemand shifts away from import-intensive goods. World efficiencysuffers not only from the reduction in heavily taxed imports but alsofrom the shift in the composition of exports away from the heavilytaxed agricultural exports. Both the foreign sector and the domesticeconomy share in the worldwide welfare loss, which is a little over2 percent of additional real tax revenue (Table 4, Parts C and D).

The domestic distributional effects differ from the equity effectsof zero rating the investment good also (Table 4, Part A). Ruralhouseholds gain while the urban sectors lose. Because the relativeexpenditure shares on the agricultural good vary less between householdsthan savings shares do, backward shifting becomes even more important.Own-account and informal rural households gain 0.08 percent and0.11 percent of their real incomes, respectively, while all other house-holds lose. Their real income losses vary between 0.01 percent(informal urban household) to almost 0.10 percent (blue- and white-collarhouseholds).

The following conclusions emerge on the effects of a differentialrate structure for domestic consumption taxation on equity, industrialstructure, international trade, and efficiency.

a. Differential rates are partly shifted backwards, particularlywhen supply is inelastic, intermediate and foreign demand is inelastic,final domestic demand is elastic, and the good with a differential rateis nontradable and primarily demanded by domestic consumers. Moreover,backward shifting is more powerful than forward shifting in affectingthe income distribution because earnings proportions vary more amonghouseholds than consumption proportions do. Thus, differential ratescan be more effective in altering the distribution from the source sidethan from the user's side.

b. Lower rates for a particular good benefit the sector producingthat good when the tax is shifted backwards (see above). The otherindustries suffer. Tradable industries suffer from the cost side whenthey depend on a low-rate nontradable good that is inelasticallysupplied for their intermediate inputs. Nontradable industries sufferfrom the demand side when their output is highly substitutable by thelower rate good while depending on final domestic consumption demand.The more tradable and the more elastically supplied the low-rated goodis, the more the nontradable sector suffers.

- 23 -


- 24 -

c. Lower rates on domestic demand for exportables or goodsproduced by export-dependent sectors act as implicit export taxes bycrowding out foreign demand. Lower rates on domestic demand for importsubstitutes or goods produced by import-dependent sectors act as importsubsidies and benefit international trade.

d. In countries with significant trade taxation, the interactionof the rate structure of domestic consumption taxes with the distortionsinduced by trade taxes is more important for the efficiency objectivethan the distortions introduced by the differential structure itself.A differential rate structure ttiat enhances international trade—with,for example, relatively high rates on domestic demand for nontradablesand exportables—can benefit the efficiency objective.

5. Exempting agriculture

Tnis subsection compares exempting the agricultural good witn zerorating it. All the effects of the exemption for agriculture are measuredwith respect to the effects of the zero rate for this sector. _!/ Theproducer prices of the domestic goods, except for the agricultural goodand the consumer good, decrease relative to world prices (Table 2,Part B). The real exchange rate—domestic producer prices in terms ofthe foreign good—falls because exempting the agricultural sector hurtsexports (see further).

The relative supply price of the agricultural good increases owingto the rising after-tax costs of intermediate inputs, which are taxed.The rising price of the agricultural good raises the input costs forthe consumer goods sector, thereby increasing the supply price ofconsumer goods.

Exempting instead of zero rating the agricultural sector hurts theagricultural and the consumption-goods sectors because of the risingcost of intermediate inputs for these sectors (Table 3, Part A). Theintermediate-goods industry, which is import dependent, suffers fromthe larger cost of both agricultural and foreign intermediate inputs.The import-competing capital goods sector benefits from the decliningreal, exchange rate.

International trade suffers from the exemption (Table 3, Parts Band C). Exports decline because the tax component of intermediateinputs in the exportable or export-related industries—agriculture andconsumption goods—is not rebated. Imports decline because the realexchange rate drops.

77This comparison amounts to the difference between the fourth andthird columns of Tables 2-4.


- 25 -

The decline in taxed imports aad agricultural exports hurtswelfare—both domestically and worldwide (Table 4). Exempting agricul-ture instead of zero rating it, hurts total and domestic real income by2.2 percent and 1.7 percent of additional real tax revenue, respectively.

