Trends in Taxation in Developing Countries

PROBLEMS OF TAXATION in developing countries have been receiving special and increasing attention in recent years, in the context of world concern for accelerating the pace of their economic growth under conditions of stability. Much of the theoretical and applied work has been related to the tasks of constructing productive and equitable tax systems suited to the peculiar economic and social conditions prevailing in various less developed countries. While the emphasis on the public sector has varied as between developing countries, the universal desire for rapid economic and social development has led most of them to accord an expanding role to government activities. Given the comparatively limited amounts of resources that it is ordinarily possible and prudent to obtain from abroad and from domestic borrowing and nontax revenues, most of the developing countries have felt the need to increase tax revenues. The willingness and ability of governments to respond to this need has been an important determinant of their capacity to carry forward development while maintaining economic stability. In working out either a development plan or a stabilization program, a major question that nearly always arises is how much tax revenue the country can reasonably expect to raise and from what sources this should come. In connection with stabilization programs, the answer to this question has an important bearing on the role to be played by other elements of financial policy and, often, on the likely success of the overall program.

Abstract

PROBLEMS OF TAXATION in developing countries have been receiving special and increasing attention in recent years, in the context of world concern for accelerating the pace of their economic growth under conditions of stability. Much of the theoretical and applied work has been related to the tasks of constructing productive and equitable tax systems suited to the peculiar economic and social conditions prevailing in various less developed countries. While the emphasis on the public sector has varied as between developing countries, the universal desire for rapid economic and social development has led most of them to accord an expanding role to government activities. Given the comparatively limited amounts of resources that it is ordinarily possible and prudent to obtain from abroad and from domestic borrowing and nontax revenues, most of the developing countries have felt the need to increase tax revenues. The willingness and ability of governments to respond to this need has been an important determinant of their capacity to carry forward development while maintaining economic stability. In working out either a development plan or a stabilization program, a major question that nearly always arises is how much tax revenue the country can reasonably expect to raise and from what sources this should come. In connection with stabilization programs, the answer to this question has an important bearing on the role to be played by other elements of financial policy and, often, on the likely success of the overall program.

I. Introduction

PROBLEMS OF TAXATION in developing countries have been receiving special and increasing attention in recent years, in the context of world concern for accelerating the pace of their economic growth under conditions of stability. Much of the theoretical and applied work has been related to the tasks of constructing productive and equitable tax systems suited to the peculiar economic and social conditions prevailing in various less developed countries. While the emphasis on the public sector has varied as between developing countries, the universal desire for rapid economic and social development has led most of them to accord an expanding role to government activities. Given the comparatively limited amounts of resources that it is ordinarily possible and prudent to obtain from abroad and from domestic borrowing and nontax revenues, most of the developing countries have felt the need to increase tax revenues. The willingness and ability of governments to respond to this need has been an important determinant of their capacity to carry forward development while maintaining economic stability. In working out either a development plan or a stabilization program, a major question that nearly always arises is how much tax revenue the country can reasonably expect to raise and from what sources this should come. In connection with stabilization programs, the answer to this question has an important bearing on the role to be played by other elements of financial policy and, often, on the likely success of the overall program.

The assessment of actual and potential tax performance of any country is a matter of judgment that should be based on a consideration of the stage of development and structure of the economy and should also take account of national traditions and relevant special circumstances. Systematic comparison with other countries can be illuminating, and a number of recent empirical studies facilitate that. Some of these studies 1 have attempted to discover, through systematic statistical analysis, the underlying economic and other factors that could be said to account for differences in tax ratios and tax structures among developing countries. A second branch of empirical studies is represented by broad surveys of the experiences of regional groups of developing countries, which have sought to highlight the major changes in the tax structures that have accompanied developmental efforts.2

The nature and objectives of this study are somewhat similar to those of the earlier empirical studies, but there are some differences. For one thing, the countries have been purposively chosen with a view to giving fair geographic representation (to the extent made possible by the availability of data) and to making interregional comparisons.3 Second, the objective here is not merely to provide “scientific” explanations of tax ratio differences but also to provide a qualitative appraisal of trends that may be useful for the formulation of policy. Accordingly, an attempt is made in the paper to integrate statistical analysis, interpretative review, and qualitative appraisal of major trends. While it is not intended to imply that the proper policy for all developing countries is to increase the share of income channeled into the public sector, it is believed that an inquiry of this nature, besides throwing some light on the underlying forces at work, would be found useful in evaluating actual and potential performance in various cases.

The plan of the paper is as follows: Section II indicates the scope and coverage and discusses the significance of some of the crucial magnitudes that will be used throughout most of the paper. Section III presents the basic figures on tax levels and composition, and outlines the major trends in taxation in a selected group of developing countries over the period 1953–55 through 1966–68. This section also attempts to highlight the major similarities and differences in the tax structures of a somewhat larger group of developing countries in the period 1966–68. Section IV is devoted to a statistical analysis of the cross-section data for the period 1966–68 with a view to identifying economic characteristics that bear significantly on tax ratios in developing countries. Section V explores the possibility of using the statistical results to derive indices of relative tax performance. Section VI summarizes the main conclusions of the study.

II. Coverage and Scope

To study the recent trends in taxation in developing countries, tax data were collected for a representative group of countries for the beginning and end of a 15-year period.4 In order to minimize the influence of fortuitous factors, 3-year averages have been used at both ends of the period. Tax data for the years 1953–55 and 1966–68 were collected for a sample of 30 developing countries.5 Other related data necessary for analysis, such as national income data and import and export ratios, could be collected for only 27 of these countries for the earlier period. For the later period, tax and economic data were collected for 50 countries. Hence, the analysis of the tax changes over the 15-year period covers 27–30 developing countries, whereas the cross-section analysis of the data for 1966–68 covers all the 50 countries except for certain computations. A few comparisons are made with an additional group of 16 developed countries.

It is necessary here to mention a word about the scope of the revenue data. The objective was to include tax revenues collected or raised by all levels of government in a country—central, provincial or state, and local. Unfortunately, satisfactory data on local taxation were not available in several countries.6 In some instances, however, the share of the local governments was known to be so small that its exclusion would not make any significant difference to the total picture. A dividing line of exclusion had to be drawn somewhere. It was felt that if local taxes accounted for 10 per cent or less of the total the central (and state) taxes could be fairly said to represent the total tax situation in a country. However, where only central taxes are included, it must be remembered that there is some understatement of tax levels. Table 16, in the Statistical Appendix, indicates the levels of goverment included as well as the exact years covered at either end, for each country. The note appended to the table describes briefly the sources and limitations of the data used in the study.

Any discussion of trends in taxation in a group of countries will necessarily have to concentrate on comparisons of levels and composition of taxation and the changes thereof. The level of taxation in a country is usually judged in terms of the ratio of “taxes” to some measure of national product. What elements should be included in the numerator and the denominator of the ratio depend on what aspect of the governmental role one wishes the ratio to reflect or signify and what one intends to show by international comparisons of tax levels. The general practice has been to use the ratio of all taxes and tax-like charges (including gross social security contributions from the private sector) to gross national product (GNP) at market prices. Many writers consider this an important index of the size of the public sector, although it is not always made clear what dimension of the public sector size the ratio is supposed to measure.

First of all, it is obvious that the tax/GNP ratio does not indicate anything about the proportion of national output generated in the public sector, i.e., the role of government as a producer; for this purpose, it would be necessary to employ the ratio of value added in the public sector to total value added. Second, the ratio is not a reliable index of the relative importance of the public sector as a final purchaser of goods and services. On the one hand, some part of the revenue is returned to the private sector in the form of transfer payments, and on the other hand, real expenditure is often partly financed through borrowing. Third, if the purpose is to get an idea of the share of the public sector in disposable income, the best measure is the ratio of “net revenue” (i.e., total revenue minus all government transfer payments) to GNP.7 This net revenue is said to measure “the net amount by which discretion over spending is transferred from private to government hands.” 8

What the overall tax/GNP ratio shows is the proportion of national income that is “compulsorily” transferred from private hands into the government sector for public purposes. As such, the ratio gives an idea of (a) the division of responsibilities between the public and private sectors and (b) the degree of control that the government can potentially exercise over the disposition of purchasing power in the economy.

