Tax Policy for Emerging Markets: Developing Countries · In developing countries with emerging markets, and especially in those that aim at becoming integrated with the international

Forum on Tax Policy in Emerging Democracies

299

National Tax JournalVol. LIII, No. 2

Abstract - This paper discusses important tax policy issues facingdeveloping countries today. It views tax policy from both the mac-roeconomic perspective, which focuses on broad questions such asthe level and composition of tax revenue, and the microeconomicperspective, which focuses on certain design aspects of selected majortaxes, such as the personal income tax, the corporate income tax,the value-added tax, excises, and import tariffs. It provides a re-view of the role of tax incentives in these countries, and identifiessome policy challenges posed by the globalization of the worldeconomy.

INTRODUCTION

The study of tax policy is concerned with the design of atax system that is capable of financing the necessary level

of public spending in the most efficient and equitable waypossible. In developing countries with emerging markets, andespecially in those that aim at becoming integrated with theinternational economy, tax policy must play a particularlysensitive role. In these countries, the tax system should: (1)raise enough revenue to finance essential expenditures with-out recourse to excessive public sector borrowing; (2) raisethe revenue in ways that are equitable and that minimize itsdisincentive effects on economic activities; and (3) do so inways that do not deviate substantially from internationalnorms.

In developing countries, the establishment of effective andefficient tax systems faces some formidable challenges. Thefirst of these challenges is the structure of the economy thatmakes it difficult to impose and collect certain taxes. The sec-ond is the limited capacity of the tax administration. The thirdis the paucity, or the poor quality, of basic data. Finally, inmany developing countries the political set up is less ame-nable to rational tax policy than in advanced countries. Itwould take too much space to discuss these challenges inany detail, so a few comments on each of them will have tosuffice.

Developing countries are often characterized by: a largeshare of agriculture in total output and employment; largeinformal sector activities and occupations; many small es-tablishments; a small share of wages in total national income;

Tax Policy for Emerging Markets:Developing Countries

Vito Tanzi &Howell H. ZeeInternational MonetaryFund, Washington, DC20431

NATIONAL TAX JOURNAL

300

a small share of total consumer spendingmade in large, modern establishments;and so on. All these characteristics reducethe possibility of relying on certain mod-ern taxes, such as personal income taxesand, to a much lesser extent, on value-added taxes. They also reduce the possi-bility of achieving high tax levels.

In part as a consequence of the struc-ture of the economy, and in part as theresult of low literacy and low humancapital, it is difficult to combine all theingredients that make for a good tax ad-ministration. When the staff of the tax ad-ministration is not well educated and welltrained, when resources to pay goodwages and to buy necessary equipmentare not there, when the taxpayers havelimited ability to keep accounts, when theuse of telephones is limited and the mailis not reliable, it is difficult to create anefficient tax administration. As a conse-quence, countries often develop tax sys-tems that allow them to exploit whateveroptions they have rather than developmodern and efficient tax systems. Oneconsequence of this situation is that manydeveloping countries often end up withtoo many small tax sources, too heavy areliance on foreign trade taxes, and a rela-tively insignificant use of personal incometaxes.

Because of the large role played byinformal activities; because of limited re-porting requirements; because many ac-tivities are not carried out by modernestablishments; and because of financiallimitations, statistical and tax offices havedifficulties in generating reliable and de-tailed statistics. This paucity of reliabledata makes it difficult for policy makersto assess the potential impact of majorchanges to the statutory tax system. Ifmade, such changes are rarely accompa-nied by easily quantifiable impacts on tax

revenue. Given the often precarious fis-cal situation of developing countries, thisuncertainty would expose them to poten-tially serious fiscal difficulties. As a con-sequence, marginal changes are often pre-ferred over major structural changes, evenwhen the latter would be clearly prefer-able. This perpetuates the inefficient taxstructures.

Finally, it is well known that develop-ing countries tend to have income distri-butions that are much less even thanindustrial countries. In the former, Ginicoefficients that exceed 0.5 are not rare.1

This highly uneven income distributionhas two implications: first, that to gener-ate high tax revenue, the top deciles wouldhave to be taxed significantly more pro-portionally than the low deciles; second,that economic and often political poweris concentrated in the top deciles so thatricher taxpayers are able to prevent taxreforms that would affect them negatively.This partly explains why personal incometaxes and property taxes have been verylittle exploited in these countries and whytax incidence studies of developing coun-tries have found that tax systems rarelyachieve effective progressivity.

In conclusion, in developing countries,tax policy is often the art of the possiblerather than the pursuit of the optimal. Itis, thus, not surprising that economictheory, and especially the optimal taxationliterature, have had relatively little impacton the formulation of tax systems in de-veloping countries.2

This paper discusses some of the impor-tant tax policy issues facing many devel-oping countries today. It draws on theextensive first-hand experience with pro-viding tax policy advice to these countriesby the International Monetary Fund(IMF), with which both authors of the pa-per are affiliated. It is organized along the

1 A new comprehensive data set on income distribution in developing countries has recently been compiled byDeininger and Squire (1996).

2 For a sympathetic discussion (largely focused on developed countries) of how the optimal taxation literaturecould be used as a guide for tax policy formulation, see Heady (1993).


301

lines of specific policy issues, rather thanindividual country practices and experi-ences.3 Space limitations dictate that theissues covered are necessarily selective;4

the selection criteria are guided more bythe issues’ general relevance to the devel-oping countries as a group than by theirimportance to only some of these coun-tries. For example, issues related to the taxassignment and revenue sharing betweenthe central and local governments are notaddressed here, even though they may bepivotal in tax policy deliberations in somedeveloping countries.

For organizational purposes, this paperviews tax policy from two perspectives:the macroeconomic perspective, whichfocuses on broad questions such as thelevel and composition of tax revenue, iscovered in the next section; and themicroeconomic perspective, which fo-cuses on certain design aspects of selectedmajor taxes, is provided in the third sec-tion. The fourth section briefly discussessome policy challenges ahead for devel-oping countries. This is followed by theconclusion.

LEVEL AND COMPOSITION OFTAX REVENUE

From a macroeconomic perspective,aspects of a tax system that are of particu-lar interest to policy makers in develop-ing countries are (1) whether the existingoverall tax level (usually expressed as aratio of tax revenue to gross domesticproduct (GDP) is appropriate, and (2)

given a particular level, whether the ex-isting composition of tax revenue (usuallyin terms of income relative to consump-tion taxation) is desirable.5 This intereststems in part from the widely-held beliefthat the welfare costs of resource misallo-cation (both intra- and intertemporally)increase with increased taxation, and that,in the choice between taxing income andconsumption, the latter is the lesser evilin affecting long-run growth. However, italso originates from the question of whatlevel of public spending is desirable for adeveloping country at a given incomelevel.

The vast literature on optimal tax theoryprovides little practical guidance on thechoice of the overall level of taxation. Theliterature is a bit more helpful in the choicebetween income and consumption taxa-tion, but even here its value is limited.6

Following brief reviews of theoretical con-siderations about the tax burden and rev-enue composition, the revenue situationin developing countries is assessedagainst that in developed countries andpolicy implications are drawn from it.7

Level of Tax Revenue

Theoretical considerations

The primary reason why optimal taxtheory has little to say about choosing theoverall tax burden for an economy is thatmuch of this theory has been developedto suggest the optimal structure of taxesin a static context to raise a given tax bur-

3 Much of the material from which the present paper is drawn is contained in confidential IMF technical reportsprepared at the request of country authorities. Most of the country-specific references have, therefore, beensuppressed, except in those few cases involving references to factual information that is readily available inthe public domain.

4 The literature on taxation and development is voluminous. See, for example, Bird (1992); Bird and Oldman(1990); Newbery and Stern (1987); and Tanzi (1991). For a recent survey of tax issues in developing countriesfrom a somewhat different perspective from the present paper, see Burgess and Stern (1993).

5 From time to time, policy makers are concerned about tax revenue in relation to short-run budgetary imbal-ances. Such concerns are country-specific and will not be addressed here.

6 Much of the theoretical and empirical literature in this area has been surveyed recently in Tanzi and Zee (1997).7 A detailed comparative study of level and composition of tax revenue between developed and developing

countries can be found in Zee (1996).


302

den. Thus, the theory traditionally has notintegrated the expenditure side of thebudget in its analysis. To relax this as-sumption in a meaningful way, for pur-poses of normative policy prescription,would necessitate the explicit modelingof the benefits from the public expendi-tures to be financed by tax revenue. Inother words, determining the optimal taxlevel is conceptually equivalent to deter-mining the optimal level of governmentexpenditure. While several recent theoreti-cal attempts have been made to addressthis issue in an integrated framework ofexpenditure and taxation,8 the results sofar have been rather abstract and highlymodel dependent. Therefore, they cannotprovide practical policy guidance.

International comparisons

Lacking a clear prescription fromtheory, an alternative approach to assess-ing whether the level of the overall taxlevel in a developing country is “appro-priate” has been to compare it to the av-erage tax burden of a representative groupof both developing and developed coun-tries, taking into account some of thesecountries’ characteristics. It is obvious thatthis is a statistically based approachwhich, though popular and at times use-ful, does not have a theoretical founda-tion.