6. Exempting services

This subsection continues to analyze tne effects of exemptions.It compares the effects of a consumption VAT that exempts the servicesector with the broad-based consumption VAT discussed in subsection 3. I/

The consequences for the relative commodity prices are thefollowing (Table 2, Parts B and G). The producer prices and the demandprices of the nonexempt goods decrease because of a rising real exchangerate. The producer price of services increases due to rising after-taxcosts. Because, in contrast to the broad-based VAT, no taxes are leviedon the producer price of services, the demand price of services drops.

In contrast to the exemption of agriculture, the exemption ofservices negatively affects the profitability of the exempted sector(Table 3, Part A). The service sector suffers from its exemption fortwo reasons. First, the production costs in an exempted sector risebecause its intermediate inputs are taxed. Because the share of inter-mediate input costs in total production costs is large in the servicesector, its costs increase substantially. Second, elastic foreign anddomestic intermediate demands shift these increased costs backwards tothe service sector. Thus, the service sector benefits when it isincluded in the VAT net, which illustrates the self-enforcing characterof a VAT that is based on the credit system.

The exemption affects the nonexempted industries as follows. Thedeclining real exchange rate hurts the import-dependent sectors whileit benefits the nonexempted tradable sectors. The exportable and export-dependent sectors gain relative to the import-competing sectors becausethey are less import dependent.

In analogy to the exemption of the agricultural good, the exemptionof services hurts exports, imports, and efficiency (Table 3, Parts Band C; and Table 4, Parts C and D). However, efficiency suffers lessfrom exempting services than it does from exempting agriculture.Exempting services instead of agriculture raises aggregate real incomeby 2.7 percent of additional real tax revenue. (Compare the fourth andfifth columns in Table 4, Part D.) This is a surprising result whenonly domestic demand distortions are considered because domestic taxesimpose higher tax rates on agricultural commodities than on services

T7This comparison amounts to the difference between the fifth andthe second columns of Tables 2-4.


- 26 -

(Table 1, Part A). Thus, reducing the tax on the agricultural goodmakes the rate structure more uniform and—with equal demand elas-ticities—reduces domestic-demand distortions. The general equilibriumframework, however, incorporates the distortions induced by trade taxes.It reveals that exempting agriculture aggravates these distortions moreseverely than exempting services because exempting agriculture hitsheavily taxed imports and agricultural exports more severely. Theresulting efficiency loss from exempting agriculture more than offsetsthe efficiency gain from reduced domestic-demand distortions.

The following conclusions emerge on the consequences of exemptinga production sector from consumption taxation for efficiency and indus-trial structure:

a. From an efficiency point of view, trade distortions strengthenthe case against exemptions and in favor of a broad tax base. Taxingintermediate inputs in an exempted sector implies an implicit tax ondomestic production which—in contrast to a domestic consumption tax—not only hits domestic demand but also foreign demand. Therefore,taxing intermediate transactions through an exemption not only reducesdomestic production efficiency but also aggravates trade distortions.

b. The exempted sector suffers from the exemption when it (1) hasa low value-added share in gross production, (2) is a tradable sector,(3) depends on elastic intermediate demand (share of final demand inoutput is low), or (4) faces inelastic domestic consumption demand.

The nonexempt tradable sectors suffer when an exempted nontradablegood is an important element in their input costs. They benefit,however, when the exemption of a tradable sector decreases the realexchange rate. The nonexempt nontradable sectors that are importdependent suffer when the exemption increases the relative cost of theforeign good.

7. Existing domestic taxes on goods and services

This subsection analyzes the effects of raising the rates ofexisting domestic taxation on goods and services—the business tax andexcise taxes. The effects are measured with respect to the effects ofthe income-based VAT. JY

J_/ These relative effects can be interpreted as the reverse effectsof a tax package which reduces cascading and rate differentiation.This tax package consists of a cut in the rates of excises and thebusiness tax compensated for by the introduction of an income-based VATsuch that public consumption is unaffected. The comparison in tnis sub-section amounts to the difference between the sixth and first columnsin Tables 2-4.