The overall tax/GNP ratio may also be said to indicate tax burden in a somewhat different sense, which is often considered significant for purposes of policy. Individuals’ reactions to taxes and public expenditure benefits are generally found to be asymmetrical: Taxes are felt to be burdens because there is no close association in the minds of people between taxes and benefits made possible by them; and therefore tax-payers’ resistance and reactions (disincentive effects) tend to increase with the level of taxation, irrespective of the uses to which revenue is put. Furthermore, as one authority notes, “The relative distribution of tax payments and benefits often re-inforces this tendency for obviously taxes mainly fall on currently producing households but benefits are to a considerable extent paid to non-producers. …”9 A high tax/GNP ratio is therefore generally associated with increasing taxpayers’ resistance and disincentive effects. These consequences of a high tax/GNP ratio and the related concept of tax burden (which is really a composite of the notions of “sacrifice” and injurious economic effects) are particularly relevant when considering questions of “taxable capacity” and “tax effort”

As one of the main objectives of taxation in developing countries is the transfer of real resources to the public sector without causing inflation, it is of interest to those concerned with problems of development—policymakers in developing countries, international organizations, and donors of foreign aid—to know what progress different countries have made in mobilizing resources for development through their tax systems. And the question has often to be faced in practice whether a given country could not, if it wanted to, raise more taxes without seriously “burdening” the economy. In this connection a comparison may be made with the performance of other developing countries, with necessary adjustments for differences in relevant circumstances. For these purposes it has seemed appropriate to work with the tax/GNP ratio.10

The numerator of the ratio, i.e., the total of taxes, includes also royalties on minerals, fiscal monopoly profits, and profits of marketing boards. Given the purpose at hand, it is a moot question whether social security contributions should be regarded as taxes. There are two major countervailing considerations here. On the one hand, social security systems in the early stages are often used as instruments for mobilizing resources for development, and when comparing tax levels in developing countries it seems proper to include at least the contributions by the private sector. On the other hand, there is present in social security contributions (at least those by the employees) a much larger element of direct quid pro quo than for most taxes; and in some countries social security payments form an appreciable proportion of contributions. For these reasons, it would not be altogether fair to include the contributions in the total for considering tax effort in the context of development. Certain compromises, such as including only the contributions by private employers or the net contributions (gross private contributions minus payments), are conceivable. In point of fact, limitations of data have determined the course adopted here.

Data on social security contributions were not available for some countries that are known to have social security systems. Second, for certain other countries, no satisfactory information on actual contributions could be gathered, as only the estimates of investible surplus were available. Besides, for lack of detailed information it was not possible in all cases to exclude contributions by the government as an employer. Therefore, it was decided to work in the main with the total excluding social security contributions. However, two alternative totals are presented so that the comparison of tax levels may be made in both ways. In the rest of the paper, the term “tax ratio” is used, unless otherwise specified, to mean the ratio of total taxes as defined above (excluding social security contributions) to GNP at market prices.

III. Major Trends in Taxation, 1953–55 to 1966–68

Overall tax ratios

The major trends in taxation in the selected developing countries will be discussed in this section mainly in terms of tax ratios and composition of tax revenues. These two aspects of change are, of course, interrelated; a significant change in the tax ratio, for instance, almost invariably involves, or is accompanied by, a shift in the composition of taxes. The changes in levels and composition will, however, be described separately with a view to identifying the elements that are to be considered in any analysis of their interrelationship.

In general, tax revenue has been growing faster than GNP in the developing countries with the consequent rise in the tax ratios. Table 1 gives the tax ratios in the periods 1953–55 and 1966–68 for 27 developing countries, as well as certain measures or indices of changes in the ratios between the two periods. The countries are arranged in descending order of income elasticity of total taxes,11 mainly for the reason that this index (as explained below) is for many purposes the most meaningful basis for cross-country comparisons of such changes.

Table 1.

Selected Developing Countries: Ratios of Tax Revenue to Gross National Product, Marginal Tax Rates, and Income Elasticities of Total Tax Revenue, 1953–55 and 1966–68 1

article image
Source: See note in the Statistical Appendix, page 326.

Relate to total taxes excluding social security contributions.

The countries are ranked according to the income elasticity of total tax revenue.

Absolute change in tax revenue divided by absolute change in GNP.

Percentage change in tax revenue divided by percentage change in GNP. This measure, it may be noted, has been calculated by relating changes in actual tax revenues to changes in income, and hence differs from the coefficient of built-in elasticity, which is calculated with respect only to the “automatic” increase in tax revenue in response to economic growth or increase in income. Here the interest is more in the total relative increase in tax revenue, whether owing to economic growth, or changes in the tax structure and rates, or improvements in compliance and enforcement.

The tax ratio has increased between the two periods in more than two thirds of the countries in the sample; in some countries the tax ratio has fallen, but, given the limitations of data, the fall could be considered appreciable only in Ghana, where it has fallen by more than one fifth. In this case there seem to have been special circumstances explaining the marked deviation from the general trend. While there have been some instances of spectacular rise, by and large the tax ratio seems to have registered a moderate rate of increase: for the group as a whole, the (average) tax ratio increased from 11.3 per cent to 13.8 per cent between the two periods (Table 2). The coefficient of variation has changed but slightly, indicating that the relative dispersion in tax ratios has remained more or less the same. However, the distribution of countries by tax ratios has become less skewed in the latter period (Chart 1).

Table 2.

Selected Developing Countries: Means and Measures of Dispersion of Ratios Relating to Taxation

article image

Calculations based on Table 1.

Calculations based on Table 3.

Average of the income elasticity of total tax revenue for 27 countries. The overall income elasticity of taxes for a country is the sum of the elasticities of the individual taxes weighted by their respective shares in the total in the first period. The mean of these overall elasticities (1.4) is also thus implicitly weighted. Therefore, it does not “tally” with the mean values of the elasticities of individual taxes given in the table.

Chart 1.
Chart 1.

Selected Developing Countries: Frequency Distribution by Tax Ratios, 1953–55 and 1966–68

Citation: IMF Staff Papers 1971, 002; 10.5089/9781451947335.024.A002

The marginal tax rate gives us an idea of the proportion of additional national product that the government has been able (or has thought fit) to divert to the public sector. This proportion is seen to have varied from as high as 25.6 per cent in Tunisia to only 7.5 per cent in Indonesia. The average of this rate for the entire sample is 14.9 per cent (Table 2).12 This means that the developing countries, starting with a tax ratio of about 11 per cent in the early 1950s, had on average taxed away about 15 per cent of the additional income. Some countries, however, had much higher marginal tax rates: Tunisia, the Democratic Republic of Congo, Guyana, Morocco, Brazil, and Chile all collected 20–26 per cent of the additional income in taxation; with the exception of Chile and Morocco, these countries had tax ratios well above the average in 1953–55.

The concept of marginal rate of taxation is useful in relation to a discussion on the appropriateness of the disposition of additional national income among several uses in any given country. A simple comparison of marginal tax rates in different countries, however, is not a very meaningful measure of the relative efforts made by different developing countries in mobilizing resources for the public sector over any given period, as no allowance is made for differences in the average rates in the base period. For this purpose, the income elasticity of tax revenue (as defined in this paper) is a more satisfactory index. The income elasticity of total taxes can be looked at from two angles. On the one hand, it shows the percentage change in taxes associated with a 1 per cent change in income; on the other hand, it expresses the ratio between the marginal tax rate and the tax ratio of the earlier period.13 Hence, a comparison of income elasticities shows both the variations in tax changes in relation to income changes and the extent of differences between the marginal tax rate and the tax ratio in the base period. In either sense, this measure proves to be a good basis for the comparison of tax level changes in different countries.

