Helping Countries Develop
Chapter

12 Tax Policy in Developing Countries: Some Lessons from the 1990s and Some Challenges Ahead

Author(s):
Benedict Clements, Sanjeev Gupta, and Gabriela Inchauste
Published Date:
September 2004
Share
  • ShareShare
Show Summary Details
Author(s)
Michael Keen and Alejandro Simone 

One of the central challenges facing low-income and developing countries is to mobilize sufficient tax revenue to sustainably finance, when combined with whatever aid is available, the expenditures needed for growth and poverty relief—and to do so in a way that does not itself undercut those objectives by unduly worsening preexisting distortions or inequities. This chapter seeks to describe and assess the way in which developing countries were addressing these problems at the turn of the century. More particularly, it focuses on the experience of the 1990s—roughly the period since the major survey of taxation in developing countries by Burgess and Stern (1993). While experiences have naturally varied quite significantly across countries over this decade, and although 10 years is a relatively short period in the life of a tax system—indeed, that may be one of the central lessons to be drawn from the analysis in this chapter—some common themes nevertheless emerge. And some of them are troubling.

Many of the characteristics of developing countries, some of which are displayed in Tables 1 and 2 (with countries grouped by income range and region, respectively),1 make the problem of revenue mobilization particularly difficult. Informal activities are extensive (as proxied, for instance, by the low degree of monetization in the second column),2 tending to imply a relatively narrow potential tax base; the agricultural sector—hard to tax in all countries, for both practical and, in many cases, political reasons—is large; the capacity of potential taxpayers to comply with tax rules, and of the authorities to administer them, is likely to be relatively low (as suggested by relatively high rates of illiteracy); and corruption is often pervasive (final column). All these features constrain effective taxation. Moreover, many of them are themselves likely to be affected by the tax system in force. Heavy taxation of the formal sector will tend to encourage growth of the informal sector, for instance, and inappropriate tax design may provide further opportunities for corruption. While similar concerns arise in developed countries, they are an order of magnitude more significant in the developing world—posing problems for tax design in these countries to which the public finance literature has paid scant attention.

Table 1.Economic Indicators by Income Range, Early 1990s and Early 2000s
AbsoluteTrade =Illiteracy
MoneyValueValue of ExportsRate,
and Quasi-Capital andand Imports ofAgriculture,Adult TotalTransparency
NominalMoneyFinancialGoods andMeasureValue(PercentIndex,
GDP Per Capita(M2)AccountServices (Open)of DebtAddedof peopleTransparency
(U.S. dollars,aged 15International
average)(Percent of GDP)and above)(1996)
Early 2000s
Low-income countries61442.25.574.279.029.625.42.8
Lower-middle-income countries1,64256.56.684.449.715.420.63.6
Upper-middle-income countries4,14250.36.093.437.09.97.74.1
High-income countries21,38772.75.498.225.43.48.47.7
Unweighted average2
Developing countries32,26949.26.084.554.318.317.53.6
High-income countries21,38772.75.498.225.43.48.47.7
Total unweighted average7,75653.35.888.353.614.616.45.1
Memorandum
PRGF-eligible countries461736.37.380.881.029.729.22.4
Early 1990s
Low-income countries44532.06.154.9103.731.132.42.4
Lower-middle-income countries1,19454.46.672.461.718.327.43.3
Upper-middle-income countries3,74743.06.762.627.814.411.24.4
High-income countries18,41862.14.180.06.34.511.57.7
Unweighted average2
Developing countries31,93642.46.562.761.621.123.13.6
High-income countries18,41862.14.180.06.34.511.57.7
Total unweighted average6,66745.85.867.560.116.721.85.6
Memorandum
PRGF-eligible countries585633.48.162.398.031.337.22.3
Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook; and World Bank, World Development Indicators database.

Data used for early 1990s and early 2000s are averages for two years 1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

For each indicator, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Averages were calculated for PRGF-eligible countries for which data were available (see Appendix Table A6 for details).

Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook; and World Bank, World Development Indicators database.

Data used for early 1990s and early 2000s are averages for two years 1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

For each indicator, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Averages were calculated for PRGF-eligible countries for which data were available (see Appendix Table A6 for details).

Table 2.Economic Indicators by Region, Early 1990s and Early 2000s
AbsoluteTrade =Illiteracy
MoneyValueValue of ExportsRate,
and Quasi-Capital andand Imports ofAgriculture,Adult TotalTransparency
NominalMoneyFinancialGoods andMeasureValue(PercentIndex,
GDP Per Capita(M2)AccountServices (Open)of DebtAddedof peopleTransparency
(U.S. dollars,aged 15International
average)(Percent of GDP)and above)(1996)
Early 2000s
Americas23,30138.06.463.751.612.011.73.8
Sub-Saharan Africa76529.26.769.790.527.928.93.0
Central Europe and BRO32,42033.76.6104.240.215.12.03.7
North Africa and Middle East2,48676.45.071.046.814.229.14.4
Asia and Pacific1,44763.13.892.647.123.719.73.1
Small islands44,67397.69.7150.730.410.95.4
Unweighted average5
Developing countries62,26949.26.084.554.318.317.53.6
High-income countries21,38772.75.498.225.43.48.47.7
Memorandum
PRGF-eligible countries761736.37.380.881.029.729.22.4
Early 1990s
Americas22,02627.64.455.283.614.215.53.6
Sub-Saharan Africa83127.73.263.088.727.138.53.5
Central Europe and BRO33,47045.09.834.317.222.43.14.6
North Africa and Middle East1,95170.88.268.168.417.039.43.8
Asia and Pacific1,06141.56.766.356.927.924.93.1
Small islands43,22177.99.8146.524.012.38.2
Unweighted average5
Developing countries61,93642.46.562.761.621.123.13.6
High-income countries18,41862.14.180.06.34.511.57.7
Memorandum
PRGF-eligible countries85633.48.162.398.031.337.22.3
Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook; and World Bank, World Development Indicators database.

Data used for early 1990s and early 2000s are averages for two years 1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

For each indicator, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Averages were calculated for PRGF-eligible countries for which data were available (see Appendix Table A6 for details).

Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook; and World Bank, World Development Indicators database.

Data used for early 1990s and early 2000s are averages for two years 1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

For each indicator, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Averages were calculated for PRGF-eligible countries for which data were available (see Appendix Table A6 for details).

One implication of these distinctive concerns is that the link between tax policy and tax administration is especially intimate in developing countries. It has been famously claimed, indeed, that in developing countries “tax administration is tax policy” (Casanegra de Jantscher, 1990). This may be going too far3—any administration operates within the broad confines of some policy framework, even if only to abuse and even misrepresent it. But it is clear that the disconnect between what tax rules say and what actually happens can be especially great in the developing world, and that this needs to be taken into account in designing tax policy. While the distinction between tax policy and administration is particularly blurred in developing countries, it is the broad policy design that sets the stage for implementation. And it is this that is the focus of this paper.

Our coverage is necessarily selective. For instance, we shall not address the tax treatment of small and medium-sized enterprises—not because it is unimportant (indeed, the considerations above suggest that the taxation of those at the margin of formality is likely to be a key issue in many developing countries), but because it does not seem to be an area in which the 1990s saw much change. Nor do we discuss in any detail the experiences of transition countries over the 1990s (though we do include them in our data analysis, treating them in our regional analyses as a separate group): these are so distinctive as to require separate treatment (for a recent study, see Summers and Baer, 2003). This indeed is a reminder that while our focus is on broad and emerging trends, tax design must ultimately be attuned to countries’ divergent circumstances.

The chapter is organized as follows. The next section seeks to develop some broad stylized facts to guide further analysis, providing a broad overview of revenue structures in developing countries, both at the turn of the century and as they changed (or not) over the past decade. The remaining sections then focus on three particular issues that emerge as having been prominent in the 1990s, and, perhaps, likely to be important in the coming years: the continued move away from trade taxes, the spread of the VAT to the developing world, and the apparent erosion of corporate tax revenues. The final section concludes.

Levels and Composition of Revenue—Some Stylized Facts Data

Any analysis of taxation in developing countries faces the fundamental problem that reliable and comparable information is scant, both on revenues and, still more, on details of tax structure. So it is appropriate to begin with a word—and a warning—about the data that we are about to use.

The account of revenue developments in this section is based on information from Government Finance Statistics (GFS), which, though the best source, suffers from a number of serious deficiencies. In particular, the breadth of coverage of the government sector varies across countries, and many countries provide information for some years, but not for others.4 Here, we focus on central government revenues; this will be too narrow a definition for countries with significant revenue-raising at lower levels of government, but alternative approaches produce unusably small samples. The incompleteness and variability of coverage in the GFS is especially troublesome for the purposes of this paper, given the focus on identifying trends. In all the tables below, we take the same sample of countries throughout the period so as to minimize spurious compositional effects from a changing sample (such as finding a large apparent increase in reliance on total revenues by including a resource-rich country at the end of the 1990s, but not at the beginning).5 The price paid for maintaining a reasonably large sample in this way is some flexibility in the years taken to represent the early 1990s and early 2000s—our (informal) sense being that biases from any consequent time-specific effects are likely to be less significant than those from country-specific ones.6 There are, thus, many caveats to be borne in mind in interpreting the figures to which we now turn.

Where Are We Now?

The upper panel of Table 3 shows the level and composition of central government revenues (relative to GDP) at the end of the 1990s; the lower panel shows level and composition at the start. Countries here are grouped into four categories by their income level at the start of the 1990s,7 the categorization corresponding broadly to that used by the World Bank income classification of countries. Roughly speaking, the category “high-income countries” corresponds to developed countries. Appendix Table A1 shows the same figures relative to total tax revenue, with countries again grouped by income range. Appendix Table A2 shows the same information but with countries grouped regionally,8 with Appendix Table A3 again reexpressing the same figures relative to total tax revenue.

Table 3.Central Government Revenues by Income Group, Early 1990s and Early 2000s1(Percent of GDP)
Domestic Taxes on

Goods and Services
Taxes on Income, Profits,

and Capital Gains
of which:International Trade Taxes2
Nominal

GDP per Capita

(U.S. dollars,

average)
of which:Social

Security

Taxes
General

sales,

turnover,

or VAT
of which:
Total

Revenue
Tax

Revenue
Other

Revenue
TotalIndividualCorporatePayroll

Taxes
TotalExcisesTotalImport

duties
Export

duties
Property

Taxes
Early 2000s
Low-income countries61418.014.93.13.91.92.01.00.05.93.52.03.72.40.20.2
Lower-middle-income countries1,64222.016.06.13.91.82.21.30.16.44.92.13.63.50.10.3
Upper-middle-income countries4,14225.621.04.65.32.62.64.50.18.25.32.42.52.40.20.3
High-income countries21,38732.827.55.38.97.42.57.30.39.16.72.81.31.30.00.7
Unweighted average3
Developing countries42,26922.117.64.54.42.12.32.50.16.94.52.23.22.70.20.3
High-income countries521,38732.827.55.38.97.42.57.30.39.16.72.81.31.30.00.7
Total unweighted average7,75625.120.44.75.73.72.33.90.17.55.12.32.92.50.20.4
Early 1990s
Low-income countries44517.714.53.13.81.32.60.70.05.32.82.14.33.80.60.3
Lower-middle-income countries1,19421.416.35.14.31.52.91.40.25.02.92.04.44.10.20.4
Upper-middle-income countries3,74726.921.95.05.92.23.34.30.26.53.72.44.33.90.60.3
High-income countries18,41831.926.65.28.57.61.97.20.38.66.32.41.92.00.00.7
Unweighted average3
Developing countries41,93622.317.94.44.71.72.92.30.15.73.22.24.33.90.50.3
High-income countries518,41831.926.65.28.57.61.97.20.38.66.32.41.92.00.00.7
Total unweighted average6,66725.020.44.65.83.52.63.70.26.54.02.33.93.60.40.4
Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook.