- 27 -

Trie relative earnings of the production factors (the relativeprofitability of sectoral capital and the real wages) indicate backwardshifting of both the increased costs of intermediary inputs and thedifferential tax rates on output _!/ (Table 2, Part D; and Table 3,Part A). The intermediate and consumer goods sectors, as well as agri-culture, suffer from existing domestic taxation while the capital goodsand service sectors gain.

Export-oriented sectors with high intermediate-cost shares (consumerand intermediate goods) suffer from cascading. These sectors cannotshift the rising cost of intermediate inputs forward because they faceelastic foreign demand. Thus, by decreasing the profitability ofcapital in the export sectors with low value-added shares, cascadingimpedes the development of a competitive export sector. It also imposesan arbitrary burden on the production factors that are employed inthese sectors.

International trade suffers from taxing intermediate transactions(Table 3, Parts B and C). Exports in all sectors except agriculturedecline because, analogous to an exemption, taxing intermediate trans-actions raises the cost of domestic production. Because the export-oriented sectors (consumer goods and intermediate goods) have highintermediate-cost shares, taxing intermediate transactions hurts exportsparticularly hard. Agricultural exports increase since both domesticintermediate demand from exportable industries and domestic finaldemand decrease owing to the high tax rate on domestic demand. Thus,cascading does not necessarily hurt exports in each industry. Inparticular, export-related industries with inelastic supply may increasetheir direct exports when taxes on domestic intermediate transactionsdecrease intermediate demand from export industries. Turning to imports,import demand for intermediate goods declines because of the tax onintermediate transactions. In addition, import demands suffer from thedeclining real exchange rate.

World efficiency decreases by about 3.5 percent of additionalpublic utility (Table 4, Parts C and D). About 65 percent of thisdecline is attributable to a fall in imports, which hurts the efficiencyof international trade. The other 35 percent is almost entirelyaccounted for by declining efficiency in domestic production. Domesticproduction efficiency suffers because the tax on intermediate inputsreduces intermediate transactions for which marginal benefits exceedmarginal costs.

I/ Table 1, Part A contains these differential tax rates.


- 28 -

8. Import tariff

This subsection compares the effects of a uniform import tariffwith the effects of the income-based VAT, which was discussed in sub-section 3. \J

The effects on the relative producer prices, capital earnings, andwages are as follows (Table 2, Parts B and D; and Table 3, Part A).The producer prices of import substitutes (capital goods) and goodsproduced by import dependent industries (intermediate goods, infrastruc-ture) rise relative to the producer prices of exportables (consumergoods). The profitability of the exportables (consumer goods) suffersfrom the rising real exchange rate, while the capital earnings in theimport-dependent industries suffer from the rising costs of the foreigngood. The profitability in the intermediate goods industry declinesmost severely because it is import dependent and also produces anexportable. The relative capital earnings in this sector decline by0.3 percent for a 1 percent increase in the import tariff. The industryproducing import substitutes (capital goods) benefits from the protec-tion that is provided by the tariff. Its relative profitabilityincreases by almost 0.4 percent for a 1 percent increase in the tariff.

The decline in international trade severely hurts world efficiency(Table 4, Parts C and D). The incremental welfare cost amounts toabout 22 percent of additional tax revenue. This relatively large lossin efficiency reflects the revenue losses that result from the erosionof the tax base. There is, however, a trade-off between national andworld welfare because the national economy can shift the welfare costto the foreign household. Table 5, which contains the welfare effectswhen the export elasticities are 20.0, indicates that when the exportelasticities are larger, the national economy shares in the worldwidewelfare loss.

9. Export tax on agricultural products

The discussion of the export tax on agricultural products will beshort. The changes that the export tax induces are large because thetax rate needs to be raised substantially in order to raise real publicconsumption by 1 percent. This is due not only to the relatively smalltax base but also to the shape of the general-equilibrium export-taxLaffer curve, which has only a small positive slope at the initialexport tax of 15 percent. Because the export tax raises the realexchange rate, it erodes not only the base of the export tax but alsothe base of the import tariffs. Table 5 shows that when the trade elas-ticities are larger, Thailand is on the downward sloping part of theLaffer curve for the export tax.