The average income elasticity of total taxes for the group as a whole is 1.4. There are five countries with income elasticity coefficients equal to or greater than 2, namely, India, Korea, Morocco, Honduras, and Paraguay; these are the countries where the marginal tax rates have been the highest in relation to the average for the base period. In this sense they are the countries that have done the most to raise the tax ratios. It could be argued, however, that those countries that already had a relatively high tax ratio did not need so much to bring about a rise in it. In fact, all the five countries referred to above (with income elasticity coefficients equal to or above 2) had tax ratios below the median (11.5) in 1953–55, while all the five countries with elasticities below unity had tax ratios equal to or above the median in 1953–55. And the rank correlation between the tax ratio of the first period and the elasticity of tax revenue is found to be fairly high (and negative).14 It can be concluded that in general countries that started out with a fairly low ratio tended to have high elasticities.15

Table 3 gives the income elasticity coefficients of major classes of taxes for different countries in the sample. As one might expect, there is considerable variation in the value of the coefficient for the several taxes in different countries. However, it may be said that most of the countries have relied relatively more on income and production taxes 16 than on international trade taxes for increasing the tax ratio. In only 44 per cent of the 25 countries in the sample is the income elasticity of the last-mentioned group of taxes greater than unity, whereas this is true of 83 per cent of the countries in respect of production taxes and 68 per cent in respect of income taxes. Correspondingly, the average income elasticity for the group is the highest for production taxes at 2.4 (Table 2); at the other extreme, the average income elasticity of international trade taxes works out only to 1.4.17 These data lend some support to the general impression that with the growth and diversification of their economies the developing countries have been turning more toward the exploitation of internal taxes with diminishing relative reliance on international trade taxes; but the decline in general has really been in the relative importance of taxes on exports. For, on average, revenue from import taxes has been growing faster than GNP, and it has been estimated that the ratio of import taxes to the value of imports has remained more or less the same.18 Since there has been a shift in many developing countries in the composition of imports toward capital goods and intermediate products that are either exempt from taxation or subject to low rates, the constant ratio of import taxes to total merchandise imports implies an increase (in general) in the average rate of tariff on consumer goods.

Table 3.

Selected Developing Countries: Income Elasticity of Major Taxes, 1953–55 to 1966–68 1

article image

Colombia and Peru (included in Table 1) are omitted from this table, as the relevant breakdown of taxes could not be obtained for the earlier period.

Tax ratios and changes in them over time are matters that are subject to control by governments. Nevertheless, it may be hypothesized that the changes in tax ratios brought about in different developing countries are partly dependent on major characteristics of the economy that have a bearing on taxable capacity. Owing to lack of data, a detailed investigation of this question and related ones could not be undertaken concerning the historical data. However, an attempt was made to discover if the elasticity coefficient or the marginal tax rate (both of which varied widely as between different countries) was systematically related to some of the major quantifiable structural characteristics of the economies.19 These two measures of tax change have not been found to be significantly correlated with such characteristics as the import/export ratio to GNP 20 and the shares of major production sectors in gross domestic product (GDP) 21 or changes in them between the two periods. Nor is there a significant correlation with the composition of tax revenues or changes in it. The marginal tax rate is seen to be highly correlated with the tax ratio of the second period (r = 0.9); but this is merely a reflection of the fact that a higher marginal rate leads to a higher tax ratio in the ensuing period.

The marginal tax rate determines whether, and how fast, the tax ratio rises in the context of an increase in income over time. It may be hypothesized that the higher the per capita income, the easier it is to achieve a high marginal tax rate and, further, that the average tax rate of the base period is likely to influence the marginal rate insofar as it is relatively easy with a given tax system to take out the same fraction from the incremental income as that from the original income base. Besides, a preliminary comparison of the changes in tax ratios in the developing countries (in the sample) indicates a strong and positive influence of the mining sector. However, statistical testing did not show any connection between the marginal tax rate and the level of per capita GNP. It was found that the tax ratio of the base period and the share of mining in GDP could explain about 39 per cent of the variation in the marginal tax rate.22 The results obtained suggest that during the particular period covered by the study deliberate effort on the part of the governments accounted for a substantial part of the variations in the marginal tax rate.

Composition of tax revenues

The composition of tax revenues or the relative importance of specified individual taxes in the tax system may be considered in terms of either the percentages of different taxes in total tax revenue or the ratios of these taxes to GNP. The first alternative brings out more clearly the differences in the patterns of taxation between different countries irrespective of the level of total taxation, whereas the second throws light on the proportions of GNP collected through different tax forms.23 The changes in the composition of revenue will be considered here mainly in terms of the ratios of individual (groups of) taxes to total tax revenue; the ratio-to-GNP approach is adopted for analyzing the cross-section data for the second period.

Table 4 gives the ratios of major categories of taxes to total tax revenue in 1953–55 and 1966–68 for a sample of 30 developing countries. The classification presented is related to the major bases of taxation, such as income and international trade, and has been partly determined by the extent of the breakdown available.24 The grouping of taxes into direct and indirect taxes follows the lines of conventional classification.25 It may be noted that social security contributions have been excluded; if it is held that they are to be regarded as direct taxes, an understatement of the latter’s share might be inferred.

Table 4.

Selected Developing Countries: Composition of Tax Revenue, 1953–55 and 1966–68 1

(In per cent of total)

article image
article image

Excluding social security contributions.

Countries are ranked according to the income elasticity of total tax revenue with the exception of the last five countries (Ethiopia, Iran, Nepal, Somalia, and the United Arab Republic), for which GNP data are not available for the first period. I signifies first period (1953–55), and II, the second period (1966–68).

Unclassified because of lack of information.

Mostly tertib.

Includes sales tax on imports.

Mostly canoun.

Included in income taxes.

Mostly internal, indirect taxes.

In computing the averages, the countries for which information on a particular tax share is not available have been excluded. Hence, the sum of the averages of individual shares will not add to 100.

The simple arithmetic means of the percentage shares for the sample as a whole indicate a noticeable shift from international trade taxes to taxes on production and internal transactions. In the sphere of direct taxes, the shift has been from property and poll taxes to taxes on income. These trends conform to the general theoretical argument that with the gradual commercialization and modernization of the economy it should be possible for the developing countries to shift from personal and crude property taxes to taxes on the incomes of individuals and corporations and that, concurrently, with diversification of domestic production, sales and excise taxes would tend to gain at the expense of import and export taxes. What is noteworthy, however, is that the fall in the share of total international trade taxes has been much greater than the fall in that of import taxes. In other words, it is the fall in the average share of export taxes—from 9.6 per cent to 7.3 per cent of the total—that really accounts for the decline in the relative importance of taxes on international trade.26 This in turn seems to have been due largely to the relative fall or stagnation in the prices of staple agricultural exports in the intervening years of the period chosen for study. However, a systematic analysis of the reasons behind the decline in export taxes would involve consideration of other general issues, such as the role of exchange systems, that are outside the scope of this study.

The share of taxes on income is shown to have increased on average by 1.6 percentage points. However, the share of total direct taxes has fallen slightly. It is not merely that the average share for the sample has fallen; the fall has taken place in a majority (18) of countries. On an analogy with the course of evolution in the now developed countries, it is generally argued that with the growth of the economy and the rise in income the share of direct taxes, which could be levied more in accordance with the individual’s ability to contribute, would begin to rise. Such a transformation of the tax systems is yet to come; during the period considered, the countries have only reduced the reliance on taxes on international trade, which, because of the variability of export tax revenues, has been in many cases the cause of instability in government revenues. Also, it would seem that modern forms of income and property taxation leviable on the generality of the population could gain in importance only after the stage of development and the per capita income have reached a certain level.27 As it is, the share of income taxation would have been much lower in some of the developing countries but for the operation of large companies engaged primarily in the export of minerals.28

Where the patterns of change are markedly divergent, simple averages are likely to be misleading. It is necessary to check in how many, or in what percentage of, countries the indicated changes have in fact taken place. On the side of increase, the most pervasive change concerns production taxes: in 20 of the 30 countries, the share of this group of taxes rose. Next come income taxes, whose share increased in 53 per cent of the sample countries. Unfortunately, data could not be obtained for all countries on the share of import taxes. In 13 of the 23 countries for which the necessary breakdown is available, the share of import taxes increased. Thus, the relative importance of taxes on imports seems to have increased in a majority of countries during this period.