Data used for early 1990s and early 2000s are averages for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

European Union countries do not report statistics on international trade taxes to Government Finance Statistics.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

See Table A5 for details.

Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook.

Data used for early 1990s and early 2000s are averages for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

European Union countries do not report statistics on international trade taxes to Government Finance Statistics.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

See Table A5 for details.

The snapshot picture at the end of the century is relatively straightforward and familiar. As is well known, tax ratios—sometimes referred to, for no very good reason, as “tax effort”—decrease, all else equal, with the level of GDP. Although it is not shown here, they also increase with the level of openness.9

In terms of structure, developing countries rely heavily on indirect taxes (about 40 percent of their tax revenue, as a rule of thumb), and derive about one-fourth of their tax revenue from income taxes (personal and corporate), one-fifth from trade taxes, and about one-sixth from payroll and social security. Property taxes are insignificant. For developed countries, indirect taxes are less important and income taxes more important (each about one-third of total tax revenue); social security and payroll is far more important (rather less than one-third), trade taxes are much less important; and property taxes, though more important than in developing countries, make only a very modest contribution.

What Happened in the 1990s?

This chapter is less concerned with the snapshot just given, however, than with developments over the 1990s. Consider first the experience of developing countries as a whole.

In terms of overall revenue performance—total revenues and total tax revenues relative to GDP—the story is of the dog that didn’t bark. This was a period in which the general objective of (or at least conventional advice to) many low-income countries was to increase their tax collections in order to finance growth-enhancing and poverty-reducing expenditures. There were exceptions, of course: Kenya, for example, has deliberately sought to encourage development by reducing its tax ratio. And certainly the efficiency and growth-friendliness of a tax system can be enhanced without necessarily increasing revenue (by, for instance, shifting away from more distortionary taxes (such as those on trade) to less (perhaps a clean VAT). Nevertheless, it is striking that there was no general increase in tax ratios for developing countries as a whole: indeed, the average tax ratio for these countries fell slightly, though the safer characterization is one of stagnation.

There are, however, some clear general changes in the composition of tax revenues. One is a shift away from trade taxes and (to a lesser extent) toward indirect taxation, especially general sales taxes (of which the VAT is the leading but not the only type). This would no doubt have been widely predicted at the start of the 1990s as a continuation of the process of trade liberalization. Even more intriguing, however, is the reduction in revenues from the corporate income tax—these fell, for developing countries as a whole, by about one-fifth. Here the contrast with the developed countries in the sample is especially marked: despite widespread concern at the erosion of corporate taxes as a result of intensified international tax competition, developed countries actually experienced an increase in corporate tax revenues both relative to GDP and as a share of total tax revenue.

This overall picture masks, however, rather different experiences at different levels of development and in different regions. The tax ratio in the poorest countries was essentially stagnant: the average increased by 0.4 point of GDP, but this is both small and largely owing to an outstanding measured increase in Estonia, a single country in unusual economic (and statistical) circumstances.10 In the two poorest regions (sub-Saharan Africa and Asia-Pacific), and in our sample of PRGF-eligible countries, the picture is again one of stagnation, but with tax ratios, if anything, falling (see Appendix Table A2). The effects in some countries were quite marked: in Côte d’Ivoire, for example, tax revenue fell by about 3 percentage points,11 and in Sri Lanka it fell by about 5 percentage points (with total revenue also falling in both cases, so that this does not simply reflect an offset to increased nontax revenues). Experience has thus been very variable among the poorest countries—but with no very clear overall tendency to enhanced domestic revenue collection. It is also in these poorest countries that some of the compositional changes mentioned above (the shift away from trade and toward indirect taxes) are least marked. It is also noticeable, however, that these countries did experience a significant loss of corporate tax revenues: relative to GDP, they fell by about one-fourth.

In contrast, while tax ratios changed relatively little among lower-middle-income countries, they tended, if anything, to fall. This masks a striking compositional change. In these countries, a spectacular increase in indirect tax revenues (by about 25 percent)—especially from sales taxes—more than offset a marked reduction in revenue from trade taxes. Revenue from the personal income tax (PIT) also tended to rise. Yet total tax revenues fell as a result of reduced revenue from the corporate income tax. The experience in upper-middle-income countries—such as Brazil—was broadly the same, except that there the reductions in trade taxes were matched by increases in indirect taxes (increasing revenue presumably being less of a concern in these countries).

One other potentially significant source of revenue should also be considered, though it is not recorded in the standard statistics used above. This is seignorage: the command over resources that governments enjoy by virtue of their ability to issue base money. Table 4 shows that this was indeed an important source of revenue for developing countries in the early 1990s: about 4 percent of GDP, or 22 percent of (seignorage-exclusive) tax revenue. By the end of the decade, the widespread reduction in inflation rates was associated with a reduction in seignorage of more than one-third.12 This fall is large enough to produce a significant overall reduction in revenues in developing countries. There are certainly considerable macroeconomic benefits from this reduction in inflation rates. But the important point here is that seignorage can be seen as a tax like any other, in that it creates inefficiencies and inequities to be weighed against the revenue it yields—and, whatever the efficiency gains, that revenue has fallen noticeably over the past decade.

Table 4.Seignorage as a Share of GDP by Income Group, Early 1990s and Early 2000s1
Nominal

GDP per Capita

(U.S. dollars,

average)
Seignorage

(Percent

of GDP)
Early 2000s
Low-income countries6143.0
Lower-middle-income countries1,6422.0
Upper-middle-income countries4,1422.2
High-income countries21,3871.7
Unweighted average2
Developing countries32,2692.4
High-income countries421,3871.7
Total unweighted average7,7562.3
Early 1990s
Low-income countries4454.4
Lower-middle-income countries1,1943.7
Upper-middle-income countries3,7473.9
High-income countries18,4181.4
Unweighted average2
Developing countries31,9364.0
High-income countries418,4181.4
Total unweighted average6,6673.5
Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook; and staff calculations.

Data used for early 1990s and early 2000s are average for two years—1991-92 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction. Seignorage is calculated as the money growth rate multiplied by money stock (M1) divided by nominal GDP. The hyperinflation episodes of Brazil and Nicaragua in the early 1990s are ignored, and replaced by estimates from normal times.

For each region, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Comparable measures of M1 were not available for most European countries.

Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook; and staff calculations.

Data used for early 1990s and early 2000s are average for two years—1991-92 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction. Seignorage is calculated as the money growth rate multiplied by money stock (M1) divided by nominal GDP. The hyperinflation episodes of Brazil and Nicaragua in the early 1990s are ignored, and replaced by estimates from normal times.

For each region, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Comparable measures of M1 were not available for most European countries.

Summing up, experiences differ greatly when one looks behind the group averages to the experience of individual countries. Nevertheless, four general conclusions seem to characterize the experience of developing countries in the 1990s:

  • revenue has been essentially stagnant, at best, in the poorest countries and regions of the developing world—taking seignorage into account, it has generally fallen;

  • revenues from general sales taxes (in practice often meaning the VAT) have increased markedly, albeit less in the poorest countries and regions than elsewhere;

  • trade tax revenues have fallen significantly, though least in the poorest countries; and

  • corporate tax revenues have declined, except in the Americas (where, especially given the relatively high average level of income, receipts from this source have long been noticeably low) and in small island economies (perhaps reflecting to some extent tax haven activities).13

We do not attempt a full explanation of the first of these observations, which would need to be rooted in the diversity of country experiences. Instead, the rest of the chapter considers in more detail the last three of these developments, which are likely to prove important in future tax design—and which may also hold some clues for explaining, and starting to escape, the stagnancy of total revenues.

The Spread of the VAT

The most significant change in the tax structure of the typical developing country during the 1990s has been the adoption, under a variety of names, of the VAT14 Having swept through Latin America and Western Europe from the late 1960s, the tax made remarkable inroads into the developing world over the past decade. In 1990, about 30 percent of developing countries in our sample had a VAT; by the end of the century, this figure had risen to about 75 percent. More detail on the rise of the VAT in the developing world is shown in Table 5. Among both low-income and lower-middle-income countries, the number of countries with a VAT more than doubled. To some extent, this reflects the unusual experience of the CIS countries, almost all of which adopted the VAT very quickly at independence. But there is much more to it than that. In sub-Saharan Africa, in particular, the number of countries with a VAT increased from two to nine; by the end of the 1990s, only 40 percent of the sub-Saharan African countries in our sample for the region did not have a VAT. These figures somewhat overdramatize the effective change: many pre-VAT sales tax systems had VAT-like features (for instance, some limited degree of crediting, perhaps among a “ring” of firms), and all VATs fall short, in practice, of the textbook ideal (perhaps excluding the retail stage, or providing only imperfect credits for purchases of capital goods). Nevertheless, the structural change is clearly profound.

Table 5.Number of Countries with VAT, by Region and Income1
By Region
Americas2Sub-

Saharan

Africa
Central

Europe

and

BRO3
North

Africa

and

Middle East
Asia

and

Pacific
Small

Islands4
Total

Number of

Developing

Countries

with VAT

in Sample
Early 2000s16914511257
Early 1990s132134023
By Income1
Low-IncomeLower-

Middle-Income
Upper-

Middle-Income
High-Income
Early 2000s17172326
Early 1990s761018
Sources: International Bureau of Fiscal Documentation, 2003, and Pricewaterhouse Coopers, Corporate Taxes 2003-04, Worldwide Summaries.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

Sources: International Bureau of Fiscal Documentation, 2003, and Pricewaterhouse Coopers, Corporate Taxes 2003-04, Worldwide Summaries.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

This remarkable spread of the VAT to the developing world has been little studied—indeed, the same is true of its spread in the developed world. Little is known, for instance, of the political economy behind the adoption of what is often an extremely unpopular tax—though there is some evidence15 that, apart from possible efficiency gains (discussed below), an important role has been played both by regional demonstration effects (whether as mere fashion, or the consequence of yardstick competition as influential groups observe the functioning of VATs nearby), and by participation in IMF programs (which may, in turn, proxy a deeper fiscal malaise).

