Abstract
Tax evasion is universal. It depends on the economic and tax structures, types of income, and social attitudes. The theory of tax evasion has limitations since it rests solely on attitudes toward risk, with full information regarding the tax administration’s behavior. Methodologies for estimating tax evasion include estimating the underground economy and comparing declared taxes with potential tax revenue calculated from national accounts. Measures to address tax evasion include use of withholding, presumptive and minimum taxes, selective auditing, penalties, and cross checks between taxes.
Tax evasion is a universal phenomenon. It takes place in all societies, all social classes, all professions, all industries, and all economic systems. Two thousand five hundred years ago, Plato was writing about it, and the aging Ducal Palace of Venice has a stone with a hole in it, through which people once informed the republic about tax evaders. The only surprise is how little attention this phenomenon has received, especially in the United States, until recent years. For example, there is no reference to it in the index to Richard Goode’s (1964) classic Individual Income Tax, nor in Richard Musgrave’s (1959) The Theory of Public Finance, nor in Joseph Pechman’s (1966) Federal Tax Policy. These authors either did not think that tax evasion was important or opted to ignore it.
In recent years, growing attention has been paid to tax evasion. In the United States, awareness may have started with the political problem of the rising fiscal deficit, which some began to argue could be solved by reducing the so-called tax gap, rather than by raising tax rates or cutting public spending.1 Because of its policy of reducing tax rates and its inability to reduce public spending, the Reagan administration, in particular, promoted the idea that the fiscal deficit could be lowered by reducing or eliminating the tax gap. In other countries, concern over tax evasion was in part prompted by a growing preoccupation with horizontal equity. The realization that people with similar incomes often end up paying very different taxes because of different opportunities for tax evasion has led many governments to worry about the implications of tax evasion. Also, a growing concern about underground economic activities, which are often the other face of tax evasion, and how these activities affect economic policies led in the 1980s to increasing interest in the extent of tax evasion.2 This is certainly true in Latin America, where the authorities, after introducing major tax policy reforms, have devoted more resources to measuring and reducing the evasion of both income and consumption taxes.3
In the paper that follows, we survey some of the sources of tax evasion and recount how economists have attempted to provide a theoretical underpinning to them. The discussion also addresses the limitations of this theoretical literature. Next we review the role of tax administration and sanctions in limiting tax evasion and briefly mention the relation between society at large and tax evasion. An Appendix lists some of the methodologies developed for estimating tax evasion.
I. Theoretical Underpinnings
The theoretical literature on tax evasion has progressed in many directions. An August 1993 conference of the International Institute of Public Finance, held in Berlin, revealed that many of the advances in the related fields of tax evasion, corruption, and the underground economy have been made using the theory of constrained maximization under uncertainty, game theory, probit analysis, catastrophe theory, and the like. However, the assumptions and results remain relatively restrictive. This section, after describing some of the sources and implications of tax evasion, attempts to assess the main thrust of the literature.
Sources and Implications of Tax Evasion
Tax evasion is practiced in different forms. Tax evaders may not declare income; may underreport income, sales, or wealth; may overreport deductible expenses; may smuggle goods or assets; or may undertake some other deception. The variety of tax evasion is truly remarkable, and taxpayers are always finding new ways to purposely reduce their tax burden.4 Many authors have reviewed these matters, including Sisson (1981) and Richupan (1987), and more recently Cowell (1990) and Webley and others (1991). Notably, empirical findings are often confidential in nature, as they result from IMF technical assistance to a country or the work of a ministry of finance.
The opportunity for tax evasion varies across sectors, which may lead to social turmoil. In Italy, for example, salaried workers have publicly demonstrated in large numbers for a reduction in tax evasion by independent professionals and other groups. In general, tax evasion is easier for independent contractors, professionals (such as doctors, lawyers, and architects), and those engaged in agricultural activities. There is increasing evidence that multinational enterprises can also reduce their tax burden through judicious transfer pricing.5
Tax evasion has much to do with the structure of an economy. The more atomized is production, the more likely it is that tax evasion will flourish. A country where much of the production takes place in large enterprises is unlikely to have much tax evasion. However, a country where much of the economic activity takes place in small shops and small farms, or is conducted by single individuals, is likely to experience a lot of evasion.