On the side of decrease, the commonest or most general trend is the fall in the share of international trade taxes, which has generally been brought about through a fall in that of export taxes.29 And although the average share of total direct taxes in the sample as a whole did not decline significantly, as already indicated, a decline in its importance occurred in a majority of countries.

Table 5 provides information on how the increases in the shares of particular taxes have been associated with the decreases in shares of other taxes. Five major groups of taxes are considered for this purpose. The figures in parentheses give the total number of countries in which increases or decreases in the shares of particular taxes have taken place and figures in the columns (rows) indicate in how many cases (out of the relevant total) decreases (increases) in the shares of other taxes have occurred. The change of the highest occurrence is the increase in the share of production taxes. In 17 of the 20 countries in which this increase has taken place, the share of international trade taxes has fallen; conversely, in 17 of the 19 countries in which the share of international trade taxes has fallen, the share of production taxes has risen. The second noteworthy feature of change is a rise in the share of income taxes in 11 of the same 19 countries with a fall in that of international trade taxes. Third, there are no countries where the share of import taxes has fallen with a rise in the share of international trade taxes; on the other hand, in 4 of the 13 countries with a rise in import tax share, the share of total international trade taxes has fallen. It may be concluded that in a majority of countries there has been a rise in the share of production and income taxes accompanied by a fall in the share of international trade taxes. As surmised from average trends, this fall has occurred mainly through a relative decline in the taxes on exports; the share of import taxes has risen in about 50 per cent of the countries.30

Table 5.

Distribution of Countries by Associated Tax Share Changes1

(Number of countries)

article image
Source: Table 4.

This table is not a proper matrix; there are no meaningful row or column totals. The figures in parentheses are introduced to help to gauge the relative proportions of the various combinations of changes.

The increase in the share of production taxes has been brought about both through the more intensive use of excises and by the adoption and expansion of sales taxes.31 The relative shares of both these taxes have generally increased between the two periods (Table 11, in the Statistical Appendix). The widespread use of sales taxes may be said to have been one of the important developments during the period covered by the study. Whereas the average share of excises has increased from 17.9 per cent to 19.6 per cent of the total, that of sales taxes has increased from 6.9 per cent to 12.9 per cent. And in 11 of the 30 sample countries in Table 4, sales taxes of one kind or another were introduced after the first period.

The direction of changes in the composition of tax revenue in the selected countries is further brought out in Chart 2, which gives the frequency distribution of the percentage shares of three important groups of taxes. The shape of distribution of countries by the share of income taxes has remained practically the same, except that the proportion of countries in which revenue from income taxes ranges from 0 per cent to 10 per cent has fallen slightly and the proportion of countries in the range of 21 per cent to 30 per cent has increased. In respect of the share of international trade taxes, the distribution is much less skewed in the second period than in the first: the countries are now more evenly distributed among the different ranges. This has come about as a result of some countries increasing their reliance on this group of taxes while others with high reliance in the early period were reducing it. The trend is seen to have been in the opposite direction in respect of production taxes. In the first period, the countries were fairly evenly distributed among the different ranges and the 11 per cent to 20 per cent range contained the highest proportion; in the second period, the middle and top ranges expanded and the latter emerged as the modal class. This change is merely the reflection of the fact that several countries increased their reliance on internal indirect taxes.

Chart 2.
Chart 2.

Selected Developing Countries: Frequency Distribution of Shares of Major Tax Groups, 1953–55 and 1966–68

Citation: IMF Staff Papers 1971, 002; 10.5089/9781451947335.024.A002

The changes in the composition of tax revenue reflect the differing rates at which the various taxes have been growing, whereas the marginal shares of those taxes indicate their relative contributions to increases in tax revenue and therefore to overall elasticity (Table 13, in the Statistical Appendix). The contribution of production taxes is seen to be by far the largest in a majority of countries. On average, this group of taxes contributed 40 per cent of the increments in tax revenue. Taxes on income and import taxes together contributed on average another 50 per cent of the increase. These proportions reveal the sources of overall elasticity. But, as can be surmised from Table 13, the level of overall elasticity is not systematically related to the magnitude of relative contributions of any particular taxes. On the other hand, some interrelationship between the absolute changes in individual and overall tax ratios may be discerned.32 In countries where the overall tax ratio has fallen, there has generally been a decline in the ratios of income taxes and/or of international trade taxes (Table 14, in the Statistical Appendix). In these countries, as there has generally been an increase in the ratio of production taxes and no significant change in the ratios of other taxes, the decline in overall tax ratios could be legitimately traced to the fall in the ratio of taxes on income and on international trade. The fall in the ratio of international trade taxes has been in general more significant than that in the ratio of income taxes and, in most cases, has really been brought about through a fall in the ratio of export taxes—Costa Rica being the most noticeable exception. In countries where tax ratios have risen, the ratio of income taxes has generally increased without any significant fall in the ratio of international trade taxes. The ratio of production taxes has increased in most countries in the sample and fallen appreciably in none of them. It would thus be seen that the developing countries in general have been attempting to raise tax revenues through the extension of internal production taxes and that the tax ratio itself rose in those countries where, in addition, the ratio of income taxes could be increased; conversely, the tax ratio fell in those countries where the increase in the ratio of production taxes was more than compensated by the decrease in the ratio of income taxes or international trade taxes or both.

The broad conclusions that can be said to emerge from the analysis of changes in the composition of tax revenues are that (a) the most important change is the increase in the share of production taxes that took place in most countries; (b) the relative importance of international trade taxes has generally declined; (c) this decrease has come about more through the decline in the share of taxes on exports (including marketing board profits) than through the fall in that of import taxes; (d) the ratio of production taxes (to GNP) has increased in the generality of countries, irrespective of the direction of change in the overall tax ratio, whereas the ratio of income taxes and the overall tax ratio have generally changed in the same direction; and (e) where tax ratios have fallen, the decline is due most frequently to the fall in the ratio of income and international trade taxes.

Variations in tax levels and composition—a cross-section of 50 developing countries (1966–68)

The foregoing discussion of changes in tax ratios and in the composition of taxes has been with reference to the smaller sample, 27–30 countries. A more rigorous statistical analysis of changes between the periods has been precluded by the nonavailability of the requisite data on related economic characteristics for all the countries in the sample for the earlier period. Hence, an attempt is made to study tax ratio differences through a cross-section analysis of the sample of 50 countries for which more satisfactory and comprehensive data could be collected for the later period only. Although such an analysis does not directly throw light on trends over time, it is a useful aid in understanding how the tax characteristics of developing countries are related to their economic characteristics and vary with the stage of development.

Table 6 gives the ratios of major categories of taxes as well as total taxes to GNP in 50 developing countries arranged according to (rising) per capita GNP in U. S. dollars. With the addition of Zambia and Nepal (at either end) the range of variation of the overall tax ratio is widened compared with the smaller sample for the same period; but the coefficients of variation are not very different for the two samples.33 The tax ratios for the 50 developing countries in the sample are all under 30 per cent, and the average ratio amounts to 14.0 per cent. The inclusion of social security contributions does not make a significant difference to the general picture, although the level of taxation in a few countries would be shown to be substantially higher: the average tax ratio (inclusive of social security contributions) for 45 countries works out to 14.6 per cent; and the modal class is found to be the same as for the distribution by the tax ratios exclusive of social security contributions (namely, 11–15 per cent).34 For reasons given earlier, the ensuing discussion also will be related mainly to the level of taxes excluding social security contributions.