Much better understood are the arguments of principle deployed by the advocates and critics of the VAT. Broadly speaking, the principal merits claimed for the VAT are the prospect of securing revenue—as a consequence of tax being levied on all transactions—more securely than do retail sales taxes (under which all tax is lost if there is evasion at the final stage) and without the distortion of production decisions and non-transparency associated with cascading turnover taxes. Critics, on the other hand, have argued that the VAT imposes unreasonable bookkeeping requirements on small traders, tends to have adverse distributional effects, and—a concern heard increasingly loudly—is, by virtue of its allowing for substantial tax refunds, worryingly vulnerable to fraud.

This is not the place to enter the detail of these debates, which are discussed at length in Ebrill and others (2001) and, with a more skeptical view of the VAT, by Stiglitz (2003). It should be noted, however, that the important question is whether the concerns of critics can be adequately addressed by proper design of the VAT rather than by its wholesale rejection in favor of some (often unspecified) alternative. In some respects, the design issues raised by critics of the VAT are relatively straightforward to address. For example, small traders can be kept out of the VAT system with relatively little loss of revenue—given the apparently ubiquitous concentration of potential tax base among a relatively small number of traders—by setting a reasonably high threshold, though the distortions that this in itself creates by placing small traders at a tax advantage further complicates the design issue (Keen and Mintz, 2004). Other trade-offs are more difficult.

In particular, as noted by Ebrill and others (2001) and stressed by Stiglitz (2003), the limited effectiveness of personal income taxes in developing countries implies a stronger case for differentiating rates of indirect taxation across commodities in developing than in developed countries. At the same time, however, the administrative and compliance problems associated with differential rates of VAT—including the increased likelihood that traders will become entitled to refunds (using highly taxed inputs to produce lightly taxed outputs)—points in exactly the opposite direction, toward a stronger case for a single VAT rate in developing countries than in developed. It may also be that the governance weaknesses of developing countries make them more vulnerable to lobbying for reduced rates against which adherence to a single rate provides some protection.16 And, indeed, a feature of the VATs adopted by developing countries over the past decade is that they had, by and large, a single positive rate.17 While that suggests that the administrative and compliance arguments have, in practice, won the day, the degree of differentiation in indirect tax rates is far greater in developing countries than the appearance of a single rate VAT might imply. Considerable differentiation is, in practice, generally achieved by the exemption of basic commodities under the VAT,18 a compromise between setting a differentially low statutory rate and the collection difficulties that would create. Moreover, excise taxes on particular commodities directly introduce some differentiation, often to a larger extent—by imposing such taxes on a wider range of items—than is common in developed countries.

Whether the VAT has reduced or increased the progressivity of the overall tax systems in developing countries is ultimately an empirical question (and one that encounters the wide range of conceptual issues associated with any such question of incidence).19 A number of studies, including Younger and Sahn (1998) and the analysis for Ethiopia by Muñoz and Cho (2004), conclude that the impact of the VAT is often moderately progressive, and sometimes more so than the taxes it replaces.

Other aspects of the spread of the VAT are even harder to quantify. While there is some evidence on the administration and compliance costs of the VAT in developed countries, for instance (such as Cnossen, 1994), there is almost no hard empirical evidence on this or on the costs associated with alternative taxes in developing countries. Moreover, while such collection costs are fairly straightforward to measure,20 the main administrative argument for the VAT is so fundamental as to defy precise definition or measurement. For, caricaturing somewhat, the (often unspoken) response to the complaint that developing countries find the VAT hard to implement is “good, they are supposed to.” This is because adoption of the VAT is often intended to spearhead a fundamental change in how taxes are collected, in particular by introducing methods of self-assessment—that is, self-declaration of liability by the taxpayer supplemented by risk-based audits—that can then be applied to other taxes. In particular, the income tax might in this way be transformed into something more than a glorified withholding tax on employees of the public sector and large enterprises. Tax collection thus moves away from the face-to-face contact that is so conducive to bribery and extortion. More generally, the VAT can serve as a catalyst for broader changes in the way countries conduct their tax business, for instance, with a move away from tax-based organizational structures to functional- (or taxpayer-segment-) based ones. This perspective can change the way one thinks about the VAT. Rather than a distraction from strengthening the income tax, for example, the VAT becomes a strategy for its ultimate improvement. While there are relatively few signs of such strengthening happening yet, moving to this second stage of reform has always been seen as a task for years rather than months. Whether this strategy for wider tax reform—and, indeed, for governance reform more broadly—succeeds or not may in time be the most important question to be asked about the spread of the VAT to the developing world.

In the meantime, there are some simpler empirical questions that can and should be addressed. The most basic of these is whether the VAT has, in fact, delivered the efficiency gains claimed by its advocates. It is a telling observation that only a handful of countries have ever removed a VAT, and of these only one—Grenada—appears adamantly opposed to its reintroduction in any guise. Revealed preference thus suggests that adoption of the VAT has, at the least, not been widely perceived as a serious mistake.

A more systematic approach to this issue is developed by Ebrill and others (2001) and Keen and Lockwood (2003). This is to exploit the prediction of theory that if adoption of the VAT reduces the marginal social cost of raising tax revenue (this cost taking into account both efficiency and distributional effects), then—assuming, as is commonly supposed, that the income elasticity of public expenditure exceeds unity—countries with a VAT should have a higher ratio of total tax revenue to GDP, all else equal, than do countries without. Empirically, this boils down to including in a standard “tax effort” regression—relating the tax ratio to GDP per capita, openness, and other usual suspects—a dummy representing the presence or absence of a VAT.21 The broad empirical conclusions are the same for both the cross-section analysis in Ebrill and others (2001) and (though less marked) the panel data used by Keen and Lockwood (2003). Broadly speaking, the revenue gain associated with a VAT increases with the level of real income and decreases with the level of openness (the former effect, perhaps, proxying the lesser importance of hard-to-tax sectors, such as agriculture, in more developed economies; the latter perhaps reflecting the ease of raising trade taxes, an alternative source of revenue).

The results leave open the possibility that in very open developing countries, adoption of the VAT may actually have been associated with a reduction in overall revenues, implying—given the maintained assumption on the demand for public expenditure—that adoption increased rather than reduced the marginal social cost of public funds. In principle, it is possible to make this calculation for any particular country. In practice, however, the coefficients are not well enough determined to do so with great confidence. What these results do suggest, however, is that it is not as easy for poor and open economies to show clear efficiency gains from the VAT as its advocates might have hoped.

For reasons touched on above, thinking of the VAT as an all-or-nothing matter—while a reasonable (and, in practice, inescapable) simplification for statistical analysis—is not a good basis for policymaking. Indeed, experience in developing countries has demonstrated the need to think of a VAT as work in progress. Many countries still struggle, in particular, to find an appropriate balance between providing speedy refunds to exporters—essential if the tax is not to function in part as a de facto export tax—and ensuring adequate protection against fraud (Ebrill and others, 2001). This, in turn, is but the most obvious example of the more general control problem implied by the observation that “… a supplier’s invoice (or export certificate) in effect constitutes a check drawn on government [and so] constitutes a tempting target for those who would loot the treasury” (Bird, 1993).

Policy and administrative improvements for the VAT are closely aligned. In policy terms, for instance, exemptions are anathema to the logic of the VAT—in a way that multiple rates of VAT, in particular, are not—since they imply a potential for production inefficiencies. Yet removing those exemptions would amplify the control difficulties. In pure policy terms too, understanding of what can be achieved under the VAT continues to improve. More has been learned, for instance, of how financial services can be brought properly into the tax net (though whether one wants to is a different issue),22 and how the VAT might be made to function as a lower-level tax within a federal system.23 In developed countries, such structural improvements are likely to dominate the VAT agenda in the coming years; in the developing world, in contrast, the primary tasks continue to be unwinding inappropriate exemptions and ensuring proper functioning of the refund and credit mechanisms—a key part of the wider reform of ways of doing tax business that, as mentioned above, was a primary motive for its introduction.

Shifting Away from Trade Taxes

The 1990s saw a continuation in the shift away from trade taxes as a source of revenue in developing countries. But the shift was particularly modest in the poorest of countries. In sub-Saharan Africa, most noticeably, there was effectively no change in the reliance on trade taxes, which continued to account for about one-third of all tax revenues. In all other regions, the decline was marked. This decline in reliance on trade tax revenues seems to have been the product of deliberate trade liberalization policies. Table 6 shows that collected rates of trade taxation—the ratio of tariff and export tax revenue to the value of imports plus exports24—fell in all income groups and regions, though least in sub-Saharan Africa, North Africa, and the Middle East. While openness increased over the 1990s, as can be seen from Table 1—perhaps partly in response to these and other measures of trade liberalization—it did not do so by enough to actually increase revenues.

Table 6.Effective Rate of Trade Taxation, Early 2000s and Early 1990s1(Percent of GDP unless noted otherwise)
Early 2000sEarly 1990s
Low-income countries5.68.1
Lower-middle-income countries4.55.9
Upper-middle-income countries2.65.4
High-income countries1.33.1
Unweighted average
Developing countries24.26.5
High-income countries1.33.1
Region
Americas2.84.9
Sub-Saharan Africa8.19.0
Central Europe and BRO31.44.7
North Africa and Middle East3.35.9
Asia and Pacific2.76.3
Small islands6.910.1
Unweighted average
Developing countries24.26.5
High-income countries1.33.1
Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook; and World Bank, World Development Indicators database.

Trade tax revenue divided by the value of exports plus imports.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Baltics, Russia, and other countries of the former Soviet Union.

Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook; and World Bank, World Development Indicators database.

Trade tax revenue divided by the value of exports plus imports.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Baltics, Russia, and other countries of the former Soviet Union.

Despite the setback to the Doha Round, it seems likely that trade liberalization will continue in the years ahead. This raises important challenges for tax policy. Trade liberalization does not, of course, always reduce revenue from trade taxes. The reduction of tariffs set, for protective purposes, above revenue-maximizing levels, the tariffication of quotas, elimination of exemptions, reduction of tariff peaks, and improvement of customs procedures can all liberalize trade—in the sense of reducing policy-induced distortions to trade flows—while actually increasing revenues. The point is argued forcefully, and shown to be more than a theoretical nicety, by Ebrill, Stotsky, and Gropp (1999). But fully free trade means no trade-related taxes. Thus, there must come a point at which further liberalization does reduce revenue from trade taxes. Whether that point has been met can only be answered on a country-specific basis.25 Nevertheless, the evidence cited above suggests that in many developing countries it has.