This approach became quite fashion-able, especially in the 1960s and 1970s. Itwould correlate the ratio of tax revenueto GDP (the dependent variable) for alarge group of countries against severalindependent variables, for the same coun-tries, that could be expected to influencethe tax ratio. Variables often used in thesestudies are per capita income, share ofagriculture output in GDP, share of min-eral exports in GDP, the openness of the

economy (measured by the share of im-ports and exports in GDP), the ratio ofmoney to GDP, and other variables. Whensolved with data for a specific country, theestimated regression equation provides ahypothetical tax ratio for that country. Thistax ratio is then compared with thecountry’s actual tax ratio.9

The comparison between the tax ratioestimated from the equation and the ac-tual tax level for the country indicateswhether, in comparison with other countries,and taking into account its own characteris-tics, the country’s tax level is above or be-low the expected one. This derived taxratio has been interpreted to reflect thedegree of tax effort that the country ismaking. As noted earlier, such a statisti-cal approach has no theoretical basis andshould not be interpreted to indicate the“optimal” tax burden for any country.Nevertheless, its use was (and at times stillis) popular and useful because the derivedratio could be used to convince a reluc-tant minister of a low taxed country (suchas Guatemala), or to provide a justifica-tion for a determined minister, to increasehis/her country’s tax burden. Such anapproach has been convenient both in es-tablishing a benchmark by which acountry’s tax level could be judged againstthe norm of its peers and in anticipatinglikely future developments as its economybecame more advanced. In fact, the regres-sions typically indicated that, ceterisparibus, a higher per capita income wasaccompanied by a higher ratio of tax rev-enue to GDP.

Table 1 provides some comparative in-formation, over the most recent decade forwhich data are available, on the tax levelsin OECD countries and in a representa-tive sample of developing countries—disaggregated by broad geographical

8 Such attempts have become increasingly fashionable since the advent of the endogenous growth literature. Aparticularly well-known example is Barro (1990). Turnovsky (1996) provides a more elaborate model of si-multaneous determination of optimal tax and expenditure.

9 The literature is very extensive. See, for example, Bahl (1971); Tait, Gratz, and Eichengreen (1979); and Tanzi(1992) for a more recent example.


303

regions.10 It shows that, for the period1985-87, the average total tax level in thedeveloping countries was about 17.5 per-cent of GDP. The variation in this averageacross geographical regions was fairlynarrow, with countries in Africa exceed-ing the sample average by about 2 per-centage points of GDP (the variationwithin each group of countries was muchgreater). In contrast, the average total taxlevel in the OECD countries in the sameperiod was more than twice as high (36.6percent of GDP), although there was a sig-nificant variance across the three OECDsubcountry groups: the average total taxlevel in the European subgroup was onthe order of 8 percentage points of GDPhigher than that in the American and Pa-cific subgroups. At the same time therewas much less variance within eachOECD group than within the group of thedeveloping countries. Essentially all of the

foregoing comparative observations areequally applicable to the tax revenue datafor the period 1995-97, during which theaverage total tax level showed only a mar-ginal increase for all subcountry groupsrelative to the earlier period.

As already mentioned, numerous stud-ies have attempted to identify the deter-minants of the level of taxation. One ofthe most commonly used determinantshas been per capita income, usually ongrounds that economic developmentwould bring about both an increased de-mand for public expenditure (Tanzi, 1987)and a larger supply of taxing capacity tomeet such demands (Musgrave, 1969).These considerations suggest—with gen-erally strong empirical support—that apositive correlation exists between tax lev-els and economic development. They alsosuggest, in theory, that the direction of cau-sation tends to run from development totax levels, and not the other way around.This is important because the commonnotion that higher tax levels would gen-erate larger distortions—and would thusbe detrimental to growth—is then not nec-essarily contradicted by the observed cor-relation between tax levels and develop-ment.11

The main policy implication of theabove discussion for developing countriesis that economic development wouldmore often than not generate additionalneeds for tax revenue to finance the risein public expenditures while at the sametime increasing the countries’ ability toraise revenue to meet those needs. It isthus important to focus more on the waysthe revenue is utilized, and perhaps on thetax structure, than on the level of taxation

10 Data for the OECD countries do not include the five recent OECD members: the Czech Republic, Hungary,Korea, Mexico, and Poland. Korea and Mexico are included in the developing country sample, which coversa total of 38 countries in Africa (8), Asia (9), Middle East (7), and Western Hemisphere (14). In both Table 1 andTable 2 (shown below), data presented are unweighted. While not shown, weighted average data (usingcountry GDP in U.S. dollars as weights) convey broadly the same comparative picture.

11 In any case, much of the available econometric evidence on the relationship between tax levels and per capitaincome growth has not been very robust, due largely to the difficulties in disentangling the growth effects ofother relevant variables from taxation. See, for example, Easterly and Rebelo (1993) and Levine and Renelt(1992).

TABLE 1COMPARATIVE LEVELS OF TAX REVENUE,

1985–97(IN PERCENT OF GDP)

OECD Countries (1)AmericaPacificEurope

Developing Countries (2)AfricaAsiaMiddle EastWestern Hemisphere

36.630.630.738.2

17.519.616.116.517.6

1985–7

37.932.631.639.4

18.219.817.418.118.0

1995–7

Sources: Revenue Statistics (OECD); and GovernmentFinance Statistics (IMF).

(1) Excludes the Czech Republic, Hungary, Korea,Mexico, and Poland.

(2) A sample of 8 African countries; 9 Asian countries;7 Middle Eastern countries; and 14 WesternHemisphere countries.

NAT

ION

AL TA

X JO

UR

NA

L

304

TABLE 2COMPARATIVE COMPOSITION OF TAX REVENUE, 1985–97

(IN PERCENT OF GDP)



Memorandum items:



Income/Consumption Taxes Corporate/Personal Income Taxes

1985–7 1995–7

(In percent)

1985–7 1995–7

1.21.82.31.1

0.50.50.60.50.4

13.914.017.113.3

4.96.35.74.73.7

2.82.53.92.7

2.82.93.54.31.8

11.311.413.211.0

1.73.12.11.01.0

11.37.67.5

12.4

10.311.79.59.1

10.6

6.03.42.36.8

2.33.21.91.52.6

Income Taxes

SocialSecurity

1995–7

Of Which:

Total Corporate Personal

Consumption Taxes

Of Which:

Total General Excises Trade

Income Taxes

SocialSecurity

1985–7

Of Which:

Total Corporate Personal

Consumption Taxes

Of Which:

Total General Excises Trade

3.82.23.74.0

2.62.32.52.43.0

0.70.60.80.7

4.25.73.64.43.7

8.85.82.8

10.1

1.20.40.11.22.4

14.215.416.313.7

5.26.96.25.03.7

3.13.04.32.9

2.62.43.03.22.3

10.812.311.410.6

2.23.93.01.31.0

11.47.08.4

12.4

10.511.69.7

10.310.6

6.63.74.37.3

3.63.83.11.54.8

3.62.02.64.0

2.42.32.23.02.3

0.30.30.60.3

3.55.12.74.32.6

9.56.13.5

10.8

1.30.50.31.12.5

1.22.21.91.1

0.50.60.60.50.4

0.20.20.30.2

1.60.91.64.31.8

0.30.20.40.3

1.20.61.02.52.3

Sources: Revenue Statistics (OECD); and Government Finance Statistics (IMF).

(1) Excludes the Czech Republic, Hungary, Korea, Mexico, and Poland.(2) A sample of 8 African countries; 9 Asian countries; 7 Middle Eastern countries; and 14 Western Hemisphere countries.


305

per se. Given the complexity of the devel-opment process and the way that taxationmight impact on it, it is doubtful that, forpolicy purposes, the concept of an “opti-mal” level of taxation that is robustlylinked to the different stages of a country’seconomic development could ever bemeaningfully derived for any country.

Composition of Tax Revenue

Theoretical considerations

Important issues in any discussion ofrevenue composition involve, first, thetaxation of income relative to that of con-sumption, and, second, under consump-tion taxation, the taxation of imports vis-à-vis domestic consumption. Considerfirst the issue of the optimal income-con-sumption tax mix. In evaluating the rela-tive merits of these two tax bases, bothefficiency and equity considerations arecentral to the analyses, especially in de-veloping countries given their high Ginicoefficients. However, the theoretical lit-erature has tended to focus on the former.