Table 6.

Selected Developing Countries: Major Categories of Taxes, 1966–68

article image
article image

Social security system exists but in terms of contributions is not very significant.

Estimate.

With the exception of Venezuela, all the countries in the sample have a per capita income below US$800. In view of the wide coverage, the sample may be said to constitute the majority of the low-income countries. In spite of the general increase in tax levels in developing countries indicated by the earlier analysis, the average level of taxation in the less developed countries continues to be much below that in the developed, high-income countries. The average ratio in a group of 16 developed countries of Europe and North America amounts to 25 per cent (in 1966–68) and most of these countries raise more than 20 per cent of GNP in taxes exclusive of social security contributions (Table 15, in the Statistical Appendix). Whereas a typical developing country raises less than 15 per cent of GNP in taxes, a developed country raises about 25 per cent. The difference is seen to be even greater if total taxes are taken to include social security contributions. The average ratio of taxes (defined in this way) to GNP in the developed countries is more than double that in the developing countries (32: 15). This much greater difference is obviously due to the fact that the former have very much more developed and comprehensive social security systems; social security contributions by themselves, on average, account for about 7 per cent of GNP in these countries.

One of the factors distinguishing the developed from the developing countries is, of course, the wide difference in per capita income levels. When a comparison is made between the two groups of countries, per capita income differences are clearly seen to be associated with tax level differences. It is generally argued that high per capita income levels lead to high tax ratios because of both the correspondingly high capacity to pay taxes and the high income elasticity of demand for public goods. These factors, however, seem to exert a decisive influence only after per capita income has risen far above the level of bare subsistence, or rather after the entire economy has been transformed and modernized. There are fairly wide income differences between the developing countries themselves. Argentina, for example, has 14 times the per capita income of Rwanda, and Mexico nearly 6 times that of India.35 Do the less poor among the developing countries tend to have a higher tax ratio than the poorer ones? This question is considered more systematically in the next section. For the moment, it may be pointed out that the average tax ratio for the countries in the upper half of the per capita income scale (i.e., with per capita income above US$200) works out to 15.5 per cent, which is distinctly higher than the average ratio for those in the lower half at 12.3 per cent. However, the top 16 countries having per capita incomes above US$300 have an average tax ratio of 14.8 per cent, which is lower than the average ratio for the upper half (15.5 per cent). This indicates that the countries in the per capita income range of US$200–300 have on the average higher tax ratios than those in the higher ranges. This is inconsistent with the thesis that the tax ratio would rise systematically with per capita income. But if the total of taxes inclusive of social security contributions is considered, the average tax ratio for the countries with per capita income above US$300 is found to be the highest for all groupings at 17.0 per cent. Social security taxes can therefore be said to be employed relatively more by the less poor among the developing countries.

Table 6 gives also the breakdown of the tax ratios into their major components. The “decomposition” of the tax ratios gives an idea of the levels of different forms of taxation in relation to GNP in the selected countries. The major interesting facts brought out by the table may be summarized briefly. First, taxes on property are not of much significance; they constitute less than 1 per cent of GNP in the vast majority of countries. It is possible that their share would have been shown to be somewhat higher if local government revenues had been included in all cases. The only firm conclusion that could be drawn is that modern forms of property taxation at the central/state government level have not been developed in most of the countries. Second, the share of income taxes is not insignificant even in the poorer countries in the sample. Some countries with per capita incomes of less than US$100 are found to raise 2–3 per cent of GNP through the taxation of incomes. However, this form of taxation brings in less than 2 per cent of GNP in more than one third of the countries in the sample. Third, the average ratios of international trade taxes or production taxes to GNP are not very different for the richer half and the poorer half of the sample. The difference in the overall tax ratio between the countries in the upper half of the income scale and the rest arises from the differences in the proportions of GNP that the two groups raise through income taxation, as seen below.

article image

It is evident that the less poor countries raise on average a distinctly higher percentage of GNP through income taxation, and this has partly helped them to achieve a higher overall tax ratio.36 But the favorable influence of mineral production seems even more important. A number of countries in the top 50 per cent are noted for the large share of mining in GDP. Indeed, the countries in this category, such as Zambia, Venezuela, Iran, Chile, Trinidad and Tobago, and Guyana, have proved to be the leading users of income taxation.

Direct taxes bring in only a minor part of revenue in most of the developing countries, their average share being about 30 per cent of total. It is well known that the share of this group of taxes is generally much higher in developed countries. Indeed, it is generally expected that economic growth with its concomitants of urbanization, proliferation of salary incomes, and the development of intangible forms of property would gradually bring about the ascendancy of modern forms of income and property taxation. It has been found, however, that the average share of direct taxes fell slightly between the two periods in the countries in the small sample. But a majority of countries in this sample had low per capita incomes (below US$300) during the second period. A study of the cross-section data for the large sample, which includes several countries with per capita incomes above US$300, might be better suited for finding out how the share of direct taxes tends to vary with differences in per capita income and the associated differences in economic conditions.

To test the hypothesis that direct taxes are positively related to per capita income, first the entire sample was considered and then the sample was divided into two groups: countries with per capita income below US$200, and the remainder.37 In each case, the degree of relationship was sought to be measured by a regression of per capita income on the share of direct taxes.38 The scatter diagrams relating to the three groups are shown in Charts 3, 4, and 5, respectively.

Chart 3.
Chart 3.

Selected Developing Countries: Relation Between Share of Direct Taxes and Per Capita Gross National Product, 1966–68

Citation: IMF Staff Papers 1971, 002; 10.5089/9781451947335.024.A002

Chart 4.
Chart 4.

Selected Developing Countries: Relation Between Share of Direct Taxes and Per Capita Gross National Product of, Less Than US$200, 1966–68

Citation: IMF Staff Papers 1971, 002; 10.5089/9781451947335.024.A002

Chart 5.
Chart 5.

Selected Developing Countries: Relation Between Share of Direct Taxes and Per Capita Gross National Product of, More Than US$200, 1966–68

Citation: IMF Staff Papers 1971, 002; 10.5089/9781451947335.024.A002

For the sample as a whole, the share of direct taxes is found to be positively related to per capita income, which turns out to be a significant determinant explaining about 25 per cent of the variation in the direct taxes share. Similarly, in countries with per capita incomes above US$200, a significant association between the share of direct taxes and per capita income is seen to exist. On the other hand, there is no evidence of such a relationship for the group of very poor countries with per capita incomes below US$200;39 indeed, Chart 4 shows clearly that it will be too farfetched to try to establish a relationship between per capita income and the share of direct taxes in these countries. It is legitimate to conclude that (a) the share of direct taxes does not rise with per capita income until a certain minimum level of income has been passed; (b) after this level has been reached, countries are able and likely to raise the share of direct taxes with increases in per capita income; and (c) the positive relationship between per capita income and the share of direct taxes holds if the very poor and less poor among the developing countries are taken together, just as it is found to hold when developing and developed countries are considered together.40

Regional variations in levels and composition of taxation

The sample of 50 developing countries has been drawn from various geographic regions of the world. It is likely that countries within a given region have certain similarities in matters of taxation because of similarities in economic conditions, “demonstration effect” as between neighboring countries, etc. It is beyond the scope of this paper to make detailed comparisons of the tax structures of countries within different regions. However, to gain some idea of regional variations, the sample is divided into five geographic groups 41 and an attempt is made to compare average or representative tax structures of the different regions.

The average tax ratios range from 16.1 per cent for the Middle East and North Africa to 11.6 per cent for Asia and the Far East (Table 7).

Table 7.

Selected Developing Countries: Regional Variations in the Levels of Taxation, 1966–68 1

(In per cent of GNP)

article image

Calculations based on Table 6; see footnote 9 to Table 4 for method of computing averages.

Excluding social security contributions.