One kind of trade reform that is becoming increasingly important in developing countries, though not necessarily one to be welcomed as liberalizing trade,26 is the entry into bilateral and regional integration agreements. Many such agreements have already been entered into, but with effects phased in over the coming years (and the most revenue-costly measures back-loaded). This is the case, for instance, with the series of association agreements reached between the European Union and countries of North Africa and the Middle East. The formation of regional customs unions among developing countries (such as the East African Community (EAC) of Kenya, Uganda, and Tanzania) can also have significant revenue effects even when trade between the participating countries is limited; the loss of revenue from the removal of internal tariffs may be modest, but that from the agreed common external tariff (affecting revenue derived from third-country imports) may be more substantial.27

The loss of revenue from trade taxes raises an obvious and critical question for tax policy: how is it to be recovered? While some countries may be prepared to absorb this loss as a permanent reduction in the size of government, the need for revenue in developing countries, especially the poorest of them, makes the issue a real one. Indeed, without a coherent strategy to compensate for this revenue loss, the process of trade liberalization in developing countries—widely, if not quite universally, seen as key to enhancing their growth prospects, and, in any event, likely to be the quid pro quo for the reduction of protection in the developed world that they need even more urgently—may be jeopardized.

In principle, recovering revenue from alternative sources should not be difficult. One conceptually simple strategy is to match tariff reductions by exactly offsetting increases in taxes on domestic consumption. This preserves the gain in production efficiency from the closer alignment of prices faced by producers to world prices as a consequence of the tariff reduction; leaves prices faced by consumers unchanged so that the combined reform has no effect on their welfare; and increases the government’s total revenue, since this is now collected not only on imports but also on the wider base of domestic consumption. This increase in public revenues could, in turn, be used to compensate those producer groups that lose as a consequence of the tariff reduction, and/or to reduce consumption taxes, and so ensure that consumers also end up strictly better off as a consequence of the reform. For a small, competitive economy importing only final goods, this simple strategy thus has extremely attractive welfare properties.28

Matters are less straightforward when, for instance, the importation of intermediate goods or the exercise of domestic market power—both important features in developing countries—are recognized. Nevertheless, the basic consequence of effective trade liberalization is an increase in the value of national output at world prices and, hence, an increase in the potential tax base. In that sense, successful trade reform itself creates additional resources from which it should be possible to more than recover any reduction in revenue from trade taxes. Moreover, it seems clear that there is a potentially important role here for indirect taxation, both general sales and excise taxes, in doing this. These instruments have both the theoretical attractions just described and the practical advantage that they are largely collected at the border—it is not uncommon for more than half of all VAT revenue to be collected at the border29—and so can be enforced by exactly the same machinery that is used to collect trade tax revenues. So far as borders provide an attractive tax handle, this can be applied to indirect taxes as well as to customs revenues.

The theory is straightforward. What about practice? Have countries in fact managed to recover reductions in trade tax revenues from other sources? The figures in Table 3 show that, on average, all income groups among the developing countries indeed managed to raise indirect tax revenue by about as much as trade tax revenue fell—but, as seen in the earlier section on levels and composition of revenue, many ended up with reduced total revenue because of a decline in corporate tax receipts.

It is important, however, to look behind these group averages, which may reflect the exceptional performance in a few countries or the impact on the level and composition of revenues of other developments (such as a growth in real incomes) over the decade. Analysis of panel data paints a gloomy picture. Using panel data for 125 countries for the period 1975-2000 (corrected to some degree for the tendency, mentioned earlier, of some developing countries to record as trade taxes other taxes), Baunsgaard and Keen (2003) find that while developed countries rarely had difficulty recovering lost trade tax revenues, developing countries did.30 Dividing the latter into two groups, they find that for each US$1 of trade tax revenue forgone, the better-off developing countries tended to recover, eventually, about 40 cents. For the less well-off, however, the recovery tended to be effectively zero.

There is significant cause for concern in these results. It seems that developing countries, especially the poorest and those in Africa, have not found it easy to deal with the revenue consequences of past trade liberalization. This is not to say that they have been harmed by such liberalization, since there are potential efficiency and growth gains to offset any revenue loss. It does suggest, however, that from the fiscal perspective more attention needs to be paid to the sequencing of trade reform and the strengthening of the domestic tax system. Here, for the reasons described above, improving the design and administration of indirect taxes is likely to be particularly important. Jordan, for instance, managed a significant switch away from trade taxes (about 4 percentage points of GDP) and toward indirect tax revenues during the 1990s, while maintaining overall tax revenue broadly constant; in Egypt, on the other hand, the overall tax ratio has for the past 20 years or so tracked a decline in trade taxes—except for one episode of reform in the early 1990s (when it moved its sales tax significantly in the direction of a VAT). But the issue is not simply one of introducing a VAT: statistically, Baunsgaard and Keen (2003) find that the presence of a VAT makes only a slight difference (if any) to the ease with which countries adjust to the loss of trade tax revenue. This may well reflect the heterogeneity of VATs stressed earlier, and there is, no doubt, more to be learned by studying the experiences of individual countries. What does seem clear, however, is that many developing countries have reached a point at which further trade liberalization would likely have significant revenue consequences, and that the domestic tax system needs to do a better job of coping with this than, in too many cases, it has done in the past.

Decline of the Corporate Income Tax

One of the most striking conclusions to emerge from the overview in the second section was the sharp contrast in the performance of corporate tax revenues in developed and developing countries; in the former, these rose noticeably, both relative to GDP and as a share of total tax revenues, while in the latter, they fell in both senses (Tables 3 and A1). In regional terms, the decline was most marked in the countries of central and eastern Europe, but it was also evident in sub-Saharan Africa and Asia-Pacific (whereas the share of corporate tax revenues rose in the Americas and among small islands) (Tables A1 and A2).

The development of corporate tax revenues in the developed world, especially in the OECD, has received much attention in recent years, reflecting the concern, which began to emerge in the latter 1980s, that increased ease of capital movements would intensify international tax competition to such an extent that revenue from capital income taxation—of which corporate tax is a prominent (and relatively easily measured31) component—would be reduced. A comprehensive review of the OECD experience is provided, for example, by Devereux, Griffith, and Klemm (2002). Figure 1 shows that statutory rates of corporation tax in developed economies have fallen substantially over this period, a trend that shows no sign of stopping. There is, moreover, increasing evidence (beyond the anecdotal, which abounds) that these developments have reflected not merely fashion but strategic interaction in tax-setting: that is, international tax competition (see, for instance, Besley, Griffith, and Klemm, 2001). As statutory rates have fallen, however, corporate tax revenues in the OECD have, broadly speaking, held up—or, at least, not declined precipitately. Thus, the reduction in statutory rates has generally been offset by an expansion of the tax base, with the latter, in turn, seen as reflecting a range of base-broadening measures (limiting depreciation and other allowances against corporation tax). From the 1984 U.K. reform through to the German tax reform of 2001, the archetypal corporate tax reform in developed countries has been one of rate reduction accompanied by base broadening. This has posed the theoretical challenge of explaining why international tax competition might take the form of a simultaneous reduction in statutory reduction and expansion of the base, one argument being that this is a way to attract highly mobile paper transactions—using transfer pricing and financial structuring—while continuing to extract surplus from relatively immobile domestic investments (Devereux, Griffith, and Klemm, 2001).

Figure 1.Average Statutory Corporate Rates, 1990-2002

(Percent)

Source: World Tax Database (Ann Arbor, Michigan: University of Michigan).

Figure 1 also shows that, as in the developed world, developing countries saw quite marked reductions in statutory rates of corporation tax in the 1990s. Tables 7 and 8 provide more information on developments in corporate taxation over this period (the former by income group, the latter by region). The reduction in statutory rates of corporation tax is typically greatest in the better-off developing countries (about 8 points, or one-fourth of the initial level), and quite moderate in the poorest developing countries (about 4 percentage points, or one-tenth of the initial level). Figure 2 shows that average statutory rates have fallen in all regions of the developing world, with especially noticeable reductions in sub-Saharan Africa and in North Africa and the Middle East. The downward pressure on statutory tax rates, much remarked on in the developed world, has thus been just as marked—perhaps more so—in the developing world. The dispersion of statutory tax rates32 has also fallen within all regions. Interestingly, while the dispersion of rates within regions has fallen within regions, it has not fallen across them (as can be seen from Figure 1). This dispersion, combined with both still-high statutory taxation and the prospect of continued reductions in developing countries, suggests that the process of rate reduction in the developing world is likely to continue.

Table 7.Corporate Tax Revenues, Rates, and Bases by Income Group, Early 2000s and Early 1990s1(Percent of GDP unless noted otherwise)
Average Statutory
CorporateCorporateAverage
TaxRate2Corporate
Revenues(Percent)Tax Base
Early 2000s
Low-income countries2.034.65.8
Lower-middle-income countries2.231.77.4
Upper-middle-income countries2.629.88.7
High-income countries2.532.68.3
Unweighted average3
Developing countries42.331.87.5
High-income countries2.532.68.3
Total unweighted average2.332.17.7
Early 1990s
Low-income countries2.638.57.7
Lower-middle-income countries2.937.08.6
Upper-middle-income countries3.337.78.6
High-income countries1.935.85.5
Unweighted average3
Developing countries42.937.88.3
High-income countries1.935.85.5
Total unweighted average2.637.17.4
Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook; and University of Michigan, World Tax Database.

Data used for early 1990s and early 2000s are average for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Since average corporate rates and corporate tax revenues are from different sources, this resulted in a reduction of the effective sample size.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook; and University of Michigan, World Tax Database.

Data used for early 1990s and early 2000s are average for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Since average corporate rates and corporate tax revenues are from different sources, this resulted in a reduction of the effective sample size.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Table 8.Corporate Tax Revenues, Rates, and Bases by Region, Early 2000s and Early 1990s1(Percent of GDP unless noted otherwise)
Corporate

Tax

Revenues
Average Statutory

Corporate

Rate2

(Percent)
Average

Corporate

Tax Base
Early 2000s
Americas31.929.26.4
Sub-Saharan Africa1.636.55.1
Central Europe and BRO41.828.26.8
North Africa and Middle East3.235.610.1
Asia and Pacific2.931.89.9
Small islands53.735.09.6
Unweighted average6
Developing countries72.331.87.5
High-income countries2.532.68.3
Memorandum
PRGF-eligible countries81.834.95.6
Early 1990s
Americas31.232.74.6
Sub-Saharan Africa2.043.65.3
Central Europe and BRO44.833.014.3
North Africa and Middle East3.147.17.0
Asia and Pacific3.435.510.2
Small islands53.435.011.1
Unweighted average6
Developing countries72.90.48.3
High-income countries1.937.85.5
Memorandum
PRGF-eligible countries82.20.47.1
Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook; and University of Michigan, World Tax Database.

Data used for early 1990s and early 2000s are average for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Since average corporate rates and corporate tax revenues are from different sources, this resulted in a reduction of the effective sample size.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

See Table A5 for details.

Averages were calculated for PRGF-eligible countries for which data were available (see Appendix Table A6 for details).

Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook; and University of Michigan, World Tax Database.