JY This comparison amounts to the difference between the seventhand first columns of Tables 2—4.


Table 5. Welfare Effects and Export Elasticities

(As percentage of real change in public consumption)

VATIncome Consump- Agriculture Agriculturebased tion based zero rate exempt

Existing Uniform ExportServices Domestic Import Taxexempt Tax Tariff Agriculture

A. Base case (export demand elasticities = 6.0)

National gainAggregate gain

B. Protectionist case

6.55 6.56-2.48 -0.80

(export demand elasticities

National gain 11.72Aggregate gain -2.34

C. Nonprotectionist case (export

National gainAggregate gain

4.69-2.34

5.08-2.98

= 2.0)

7.03 10,16-1.56 -3.13

demand elasticities = 20.0)

6.25 3.13-3.13

3.37-5.21

10.16-4.96

0.78-5.47

5.56 3.48 18.92 -536.32-2.45 -6.96 -22.41 -726.96

10.16 14.06 94.53 65.63-3.13 -7.03 -13.28 -56.25

3.91 — -7.81 145.31 I/-1.56 -5.47 -25.00 157.81


\J In the nonprotectionist case, the export tax on agriculture decreases tax revenue. Thus, the export tax has todecrease in order to raise public consumption by 1 percent.


- 30 -

The export tax is inefficient; i't not only imposes an incrementalworldwide welfare cost of more than 700 percent of additional publicconsumption, but also decreases national welfare by more than 500 per-cent of tax revenue. This decline in national welfare indicates theimportance of second-best considerations: the initial export tax isbelow the optimal export duty of 17 percent derived from the inverseelasticity rule, which is based on first-best analysis.

Concerning the equity effects, the export tax lowers domestic foodprices and the real exchange rate. Consequently, it benefits nontaxedexportables and the urban households.

IV. Policy Conclusions

This section summarizes the major policy conclusions on efficiency,revenue, equity, and international policy coordination. It discussesthe conclusions for tax policy in general and stabilization tax packagesin particular.

1. Tax instruments interact with other distortions in the "second-best" economy, and these interactions are important from an efficiencypoint of view. An "efficient" tax package that accounts for otherdistortions in a heavily distorted economy differs from an efficienttax package that ignores these other distortions.

Therefore, efficient stabilization tax packages should takeexisting trade distortions into account. Many developing countriesrely on trade taxes, which distort international trade. Moreover, inmany of these countries, fixed prices and distortionary policies, suchas overvalued exchange rates, impose implicit taxes that induce addi-tional distortions, including trade distortions. _!/ This study hasshown that, in the presence of trade distortions, efficient tax packagesshould avoid implicitly taxing exports. Exports are implicitly taxedthrough taxes on domestic production, which include domestic and importtaxes on intermediate transactions. Thus, efficient tax packages incountries with significant trade distortions should focus on broad-based domestic consumption taxes, thereby avoiding taxes on intermediatetransactions. These observations indicate that zero rating a sector for

J7 Dervis, de Melo, and Robinson (1982), for example, show in ageneral equilibrium study on Turkey that overvalued exchange rates andimport rationing induce relatively large efficiency losses, particu-larly when accompanied by rent seeking. These results suggest that inmany developing countries the efficiency losses from implicit taxesdominate the efficiency costs from explicit taxes. For a more generalway to model "implicit" taxes in a general equilibrium framework, seeCornielje and Keller (1984).


- 31 -

domestic taxes is more efficient than exempting it from these taxes—particularly when the sector is export-related, exportable, or import-related. In addition, this study has shown that, in the presence oftrade distortions, a differential rate structure for domestic taxes,with higher rates for nontradables and exportables. and. lower ratesfor import competing and import dependent sectors, is more efficientthan a uniform rate structure; ignoring explicit and implicit tradetaxes, a uniform rate structure seems more efficient.

Another illustration of the importance of second-best considera-tions concerns optimal export taxation. Export taxes that raise realnational income in a country at the expense of real income of the restof the world (or world efficiency) may be considered country efficientin the absence of other trade distortions. The same export taxes,however, may not only hurt world efficiency more severely but may alsoreduce real national income in the country concerned in the presence ofother trade distortions.