The conventional belief that taxing in-come entails a higher welfare (efficiency)cost than taxing consumption is primarilybased on the observation that the incometax consists of two broad components: alabor tax and a capital tax. Since the labortax is equivalent to a tax on consumptionin an intertemporal framework, the in-come tax gives rise to an additional dis-tortion—on savings—that is absent from

the consumption tax. It turns out that, inthe traditional neoclassical growth model,the length of the consumer’s planninghorizon plays a crucial role in the theo-retical ambiguity of the relative superior-ity of the consumption tax. If savingdecisions are based on life-cycle consid-erations, the optimal mix of income andconsumption taxes would depend entirelyon the relevant elasticities, i.e., of laborsupply and savings.12 If, however, theplanning horizon is infinite, then the op-timal tax on capital would in fact be zeroin the long run.13

The analytical picture would get morecomplex and the results more ambiguousif human capital—the crucial ingredientin the new endogenous growth litera-ture—is brought into the analysis. In gen-eral, the nature and process of humancapital accumulation, i.e., whether its ac-quisition is thought to require time (fore-gone wages), physical capital, even hu-man capital itself, or some combinationof all three, will ultimately have a bearingon the relative welfare costs of income andconsumption taxation. The upshot of theabove theoretical considerations is that,while taxing (physical) capital may welldepress (physical) capital accumulation,it, like taxing consumption, could have animpact on human capital accumulationand other variables through a web of com-plex interactions, rendering the relativewelfare costs of the two taxes a priori un-certain.14 Such considerations also under-score the importance in tax policy delib-

12 It is not uncommon to encounter arguments for relatively heavy consumption taxation on the basis that theelasticity of labor supply—at least for the group of prime male workers—is low. It must be noted, however,that the cited inelasticity usually refers to the uncompensated labor supply curve. The compensated elastic-ity—the concept relevant for measuring welfare costs—is typically much higher. Moreover, there is a greatdeal of uncertainty about the magnitude of the interest elasticity of savings, even for developed countries. Fordeveloping countries, data limitations have generally hampered empirical investigations on this issue. Forexample, the study by Masson, Bayoumi, and Samiei (1998), covering a large sample of developing countries,has found an insignificant and nonrobust relationship between the real interest rate and the private savings/GDP ratio.

13 The life-cycle results are established in Atkinson and Sandmo (1980), and results from the infinite-horizonmodel are derived in Chamley (1986). It could be optimal to tax capital in the life-cycle model because theintergenerational excess burden of a tax on capital is not fully captured in such a framework.

14 A survey of the literature on human capital accumulation and growth is beyond the scope of this paper. On atextbook treatment, see Barro and Sala-i-Martin (1995).


306

erations of focusing on the impact of bothincome and consumption taxes at leastequally on human and on physical capi-tal accumulation through their variousdesign aspects (e.g., tax credits or exemp-tions targeted for expenditures that areconducive to human capital formation).In developing countries, the impact oftaxation on physical capital accumulationhas traditionally received the lion’s shareof attention, which in turn has led to theexcessive use of tax incentives for its pro-motion (further discussed below).15

Another concern in the choice betweentaxing income and taxing consumptioninvolves their relative impact on (vertical)equity. This concern is particularly impor-tant in view of the uneven income distri-bution in developing countries. Tradition-ally, it has been thought that taxingconsumption is inherently more regres-sive than taxing income, since it is admin-istratively infeasible to effectively imple-ment, on a broad scale, graduated tax rateson consumption.16 Two lines of researchhave, however, cast doubt on this conclu-sion. First, the traditional form of the con-sumption tax, i.e., taxing consumption asit takes place (such as a value-added tax(VAT) or retail sales tax), has been foundto be far less regressive than commonlythought when viewed from a life-cyclerather than a static perspective.17 Second,at least in theory, consumption can betaxed on the same graduated basis as in-come, by allowing unlimited deductionsfrom income of savings.18 But such a tax

is likely to pose tremendous administra-tive difficulties in most developing coun-tries, as net savings during a tax periodeligible for deduction must be tracked andreported to the tax authorities. Two coun-tries that, following Kaldor’s recommen-dations, experimented with this tax about40 years ago (India and Sri Lanka) aban-doned it soon after its introduction. InSweden, a commission (the Lodin Com-mission) that studied in detail the possi-bility of introducing a personal expendi-ture tax concluded that such a tax couldnot be administered. The upshot of theabove discussion is that the equity con-cerns of the traditional form of taxing con-sumption are probably overstated, and fordeveloping countries attempts to addresssuch concerns on the consumption taxside would in any case be either ineffec-tive or administratively infeasible.

Turning to the issue of taxes on imports,the traditional heavy reliance on importduties as a convenient tax handle by de-veloping countries implies that loweringtariff rates—necessitated perhaps by theirdesire to join the World Trade Organiza-tion, to participate in regional trading ar-rangements such as the ASEAN in Asiaand NAFTA or Mercousur in WesternHemisphere, or simply to conclude bilat-eral trading agreements with developedcountries—could have significant eco-nomic and revenue consequences in thesecountries. First and foremost, tariff reduc-tions, when properly structured (see fur-ther discussions below) and not accom-

15 For small, open countries, an additional relevant consideration is clearly the difference in the relative mobilitybetween capital and labor across national boundaries. In this context, the extent to which capital income canbe taxed in these countries is at least partly dependent on how such income is taxed elsewhere in the sameregion of the world.

16 A limited application of differential consumption taxation is certainly feasible and in fact is widely practiced.There is, however, compelling evidence suggesting that such a practice is ineffective in achieving equity ob-jectives, since both the rich and the poor consume (albeit in different proportions) the same goods that arebeing taxed differentially. For a forceful statement of this point in the context of an actual tax reform programin a developing country, see Republic of South Africa (1994).

17 See a series of studies by Metcalf, e.g., Metcalf (1994). These results are, however, generally more relevant foradvanced than developing countries.

18 This is the idea lying behind, for example, the so-called USA (unlimited savings allowance) tax that has beenproposed in the United States recently (see Seidman, 1997). It is also broadly the idea behind Kaldor’s (1955)expenditure tax.


307

panied by other increases in explicit orimplicit trade barriers, would lead in gen-eral to lower levels of both nominal andeffective protection. Secondly, tariff reduc-tions could also result in a significant lossin budgetary revenue, at least in the shortrun before the volume of imports has hadtime to respond.19

While reducing protection of domesticindustries from foreign competition is aninevitable consequence or even the objec-tive of any trade liberalization program,reducing budgetary revenue could be anunwelcome by-product of the programthat needs to be addressed.20 Feasiblecompensatory revenue measures underthe circumstances almost always involveincreasing domestic consumption taxes;21

rarely would increasing income taxes beconsidered as a viable option on groundsof both policy (on account of their per-ceived negative impact on investment)and administration (on account of theirrevenue effects being less certain andtimely than consumption tax changes).22

International comparisons

Table 2 provides a breakdown of totalrevenue, for the same countries and overthe same periods as Table 1, into majorcategories of taxes: income taxes (withsubcategories of corporate income tax(CIT) and personal income tax (PIT)); con-

sumption taxes (with subcategories ofgeneral23 , excises, and trade taxes24 ); andsocial security taxes. For the entire sampleof developing countries, the ratio of in-come to consumption taxes was about 0.5in 1985–7, compared to 1.2 for OECDcountries. For either group of countries,this ratio showed almost no change a de-cade later.

Another notable difference betweendeveloped and developing countries is theratio of CIT to PIT. Developed countriesraised about four times as much revenuefrom the PIT as from the CIT, while thedeveloping countries raised more revenuefrom the CIT than from the PIT. Undoubt-edly, the difference in wage income, in thesophistication of the tax administration,and in the political power of the richestdeciles between the two country groupsare primary contributing factors to thelarge difference between them in the rela-tive importance of the CIT and the PIT asrevenue sources. Finally, as expected, rev-enue from trade taxes has been signifi-cantly higher—though falling—in devel-oping countries (4.2 percent of GDP in1985–7 and 3.5 percent of GDP in 1995–7)than in developed countries (less than1 percent of GDP in both periods).

While it is difficult to draw clear-cutnormative policy prescriptions from theabove international comparisons as re-

19 In some cases, the revenue impact of tariff reductions, even in the short run, could be moderated to varyingdegrees if accompanied by, for example, the tariffication of quotas, or the imposition of some minimum tariffon imports previously exempted from duties, as part of an overall trade liberalization program. See the de-tailed study by Ebrill, Stotsky, and Gropp (1999).

20 Some developing countries have taken tariff reductions in a required trade liberalization program as an op-portunity to lower the overall level of taxation. Such cases are, however, few and far between, since the abilityto do so would typically necessitate a commensurate reduction in expenditures to avoid endangering thebudgetary position.

21 Keen and Ligthart (1999) have recently shown that, if an underlying tariff reform improves production effi-ciency, replacing the tariffs with domestic consumption taxes would raise welfare in a small open economy.

22 Increasing income taxes in the form of reducing distortive tax incentives would, however, arguably be a desir-able policy measure (see discussions below).

23 Included in this subcategory are the VAT; all other VAT-like taxes that sometimes go by different names, suchas goods and services taxes or general sales taxes; and other broad-based single- and multi-stage sales taxes (ifthey exist).

24 Trade taxes are mostly import tariffs; export tariffs have become relatively insignificant in recent years. Untilthe early 1980s, however, they had been important in some countries and especially in some Latin American(e.g., Argentina) and African (e.g., Côte d’Ivoire) countries.