Data used for early 1990s and early 2000s are average for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Since average corporate rates and corporate tax revenues are from different sources, this resulted in a reduction of the effective sample size.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

See Table A5 for details.

Averages were calculated for PRGF-eligible countries for which data were available (see Appendix Table A6 for details).

Figure 2.Statutory Corporate Tax Rates of Developing Economies by Region, 1990-2002

(Percent)

Source: World Tax Database (Ann Arbor, Michigan: University of Michigan).

1 Including Russia, the Baltic countries, and other countries of the former Soviet Union.

This reduction in statutory rates in developing countries—which has been little noted in the literature—partly accounts for the reduced revenue yield of corporate tax in developing countries. This could, in principle, have been offset to some degree by expansion of the base, whether due to supply-side effects or to policy measures to limit allowances. But, in fact—and in complete contrast to developed countries—Table 7 shows that the exact opposite has happened: the corporate tax base, which increased in the developed world (by enough, as we have already seen, to more than offset rate reductions), has fallen, and noticeably so, in developing countries. The regional breakdown in Table 8 shows that much of the action here comes from central and eastern Europe, where, under circumstances that were clearly exceptional (at least in the first half of the decade or so), the corporate tax base fell by more than 50 percent. In the Americas and in North Africa and the Middle East, the corporate tax base has actually risen. In all other regions—including sub-Saharan Africa, Asia, and the Pacific—the impression is of stagnation at best. Thus, while corporate tax reform among developed countries has been (to use the unavoidable cliché) rate reducing and base broadening, in the developing world it has been rate reducing but also base reducing (or, at best, base neutral).

The poor performance of the corporate tax base in developing countries could reflect wider structural changes in the income share of the corporate sector, or simply the weakness of hoped-for supply-side effects. But there is also evidence that it reflects quite widespread base-narrowing policy reform. Figure 3 shows the spread of a variety of tax incentives, for a sample of 40 developing countries for which sufficient information can be gleaned (from tax guides) for both the start and end of the decade: tax holidays (widely regarded as the most pernicious form of incentive33), reduced statutory rates for particular sectors or regions, direct tax breaks for exporters (WTO-inconsistent for all but the poorest countries) and free-trade zones.34 All become more common. The proportion of the sample offering tax holidays rose from 45 percent to about 60 percent, for instance, while the growth of the other incentives was even more dramatic: reduced rates were available in about 40 percent of the sample at the start of the period, for example, and in 60 percent by the end; tax breaks for exporters and free-trade zones increased from 33 percent and 18 percent of the sample, respectively, to about 45 percent. In short, and with the exception of free-trade zones and tax breaks for exporters, each of these incentives was initially offered only by a minority of countries in the sample, but by a majority at the end.

Figure 3.The Spread of Tax Incentive in Developing Countries, 1990 and 2001

(Percent of developing countries in sample)

Source: Corporate Taxes, 2003-04, Worldwide Summaries (PricewaterhouseCoopers).

Moreover, Table 9 shows that the spread of tax incentives has been especially marked in the poorest of the developing countries. Among the very lowest-income countries, in particular, the proportion offering tax holidays increased by 75 percent, to about 78 percent of countries. In the lower-middle-income countries, on the other hand, the incidence of holidays remained broadly the same, while that of tax breaks for exporters and free-trade zones increased particularly noticeably. There are also important differences in experience across regions, as can be seen by the transition matrices—by region and incentive type—shown in Appendix Table A7. In sub-Saharan Africa, although a small sample size (seven countries) limits the firmness of any conclusion, it is striking that whereas only one country offered tax holidays at the start of the decade, all did by its end.35 And, in the Americas, although two countries (from a more respectable sample size of 15) introduced holidays during the 1990s, rather more (four) removed them. In North Africa and the Middle East, in contrast, there was only a very slight increase in incentives.

Table 9.Tax Incentives in Developing Countries1(Percent of total number of countries in category)
TaxReducedInvestmentTax Breaks
HolidaysCorporate RateAllowancefor ExportersFree-trade Areas
EarlyEarlyEarlyEarlyEarlyEarlyEarlyEarlyEarlyEarly
Countries1990s2000s1990s2000s1990s2000s1990s2000s1990s2000s
Low-income44.477.844.466.711.111.144.477.833.355.6
Sub-Saharan AfricaKenya
Zambia
Zimbabwe
Central Europe and BRO2Estonia
Asia and PacificChina
India
Indonesia
Pakistan
Philippines
Lower-middle-income58.358.333.333.30.00.033.350.016.741.7
AmericasGuatemala
Dominican Republic
Ecuador
Colombia
Paraguay
Peru
Sub-Saharan AfricaCameroon
Congo, Rep. of
North Africa and Middle EastMorocco
Egypt
Asia and PacificPapua New Guinea
Thailand
Upper-middle-income36.842.142.168.426.321.126.321.110.542.1
AmericasJamaica
Costa Rica
Panama
Venezuela, Rep. Bol.
Chile
Brazil
Uruguay
Mexico
Argentina
Sub-Saharan AfricaMauritius
South Africa
Central Europe and BRO2Hungary
Russia
Lithuania
Latvia
North Africa and Middle EastIran, I.R. of
Oman
Asia and PacificMalaysia
Small islands3Malta
All developing countries45.057.540.060.015.015.032.545.017.545.0
Total number of countries in sample40.040.040.040.040.040.040.040.040.040.0
Total number of low-income countries9.09.09.09.09.09.09.09.09.09.0
Total number of lower-middle-income countries12.012.012.012.012.012.012.012.012.012.0
Total number of upper-middle-income countries19.019.019.019.019.019.019.019.019.019.0
Source: Pricewaterhouse Coopers, Corporate Taxes, 2003-2004, Worldwide Summaries; and IMF staff compilation.

For transition countries data are for 1993 instead of 1990.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

Source: Pricewaterhouse Coopers, Corporate Taxes, 2003-2004, Worldwide Summaries; and IMF staff compilation.

For transition countries data are for 1993 instead of 1990.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

It is by no means clear why the thrust of corporate tax policy reform has been to broaden the base in developed countries but to narrow it in the developing world, especially in the poorest regions. Certainly the standard advice to developing countries has been against the kind of base erosion seen above. In terms of the explanation for the trend in developing countries above, perhaps they have a smaller immobile base, offering less of a counterweight to the incentive to narrow the base to attract mobile inward investment. However, this is somewhat difficult to reconcile with the observation (from Table 7) that, at the start of the 1990s, the corporate tax base (relative to GDP) was actually higher in developing countries than in developed. Perhaps the explanation of a greater proclivity toward corporate tax incentives in developing countries lies in institutional structures more vulnerable to the exercise of influence by interest groups, including, not least, foreign multinationals. Why international tax competition appears to be taking such different forms in developed and developing countries—with a stronger adverse effect on revenue in the latter—thus remains something of a puzzle, clearly in need of explanation.

Should the apparent erosion of the corporate tax in developing countries be regretted or welcomed? It is a fairly robust result of optimal tax theory that economies that are small in world capital markets should set a marginal effective corporate tax rate—that is, a rate on the investment that just breaks even after tax—of zero. This is essentially a production efficiency argument, an intertemporal analogue to that for the absence of trade taxes in developing countries:36 small economies should trade at world prices for capital, just as they should for atemporal commodities. This suggests that the developing world may simply be moving toward what is now (even if it was not when capital was less mobile) an efficient tax structure.

That, however, seems too optimistic a view. To the extent that the corporate tax bears on rents, it does not distort investment decisions, and so is not optimally zero. It may be that part of the reduction in base and revenues simply represents a loss of rents, either (through bad policy) location-specific rents that could in principle be taxed at up to 100 percent without driving investment abroad or rents (through competition for footloose firms) that can be taken in any of a variety of countries. In either case, the result is the loss of an attractive source of revenue. The corporate tax is also traditionally seen as playing an important role as a backup to the income tax, preventing avoidance by accumulating earnings subject only to capital gains tax (often absent or ineffective in developing countries). There is little evidence in Tables 3 and A1-A3, however, that the personal income tax has been subject to significant erosion in the developing world, especially bearing in mind that there has in general been a marked reduction in personal income tax rates over this period—at least in the highest marginal rate37—in tandem with the reduction in statutory rates of corporation tax. Perhaps most troubling is the simple fact that developing countries have traditionally relied more heavily on corporate tax revenues than have developed: at the end of the 1990s, they still accounted for 12 percent of tax revenue in PRGF-eligible countries. This, in turn, seems likely to reflect the relative administrative ease of collecting corporate tax revenue, which is typically highly concentrated in a relatively small number of large (and perhaps also relatively honest) firms. Like the decline of trade tax revenues, the erosion of the corporate tax may thus jeopardize a convenient tax handle.

What then is the proper policy response to these pressures, apparently from intensified international tax competition, on corporate tax revenue in developing countries? Some take the view that downward pressure on tax revenues from tax competition should be welcomed as a further means of disciplining governments prone to spend wastefully, and this has indeed been an influential view in the formation of policy in the OECD. In the context of low-income and developing countries, however, the most common prescription, as noted earlier, is for an increase in revenues (even while recognizing that some would doubtless be wasted). This implies that there would then be merit in measures of cooperation intended to bolster corporate tax revenues.

Such cooperation might take a number of forms. One (fairly minimalist) possibility, for instance, is agreement on a code of conduct, proscribing and seeking to roll back particular forms of tax incentive: this might include a prohibition, for example, on the issuing of new tax holidays. A code of this sort—enforced not legally, but by peer pressure—has been adopted, for example, by the European Union38 and proposed for the West African Economic and Monetary Union and for the South African Development Community. This strategy may be particularly appealing for emerging customs unions, since the elimination of internal tariff barriers can be expected to enhance firms’ mobility between participating countries, and so increase the incentive of each country to offer more favorable tax treatment than do the others. The difficulty with this and any other form of coordination, however, is that by increasing their levels of corporate taxation those countries party to the agreement may find themselves more vulnerable to tax competition from those outside it.39 Nor is it even clear that a wide regional coverage will deal with this problem, since potential competitor countries may be geographically far removed from one another: Mauritius may compete for investment with Fiji, for example. Only a genuinely global arrangement, which remains far from political reality, avoids this problem. How vulnerable more limited regional agreements would be to competition from third countries, and (a related issue) how much political will can be mustered to enter and abide by such arrangements, are likely to prove important issues for the future of corporate taxation not only in developed but also, perhaps especially, in developing countries. Experience with agreements of this kind, for example with the CARICOM agreement on fiscal incentives, has not been wholly encouraging.

Concluding Remarks

Experiences with tax policy varied widely across developing countries in the 1990s. Nepal, for instance, managed to increase its tax ratio by over 3 percentage points while modestly reducing its reliance on trade taxes and offsetting this with increased revenue from sales taxation. In Burundi, on the other hand, the tax ratio fell by about 4 percentage points, even though trade tax revenue remained broadly constant.