Tax evasion is also strictly connected with the structure of the tax system and is likely to vary with the use of different tax bases. For example, in the case of income taxes, evasion is likely to vary between dependent and nondependent income sources, as well as between large, small, and multinational enterprises. In the case of sales taxes, it is likely to be connected with the underreporting of sales or the overreporting of purchases. In theory, at least, tax evasion is connected with the accounting concepts of tax liabilities. When a country relies on presumptive concepts of taxation, tax evasion is likely to be more limited, unless the assets on which the presumptive estimate of the tax payment is based can be hidden. The number of taxes will also influence tax evasion. At times, governments introduce additional taxes in order to neutralize the fosses connected with tax evasion. However, an increase in the number of taxes produces inefficiencies in the tax system and facilitates the search for new ways to avoid paying taxes.
The policy implications of tax evasion are quite different depending on whether evasion is an individual or a social phenomenon. A single tax evader in a country of honest taxpayers typifies the behavior of that individual only. However, a tax evader in a country where tax evasion is a national sport is a different phenomenon, in that tax evasion begins to have implications for both the horizontal and vertical equity of the tax system. It also has implications for the efficiency of the tax system and even for the market framework. For example, it is impossible to have pure competition when some of the sellers can evade taxes and others cannot. In this case, the former will be able to undersell the latter.
Tax evasion also affects the productivity of the tax system by reducing the amount of revenue that can be raised under the statutory system. It affects the attitude of citizens toward their government, often building cynicism about the role of the public sector. Often it affects even the statutory system in the sense that the tax laws begin to anticipate the tax evasion by particular groups and try to penalize it by increasing the tax rates for those groups. This often results in increased horizontal inequity since not all the taxpayers in those groups behave alike.
Theory of Tax Evasion
Since Allingham and Sandmo (1972) wrote their classic theoretical paper on tax evasion, the problem of tax evasion, as seen from the taxpayer’s point of view, has been discussed as a kind of game theory.6 In this framework, the taxpayer is faced with the decision of whether or not to evade taxes. In other words, the decision about whether to pay the tax becomes similar to playing a lottery, in that one is free either to buy or not to buy a lottery ticket. For a rational individual, the choice will be based on the expected gains or losses associated with the decision. The objective is to maximize the utility of the taxpayer.7
The benefit derived from tax evasion is related to the expected value of the money (and thus to the utility of the money) that the individual does not pay. The cost of tax evasion is connected to the probability of being caught and the consequences of this outcome. These consequences, in the Allingham and Sandmo model, are associated with fines that can considerably exceed the original tax due. Nonetheless, the probability that the individual will pay these fines depends on the probability of being caught, and that probability can be very low.
The Allingham and Sandmo theory has some important implications for tax administration. In fact, the theory implies that tax evasion can be reduced either by increasing the penalties associated with it or by increasing administrative expenses, assuming that this increase raises the probability that the tax evader will get caught.8 In an extreme interpretation of the AHingham and Sandmo theory, it has been argued that the penalties should become so high that, at the limit, the tax evader who gets caught would be hanged and the cost of administration would approach zero, as would the probability of being caught.
The theoretical and practical limitations to the literature on tax evasion have not received the attention they deserve. But some have been discussed by various writers. A first limitation has to do with risk aversion, which may vary among individuals and may depend on the level of the taxpayer’s income or wealth. In more recent theoretical advances, the taxpayer’s behavior toward tax compliance turns entirely on his or her attitude toward risk.9 For example, treating tax evasion in the context of intertemporal choice models, Banerji (1991, p. 98), concludes,
Is there a more subtle way of enforcing compliance without such elaborate calibration—by simply increasing the risk of detection for the evader, and thereby making him or her switch from the riskier asset to the safer one of declared income? Unfortunately, this plan would work with certainty only if we were willing to assume that all possible evaders in the economy had constant absolute risk aversion, i.e., that their willingness to take risks did not depend upon their level of income or consumption.
A second limitation has to do with the use of penalties applied to those evaders who get caught. In other words, many tax evaders who should be penalized do not get caught and so are not affected by the penalties. This raises the question of whether the judiciary system and the community at large are willing to penalize fully the few unlucky individuals who get caught when many more individuals are committing the same offense but are not being punished. Anecdotal evidence from many countries indicates that the judiciary system is unwilling to apply the penalties fully under these conditions. This means that one basic conclusion of the theoretical literature is unlikely to hold if the penalties actually imposed differ from those on the books.
Third, the theory assumes that the taxpayers know precisely the probability of being caught and the penalties they will endure, so that they make the cost-benefit calculations. However, tax administrations often keep this information highly confidential so that for most, if not all, taxpayers the probability of being caught is an unknown. And the penalties may be highly uncertain.10
Fourth, the theory ignores the costs in terms of embarrassment, loss of self-esteem, and social status experienced by those who get caught. These costs vary from society to society and from individual to individual. In a society where tax evasion is condoned because of the unpopularity of the government, tax evaders may be admired and the social costs associated with tax evasion may be low or even negative. In a society where tax evasion is taboo, these costs can be very high.