308

gards the income-consumption tax mix,25

a compelling positive policy implicationrevealed by the comparisons is that eco-nomic development tends to lead to a rela-tive shift in the composition of revenuefrom consumption to personal incometaxes. At any given point in time, how-ever, the important tax policy issue fordeveloping countries is not so much indetermining the optimal tax mix as in (1)spelling out clearly the objective(s) to beachieved by any contemplated shift in themix, (2) assessing the economic conse-quences of the shift—in both efficiencyand equity terms—in the most objectivemanner possible, and (3) implementingcompensatory—possibly nontax (e.g., ex-penditure)—measures, if those who arebeing made worse off by the shift are fromthe poorer deciles.

POLICY ISSUES IN SELECTEDMAJOR TAXES

In developing countries where marketsare taking on an increasingly importantrole in allocating resources, the most im-portant objective of tax policy is to mini-mize the interference by the tax system inthat allocation process, subject, of course,to revenue and redistribution require-ments. This means not only that thetax system should be as neutral in designas possible, it should also have simpleand transparent administrative rulesand procedures, so that ex ante neutralityis not negated by ex post nonneutralitydue to the inability of the tax administra-tion to enforce the tax system as designed.The following subsections highlight

some of the most important tax policyissues encountered in developing coun-tries.

Personal Income Tax

General conceptual issues relating tothe PIT have been comprehensively dis-cussed in Cnossen and Bird (1990), al-though the focus of that study is on OECDcountries. In developing countries, the is-sues of interest are typically narrower inscope, but generally require that more at-tention be paid to their administrativeimplications, given that administrativecapabilities are much more binding inthese countries than in developed coun-tries. Also, the fact that wage income isoften a small share of national income hascontributed to the difficulties in render-ing the PIT as a significant revenue source.

Rate structure

Any discussion of the personal incometax in developing countries must startwith the observation that this tax hasyielded very little revenue in most of thesecountries and that the number of indi-viduals who are subject to this tax and,especially, who are subject to the highestmarginal tax rate, is very small.

The rate structure of the PIT is often themost convenient and visible policy instru-ment for most governments in develop-ing countries to underscore their commit-ments to social justice, and hence to gainpolitical support for their policies. It is,therefore, not surprising to find that manydeveloping countries attach great impor-tance to maintaining some degree of

25 Existing econometric evidence on the relationship between the income-consumption revenue mix on the onehand, and either the growth or savings rate on the other, has been largely inconclusive. While employing taxinstruments to alter rates of return to savings may have an impact on the composition of savings, there hasbeen little conclusive international evidence that such measures (unless of a drastic nature) could signifi-cantly affect either private or national savings as a whole in the long run. For a recent review of tax effects onhousehold savings in OECD countries, see Normann and Owens (1997). A recent study by Tanzi and Zee(forthcoming) has found, however, rather strong results, in a sample of OECD countries, that increases inincome taxes reduce household savings more than increases in consumption taxes (for raising the same amountof revenue).


309

nominal PIT rate progressivity by apply-ing many rate brackets,26 and are reluc-tant to undertake PIT reforms that wouldsuggest any lessening of such commit-ments.

More often than not, however, the ef-fectiveness of nominal rate progressivityin delivering effective rate progressivityis severely undercut by the very high per-sonal exemption which may amount toseveral times the country’s per capita in-come (see Table 3 for illustrative cross-country comparative information), and bythe typical plethora of exemptions anddeductions commonly found in develop-ing countries which benefit those withhigh incomes (e.g., the exemption of capi-tal gains from tax, generous deductionsfor medical and education expenses, thelow taxation of financial income). Tax re-lief provided in the form of deductions isparticularly egregious under highly pro-gressive nominal PIT rates because its

value (in terms of implied tax savings) in-creases with the rate bracket the taxpayeris in. Experiences with PIT reforms in de-veloping countries (as well as in developedcountries, for that matter) tend to suggest,quite compellingly, that the effective rateprogressivity could be improved by reduc-ing the degree of nominal rate progressivityand the number of rate brackets, and re-ducing exemptions and deductions. In-deed, any reasonable equity objectivewould not require more than a few mod-erately progressive nominal rates in the PITrate structure. If political constraints pre-vent a meaningful restructuring of rates, asubstantial improvement in equity couldstill be achieved by replacing PIT deduc-tions with tax credits, which would deliverthe same benefits to taxpayers in all taxbrackets. The use of tax credits of any sig-nificant extent is, however, still very rarein developing countries (Israel is one coun-try that uses tax credits extensively).

26 It should be noted, however, that neither the degree of nominal rate progressivity nor the number of ratebrackets in developing countries could be considered excessive when compared to developed countries,although the latter countries have unmistakably moved to flatten out their PITs and have reduced the num-ber of rates during the last decade or so. For a review of PIT reform experiences in OECD countries, seeMessere (1993).

TABLE 3COMPARATIVE PIT RATES IN SELECTED DEVELOPING COUNTRIES (1)

Minimum IncomeSubject to Tax

(In Multiples of PerCapita Income

Marginal Tax Rates

Number Range

(In Percent)

Africa Kenya South Africa Tanzania Uganda Zambia Zimbabwe

Western Hemisphere Argentina Brazil Chile Costa Rica Mexico Nicaragua

4.10.82.14.21.30.5

1.63.50.21.30.19.3

67

11335

726485

10.0–32.519.0–45.07.5–35.0

10.0–30.010.0–30.020.0–40.0

6.0–33.015.0–25.05.0–45.0

10.0–25.03.0–35.07.0–30.0

Sources: Individual Taxes 1998: Worldwide Summaries (PricewaterhouseCoopers); and authors’ calculations.(1) Information generally pertains to 1997/1998.


310

The effectiveness of a high marginal taxrate is also much reduced by the fact thatit is often applied at such high levels ofincome, expressed as shares of per capitaGDP, that little income is subjected to theserates. In some developing countries (e.g.,Honduras), a taxpayer’s income must behundreds of times the per capita incomebefore it enters the highest rate bracket.

Another important issue relating to therate structure is the level of the top mar-ginal PIT rate. In some developing coun-tries, this rate exceeds the CIT rate by asignificant margin, which inevitably pro-vides strong incentives for taxpayers tochoose the corporate form of doing busi-ness purely for tax reasons. This distor-tion would exist even if the PIT and theCIT were fully integrated in the taxationof dividends (see below), because the tax-payers who are best situated to abuse thedelay in the taxation of accrued incomeby incorporating themselves are preciselythose whose sole purpose for incorpora-tion is to evade the PIT rate in the firstplace. Professionals (e.g., lawyers and ac-countants) and small entrepreneurs, forexample, can easily siphon off profitsthrough expense deductions over timeand escape the higher PIT rate perma-nently. For these taxpayers, taxes delayedare often taxes evaded. It is, therefore,good tax policy to ensure that the topmarginal PIT rate does not differ materi-ally from the CIT rate.27

Tax base

In addition to the problem of the highlevels of exemptions and deductions thattends to narrow the tax base and negatemuch of the effective progressivity of a

progressive nominal rate structure, asnoted above, it is common to find that thePITs (as well as the CITs, for that matter)in developing countries are riddled withserious violations of the two basic prin-ciples of good tax policy at the practicallevel: symmetry and inclusiveness.28 Thesymmetry principle refers to the identicaltreatment for tax purposes of gains andlosses of any given source of income, e.g.,if the gains are taxable, then the lossesshould be deductible. The inclusivenessprinciple relates to the capturing of an in-come stream in the tax net (unless it isexplicitly exempt) at some point along thepath of that stream, e.g., if a payment isexempted in the hands of the payee, thenit should not be a deductible expense inthe hands of the payer. Violating theseprinciples would, in general, lead to dis-tortions and inequities.

To be sure, there are circumstances un-der which some (limited) deviation fromthe symmetry and/or inclusiveness prin-ciples could be called for, but any suchdeviation should be justified by clearpolicy objectives.29 Violations of the prin-ciples found in many developing coun-tries are, however, often due simply toinadequacies in the PIT design. Somecommon examples include the deductibil-ity from taxable income on the one handof pension/saving contributions, interestexpenditure, capital losses, and (cash andnoncash) perquisites paid by employersto employees as a business expense; andthe exemption from tax on the other handof pension income/saving withdrawals,interest income, capital gains, and perqui-sites in the hands of employees. It is easyto see how these tax provisions could be

27 Preferably, these two rates should be equalized if there is full integration of the PIT and the CIT in dividendstaxation (further discussed below). Absent full integration, the top marginal PIT rate should technically besomewhat lower than the CIT rate.

28 It goes without saying, of course, that tax policy should also be guided by the general principles of neutrality,equity, and simplicity.

29 For example, to prevent excessive tax avoidance, many countries have found it prudent to place limits on thedeductibility of capital losses in any given year or the number of years losses of any kind can be carriedforward.


311

exploited to varying degrees by taxpay-ers to evade taxes. To address them, theremedy is invariably the elimination ofeither the deductibility or the exemption(but not both), and the choice between thetwo depends to a large extent on admin-istrative considerations.