Negligible.

The latter, with the lowest regional ratio, contains 12 countries, only 3 of which are in the lowest fourth of the countries in the per capita income scale. Similarly, Central America and the Caribbean ranks fourth in average tax ratio, but all countries in this region fall in the upper half of the income scale. The “tradition” in several countries in Asia and the Far East and in Central America and the Caribbean seems to be in the direction of low tax ratios.

The variations of the overall tax ratios as between the groups arise more from differences in the shares of income taxes, or of total direct taxes, than from those in indirect taxes, except for Central America and the Caribbean, which has the lowest ratio of indirect taxes. The relative importance of major taxes in the different regions is better considered in terms of the shares of those taxes in the total indicated in Table 8. It is seen that the Middle East and North Africa, having the highest average tax ratio, have more than the average income tax and production tax shares. Tropical Africa is found to be the most dependent on international trade taxes: over 40 per cent of tax revenue for the group is derived from these taxes against the average share of 34 per cent. Asia and the Far East has the lowest share of income taxes and, although this group derives as much as 10 per cent of its revenue from property taxes, its share of direct taxes remains the lowest among the regions. The range of (percentage) variation is seen to be the least in respect of import taxes. The developing countries in all the regions may be said to derive 20–28 per cent of their tax revenue from the taxation of imports.

Table 8.

Selected Developing Countries: Regional Averages of Major Categories of Taxes, 1966–68

article image

For method of computing averages, see footnote 9 to Table 4.

It is of interest to examine how the different tax structure characteristics are interrelated in various regions. For this purpose, the rankings of the regions by overall tax ratios and also by specified tax shares are given in Table 8. In Asia and the Far East, a low overall tax ratio ranking is combined with a low direct tax share ranking, and conversely for the Middle East and North Africa; but in Central America and the Caribbean the highest ranking in respect of direct tax share is combined with low (fourth) total tax ratio ranking. However, it is obvious that but for Central America and the Caribbean there is close correspondence between overall tax ratio ranking and income, as well as direct tax share ranking. On the other hand, the mix of indirect taxes (i.e., international versus internal) seems to vary with reference to other circumstances, there being little relationship between tax ratio ranking and production tax share ranking.

IV. Determinants of Tax Ratios in Developing Countries

As indicated earlier, government revenue share, or the tax ratio, has been widely regarded as an index of the size of the public sector, and several attempts have been made in the literature to explain the variations in the size of the public sector as reflected in tax ratio differences between different countries as well as over time.42 These attempts have consisted in trying to discover if there is any systematic relationship between measurable economic characteristics including the level of development, on the one hand, and tax ratios, on the other hand. Generally, the studies have been in the nature of cross-section analysis of a sample of countries: some of the samples studied consisted of developed and developing countries; others were drawn only from developing countries. The practice has often been to study first a sample consisting of both developed and developing countries and then to break it down into two groups on the basis of some cut-off line.43 Apart from the study by Martin and Lewis, all the other studies referred to have attempted, through cross-section analysis,44 to measure the extent of variations in tax ratios (or revenue/GNP ratios) explained by different economic (and in some instances, also sociopolitical) factors. It is not intended here to review these studies in detail. A brief account will, however, be given as an introduction to the exposition of the present exercise and the subsequent discussion of a basis for assessing tax performance by developing countries.

A résumé of earlier studies

In perhaps the earliest systematic statistical analysis, Williamson used per capita income as an indicator of the stage of development and fitted an exponential function to data for a sample of 33 developed and developing countries.45 His result indicated that there was a significant positive relationship between the revenue ratio and per capita income, although differences in the revenue share were less pronounced than those in per capita income. In a study relating to 20 countries classified as less developed, Plasschaert46 used per capita income and import/GNP ratio as determinants to explain variations in the ratio of government revenue to GNP. While the import ratio turned out to be significant 47 both when used alone and when used in conjunction with per capita income, per capita income did not emerge as a significant determinant of the revenue ratio. Likewise, Hinrichs found that “for less developed countries with per capita income below $300, ‘openness’ (as measured by the import ratio), not per capita income, is a key determinant of government revenue shares of gross national product.” 48

Lotz and Morss, in their first paper on tax ratio analysis,49 chose a sample of 72 developed and developing countries and sought to examine the relationship between tax ratio 50 differences and differences in per capita income and degree of openness. They used the ratio of the sum of imports and exports to GNP, rather than the import ratio or the export ratio, as the index of openness. For the sample as a whole, they found both income and openness to be significant explanatory factors positively related to the tax ratio; and together they explained a high proportion of the variance. When their exercise was repeated with reference to the 52 low-income countries in their sample, the explanatory factors remained significant, but the degree of explained variance was reduced substantially.51

In their later paper,52 the authors introduced additional variables to the above-mentioned analysis. The introduction of the degree of monetization as an explanatory factor resulted in a significant increase in the explained variance 53 but lowered considerably the significance of per capita income.54 They also tried to measure the effects of (a) export composition and (b) government centralization. In neither case was the result conclusive; also, they were unable to increase the explained variance substantially above that obtained by using income, openness, and monetization.55

In a recent study,56 the Secretariat of the United Nations Conference on Trade and Development (UNCTAD) has attempted an extension of the original Lotz-Morss analysis. Cross-section and time-series data were pooled for 36 developing countries for the 1950–66 period and two new explanatory factors, namely, the share of agriculture in GDP and the rate of inflation, were added to per capita income and openness. All of them came out to be significant, but the collinearity between the agricultural share and per capita income 57 required that one of them should be dropped. It was also felt that there was little; a priori reason to include the inflation factor. Therefore, the study settled for a model that incorporated only the agricultural share and openness index 58 as determinant factors and that explained 32 per cent of the variance.

There have been other studies that have attempted to take account of the influences of more intangible factors, such as sociological factors, forms of government, and geographic regions. The review given above, however, will give a fair idea of the nature and broad results of the analyses of tax ratio variance carried out to date. The following comments may be made:

a. When developed and developing countries are taken together, a close and strong relationship exists between per capita income and public sector size as reflected by the tax ratio. If tax ratio differences among less developed countries alone are considered, per capita income has been found to be less effective as an explanatory factor.59

b. Most of the studies have concentrated on factors that could possibly influence the tax ratio via their effects on the ability of the populace to pay taxes or the capacity of the governments to collect them. For example, the level of per capita income is taken to reflect the level of “surplus” over subsistence out of which taxes could be paid as well as the level of economic development, which is accompanied by an increase in the literacy rate, monetization, urbanization, etc.—all of which facilitate tax collections. The degree of openness is also brought in as a factor that facilitates tax collection directly as well as through its effects on other characteristics of the economy mentioned above. It is fair to say that in most cases only the factors assumed to operate on the supply side of funds have been taken into account.

c. Emphasis has been placed more on the level of development than on the structure of the economy, in the sense of the relative importance of different sectors of production.

d. The degree of total variation in the tax ratio explained by the different regression models has not been very high for developing countries.

Conceptually, the tax ratio or the share of national income appropriated by the government can be said to be determined by four broad groups of factors. On the side of demand for government services or activities, a distinction can be made between two groups of factors: (a) the need for services arising out of “objective” conditions and (b) the preferences of the people and the leaders as between public and private services, including the institutional arrangements (for the fulfillment of particular needs) arising from them. On the side of raising resources or supply of funds, two operative factors may be postulated: (a) the ability of the people to pay taxes and (b) the ability of the government to collect taxes. In this connection, two important facts have to be noted. First, the ability of the population to pay taxes is not independent of the types of services provided, as has been shown in the long discussion on “taxable capacity” in the traditional public finance literature. Therefore, certain initial adjustments in tax ratios would have to be effected, strictly speaking, before making intercountry comparisons.60 Second, whereas in the case of private demand there is in the minds of individuals a direct interaction between needs and capacity, in the case of demand for public goods “the governmental authorities” 61 form the transmission mechanism and have to act to bring to bear the needs and preferences on the potential supply of resources. Thus, the needs and preferences for public goods may be said to affect the tax ratio through the willingness to tax.