Finally, many countries rely on means testing, based on declared income, for determining access to many government benefits, such as food stamps, free health care, free education or scholarships, and so forth. Therefore, the advantages from tax evasion may far exceed those measured by the nonpayment of the tax.
Role of Penalties and Amnesties
A few comments on the penalties themselves would be appropriate. Some of these comments have relevance for the theoretical literature on tax evasion. The higher are the penalties, the more probable it is that they will not be applied. If the high penalties lead to a reduction in the cost of administration, this reduces the probability of detection and thus the number of cases requiring the imposition of penalties. Many societies feel uncomfortable about singling out and punishing particular individuals, almost by lottery, when many others have committed the same offenses. Second, for the penalties to be effective, they must be applied quickly.
A penalty that is delayed for years, because of appeals on the part of the taxpayer, is unlikely to become as effective a deterrent as one that is applied immediately. In some legal systems, for example, the Italian and Tunisian ones, it has been possible to postpone for many years, through appeals, the application of the penalties.11 Moreover, the impact of penalties on tax compliance may not always be great. For example, using a model of varying attitudes toward risk and applying econometric estimation techniques to Mexican data for 1982–89, Dunn (1992, p. 14) concludes
… large changes in the odds of being detected and the penalty for illegal evasion are required to even modestly alter compliance … a doubling of the fines for tax evasion would increase declared taxable income by about 10 percent. Similarly, a large increase in the number of audits would achieve only a modest rise in compliance.
Of course, in the prepenalty period, the appeal may be successful or a tax amnesty may come along.12 Appeals mechanisms and tax amnesties considerably confuse the theory, which assumes that the probability of the penalty being applied and the penalty itself are known and precisely defined. The theory is also affected by the existence of administrative corruption. If the individual who gets caught can bribe some tax officials, and if the bribe is less than the penalty, then the theory becomes ambiguous. The tax amnesties used in some countries also have important implications for tax evasion because in many ways they encourage tax evasion, at least over the longer run, and by so doing they have an impact on the equity of the tax system, tax revenue, and the tax administration. For example, using a game-theoretic approach to the analysis of tax amnesties, Stella (1989, p. i) concludes
… while in general it may be correct to impose a reduced penalty on individuals who voluntarily disclose tax evasion, short-lived amnesties of the type most frequently observed in practice are unlikely to generate significant revenue when judged against the potential danger of reducing future tax compliance.
Also, analyzing the sustainability of the revenue intake from tax amnesty programs in different countries, including Argentina, Colombia, and India during the 1980s, Uchitelle (1989, p. 53) concludes
… most of the programs have not led to a widening of the overall tax base, and many have failed to produce even very large one-time revenue gains.
II. Tax Administration and Tax Evasion
The tax administration of a country plays an important role in the extent to which tax evasion prevails. To the best of our knowledge, the theory of the firm has not yet been applied to the activities of a tax administration.13 But, a tax administration is not very different from a firm, though it should be compared to a monopolistic firm. The tax administration has a given budget assigned to it by the state and with this budget its task is to maximize an output—tax revenue—taking into account certain important constraints. The allocation of resources within the tax administration is obviously important for determining this output. Under optimal conditions the tax administration would not be able to increase output by shifting resources across various activities, such as assessment, collection, and auditing.14
Size and Targeting of Administrative Resources
Some of the constraints on the tax administration are imposed by tax policy; others are objectives that the tax administration needs to take into account (such as the equitable treatment of taxpayers). How much revenue a country should allocate to the administration of taxes remains a largely unexplored subject. There is a remarkable variance among countries in both the amount of resources allocated to tax administration as a share of national income and the amount as a share of total tax collection by the tax administration.15 It should not be concluded that in either case a low share is necessarily good. In fact, a country that wants to minimize collection costs can simply go after the taxes that are easiest to collect and collect them from the largest taxpayers. Alternatively, it could focus its activity on the largest city or cities. This behavior would condone a lot of tax evasion and generate tax revenue in a way that would be far from optimal. It would also conflict with other objectives of taxation, such as neutrality and equity.
A tax administration should be careful to minimize not only the explicit costs it bears (its collection costs) but also the costs borne by taxpayers and the economy. These latter costs do not show up in the administration’s balance sheet and often tend to be ignored. They are essentially welfare costs, compliance costs, and perhaps “good relations” costs.