Tax treatment of financial income

The tax treatment of financial income isa particularly problematic area in devel-oping countries with limited tax adminis-tration capabilities, as alternative forms ofsuch income could be easily disguised,interchanged, and otherwise arbitraged iftax provisions are not carefully written todeal with them. This is an issue that evenmost developed countries would find dif-ficult to tackle, and a comprehensive dis-cussion of it is clearly beyond the scope ofthe present paper. Here, only two particu-larly notable issues dealing with the taxa-tion of interest and dividends are high-lighted. In many developing countries,interest income, if taxed at all, is taxed foradministrative reasons through a finalwithholding tax at a rate that is generallysubstantially below both the top marginalPIT rate and the CIT rate. For taxpayerswith mainly wage income (usually mak-ing up the bulk of PIT taxpayers), thisseems an acceptable compromise betweentheoretical correctness and practical feasi-bility. For those with business income (andfor companies as taxpayers under the CIT),however, the low tax rate on interest in-come coupled with full deductibility ofinterest expenditure implies that signifi-cant tax savings could be realized throughfairly straightforward arbitrage transac-tions. Hence, it is important that the ap-plication of final withholding on interestincome be carefully targeted, e.g., finalwithholding not be applied if a taxpayerhas business income.30

The tax treatment of dividends raisesthe well-known double taxation issue. Inmost OECD countries, the double taxationof dividends is eliminated, or at least par-tially alleviated, through a variety of re-lief measures at either the corporate or theshareholder level.31 While such measuresgenerally entail administrative complica-tions that many developing countrieswould find difficult to cope with, on eco-nomic grounds not providing relief at allwould also not be a particularly desirablecourse of action to take. For most devel-oping countries, a reasonable optionwould be to either exempt dividends fromthe PIT altogether, or to tax them at a rela-tively low rate, perhaps through a finalwithholding tax at the same rate as thatimposed on interest income (if it exists).

Corporate Income Tax

Tax policy issues relating to the CIT indeveloping countries are numerous andcomplex, and are similar to those foundin many developed countries. This paperwill not attempt to provide a detailed andcomprehensive discussion of such issues,but will instead focus on two problematicareas that seem particularly prevalent indeveloping countries: multiple CIT ratesbased on a sectoral differentiation and theincoherent design of the depreciation sys-tem.

Multiple CIT rates

Developing countries (e.g., Egypt, Para-guay, Vietnam, Zambia) are more proneto having multiple rates differentiatedalong sectoral lines (including the com-plete exemption from tax of certain sec-tors, especially the parastatal sector) thandeveloped countries, possibly as a legacyof past economic regimes that emphasizedthe state’s role in resource allocation. Such

30 This does not mean that the scope of the withholding itself should be targeted; withholding as a collectiondevice (but not as a final tax) could still be broadly applied.

31 For a survey of practices in OECD countries, see OECD (1991).


312

practices are, however, clearly detrimen-tal to the proper functioning of marketforces (i.e., sectoral allocation of resourceswould be distorted by differences in thetax rates) and, therefore, not defensible ifthe government’s commitment to a mar-ket economy is real. Unifying multiple CITrates across sectors where they exist is animportant outstanding tax policy issue indeveloping countries.

Depreciation

Allowable depreciation of physical assetsfor tax purposes is one of the most impor-tant structural elements in a CIT in deter-mining the cost of capital, and thus theprofitability of investment. Designing anappropriate depreciation system is, there-fore, crucial for fostering a favorable invest-ment climate. Yet, notwithstanding thegreat importance they frequently attach topromoting investment, developing coun-tries far too often have depreciation systemsthat are complex, incoherent, restrictive,and in general not investment-friendly.

The most common shortcomings foundin the depreciation systems in developingcountries include: (1) an excessive num-ber of asset categories and depreciationrates; (2) excessively low depreciationrates; and (3) a structure of depreciationrates that is not in accordance with therelative obsolescence rates of different as-set categories. Rectifying these shortcom-ings should receive a high priority in taxpolicy deliberations in these countries.

In restructuring their depreciation sys-tems, developing countries could wellbenefit from the following guidelines: (1)under most circumstances, classifying as-sets into, say, three or four categories

should be more than sufficient, e.g., group-ing long-lived assets such as buildings atone end and fast-depreciating assets suchas commercial vehicles and computers atthe other end, with one or two categoriesof machinery and equipment in between;(2) only one depreciation rate should beattached to each asset category; (3) depre-ciation rates should generally be set higherthan the actual physical lives of the un-derlying assets to compensate for the lackof a comprehensive inflation-compensat-ing mechanism in most tax systems;32 and(4) on administrative grounds, the declin-ing-balance method—still not commonlyused in developing countries—should bepreferred to the straight-line method. Asis well known, the declining-balancemethod allows the pooling of all assets inthe same asset category and automaticallyaccounts for capital gains and losses fromasset disposals, thus substantially simpli-fying bookkeeping requirements.33

Value-added Tax, Excises, andImport Tariffs

Value-added Tax

One of the most visible tax reforms un-dertaken by developing countries duringthe past three decades has been the intro-duction of the VAT (or a VAT-like tax).Since the VAT can now be found in anoverwhelming majority of developingcountries,34 the outstanding tax policy is-sue in the domestic consumption tax areain these countries is no longer the replace-ment of cascading turnover taxes, but theproper design of the VAT and the scopeof excise taxes.

32 The basis for this is simply that, while inflation increases the replacement costs of assets, depreciation is inevi-tably computed on their historical costs.

33 A conversion from the straight-line to the declining-balance method would necessitate, of course, an upwardadjustment in depreciation rates on account of the conversion alone to maintain the same depreciation allow-ances in present-value terms. In a number of OECD countries, a switchover at some point of an asset’s lifefrom the declining-balance to the straight-line method is allowed (see OECD, 1991). For a discussion of theswitching between the depreciation methods, see Messere and Zuckerman (1981).

34 According to a recent study by Ebrill, et al. (forthcoming), the VAT (or a VAT-like tax) can be found in 116countries around the world as of September 1998.


313

While the VAT that has been adoptedin developing countries is, almost with-out exception, implemented through acredit-invoice mechanism modeled afterthe Western European countries, it fre-quently suffers from the limitation of be-ing incomplete in its application in oneform or another. It is all too common tofind, for example, that many importantsectors, most notably services and thewholesale and retail sector, have been leftout of the VAT net; or that the creditmechanism is excessively restrictive (i.e.,denials or delays in providing propercredits for the VAT on inputs), especiallywhen it comes to capital goods.35 Thesefeatures allow a substantial degree of cas-cading to remain in the system, and thusgreatly reduce the benefits from introduc-ing the VAT in the first place. Rectifyingsuch limitations in the VAT design andadministration should, therefore, be givena high priority in developing countries.

Another aspect worthy of attention isthe adoption on the part of many devel-oping countries (see Table 4 for illustra-tive cross-country comparative informa-tion) of two or more rates (including thezero rating of certain nonexport supplies).While multiple rates are likely to compli-cate the administration of the VAT, theyare politically attractive by ostensiblyserving—though not necessarily effec-tively—an equity objective. In fact, mostOECD countries also have multiplerates. Still, the administrative price for ad-dressing equity concerns through havingmultiple VAT rates is likely to be higherin developing than in developed coun-tries.

Excises

The most notable shortcoming of theexcise systems found in many developingcountries is their inappropriately broadcoverage of products—often for revenuereasons, but sometimes for reasons diffi-cult to discern as the marginal revenueraised from some excised goods (whichshould not have been excisable) couldwell be insignificant.

As is well known, the economic ratio-nale from imposing excises is very differ-ent from that for imposing a general con-sumption tax, such as the VAT. While thelatter should be broad-based so as to maxi-mize revenue with minimum distortion,the former should be highly selective, nar-rowly targeting a few goods mostly ongrounds that their consumption entailsnegative externalities on society. It so hap-pens that the list of goods typicallydeemed to be excisable on such grounds(e.g., tobacco, alcoholic, and petroleumproducts; as well as motor vehicles) arefew and usually highly inelastic in de-mand. Thus, a good excise system is in-variably characterized by its ability to gen-erate revenue (as a by-product) from anarrow base and with relatively low ad-ministrative costs.36

Import tariffs

As noted earlier, reducing import tar-iffs as part of an overall program of tradeliberation is a major policy challenge cur-rently facing a large number of develop-ing countries. From a tax policy perspec-tive, this challenge involves two concernsthat need to be carefully addressed. First,

35 If the VAT on capital goods is creditable, the VAT is known as a consumption-type VAT (the standard formfound in developed countries); if not, it is known as a production-type VAT. Even in those developing coun-tries where the VAT is ostensibly of the consumption-type, credits on capital goods are frequently grantedwith a substantial (administrative) delay. For a general discussion of the various variants of a VAT, the variousways they can be implemented, and their different economic implications, see Zee (1995). Cnossen (1998) andEbrill, et al. (forthcoming) contain useful descriptions of VAT features and experiences in a large group ofdeveloping countries.

36 While an extensive excise system with highly differentiated rates set in inverse relationship to demand elas-ticities could well be a policy implication of the optimal commodity taxation literature, in practice such asystem is rarely if ever administratively feasible—even in developed countries.