A further point to note is that the ability of the government to collect taxes is determined partly by objective, structural factors, and partly by such “volitional” factors as efficiency in administration and the nature of political leadership. The former taken together with the ability of the population to pay taxes may be said to constitute the total “capacity” factors affecting the tax ratio. The limited part of variation explained by tax ratio analyses with reference to developing countries is explained partly by the fact that they have in general sought to consider mainly the influence of “capacity” factors.

A tax effort model

Following the Lotz-Morss study and later studies, a statistical model, relating the tax ratio to certain major economic factors hypothesized to affect taxable capacity, is presented below. The model will be used as a framework for considering relative tax effort by the countries in the sample.62

Explanatory factors

On the basis of a priori reasoning and of a preliminary analysis of data, it has seemed proper to incorporate three major factors that may be held to exert a strong influence over the tax ratio, via their effect on what have been termed “capacity” factors.63 These are as follows: (1) the degree of “openness”; (2) the level of development and income; and (3) the composition of income. In what follows, a brief discussion of the choice of proxy measures reflecting these factors is given.

(1) “Openness” of the economy. Common observation as well as the results of the earlier studies in general indicates the degree of openness to be an important factor affecting the tax ratio in developing countries. The question arises in what way openness is best measured for purposes of tax ratio analysis. There are three candidates in the field: the import ratio, the export ratio, and the foreign trade ratio represented by the ratio of imports plus exports to GNP. The last may be held to be the best indicator of the relative size of the total potential base for foreign trade taxes. However, it could be argued that what is to be taken into account in this connection is not the base for foreign trade taxes as such but the total tax base and, therefore, the effects on monetization, the relative size of the organized sector, etc., flowing from a large export sector in a developing country. Large imports of consumer goods, which are generally selected for appreciably high rates of taxation,64 are made possible through large exports. Conceptually, therefore, it seems better to take the export ratio. It was also found (through correlation analysis) that the export ratio was more closely associated with the tax ratio than either the import or the foreign trade ratio.65 Hence, the export ratio is chosen as the index of openness for purposes of the present analysis.

(2) Level of income. Following earlier studies, it may be hypothesized that the level of income is a determinant of the tax ratio through its effect on the ability to pay taxes. A higher level of per capita income is also generally indicative of a higher level of development with the associated characteristics of better organization, a higher literacy rate, etc., which facilitate tax levy and collection. A higher per capita income reflecting a higher level of development may thus be held to indicate a higher capacity to pay taxes as well as a greater capacity to levy and collect them. However, the conversion of per capita incomes in local currencies into U. S. dollars at the official exchange rates for purposes of comparison leads to some inaccuracies that may distort the results.

An alternative measure or indicator of the stage of development is the proportion of income generated in the agricultural sector. Generally, a higher agricultural share (in GDP) will be associated with a lower per capita income, a larger subsistence sector, and a lower level of industrialization—all indicators of lower taxable capacity. Besides, the agricultural sector itself is notoriously difficult to tax.

There is also empirical evidence of a close relationship between the agricultural share of total employment and both the structure of the economy and the tax ratio. In their study of 74 developing countries, Adelman and Morris found that the size of the agricultural sector in terms of employment was negatively related to per capita income as well as to several other indicators of the stage of development, such as the level of education and the extent of mass communications.66 With reference to the sample in this study, the correlation analysis shows that the agricultural share of income is highly correlated (negatively) with the tax ratio.67

(3) Economic structure or the composition of income. An examination of the tax ratios in the 50 sample countries would readily reveal the influence of the share of mining in GDP. Mineral production is of key or major importance in at least six 68 of the ten countries having the highest tax ratio; and in another one—Tunisia—where phosphate production was already of some importance, the production of petroleum and natural gas has recently gained in importance with corresponding contribution to public revenues. Accordingly, a positive relationship between the mining share of income and the tax ratio has been hypothesized. Because of the heavy fixed investment associated with extractive industries, operations tend to be confined to a few large firms and as long as world demand conditions ensure high profitability, there exists a combination of taxable “surplus” and administrative ease.69 Further evidence that there is some need to consider specifically the mining share is found in the results of the Lotz-Morss and UNCTAD studies where, because no adjustment was made for differences in the mining share of income, the heavily mining-oriented countries tended to rank high in terms of tax effort.70 With reference to the present sample, the mining share shows a stronger relationship with the tax ratio than any other variable in the correlation analysis.

If the composition of income is to be fully taken into account, allowance should be made also for the manufacturing share. It is a plausible hypothesis that the share of income generated in the manufacturing sector will be positively related to the tax ratio because domestic production generates a tax base just as imports do. However, no measurable relationship between the manufacturing share and the level of the tax ratio could be found; furthermore, there is a high correlation between manufacturing share and per capita income.

The results

The foregoing discussion of the proxy measures reflecting tax capacity factors would indicate that a plausible model is one that incorporates per capita income, the export ratio, and the mining share as explanatory (capacity) factors. However, since exports are part of income, it is considered preferable to use per capita nonexport income in place of per capita income. Similarly, to avoid the overlap between export ratio and the mining share (in all countries exporting mineral products), it would be preferable to use the export ratio excluding mineral exports. When the model is tested, it is found that the mining share is highly significant, the export ratio excluding mineral exports is significant, and per capita nonexport income not significant (statistically);71 the total degree of explained variation in the tax ratio is 39 per cent.72

As already indicated, the agricultural share could be used instead of per capita income as a proxy for the level of taxpaying capacity and the stage of development. A positive advantage in the substitution is that no conversion of the measure into a common standard would be called for. When a model incorporating the export ratio, the agricultural share, and the mining share is tested it is found that the mining share is highly significant and positively related to the tax ratio, the agricultural share just falls short of significance 73 and is negatively related to the tax ratio, and the export ratio is not a significant determinant; the total degree of explained variance is 41 per cent.74 If the export ratio is dropped and only the agricultural share and the mining share are taken into account, the best statistical results are obtained.75 Such a model explains 41 per cent of the variation in the tax ratio, and both the factors included are statistically significant.

The finding that most clearly emerges from the foregoing analysis is that the mining share is a very important determinant of the tax ratio in developing countries at their present stage of development. If this factor is not include in an explanatory model and if account is taken only of per capita nonexport income and the export ratio (equation (8)—more or less in line with the original Lotz-Morss formulation), both the included factors are significant, but there is little justification for omitting an important factor that by itself explains a greater degree of variation. On the other hand, the model incorporating only the agricultural share and the mining share yields the best statistical results, but it excludes per capita income, which has normative significance for tax effort analysis; and the model also does not make allowance directly for the export factor in countries where mining is not so important. Furthermore, there is room for difference of opinion about the extent to which the relative size of the agricultural sector indicates capacity to tax, as distinguished from willingness to tax. As is well known, many developing countries have found it difficult to tax agriculture adequately, for historical and political reasons. This being so, it is debatable whether the agricultural share could be properly included in a tax effort model.

The conclusion reached here is that the model that embodies the set of hypotheses most meaningful from the point of view of economic reasoning and tax effort analysis is the one that incorporates per capita nonexport income, the export ratio,76 and the mining share (equation (7)).

In the chosen model, one of the factors—namely, per capita income—falls short of statistical significance (by conventional standards). Nevertheless, it has been decided to retain it in the model to be employed to measure taxable capacity mainly for the following reasons: First, as already mentioned, per capita income has normative significance for tax effort analysis, and although it does not pass the conventional test of significance with reference to the present sample, the clear association between per capita income and the tax ratio revealed when developed and developing countries are taken together is a strong pointer to the interrelation between it and taxable capacity.77 Second, it was found with reference to the present sample that per capita income became significant if the same model (equation (7)) was tested with the introduction of dummy variables for the regions. This suggested that some regional effect was pulling down or swamping the influence of per capita income, which, basically, did have a determining influence. Third, in terms of our a priori reasoning, an indicator of the stage of development needs to be included as a capacity factor. As indicated earlier, the choice is between per capita income and the share of agriculture in GDP. However, because of the political element brought in by the introduction of the agricultural share (referred to earlier) and because of the fact that the agricultural share also does not become statistically significant unless the export share is dropped, it has been decided to opt for the per capita income variable.