While generalizations are, thus, dangerous, the overall picture that emerges is certainly not comforting. Tax ratios in the developing world have been stagnant at best (and tending to fall once seignorage is taken into account), when the most common objective was actually to mobilize more domestic revenue. That in itself does not imply unmitigated failure: a number of countries have succeeded in keeping revenue broadly unchanged while reducing their reliance on distortionary trade taxes, which would generally be regarded as a worthwhile improvement in the efficiency of the overall tax system. There are also some clear achievements. In particular, there is evidence that the adoption of the VAT has improved the effectiveness of the tax systems of many developing countries. Nevertheless, there is cause for concern in the lackluster performance of overall revenues in low-income countries.

What also stands out is the prospect of significant challenges ahead. With the VAT now so widespread, the emphasis is likely to shift from the introduction of this tax (except, perhaps, in parts of the Middle East, where it has only recently begun to make its appearance) toward the improvement of its design and administration, tasks that may require more time than originally supposed. Other challenges are even more fundamental. Experience suggests, in particular, that better sequencing of domestic and trade-tax reform is necessary if developing countries, especially the poorest of them, are to preserve their revenues in the face of further trade reforms. And the erosion of corporate tax revenues in developing countries over the past decade suggests that international tax competition may have a far more significant impact on revenues for them than it does (or, at least, has yet had) for developed countries. This raises issues of international tax coordination that have proved both conceptually and politically thorny. There are other issues too, beyond those addressed in this chapter, likely to require attention in the coming years—not least, improving the tax treatment of small and medium-sized enterprises.

The past decade has been one of some successes. But trade reform and tax competition may provide another twist of the fiscal knife for many low-income and other developing countries, posing continuing and, in some respects, deepening problems for tax design in the years ahead.

Appendix

The appendix provides details regarding the sample of countries (region classification, income classification, PRGF sample of countries), the results discussed in the text, and the tables and figures of the chapter.

Table A1.Central Government Revenues by Income Group, Early 2000s and Early 1990s1(Percent of total tax revenues)
Domestic Taxes on

Goods and Services
Taxes on Income, Profits,

and Capital Gains
of which:International

Trade Taxes2
Nominal

GDP per Capita

(U.S. dollars,

average)
of which:Social

Security

Taxes
General

sales,

turnover,

or VAT
of which:
Total

Revenue
Tax

Revenue
Other

Revenue
TotalIndividualCorporatePayroll

Taxes
TotalExcisesTotalImport

duties
Export

duties
Property

Taxes
Early 2000s
Low-income countries614121.2100.021.226.012.613.27.00.239.423.513.224.916.41.71.4
Lower-middle-income countries1,642137.9100.037.924.211.513.58.30.440.130.913.222.721.80.91.6
Upper-middle-income countries4,142122.1100.022.125.012.212.421.30.539.325.411.311.911.50.81.4
High-income countries21,387119.3100.019.332.626.99.326.71.233.024.410.34.94.60.02.5
Unweighted average3
Developing countries42,269125.8100.025.825.112.112.914.00.439.425.812.318.315.61.11.4
High-income countries21,387119.3100.019.332.626.99.326.71.233.024.410.34.94.60.02.5
Total unweighted average7,756123.3100.023.328.017.911.519.00.736.925.011.514.212.20.81.9
Early 1990s
Low-income countries445121.4100.021.426.08.817.65.00.136.119.214.629.926.04.31.7
Lower-middle-income countries1,194131.1100.031.126.29.117.58.81.230.917.512.226.725.11.22.3
Upper-middle-income countries3,747122.7100.022.726.910.115.219.70.829.716.711.019.717.62.61.4
High-income countries18,418119.7100.019.732.028.67.027.11.032.223.69.17.37.40.12.7
Unweighted average3
Developing countries41,936124.4100.024.426.49.516.412.80.731.717.612.324.321.82.61.7
High-income countries18,418119.7100.019.732.028.67.027.11.032.223.69.17.37.40.12.7
Total unweighted average6,667122.7100.022.728.617.212.818.20.931.919.711.119.217.52.02.1
Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook.

Data used for early 1990s and early 2000s are averages for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

European Union countries do not report statistics on international trade taxes to Government Finance Statistics.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook.

Data used for early 1990s and early 2000s are averages for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

European Union countries do not report statistics on international trade taxes to Government Finance Statistics.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Table A2.Central Government Revenues by Region, Early 2000s and Early 1990s1(Percent of GDP)
Domestic Taxes on

Goods and Services
Taxes on Income, Profits,

and Capital Gains
of which:International

Trade Taxes2
Nominal

GDP per Capita

(U.S. dollars,

average)
of which:Social

Security

Taxes
General

sales,

turnover,

or VAT
of which:
Total

Revenue
Tax

Revenue
Other

Revenue
TotalIndividualCorporatePayroll

Taxes
TotalExcisesTotalImport

duties
Export

duties
Property

Taxes
Early 2000s
Americas23,30120.016.33.73.91.31.92.30.17.95.52.21.91.90.00.3
Sub-Saharan Africa76519.715.93.84.72.71.60.30.15.03.21.45.63.50.40.2
Central Europe and BRO32,42026.723.43.23.62.11.88.10.110.56.83.01.10.90.40.1
North Africa and Middle East2,48626.217.19.15.52.53.20.80.25.94.62.53.33.00.10.5
Asia and Pacific1,44716.613.23.44.62.02.90.50.05.32.71.92.11.90.20.2
Small islands44,67332.024.57.64.82.73.72.80.06.44.31.69.79.70.00.3
Unweighted average5
Developing countries62,26922.117.64.54.42.12.32.50.16.94.52.23.22.70.20.3
High-income countries21,38732.827.55.38.97.42.57.30.39.16.72.81.31.30.00.7
Memorandum
PRGF-eligible countries761719.614.84.83.41.71.81.00.05.63.41.94.53.50.30.2
Early 1990s
Americas22,02618.314.93.43.71.11.22.20.15.22.82.12.92.50.20.4
Sub-Saharan Africa83119.316.32.94.72.12.00.30.05.03.31.45.94.91.00.2
Central Europe and BRO33,47030.927.33.66.61.74.87.90.59.95.23.72.31.40.80.2
North Africa and Middle East1,95123.315.18.34.91.83.11.00.24.02.91.83.83.60.10.5
Asia and Pacific1,06117.613.64.04.51.73.40.20.04.92.22.23.43.20.30.3
Small islands43,22133.425.57.94.21.43.32.80.04.00.50.614.013.50.30.3
Unweighted average5
Developing countries61,93622.317.94.44.71.72.92.30.15.73.22.24.33.90.50.3
High-income countries18,41831.926.65.28.57.61.97.20.38.66.32.41.92.00.00.7
Memorandum
PRGF-eligible countries785619.415.24.13.51.32.20.40.05.02.42.25.54.80.60.2
Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook.

Data used for early 1990s and early 2000s are averages for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Averages were calculated for PRGF-eligible countries for which data were available (see Table A6 for details).

Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook.

Data used for early 1990s and early 2000s are averages for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Averages were calculated for PRGF-eligible countries for which data were available (see Table A6 for details).

Table A3.Central Government Revenues by Region, Early 2000s and Early 1990s1(Percent of total tax revenue)
Domestic Taxes on

Goods and Services
Taxes on Income, Profits,

and Capital Gains
of which:International

Trade Taxes2
Nominal

GDP per Capita

(U.S. dollars,

average)
of which:Social

Security

Taxes
General

sales,

turnover,

or VAT
of which:
Total

Revenue
Tax

Revenue
Other RevenueTotalIndividualCorporatePayroll

Taxes
TotalExcisesTotalImport

duties
Export

duties
Property

Taxes
Early 2000s
Americas23,301122.510022.523.98.011.414.30.548.333.613.211.711.50.01.9
Sub-Saharan Africa765123.510023.529.516.79.81.70.331.220.29.034.821.92.31.3
Central Europe and BRO32,420113.810013.815.38.97.734.50.444.729.012.84.93.91.70.3
North Africa and Middle East2,486153.710053.732.414.518.64.91.134.627.114.719.417.50.53.1
Asia and Pacific1,447125.910025.935.115.522.13.90.140.520.814.715.914.51.81.5
Small islands44,673130.910030.919.511.215.211.40.026.017.46.539.739.70.01.4
Unweighted average5
Developing countries62,269125.810025.825.112.112.914.00.439.425.812.318.315.61.11.4
High-income countries21,387119.310019.332.626.99.326.71.233.024.410.34.94.60.02.5
Memorandum
PRGF-eligible countries7617132.3100.032.322.711.412.16.50.237.822.712.730.123.51.81.2
Early 1990s
Americas22,026122.610022.624.97.18.314.50.635.118.913.819.317.01.62.6
Sub-Saharan Africa831118.110018.128.612.812.41.90.330.420.38.736.430.35.91.0
Central Europe and BRO33,470113.110013.124.06.117.629.01.736.319.113.78.55.22.80.7
North Africa and Middle East1,951155.010055.032.212.020.66.61.126.719.112.125.524.00.73.3
Asia and Pacific1,061129.010029.033.212.424.81.50.036.116.116.324.823.42.12.4
Small islands43,221130.910030.916.45.412.811.00.015.81.92.454.853.11.31.0
Unweighted average5
Developing countries61,936124.410024.426.49.316.312.80.731.717.612.324.321.82.61.7
High-income countries18,418119.710019.732.028.67.027.11.032.223.69.17.37.40.12.7
Memorandum
PRGF-eligible countries75,624127.2100.027.222.98.514.82.90.132.815.714.236.131.64.11.4
Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook.

Data used for early 1990s and early 2000s are averages for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Averages were calculated for PRGF-eligible countries for which data were available (see Table A6 for details).

Sources: IMF, Government Finance Statistics, International Financial Statistics, and World Economic Outlook.

Data used for early 1990s and early 2000s are averages for two years—1990-91 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

For each revenue classification, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Averages were calculated for PRGF-eligible countries for which data were available (see Table A6 for details).

Table A4.Seignorage as a Share of GDP, by Region1
Nominal

GDP per Capita

(U.S. dollars, average)
Seignorage

(Percent of GDP)
Early 2000s
Americas23,3011.2
Sub-Saharan Africa7652.5
Central Europe and BRO32,4203.1
North Africa and Middle East2,4863.3
Asia and Pacific1,4472.7
Small islands44,6731.9
Unweighted average5
Developing countries62,2692.4
High-income countries721,3871.7
Memorandum
PRGF-eligible countries86172.4
Early 1990s
Americas22,0263.5
Sub-Saharan Africa8312.5
Central Europe and BRO33,4706.2
North Africa and Middle East1,9514.5
Asia and Pacific1,0614.0
Small islands43,2212.5
Unweighted average5
Developing countries61,9364.0
High-income countries718,4181.4
Memorandum
PRGF-eligible countries88564.0
Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook, and staff calculations.