314

it is important to ensure that the way thenominal tariffs are reduced does not leadto unintended changes in the relative ratesof effective protection across sectors—which may come about from carelessor incoherent differences in the extentto which nominal tariff reductions areeffected on inputs and outputs. Whileeffective protection is a relatively well-established concept in economics, in prac-

tice the attention of a tariff reductionprogram is all too frequently focused onnominal tariff rates. One simple way ofensuring that unintended consequencesdo not occur would be to reduce allnominal tariff rates by the same propor-tion whenever such rates need to bechanged.37

The second concern of nominal tariffreductions is the likely short-term revenue

TABLE 4COMPARATIVE VAT RATES IN SELECTED DEVELOPING COUNTRIES (1)

Standard Rate Significant other Rate(s) (2)

(In Percent)

Africa Côte d’Ivoire Kenya Mauritius South Africa Zambia

Asia Fiji Indonesia Korea Philippines Singapore Sri Lanka Thailand

Middle East Egypt Jordan Morocco Pakistan Tunisia

Western Hemisphere Argentina Bolivia Chile Colombia Costa Rica Dominican Republic El Salvador Mexico Nicaragua Panama Peru Uruguay Venezuela

20.016.010.014.017.5

10.010.010.010.03.0

12.510.0

10.010.020.015.018.0

21.014.918.016.015.08.0

13.015.015.05.0

18.023.016.5

11.112.0———

—5.0, 20.0, 35.0

2.0, 3.5————

5.0, 25.020.0

7.0, 10.0, 14.0—

6.0, 10.0, 29.0

10.5, 27.0——

8.0, 10.0, 20.0, 35.0 45.6———

10.05.0, 6.0, 10.0

10.0—

14.0—

Source: Ebrill, et al. (forthcoming).

(1) Countries shown are those in the sample, to which Table 1 and Table 2 refer, that have a VAT (or a VAT- like tax). VAT rates (tax exclusive) shown are as of March, 1999.(2) Excludes the zero rate on exports.

37 Note that the point here is on preventing unintended changes to the relative pattern of effective protectionrates. The initial relative pattern of such rates may well be deemed inappropriate, but measures to alter itwould then be intentional.


315

loss they may entail.38 Here, the strategyshould be relatively clear-cut, and shouldinvolve three separate compensatory mea-sures to be considered in sequence: (1) re-ducing the scope of tariff exemptions inthe existing system;39 (2) compensating forthe tariff reductions on excisable importsby a commensurate increase in their ex-cise rates;40 and, finally (3) adjusting therate of the general consumption tax (suchas the VAT) to meet remaining revenueneeds.41

Tax Incentives

While granting tax incentives to pro-mote investment is common in countriesaround the world, available evidence sug-gests that their effectiveness in attractingincremental investments (above and be-yond the level that would have takenplace if no incentives were given) is oftenquestionable and that their revenue costcould well be high (e.g., tax incentives canbe abused by existing enterprises dis-guised as new ones through nominal re-organization).42 For foreign investors—the primary target of most tax incentivesin most developing countries—the deci-sion to enter a country would normallydepend on a whole host of factors, amongwhich the availability of tax incentives isonly one and frequently far from being themost important one. The existence of natu-ral resources, political and economic sta-bility, transparency of the legal and regu-latory systems, adequacy of supportinginstitutions (e.g., banking, transportation,communication, and other infrastructurefacilities), ease of profit repatriation, and

economic and skilled workforce, are usu-ally far more decisive than tax consider-ations in determining suitable investmentlocations. If these factors are favorable,and the country’s tax system is in line withinternational norms, then tax incentiveswould at best play a role at the margin ininfluencing an investor’s decision.43

Tax incentives could also be of question-able value to a foreign investor becausethe true beneficiary of the incentives maynot be the investor concerned, but ratherthe treasury of his home country. Thiscomes about because any income that isspared from taxation by the host countrycould be taxed by the investor ’s homecountry if the latter’s tax system is basedon the residence principle (i.e., the incen-tives could reduce the amount of foreigntax credits available to the investor), un-less a tax sparing clause is included in bi-lateral double tax treaties. At present,many developed countries are increas-ingly reluctant to grant such a clause intheir treaties.44

Subject to the above constraints on theefficacy of tax incentives, a conceptuallylegitimate purpose for granting them indeveloping countries is to rectify someforms of market failure, most notablythose involving externalities. An obviousexample would be incentives targeted forpromoting certain sectors, such as high-technology industries, the development ofwhich is likely to confer significant posi-tive externalities on the rest of theeconomy. By far the most compelling casefor granting targeted incentives (if at all)is, however, for meeting regional devel-opment needs of these countries, such as

38 In the long term, the revenue consequence would obviously depend on import elasticities.39 Frequently, the most practical way to do this is to impose a low minimum tariff on all imports.40 If the affected imported excisable is also produced domestically, then the increase in its excise rate should also

apply, of course, to its domestically-produced counterpart.41 Any reduction in the scope of tax incentives in general (discussed below) should, of course, always be ex-

plored as an additional compensatory measure.42 For a comprehensive treatment of tax incentives in developing countries, see Shah (1995).43 OECD (1994) reports that, while tax incentives are widely used in Asian countries, country authorities are

generally skeptical about their effectiveness if the other aforementioned factors are absent.44 The United States never grants tax sparing.


316

encouraging investment in their less-de-veloped areas, or discouraging invest-ment in their more congested areas.45

Nevertheless, not all incentives areequally suited for achieving such objec-tives, however justifiable they may be;some are simply more cost effective thanothers on both policy and administrativegrounds. Unfortunately, the most preva-lent forms of incentives found in devel-oping countries tend also to be the leastmeritorious.

Tax holidays and reduced tax rates

Of all the different forms of tax incen-tives, tax holidays are the most popularamong developing countries. While ad-mittedly simple to administer, they havenumerous shortcomings which, eventhough shared to some degree by othertypes of incentives, are particularly pro-nounced: (1) by exempting profits irre-spective of their amount, tax holidays tendto benefit an investor who expects highprofits and would have undertaken theinvestment even if there were no such in-centives; (2) tax holidays provide strongincentives for tax avoidance, as taxed en-terprises could enter into economic rela-tionships with exempt ones to shift theirprofits to the latter through transfer pric-ing; (3) the duration of tax holidays, evenif formally time-bound, is prone to abuseand extension by investors through cre-ative redesignation of existing investmentas new investment (e.g., closing down andrestarting the same project under a differ-ent name but with the same ultimate own-ership); (4) time-bound tax holidays tendto attract (if they have an impact at all)short-run projects, which typically are notas beneficial to the economy as longer-term ones. The latter may become profit-able only toward the end of the holidaysand, therefore, can make little use of suchholidays even if losses can be carried for-

ward beyond the holiday period (if lossesare not allowed to be carried forward intothe post-holiday period, tax holidayscould, under some circumstances, be adisincentive to investment46 ); and (5) therevenue cost to the budget is seldom trans-parent, unless enterprises enjoying theholidays are still required to file propertax returns, in which case administrativeresources must be devoted to activitiesthat yield no revenue, and a frequentlyalleged benefit of tax holidays would benegated: allowing investors to dispensewith dealing with the tax authorities.

Since reduced tax rates are simply amilder form of tax holidays, the formerhave the same shortcomings as the latter,only on a lesser scale.

Investment allowances and tax credits

Compared with tax holidays, these taxincentives have a number of advantages.They are, for example, a much better tar-geting instrument than tax holidays forpromoting particular types of investment,and their revenue cost is much more trans-parent and easier to control. A particularlysimple and effective way of administer-ing a tax credit system is the following.Once the amount of tax credits to be givento a qualified enterprise is determined, itwould be “deposited” into a special taxaccount (kept simply in the enterprise’stax file) in the form of a bookkeeping en-try. The enterprise qualifying for this in-centive would in all respects be treatedlike a regular taxpayer and, therefore, besubject to all applicable tax provisions andregulations, including the computation oftaxable profits and the requirement to filetax returns. The only difference would bethat its income tax liabilities would bepaid from credits “withdrawn” from itstax account until the balance is reducedto zero. If desired, such a tax account couldbe closed after a specified period (i.e., a

45 Even under these circumstances, better policy instruments (e.g., increased infrastructure investment in re-mote areas) than tax incentives could often be found to achieve the stated objectives.

46 See Mintz (1990) for a rigorous demonstration of this result.


317

sunset provision is attached to the ac-count), so that all unused tax credits aresimply allowed to expire. In this way,information on total revenue foregonebecause of the incentive over any givenperiod is readily available at all times. Fur-thermore, as the amount of tax creditsgranted to qualified enterprises is knownwith certainty in advance, it can easily beincluded in the budget as a tax expendi-ture and subject to the same scrutiny asany other type of expenditure in the bud-getary process, thus achieving a high de-gree of transparency. Explicit recognitionof tax expenditure is a practice that canbe found in an increasing number of de-veloped and developing countries, andcan greatly facilitate the reviewing bypolicy makers of the cost-effectiveness oftax incentives.