It may be reiterated that the present analysis of tax ratio variations is deliberately intended to consider only the major factors that are presumed to affect the tax ratio through the side of ability to pay and collect taxes. Several sociopolitical and other factors affecting it through what has been termed the willingness to tax have been left out. If desired, it might be possible to take account of some of these factors through the introduction of dummy variables or by the adoption of a more elaborate model. Be that as it may, implicit in the tax effort approach is the assumption that the factors affecting the willingness to tax are largely independent of the capacity factors whose effects have been estimated.

V. Evaluation of Tax Performance

While each country decides for itself what level of taxes it should impose in the light of myriad considerations, and while there is no special economic merit in a low or high level of taxes as such, the question is often raised whether a country could not, if it so desired, levy a higher level of taxes. And policymakers inevitably make comparisons with the levels of taxation and the experiences in other countries in a similar stage of development. In recent years, international organizations and donor countries have tended to make such comparisons among developing countries.

It is not uncommon for those interested in, or entrusted with, tax policy formulation to make comparisons of actual tax ratios in different countries. But a simple comparison of the tax ratios makes no allowance for differences among countries in their capacity to raise taxes. Now, the question whether a given country has the capacity to raise taxes, and if so by how much, cannot in any case be given a precise answer, and any final pronouncement could be advanced only after a detailed examination of the particular conditions of each instance. However, a prima facie case for the existence of tax potential could be made on the basis of a proper comparison with other similar countries.

The approach of Lotz and Morss 78 and, following them, of the UNCTAD study 79 has been to assume that the explanatory variables included in their equations are the major factors affecting taxable capacity and then to interpret the unexplained residuals as reflecting relative tax effort. The equation yields for each country a “predicted” or “computed” tax ratio. It could be argued that since the equation suggests the average relationship between the tax ratio and the explanatory variables affecting capacity, the predicted ratio is what the country concerned would have had if it had made the average tax effort. Lotz and Morss went on to rank countries according to the percentage differences between the actual and predicted tax ratios. The same procedure has been adopted in the UNCTAD study.

It is well to point out the limitations of this approach. First of all, there is an implicit assumption that all the major factors affecting taxable capacity have been taken care of by the included variables. (However, it may be said that even if this assumption does not hold, the comparison on the basis of adjusted tax ratios is likely to be more meaningful or accurate than that on the basis of unadjusted tax ratios.) Second, care must be taken to see that only “objective” factors that can be considered legitimately as “beyond the control” of the governments are included. Third, it is likely—indeed very probable—that rankings will change with the equations used. A new equation might be brought in to “supersede” one that holds the field at a given time. Faced with alternative equations, one may be tempted to choose that which gives the best statistical fit. This simple way out should not be adopted because what is more important is the economic rationale of the hypotheses embodied in the alternative equations. Judgment inevitably has to be exercised in deciding which of the formulations represents the best combination of economic reasoning and statistical merit. Finally, the validity of this approach is predicated on the assumption that the tax “effort” factors, which are held to be represented by the unexplained residuals, are largely independent of the “capacity” factors.

Given these qualifications and assumptions, the indices of relative tax effort derived through this method have to be interpreted with due caution. Importance is to be attached not so much to the rank occupied by a given country as to the magnitude of the ratio of actual to adjusted tax ratios for that country.80 If this index should be appreciably below unity, say, less than 0.80, a prima facie case could be said to exist for thinking that there has been less than average tax effort, and a more detailed examination of the tax system would be warranted. In any such examination, qualitative allowances could (and should) be made for any major factors particularly relevant for the given country, which have not been included in the estimating equation.

The indices of relative tax effort derived on the basis of equation (7) as well as the actual tax ratios for 49 countries in the sample are presented in Table 9.81 If the actual tax ratios are considered, 21 of the 49 countries have tax ratios above the average (14.0 per cent); if tax effort indices are taken into account, roughly about the same proportion of countries shown as making above average tax effort is not very different (25 out of 49). But the relative positions of several countries are changed considerably if ranking is done on the basis of indices of tax effort. However, there are several exceptions. Brazil, Ivory Coast, Senegal, the United Arab Republic, Tunisia, and the Democratic Republic of Congo are among the 11 countries with the highest tax ratios and also emerge as part of the top 20 per cent in terms of tax effort indices. The tax effort exercise suggests that in these countries a high tax ratio has been achieved by a fair amount of effort. At the other end of the scale, Nepal, Indonesia, Bolivia, Guatemala, Rwanda, and Pakistan have low values both for tax ratios and for tax effort indices (within the lowest ten countries in both lists). In these countries, the presumption seems strongest that there is scope for further tax effort, should they desire to have higher tax levels.82

Table 9.

Selected Developing Countries: Tax Ratios and Indices of Tax Effort, 1966–68 1

article image
article image

Somalia has been excluded from the sample, owing to the unavailability of the economic data needed for the analysis of tax effort.

Ranked according to tax ratios.

Ranked according to index of tax effort, defined as the quotient of the actual tax ratio divided by the tax ratio estimated according to equation (7). Tax effort indices are carried to three decimal places in this table in order to show the slight differences between the indices of different countries.

Strictly speaking, tax effort is a process; it takes several forms, including reform of existing taxes, improvement in administration, and introduction of new taxes. All these steps necessarily require time to plan, legislate, and implement. Countries that for historical or other reasons started out with a low tax ratio a decade or so ago might have undertaken considerable effort to raise their respective tax ratios but may not yet have reached even the average level of taxes in developing countries. Tax effort, therefore, should be considered also in the dynamic sense of comparing changes in the tax ratio over time. Thus, even if a country has a low tax effort in the static sense, the question may be asked whether it has made efforts over a period of time to increase tax revenues. For this purpose, it seems best (for reasons explained earlier) to compare the income elasticities of total taxes.

In sum, the tax performance of developing countries can be looked at or appraised (a) on the basis of indices of tax effort in the static sense and (b) on the basis of income elasticities of total tax revenue, which may be called the indices of effort in the dynamic sense. While the combination of indices raises the familiar problem of “weighting,” a comparison on the basis of any single index may be misleading, especially in view of the limitations inherent in the exercise. Besides, one could distinguish between two main purposes to consider the question of relative tax effort. The first is to get an idea of the potential for tax increase in a given country through comparison with other developing countries; for this purpose, it would be necessary to look at the static index of tax effort and the actual tax ratio. The other main purpose is to assess the tax performance of a country as part of overall development performance in the context of, say, considering the “eligibility” for further aid or the distribution of assistance among countries; for this purpose, it would be necessary to take into account both the static and dynamic indices.

Table 10 brings together the three quantitative measures of tax effort for the 27 developing countries in the small sample. Countries are ranked in the same way as in Table 1. These measures can be compared or contrasted in different ways depending upon the purpose one has in view. A few illustrations only may be offered. First, countries with a low value of static index may have high values of elasticity. There are three countries in the sample, namely, Paraguay, Honduras, and Korea, whose static index is less than 1; but they all have elasticity coefficients equal to or greater than 2 and rank high in terms of that measure. It would be fair to argue that a distinction should be made between such cases and those of Costa Rica, Trinidad and Tobago, and Ecuador, which have both low elasticity coefficients and low values of static index. And all of them except Trinidad and Tobago also have lower than average tax ratios.83

Table 10.

Selected Developing Countries: Tax Ratio, Income Elasticity of Taxes,1and Static Index of Tax Effort, 1953–55 to 1966–68

article image

Excluding social security contributions.