2. The interaction among taxes, distortions, and various othereconomic policies is important from a revenue point of view. In orderto reach the revenue objectives for a particular tax instrument, otherinstruments of economic policy need to be supportive. To illustrate,the significant base erosion caused by an import tariff (Section III,subsection 3) reflects the negative effects of an overvalued exchangerate on tax revenue in countries that heavily depend on trade taxation.An illustration of the interaction among various tax instruments isthe revenue effects of the export tax on agricultural commodities.These effects indicate that an economy that heavily relies on tradetaxes can be on the downward-sloping segment of the export-tax Laffercurve, not only because of the eroding base for the export tax itself,but also because of the additional effect of a depreciating exchangerate on the base of import tariffs.

The base erosion that is induced by trade taxes has importantimplications for the revenue estimates in stabilization tax packagesraising trade taxes in countries that already rely heavily on tradetaxes. These revenue estimates should account for the base erosionthat is induced by higher trade taxes; ignoring these effects resultsin seriously overestimating revenue.

3. From an equity point of view, the backward shifting of indirecttaxes may be as important, if not more important, as forward shifting,particularly in relation to the distribution between rural and urbansectors. To illustrate, zero rating of investment goods benefits theurban sectors, which depend on earnings from sectors producing invest-ment goods, whereas zero rating of agricultural goods benefits therural sectors.


- 32 -

Thus, rate differentiation, particularly for nontradables withinelastic supply, is ineffective in redistributing income from thedemand side. However, policies that affect factor prices can be effec-tive in redistributing income from the supply side.

4. The sensitivity of national welfare gains to the price elasticitiesof export demand reveals an important link between trade policies inindustrial countries and tax policy in the developing countries. Whenexport elasticities are large, developing countries benefit in the shortrun from tax policies (lower import taxes as well as lower explicit andimplicit export taxes, including production taxes) that simultaneouslyimprove competitiveness and world efficiency, thereby benefiting theworld economy. When export elasticities are small, however, developingcountries may suffer short-run welfare losses from tax policies thatimprove competitiveness and world efficiency. They may, therefore, bediscouraged from adopting such policies. Because removal of tradebarriers by the industrial countries will raise the export elasticitiesthat developing countries face, these nonprotectionist policies providesignificant incentives for developing countries to formulate outward-looking supply-side tax policies that benefit themselves as well as theworld as a whole.


- 33 - APPENDIX

Algebraic Presentation of tne Model

This Appendix contains an algebraic description of the model.Both the description and the model rely heavily on Keller (1980). Themodel contains 6 household sectors (including the foreign and publichouseholds), 13 production sectors (including the investment sector andthe Armington sectors), and 25 goods and factors (including theArmington goods and sector-specific capital).

1. Households sectors

Tne column-vector q™ contains the relative changes in the demands(supplies are measured negatively) of household i for all 25 goods.The household-specific demand functions are given by

Here, X is a scalar representing the relative change in revenuefrom transaction taxes, for the public household nL.is the 25-vector ofincome elasticities ,. for the nonpublic households n1 is a 25-vector con-taining zeros, and NH is a 25x25 matrix of uncompensated price elastic-ities of household i (defined according to equation (1) in Section II,subsection 3).

Tne symbol pH stands for the 25-vector of after-tax prices facinghousehold i. These prices are related to the relative changes inmarket prices, which are contained in the 25-vector pM, and to thechanges in the 8 tax . instruments, which are contained in the 8-vector t,by the 25x8 matrix T̂ :

The 25-vector q^ contains the relative changes in demands of theaggregate household sector, which is found by aggregating over the8 households:

Here, a,, stands for a diagonal 25x25 matrix with the shares of house-hold sector i in aggregate household demands on its diagonal.


- 34 - APPENDIX

Substituting (A.2) and (A.i) into (A.3), aggregate householdbehavior is described by

Here, the superscript p represents the public household.

2. Production sectors

The column-vector qp contains the relative changes in the supplies(demands are measured negatively) of firm j. The firm-specific supplyfunctions are given by

with

Here, qsj stands for the output level of firm j, i is a 25-vector thatconsists of unit elements, and Np is a 25x25 matrix of uncompensatedprice elasticities for firm j (defined according to equation (2) inSection II, subsection 3).