A system of investment allowancescould be administered in much the sameway as tax credits, achieving similar re-sults. The only substantive differencebetween the two is that, if the CIT hasmultiple rates, then a given amount of in-vestment allowances is worth more in ab-solute terms, the higher the tax rate atwhich the allowances are given. In con-trast, a given amount of tax credits isworth exactly the same regardless of theapplicable tax rate.

There are two notable weaknesses as-sociated with investment allowances ortax credits. First, these incentives tend todistort the choice of capital assets in fa-vor of short-lived ones, since a further al-lowance or credit becomes available eachtime an asset is replaced. Second, quali-fied enterprises may attempt to abuse thesystem by selling and purchasing thesame assets to claim multiple allowancesor credits, or by acting as a purchasingagent for enterprises not qualified to re-ceive the incentive. Hence, safeguardsmust be built into such an incentive sys-

tem to minimize this danger, for example,by specifying a minimum holding periodfor an asset on which the incentive hasbeen given. Any shortfall in the holdingperiod would then require the enterpriseto remit to the tax authorities a proratedshare of the allowances/credits it has al-ready claimed and utilized associatedwith the asset.

Accelerated depreciation

Providing tax incentives in the form ofaccelerated depreciation has the least ofthe shortcomings associated with tax holi-days and all of the virtues associated withinvestment allowances/tax credits—andovercomes the latter’s weaknesses to boot.Since merely accelerating the depreciationof an asset does not increase the totalallowable nominal depreciation of theasset beyond its original cost, little distor-tion in favor of short-lived assets is gen-erated,47 and neither is there much incen-tive for an enterprise to engage in the kindof tax abuse connected with investmentallowances/tax credits.

Compared with other types of tax in-centives, accelerated depreciation has twoadditional merits. First, it is generally leastcostly, as the forgone revenue (relative tono acceleration) in the early years is atleast partially recovered in subsequentyears of an asset’s life. Second, if the ac-celeration is given only temporarily, thenit could (all other things equal) induce asignificant short-run surge in investment,as investors are likely to bring forwardinvestments planned for the future to takeadvantage of the incentive.

Investment subsidies

While investment subsidies share someof the merits of investment allowances/tax credits, such as ease of targeting, theyare generally quite problematic in thatthey pose an even more serious revenue

47 It should be noted that, if the underlying structure of depreciation rates for tax purposes deviates systemati-cally from the assets’ true structure of economic depreciation, then any change to the former (e.g., acceleratingthe depreciation rates) could induce additional distortions from the standard second-best type of reasoning.


318

risk for the budget than tax holidays. Theyinvolve out-of-pocket expenditure by thegovernment up front, and they benefitnonviable investments as much as profit-able ones. In contrast, other types of in-come tax incentives are of value only tothe latter. Hence, the use of investmentsubsidies is seldom advisable.

Indirect tax incentives

Indirect tax incentives are very proneto abuse, as qualified purchases can eas-ily be diverted to buyers not intended toreceive the incentives. They are also diffi-cult to justify on policy grounds, exceptunder extremely limited circumstances.Exempting raw materials and capitalgoods from the VAT, for example, wouldmake little difference to the ultimate taxburden of the enterprise concerned, sincethe VAT on such purchases is usually cred-itable. If the objective of such incentivesis simply to relieve the enterprise of itscash flow burden of the VAT, then a bet-ter solution would definitely lie else-where, most notably in providing it withprompt VAT refunds.

Exempting capital goods and raw ma-terials used to produce exports from im-port tariffs is somewhat more justifiable,as removing such duties embedded in thecontents of exportable goods throughstandard duty drawback mechanisms isinvariably complex and imprecise. Thedifficulty with this exemption lies, ofcourse, in ensuring that the exemptedpurchases would in fact be used as in-tended by the incentive. Many countrieshave attempted to solve this problem byestablishing special export production/processing zones whose perimeters aresecured by customs controls. Imports ofcapital goods and raw materials into thesezones are free from import tariffs, but tar-iffs are imposed on all exports from thezones to the rest of the country. Establish-

ing such zones cannot, however, stampout all abuse, as zones of this type arenever completely leakage-proof (even ifphysically controlled).

Triggering mechanism

A particularly important issue concern-ing the use of tax incentives in develop-ing countries is the mechanism by whichthey can be triggered. An automatic trig-gering mechanism allows the investmentto receive the incentives automaticallyonce it satisfies certain clearly specifiedobjective qualifying criteria, such as aminimum amount of investment in cer-tain sectors of the economy. These criteriaare usually laid down either in the rel-evant laws or their implementing regula-tions. In granting the incentives, the rel-evant authorities would only undertaketo ensure that the qualifying criteria aremet. All other aspects of the investmentare irrelevant.48

A discretionary triggering mechanisminvolves approving or denying an appli-cation for incentives on the basis of sub-jective value judgement of the relevantincentive-granting authorities after takinginto account a variety of considerations,irrespective of any formally stated quali-fying criteria. If such criteria exist, theyare stated either as minimum conditionsor in very general terms, thus requiringsubjective interpretation. An extreme formof a discretionary triggering mechanismwould be one by which incentives aregranted entirely on an ad hoc, case-by-case basis.

It is sometimes argued that a discretion-ary triggering mechanism is preferable toan automatic one because the formerwould provide the incentive-granting au-thorities with more flexibility in determin-ing the merits of individual incentive ap-plications. This advantage is likely to beoutweighed in practice, however, by a

48 Note that the approval of investment and the granting of tax incentives to approved investment could be twoseparate processes. Hence, the automatic triggering of the latter (once qualifying criteria are met) does notnecessarily require that the former be implemented also on a similar basis.


319

variety of problems that are inevitablyassociated with discretion, most notablya lack of transparency in the decision-making process, which in turn coulddiscourage some potential investors andencourage corruption and rent-seeking ac-tivities. In fact, incentives have been a fer-tile ground for corrupt practices in manycountries.

If the concern about having an auto-matic triggering mechanism is the loss ofdiscretion in handling exceptional cases,then the preferred safeguard would be toformulate the qualifying criteria in as nar-row and specific a fashion as possible, sothat incentives are only granted to invest-ments which can meet the highest objec-tive and quantifiable standard of merit.Ultimately the question is one of achiev-ing an optimal balance, and on balance itis generally advisable to minimize the dis-cretionary element in the incentive-grant-ing process.

Summing up

The cost effectiveness of providing taxincentives for investment promotion isgenerally questionable. The first-beststrategy for sustained investment promo-tion consists invariably of providing stableand transparent legal and regulatoryframeworks as well as adequate support-ing institutions and facilities, and of put-ting in place a tax system that is in linewith international norms.

Some objectives, such as those associ-ated with regional development needsof a country, are more justifiable thanothers as a basis for granting tax incen-tives. Not all tax incentives are, how-ever, equally effective. Accelerated depre-ciation has the most comparative merits,followed by investment allowances or taxcredits. Tax holidays and investment sub-sidies are among the least meritorious.As a general rule, indirect tax incen-tives should be avoided, and discretionin granting incentives should be mini-mized.

CHALLENGES AHEAD

As trade barriers come down and capi-tal mobility increases around the world,the formulation of sound and effective taxpolicy poses significant challenges fordeveloped and developing countriesalike. For the latter countries, however, thechallenges will be particularly acute onaccount of their relatively more limited taxadministration capabilities and their highdependence on foreign trade taxes. Withincreased global economic integration, thechallenge posed by the need to replaceforeign trade taxes with domestic taxeswill be accompanied by intensified con-cerns about profit diversion throughtransfer pricing, and/or profit strippingthrough thin capitalization, by foreign in-vestors. Yet, anti-tax abuse provisions inthe tax laws as well as the technical train-ing of tax auditors in many developingcountries are generally inadequate to de-ter and detect such practices. A concertedeffort to upgrade these deficiencies is,therefore, of the utmost urgency.

Another tax policy challenge in a worldof liberalized capital movements is in thearea of tax competition. For most devel-oping countries, the relevant issue here istheir perceived heightening of the pres-sure as well as the temptation to broadenthe scope of general tax incentives to at-tract and compete for foreign invest-ments—to the neglect of pursuing morefundamental tax reforms in, for example,those areas of the tax system that havebeen identified earlier. The effectivenessof tax incentives—in the absence of othernecessary fundamentals—is highly ques-tionable, as already noted. Furthermore,a tax system that is riddled with such in-centives will inevitably provide fertilegrounds for rent-seeking activities. To al-low their emerging markets to take properroot, developing countries would be welladvised to refrain from relying on tax in-centives—especially of the kind that arenot carefully and narrowly targeted—as


320

the main vehicle for investment promo-tion.49

Finally, as already indicated, personalincome taxes have been contributing verylittle to total tax revenue. Apart from thestructural, policy, and administrativeconsiderations discussed earlier, the facil-ity with which income received by indi-viduals can be invested abroad in an en-vironment of globally integrated financialmarkets and thus escape domestic taxa-tion—or, for that matter, escape taxationaltogether—also has been a significantcontributing factor of this outcome. In fact,tax havens as well as special arrangementsin several developed countries, such as taxfree accounts for non-resident aliens, havecreated situations whereby, for many de-veloping countries, it is difficult to applythe concept of the global income tax topersonal income taxation. Bringing finan-cial income adequately to tax is, therefore,a daunting challenge for developing coun-tries.50 This has been a particularly seri-ous problem in several Latin Americancountries, which have largely stopped tax-ing financial income to encourage finan-cial capital to remain in the countries.