Data used for early 1990s and early 2000s are average for two years—1991-92 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction. Seignorage is calculated as the money growth rate multiplied by money stock (M1) divided by nominal GDP. The hyperinflation episodes of Brazil and Nicaragua in the early 1990s are ignored, and replaced by estimates from normal times.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

In each region, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Comparable measures of M1 were not available for most European countries.

Average were calculated for PRGF-eligible countries for which data were available (see Table A6 for details).

Sources: IMF, Government Finance Statistics, International Financial Statistics, World Economic Outlook, and staff calculations.

Data used for early 1990s and early 2000s are average for two years—1991-92 and 2000-01 for most countries. For countries for which these averages could not be calculated, some flexibility in the years taken to represent the early 1990s and early 2000s was used to avoid a significant sample size reduction. Seignorage is calculated as the money growth rate multiplied by money stock (M1) divided by nominal GDP. The hyperinflation episodes of Brazil and Nicaragua in the early 1990s are ignored, and replaced by estimates from normal times.

Regional breakdown averages include only developing countries.

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

In each region, only countries for which data are available are included in the calculation.

Developing countries are defined to be low- and middle-income countries, closely following the World Bank income classification of economies.

Comparable measures of M1 were not available for most European countries.

Average were calculated for PRGF-eligible countries for which data were available (see Table A6 for details).

Appendix Table A5.List of Countries by Region and Income
AmericasSub-Saharan

Africa
Central

Europe

and BRO1
North

Africa and

Middle East
Asia

and Pacific
Small

Islands2
European

Union3
Low-income countriesBoliviaBurundiAlbaniaBhutan
NicaraguaCongo,EstoniaChina
Dem. Rep. ofIndia
Côte d’IvoireIndonesia
EthiopiaMongolia
GuineaMyanmar
KenyaNepal
LesothoPakistan
MadagascarPhilippines
Sierra LeoneSri Lanka
ZambiaVietnam
Zimbabwe
Lower-middle-incomeColombiaCameroonBulgariaEgyptPapua NewMaldives
countriesDominican Rep.Congo, Rep. ofRomaniaJordanGuineaVanuatu
El SalvadorLebanonThailand
GuatemalaMorocco
ParaguaySyria
PeruYemen
Tunisia
Upper-middle-incomeArgentinaMauritiusAzerbaijanIranMalaysiaMalta
countriesBelizeSouth AfricaBelarusOmanKoreaSt. Vincent
BrazilCzech RepublicTurkey& Grenadines
ChileHungarySeychelles
Costa RicaKyrgyz Republic
JamaicaLatvia
MexicoLithuania
PanamaMoldova
UruguayPoland
VenezuelaRussian Federation
High-income countriesCanadaCroatiaBahrainAustraliaBahamas, TheAustria
United StatesSloveniaIsraelNew ZealandCyprusBelgium
KuwaitSingaporeIcelandDenmark
United ArabFinland
EmiratesFrance
Germany
Greece
Ireland
Italy
Luxembourg
Netherlands
Norway
Portugal
Spain
Sweden
Switzerland
United Kingdom
Source: IMF staff compilation. The regional classification follows that in Ebrill and others (2001).

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

Plus Norway and Switzerland.

Source: IMF staff compilation. The regional classification follows that in Ebrill and others (2001).

Baltics, Russia, and other countries of the former Soviet Union.

Island economies with population of under 1 million.

Plus Norway and Switzerland.

Table A6.Sample of PRGF-Eligible IMF Members
AlbaniaGuineaNepal
AzerbaijanIndiaNicaragua
BhutanKenyaPakistan
BoliviaKyrgyz RepublicPapua New Guinea
BurundiLesothoSierra Leone
CameroonMadagascarSt Vincent and Grenadines
Congo, Dem. Rep. ofMaldivesVanuatu
Congo, Rep. ofMoldovaVietnam
Côte d’IvoireMongoliaYemen
EthiopiaMyanmarZambia
Source: IMF staff compilation.
Source: IMF staff compilation.
Table A7.Tax Incentive Transition Matrices, 1990 and 20011
RegionCountriesTax HolidaysReduced

Corporate Rates
Investment

Allowance
Tax Breaks

for Exporters
Free-Trade

Areas
Asia and PacificChinaEarly 2000sEarly 2000sEarly 2000sEarly 2000sEarly 2000s
IndiaYesNoTotalYesNoTotalYesNoTotalYesNoTotalYesNoTotal
IndonesiaEarlyYes505EarlyYes303EarlyYes202EarlyYes505EarlyYes303
Malaysia1990sNo1121990sNo1341990sNo0551990sNo2021990sNo224
Papua New GuineaTotal61Total43Total25Total70Total52
Philippines
Thailand
Small islands2MaltaEarly 2000sEarly 2000sEarly 2000sEarly 2000sEarly 2000s
YesNoTotalYesNoTotalYesNoTotalYesNoTotalYesNoTotal
EarlyYes101EarlyYes101EarlyYes101EarlyYes101EarlyYes000
1990sNo0001990sNo0001990sNo0001990sNo0001990sNo011
Total10Total10Total10Total10Total01
North Africa andEgyptEarly 2000sEarly 2000sEarly 2000sEarly 2000sEarly 2000s
Middle EastIranYesNoTotalYesNoTotalYesNoTotalYesNoTotalYesNoTotal
MoroccoEarlyYes404EarlyYes202EarlyYes000EarlyYes303EarlyYes011
Oman1990sNo0111990sNo1231990sNo0551990sNo1121990sNo134
PakistanTotal41Total32Total05Total41Total14
AmericasArgentinaEarly 2000sEarly 2000sEarly 2000sEarly 2000sEarly 2000s
MexicoYesNoTotalYesNoTotalYesNoTotalYesNoTotalYesNoTotal
BrazilEarlyYes044EarlyYes202EarlyYes202EarlyYes213EarlyYes303
Chile1990sNo29111990sNo58131990sNo013131990sNo111121990sNo4812
ColombiaTotal213Total78Total213Total312Total78
Costa Rica
Dominican Rep.
Ecuador
Guatemala
Jamaica
Panama
Paraguay
Peru
Uruguay
Venezuela
Central EuropeHungaryEarly 2000sEarly 2000sEarly 2000sEarly 2000sEarly 2000s
and BRO3EstoniaYesNoTotalYesNoTotalYesNoTotalYesNoTotalYesNoTotal
LatviaEarlyYes123EarlyYes314EarlyYes011EarlyYes000EarlyYes000
Lithuania1990sNo2021990sNo1011990sNo0441990sNo0551990sNo325
RussiaTotal32Total41Total05Total05Total32
Sub-Saharan AfricaCameroonEarly 2000sEarly 2000sEarly 2000sEarly 2000sEarly 2000s
CongoYesNoTotalYesNoTotalYesNoTotalYesNoTotalYesNoTotal
KenyaEarlyYes101EarlyYes314EarlyYes000EarlyYes101EarlyYes000
Mauritius1990sNo6061990sNo2131990sNo1671990sNo2461990sNo257
South AfricaTotal70Total52Total16Total34Total25
Zambia
Zimbabwe
OverallEarly 2000sEarly 2000sEarly 2000sEarly 2000sEarly 2000s
YesNoTotalYesNoTotalYesNoTotalYesNoTotalYesNoTotal
EarlyYes12618EarlyYes14216EarlyYes516EarlyYes12113EarlyYes617
1990sNo1111221990sNo1014241990sNo133341990sNo621271990sNo122133
Total2317Total2416Total634Total1822Total1822
Sources: Pricewaterhouse Coopers, Corporate Taxes, 2003-04, Worldwide Summaries; and IMF staff compilation.

For most countries early 1990s means 1990 and early 2000s means 2001. For transition countries data for 1993 are used instead of 1990 and 2001 instead of 2000.

Island economies with population of under 1 million.

Baltics, Russia, and other countries of the former Soviet Union.

Sources: Pricewaterhouse Coopers, Corporate Taxes, 2003-04, Worldwide Summaries; and IMF staff compilation.

For most countries early 1990s means 1990 and early 2000s means 2001. For transition countries data for 1993 are used instead of 1990 and 2001 instead of 2000.

Island economies with population of under 1 million.

Baltics, Russia, and other countries of the former Soviet Union.

Table A8.Statutory Rates of Personal Income Tax in Developing Countries1
Averages
Late 1990s2
Low-income countries333.3
Lower-middle-income countries433.3
Upper-middle-income countries530.4
Early 1990s2
Low-income countries641.3
Lower-middle-income countries742.6
Upper-middle-income countries835.9
Source: University of Michigan, World Tax Database.

Top income tax rates for 2000 and 2001 are not available.

Data used for countries of the former Soviet Union for the early 1990s are not from years 1990 and 1991, but the earliest available observation.

Excluding Bhutan, Burundi, Ethiopia, Guinea, Lesotho, Madagascar, Malaysia, Moldova, Myanmar, Sierra Leone, Nepal, and Sri Lanka.

Excluding Albania, Belarus, Jordan, Maldives, Tunisia, Turkey, St. Vincent and Grenadines, and Syria.

Excluding Belize, Croatia, Czech Republic, Hungary, Lebanon, Mexico, Poland, and Seychelles.

Excluding Albania, Bhutan, Burundi, Ethiopia, Guinea, Lesotho, Madagascar, Malaysia, Myanmar, Nepal, Sierra Leone, Slovenia, and Vietnam.

Excluding Bulgaria, Cameroon, India, Jordan, Lebanon, Maldives, Morocco, Tunisia, Yemen, and Vanuatu.

Excluding Azerbaijan, Belarus, Czech Republic, Hungary, Korea, Kyrgyz Republic, Mexico, Moldova, Poland, St. Vincent and Grenadines, Seychelles, and Turkey.

Source: University of Michigan, World Tax Database.

Top income tax rates for 2000 and 2001 are not available.

Data used for countries of the former Soviet Union for the early 1990s are not from years 1990 and 1991, but the earliest available observation.

Excluding Bhutan, Burundi, Ethiopia, Guinea, Lesotho, Madagascar, Malaysia, Moldova, Myanmar, Sierra Leone, Nepal, and Sri Lanka.

Excluding Albania, Belarus, Jordan, Maldives, Tunisia, Turkey, St. Vincent and Grenadines, and Syria.

Excluding Belize, Croatia, Czech Republic, Hungary, Lebanon, Mexico, Poland, and Seychelles.

Excluding Albania, Bhutan, Burundi, Ethiopia, Guinea, Lesotho, Madagascar, Malaysia, Myanmar, Nepal, Sierra Leone, Slovenia, and Vietnam.

Excluding Bulgaria, Cameroon, India, Jordan, Lebanon, Maldives, Morocco, Tunisia, Yemen, and Vanuatu.

Excluding Azerbaijan, Belarus, Czech Republic, Hungary, Korea, Kyrgyz Republic, Mexico, Moldova, Poland, St. Vincent and Grenadines, Seychelles, and Turkey.