Pp stands for the 25-vector of after-tax prices facing firm j. .These prices are related to the tax instruments by the 25x8 matrix Tp:

The zero-profit condition for firm i is described by

Here, ci is the 25-vector of cost shares of f i rm j.


- 35 - APPENDIX

The 25-vector qp contains the relative changes in supplies of theaggregate production sector, which is found by aggregating over the13 firms:

Here, ap represents a diagonal 25x25 matrix with the shares of firm jin aggregate firm supplies on its diagonal.

Using (A. 8), (A. 9), (A. 10), and (A.ll), aggregate firm behavior isdescribed by

In order to arrive at a closed-form solution for the aggregatefirm sector, the 13-vector qg, which contains the output levels of thefirms, is eliminated from (A.12) to arrive at 25 - 13 = 12 independentequations in qp Together with (A. 13) these equations yield 25 indepen-dent equations for the aggregate production sector.

with

and

is the 25-vector with shares of firm j in aggregate firm supplies.


- 36 - APPENDIX

3. Equilibrium

Combining the equations for the aggregate household and productionsectors with the equilibrium condition

and fixing the world price of the foreign good, the price-vector p^and total tax revenue X can be solved for.


- 37 -

References

Armington, P., "A Theory of Demand for Products Distinguished by Placeof Production," Staff Papers, International Monetary Fund (Washington,D.C.), Vol. 16 (July 1969), pp. 159-78.

Beveridge, W.A., and M.R. Kelly, "fiscal Content of Financial ProgramsSupported by Stand-By Arrangements in the Upper Credit Tranches,1969-78," Staff Papers, International Monetary Fund (Washington, D.C.),Vol. 27 (June 1980), pp. 205-49.

Bovenberg, A.L., "The General Equilibrium Approach: Relevant for PublicPolicy?" Paper presented to the 41st Congress of the InternationalInstitute of Public Finance, Madrid, Spain (August 1985).

, and W.J. Keller, "Nonlinearicies in Applied General EquilibriumModels," Economics Letters, Vol. 14 (February 1984), pp. 53-59.

Cooper, R., and K. McLaren, "The Orani-Macro Interface: An IllustrativeExposition," The Economic Records, Vol. 59 (June 1983), pp. 166-79.

Cornielje, O.J.C., and W.J. Keller, "Exploring Malinvaud-type Disequilib-rium Models Using Virtual Taxes," Discussion Paper, Free UniversityAmsterdam (Amsterdam, 1984).

Dervis, K. , J. de Melo, and S. Robinson, General Equilibrium Models forDevelopment Policy (New York.: Cambridge University Press, 1982).

Devarajan, S., and H. Sierra, "Growth Without Adjustment: Thailand,1973-1982," Research Paper, World Bank (Washington, D.C. , 1985).

Ebrill, L.P., "The Effects of Taxation on Labor Supply, Savings, andInvestment in Developing Countries: A Survey of the Empirical Litera-ture," International Monetary Fund (Washington, D.C.), DM/84/23 (1984).

Gandhi, V.P. , "Vertical Equity of General Sales Taxation in DevelopingCountries," International Monetary Fund (Washington, D.C.), DM/79/52(1979).

Goldstein, M., and M.S. Khan, "The Supply and Demand for Exports: ASimultaneous Approach," Review of Economics and Statistics, Vol. 60(1978), pp. 275-86.

Johansen, L., A Multi-Sectoral Study of Economic Growth (Amsterdam:North-Holland, 1960).

Keller, W.J., Tax Incidence; A General Equilibrium Approach (Amsterdam:North-Holland, 1980).


- 38 -

Shoven, J. , and J. Whalley, "Applied General Equilibrium Models ofTaxation and International Trade: Introduction and Survey," Journalof Economic Literature, Vol. 22 (September 1984), pp. 1007-51.

Tanzi, V. , "Quantitative Characteristics of the Tax Systems of DevelopingCountries," International Monetary fund (Washington, D.C.), DM/83/79(1983).


Documents

Ij - IMF eLibrary