CONCLUDING REMARKS

In this paper we have discussed severaltopics related to tax policy in developingcountries. Those among them which, asemerging markets, are attempting to be-come fully integrated with the worldeconomy, will face particularly significantchallenges. These countries will probablyneed a higher tax level, because of theneed to pursue a government role closerto that of the industrial countries that havetwice the tax burden. They will need toreduce sharply their reliance on foreigntrade taxes. And they will have to do sothrough tax policy changes that do notcreate strong disincentives. This will be

particularly important in their attempt toraise more revenue from the personal in-come tax. To meet these challenges suc-cessfully, policy makers in these countrieswill not only need to get their policy pri-orities right, they will also need to havethe political will to implement reformsthat are often difficult but absolutely nec-essary. It is also necessary to comment thatthe tax administrations must be strength-ened to accompany the needed policychanges.

Acknowledgments

The views expressed are those ofthe authors and do not necessarily repre-sent those of the IMF. Useful com-ments from Michael Keen and JohnYinger, and assistance from AsegedechWoldeMariam for compiling the tax dataused in the paper, are gratefully acknowl-edged.

REFERENCES

Atkinson, Anthony B., and Agnar Sandmo.“Welfare Implications of the Taxation of Sav-ings.” The Economic Journal 90 No. 359 (Sep-tember, 1980): 529–49.

Bahl, Roy W.“A Regression Approach to Tax Effort andTax Ratio Analysis.” IMF Staff Papers 18 No.3 (November, 1971): 570–612.

Barro, Robert J.“Government Spending in a Simple Modelof Endogenous Growth.” Journal of PoliticalEconomy 98 No. 5, Part 2 (October, 1990):S103–25.

Barro, Robert J., and Xavier Sala-i-Martin.Economic Growth. New York: McGraw-Hill,1995.

Bird, Richard M.Tax Policy and Economic Development. Balti-more: The Johns Hopkins University Press,1992.

49 This may well require a coordinated multilateral approach, at least on a regional basis.50 For more detailed discussions of these issues, see Tanzi (1995) and Zee (1998).


321

Bird, Richard, and Oliver Oldman (eds.).Taxation in Developing Countries, Fourth Edi-tion. Baltimore: The Johns Hopkins Univer-sity Press, 1990.

Burgess, Robin, and Nicholas Stern.“ Taxation and Development.” Journal of Eco-nomic Literature 31 No. 2 (June, 1993): 762–826.

Chamley, Christophe.“Optimal Taxation of Capital Income in Gen-eral Equilibrium with Infinite Lives.”Econometrica 54 No. 3 (May, 1986): 607–22.

Cnossen, Sijbren.“Global Trends and Issues in Value AddedTaxation.” International Tax and Public Fi-nance 5 No. 3 (July, 1998): 399–428.

Cnossen, Sijbren, and Richard M. Bird (eds.).The Personal Income Tax: Phoenix from theAshes? The Netherlands: North-Holland,1990.

Deininger, Klaus, and Lyn Squire.“A New Data Set Measuring Income In-equality.” The World Bank Economic Review10 No. 3 (September, 1996): 565–91.

Easterly, William, and Sergio Rebelo.“Fiscal Policy and Economic Growth: An Em-pirical Investigation.” Journal of Monetary Eco-nomics 32 No. 3 (December, 1993): 417–58.

Ebrill, Liam, Michael Keen, Jean-Paul Bodin,and Victoria Summers.

Recent Experience with the Value-Added Tax.Washington, D.C.: International MonetaryFund, forthcoming.

Ebrill, Liam, Janet Stotsky, and Reint Gropp.Revenue Implications of Trade Liberalization.Occasional Paper No. 180. Washington,D.C.: International Monetary Fund, 1999.

Heady, Christopher.“Optimal Taxation as a Guide to Tax Policy:A Survey.” Fiscal Studies 14 No. 1 (February,1993): 15–41.

Kaldor, Nicholas.An Expenditure Tax. London: George Allenand Unwin, 1955.

Keen, Michael, and Jenny E. Ligthart.“Coordinating Tariff Reduction and Domes-tic Tax Reform.” IMF Working Paper WP/99/93. Washington, D.C.: International Mon-etary Fund, 1999.

Levine, Ross, and David Renelt.“A Sensitivity Analysis of Cross-CountryGrowth Regressions.” American EconomicReview 82 No. 4 (September, 1992): 942–63.

Masson, Paul R., Tamim Bayoumi, andHossein Samiei.

“International Evidence on the Determi-nants of Private Saving.” The World Bank Eco-nomic Review 12 No. 3 (September, 1998):483–501.

Messere, Ken C.Tax Policy in OECD Countries: Choices andConflicts. The Netherlands: IBFD Publica-tions BV, 1993.

Messere, Ken, and Gilroy J. Zuckerman.“An Alternative Approach to DepreciationSwitches.” Accounting Review 56 No. 3(1981): 642–52.

Metcalf, Gilbert E.“Life-Cycle Versus Annual Perspective onthe Incidence of a Value-Added Tax.” TaxPolicy and the Economy 8, edited by James M.Poterba. Cambridge: The MIT Press, 1994.

Mintz, Jack M.“Corporate Tax Holidays and Investment.”The World Bank Economic Review 4 No. 1(January, 1990): 81–102.

Musgrave, Richard A.Fiscal Systems. New Haven: Yale UniversityPress, 1969.

Newbery, David, and Nicholas Stern (eds.).The Theory of Taxation for Developing Coun-tries. New York: Oxford University Press,1987.

Norman, Göran, and Jeffrey Owens.“Tax Effects on Household Saving: Evidencefrom OECD Member Countries.” In Promot-ing Savings in Latin America, edited byRicardo Hausmann and Helmut Reisen.Paris: OECD, 1997.

OECD.Taxing Profits in a Global Economy: Domesticand International Issues. Paris: OECD,1991.

OECD.Taxation and Investment Flows: An Exchangeof Experiences Between the OECD and theDynamic Asian Economies. Paris: OECD,1994.


322

Republic of South Africa.Interim Report of the Commission of Inquiry intoCertain Aspects of the Tax Structure of SouthAfrica. Pretoria: Government Printer, 1994.

Seidman, Laurence S.The USA Tax: A Progressive Consumption Tax.Cambridge: The MIT Press, 1997.

Shah, Anwar (ed.).Fiscal Incentives for Investment and Innovation.New York: Oxford University Press, 1995.

Tait, Alan A., Wilfrid L. M. Gratz, and Barry J.Eichengreen.

“International Comparisons of Taxation forSelected Developing Countries.” IMF StaffPapers 26 No. 1 (March, 1979): 123–56.

Tanzi, Vito.“Quantitative Characteristics of the Tax Sys-tems of Developing Countries.” In TheTheory of Taxation for Developing Countries, ed-ited by David Newbery and Nicholas Stern.New York: Oxford University Press, 1987.

Tanzi, Vito.Public Finance in Developing Countries.Aldershot: Edward Elgar, 1991.

Tanzi, Vito.“Structural Factors and Tax Revenue in De-veloping Countries: A Decade of Evidence.”In Open Economies: Structural Adjustment andAgriculture, edited by Ian Goldin and L. AlanWinters. New York: Cambridge UniversityPress, 1992.

Tanzi, Vito.Taxation in an Integrating World. Washington,D.C.: The Brookings Institution, 1995.

Tanzi, Vito, and Howell H. Zee.“Fiscal Policy and Long-Run Growth.”IMF Staff Papers 44 No. 2 (June, 1997): 179–209.

Tanzi, Vito, and Howell H. Zee.“Taxation and the Household Saving Rate:Evidence from OECD Countries.” BancaNazionale del Lavoro Quarterly Review, forth-coming.

Turnovsky, Stephen J.“Optimal Tax, Debt, and Expenditure Poli-cies in a growing Economy.” Journal ofPublic Economics 60 No. 1 (April, 1996): 21–44.

Zee, Howell H.“Taxation of Financial Capital in a Global-ized Environment: The Role of WithholdingTaxes.” National Tax Journal 51 No. 3 (Sep-tember, 1998): 587–99.

Zee, Howell H.“Empirics of Cross-Country Tax RevenueComparisons.” World Development 24 No. 10(October, 1996): 1659–71.

Zee, Howell H.“Value-Added Tax.” In Tax Policy Hand-book, edited by Parthasarathi Shome. Wash-ington, D.C.: International Monetary Fund,1995.

Documents

Tax Policy for Emerging Markets: Developing Countries · In developing countries with emerging markets, and especially in those that aim at becoming integrated with the international