References

    BaunsgaardThomas and MichaelKeen2003“Tax Revenue and (or?) Trade Liberalization” (unpublished; Washington: International Monetary Fund).

    BesleyTimothyRachelGriffith and AlexanderKlemm2001“Empirical Evidence on Fiscal Interdependence in OECD Countries” (unpublished; London: Institute for Fiscal Studies).

    BirdRichardM.1993“Review of ‘Principles and Practice of Value-Added Taxation: Lessons for Developing Countries,’”Canadian Tax JournalVol. 41No. 6 pp. 1222-25.

    BoadwayRobin and MichaelKeen2003“Theoretical Perspectives on the Taxation of Capital Income and Financial Services,”in The Taxation of Financial Intermediation: Theory and Practice for Emerging Economiesed. by PatrickHonohan (World Bank and Oxford University Press) pp. 31-80.

    BurgessRobin and NicholasStern1993“Taxation and Development,”Journal of Economic LiteratureVol. 31 pp. 762-830.

    Casanegrade JantscherMilka1990“Administering the VAT,”in Value-Added Taxation in Developing Countriesed. by MalcolmGillisCarl S.Shoup and Gerardo P.Sicat (Washington: World Bank).

    CnossenSijbren1994“Administrative and Compliance Costs of the VAT: A Review of the Evidence,”Tax NotesVol. 63 (June 20) pp. 1609-26.

    DevereuxMichaelRachelGriffith and AlexanderKlemm2002“Corporate Income Tax Reforms and International Tax Competition,”Economic PolicyVol. 35 (October) pp. 451-95.

    EbrillLiamJanetStotsky and ReintGropp1999Revenue Implications of Trade LiberalizationIMF Occasional Paper No. 180 (Washington: International Monetary Fund).

    EbrillLiamMichaelKeenJean-PaulBodin and VictoriaSummers2001The Modern VAT (Washington: International Monetary Fund).

    KeenMichael and Jenny E.Ligthart2001“Coordinating Tariff Reduction and Domestic Tax Reform,”Journal of International EconomicsVol. 56(March) pp. 489-507.

    KeenMichael and BenLockwood2003“The Causes and Consequences of the Uptake of VAT: An Empirical Investigation” (unpublished; Washington: International Monetary Fund).

    KeenMichael and JackMintz2004“The Optimal Threshold for a Value-Added Tax,”Journal of Public EconomicsVol. 88 (March) pp. 559-76.

    KeenMichael and DavidWildasin2004“Pareto Efficient International Taxation,”American Economic Reviewforthcoming.

    KhattryBarsha and MohanRao2002“Fiscal Faux Pas? An Analysis of the Revenue Implications of Trade Liberalization,”World DevelopmentVol. 30 (August) pp. 1431-44.

    KonradKai and GuttormSchelderup1999“Fortress Building in Global Tax Competition,”Journal of Urban EconomicsVol. 46 (July1999) pp. 156-67.

    MuñozSonia and StanleySang-Wook Cho2004“Social Impact of a Tax Reform: The Case of Ethiopia,”Chapter 13 in Helping Countries Develop: The Role of Fiscal Policyed. by SanjeevGuptaBenedictClements and GabrielaInchauste (Washington: International Monetary Fund).

    PanagariyaArvind and DaniRodrik1993“Political-Economy Arguments for a Uniform Tariff,”International Economic ReviewVol. 34 (August) pp. 685-703.

    RodrikDani1998“Why Do More Open Economies Have Bigger Governments?”Journal of Political EconomyVol. 106 (October) pp. 997-1032.

    SchneiderFreidrich and DominikEste2000“Shadow Economies: Size, Causes, and Consequences,”Journal of Economic LiteratureVol. 37 (March) pp. 77-114.

    StiglitzJoseph2003“Development Oriented Tax Policy,”presentation at International Institute of Public FinanceWashingtonNovember 6.

    SummersVictoria and KatherineBaer2003“Revenue Policy and Administration in the CIS-7: Recent Trends and Future Challenges” (unpublished; Washington: International Monetary Fund).

    TanziVito1987“Quantitative Characteristics of the Tax Systems of Developing Countries,”in The Theory of Taxation for Developing Countriesed. by DavidNewbery and NicholasStern (New York: Oxford University Press for the World Bank) pp. 205-41.

    YoungerStephenD. and David E.Sahn1998“Fiscal Incidence in Africa: Microeconomic Evidence,”CFNPP Working Paper No. 91 (Ithaca, New York: Cornell Food and Nutrition Policy Program).

    ZeeHowellJanetStotsky and EduardoLey2002“Tax Incentives for Business Investment: A Primer for Policy Makers in Developing Countries,”World DevelopmentVol. 30 (September) pp. 1497-1516

Data and definitions are discussed in the next section.

The survey by Schneider and Este (2000) of empirical work on the shadow economy—which they define as legal but untaxed/unregistered activities—tends to confirm that this is substantially larger in developing countries than in developed.

As President Clinton even more famously remarked, “it depends what you mean by ‘is’.” (We owe this happy remark to Joel Slemrod.)

There are other difficulties too. For instance, a number of countries appear to record as trade tax revenue the receipts from other taxes (notably, VAT and excises) collected at their borders.

Such effects are eliminated with respect to overall and total tax revenue (since countries are included in the sample if and only if data on these are available at both start and end of the decade), but not for the various components of tax revenue (since some countries do not report a complete breakdown of their revenues).

Details of the country coverage and regional grouping are provided in Appendix Table A5.

Thus, a country classified as “low-income” at the start of the 1990s is included in the low-income group at the end of the 1990s even if, by the World Bank standards of the time, it would then have been classified as being “lower-middle-income.” This is to avoid biases that would otherwise arise from excluding the experience of “successful” low-income countries in evaluating the performance of the group as a whole.

Details of the country groups are also in Appendix Table A5.

There is a large empirical literature on tax effort, to which we refrain from adding. See, for instance, Tanzi (1987) and, with a particular focus on the role of openness—the question being why it is that openness is associated with high revenues—Rodrik (1998). Some broad hint of these robust empirical findings can be seen from Tables 1 and 3.

The tax ratio in Estonia increased by about 5 percentage points of GDP between 1990-01 and 2000-01, but the accuracy of the figure for the start of the decade is questionable.

The sample period predates the current troubles in Côte d’Ivoire.

For completeness, Appendix Table A4 shows the same data by region.

Six of the eight small island economies in our sample were included in the OECD’s descriptive list of tax havens.

The essence of the VAT is that, in principle (from which practice often departs in important ways), tax is charged on all sales (including imports, but excluding exports), whether to consumers or producers, but producers are effectively refunded (whether explicitly or by way of a credit against the tax due on their own sales) taxes charged on their inputs. The tax thus becomes one on final domestic consumption.

See, for instance, Panagariya and Rodrik (1993).

See Table 7.2 of Ebrill and others (2001).

“Exemption” means that tax is not charged on sales but (in contrast to zero-rating) tax paid on inputs is irrecoverable.

Concerning, for instance, the extent to which indirect tax increases are passed on to consumers through higher prices rather than borne by profits or payments to labor; whether progressivity is better assessed relative to consumption rather than income (on the grounds that the former is a better indicator of lifetime welfare); and what the appropriate counterfactual is in assessing the impact of the VAT. These and other issues in assessing the distributional impact of the VAT are discussed in Chapter 10 of Ebrill and others (2001).

There are real difficulties (for instance, in the treatment of costs shared between implementation of the VAT and other taxes), but of an order commonplace in empirical work.

The purpose of this exercise is not, it should be stressed, to judge the VAT by whether it increased revenue (even if revenue is increased, national welfare could fall because the additional revenue is spent in a corrupt or wasteful manner), but rather to use any increase in revenue to infer an increase in the efficiency of the overall tax system.

This issue is surveyed in Boadway and Keen (2003).

See, for instance, the symposium on this issue in International Tax and Public Finance, December 2000.

Export taxes are still important in a few cases (such as forestry exports from West Africa), though, as Table 3 and Appendix Tables A1-A3 show, these have been far less significant than import tariffs throughout this period.

Ebrill, Stotsky, and Gropp (1999) report regressions of trade tax revenue against (inter alia) the collected tariff that imply that revenue is maximized (for their full sample of countries) at a little over 20 percent. Khattry and Rao (2002) arrive at the higher figure of 40 percent. As these authors recognize, however, the collected tariff is such an amalgam of distinct policy instruments (reflecting exemptions, rate structures, and administrative effectiveness), that these figures—even leaving aside some nonrobustness—should be taken as only very broadly suggestive of the likely impact of further tariff reductions.

In particular, such selective tariff reductions can lead to inefficient trade diversion, with imports being sourced from countries charging higher prices before tariff but lower prices inclusive of the tariff; moreover, the formation of trading blocs may impede rather than facilitate multilateral liberalization. For a taste of the issues, see the symposium on this topic in the July 1998 issue of the Economic Journal.

Indeed, low-tariff countries entering a customs union may find that their trade tax revenue increases, as seems likely for Uganda in the case of the EAC.

The arguments in this paragraph are spelled out and explored further in Keen and Ligthart (2001).

See Table 4.3 of Ebrill and others (2001).

For a smaller sample of countries, using uncorrected data and focusing on simple cross-tabulations, Khattry and Rao (2002) reach broadly similar conclusions.

The main difficulty is that imputation forms of corporate tax (which alleviate the double taxation of dividends by providing credit at a personal level with respect to taxes paid at the corporate level) in effect record as corporate tax receipts what are more accurately shareholder-level taxes on dividends. This is unlikely to be a serious concern for the analysis here, since most countries retained the same system throughout this period (the move away from integration in many developed economies—including the United Kingdom, Germany, Ireland, and Italy—generally being more recent than our data).

As measured by either the standard deviation or (except in North Africa and the Middle East) the coefficient of variation.

For reasons spelled out, for example, in Zee, Stotsky, and Ley (2002).

There is an important distinction between free-trade zones that offer only customs and indirect tax remission to exporters within the zone (this being essentially a particular way of implementing standard procedures), and those that also offer zone firms direct tax advantages of various kinds. Available information does not allow us to distinguish between these, so that the figures are for free-trade zones of all types.

However, Uganda (not in the sample) withdrew tax holidays (grandfathering those already granted) in 1997.

As such, the production efficiency argument is subject to a range of qualifications: it presumes that pure profits are fully taxed, and that there are no constraints on the other tax instruments that can be deployed. Further, qualifications arise in the international context: see Keen and Wildasin (2004). Nevertheless, this remains an important benchmark for policy design.

This is documented in Appendix Table A8.

There it is also supplemented by binding state-aid rules, which are proving even more effective in unwinding tax practices containing incentive elements.

Konrad and Schelderup (1999) establish conditions under which tax coordination among a subset of countries is welfare improving for the participants.

    Other Resources Citing This Publication