Welfare and Compliance Costs
The welfare cost per dollar collected can be defined as the excess cost to society of collecting $1 of tax revenue. These are the costs that have attracted the attention of economists. These costs have been estimated by various authors for the United States, such as Ballard, Shoven, and Whalley (1985a, 1985b), Hansson and Stuart (1987), and others. They have shown that the marginal dollar raised by the U.S. tax administration may have cost the country more than $1.50. Usher (1986) has discussed the marginal cost of taxation in the presence of tax evasion. Clearly, the above estimates indicate that the tax system is far from optimal. However, attempting to make the system optimal may raise other costs, such as administrative and compliance costs. These latter costs have received much less attention by economists. There is still no literature to deal with the administrative and compliance costs of trying to pursue “optimal” tax policies.16 However, some recent literature has been trying to assess the implications of tax evasion for optimal taxation (see Cremer and Gahvari (1993)).17
The compliance costs are more closely associated with the behavior of the tax administration and more likely to be connected with tax evasion. Compliance costs refer to the cost to the taxpayers—in terms of lost time, added stress, payments to tax accountants and lawyers, trips to the tax office, and so forth—associated with a given tax payment.18 In some countries, and for some taxes, these compliance costs can be enormous, especially if the taxpayers have to stand in line for hours, sometimes days, perhaps several times a year in order to meet their tax obligations.19 They are also likely to be extremely high when the tax laws are so complicated that the taxpayer has to rely on an expert’s advice or, in the case of enterprises, has to hire experts whose only function is to comply with the tax obligations. There have been reports from Latin America that even relatively small enterprises have sometimes had to establish sizable tax departments simply to find their way through the jungle of fiscal laws and regulations. When this situation prevails, the tendency to evade taxes rises. There is a direct and positive relationship between the size of tax evasion and the cost of compliance. It should also be noted that when firms create tax departments to comply with existing tax obligations, those same departments will be used to scrutinize the taws for any possible loopholes or for any ambiguity that might justify tax avoidance.
Public Relations
Let us now turn briefly to what could be called“good relations” costs, which are essentially the public relations activity of a tax administration. This public relations role is connected with the way in which tax administrations are organized: the number of employees and their use, the level of their salaries, the quality of their working conditions, and the controls that the tax administration has on the behavior of tax inspectors. These controls are necessary to minimize or eliminate the possibility that these inspectors, or other tax administrators, will use their positions for their own benefit.20
A tax administration that wants to improve taxpayer compliance and minimize tax evasion must be available to the taxpayer who needs information, forms, specific instructions, and so forth. It must show courtesy toward the taxpayers since resentment is likely to create a lower propensity to pay taxes. It must also show punctuality in sending refunds to those who have overpaid since a taxpayer is likely to underpay if he or she might have to wait years for a refund.
Withholding, Presumptive and Minimum Taxes, and Cross Controls
Collection systems are also important for minimizing tax evasion. There is now overwhelming evidence that evasion is minimized whenever there is withholding at source. In the United States, for example, the difference in tax evasion between independent contractors, for whom there is no withholding at the source, and dependent workers, whose taxes are withheld by their employers, is enormous. The same evidence is available for taxes on interest income and dividends.
Various countries have tried to minimize evasion by resorting to minimum taxes or to presumptive methods of taxation. In these presumptive methods now used in many countries, the government tries to assign a particular income to taxpayers based on their standard of living, the value of the houses in which they live, the value of the cars they drive, and so forth.21 It also tries to estimate, for example, the value added of a company based on its sales statistics and other criteria (employees and floor space, for example). The minimum income tax of a company or individual can be based on their gross assets, a system that has been introduced in Argentina and Mexico.
Tax administrations may also utilize various instruments to limit tax evasion. For example, cross controls between the information available to the tax administration, the social security institution, and the customs administration can play an important role. The assignment of a taxpayer identification number, to be used in this cross control, is extremely important since it facilitates the use of computers. Other such instruments include (1) the government’s ability to access the bank accounts of individuals or companies, (2) detailed audits of taxpayers, and (3) reporting requirements by employers or by those who make payments.
Social Ethics
Before leaving the section dealing with the role of the tax administration, it may be worthwhile to refer to another relationship, that between society at large and tax evasion. Tax evasion prospers when society condones it. In a society that does not condone tax evasion, the phenomenon remains isolated and concerns relatively few individuals. When society condones it, however, the phenomenon becomes much more widespread. Citizens at large should have a responsibility to prevent tax evasion. Since tax evasion is often facilitated by the acquiescence of some citizens in the tax-evading behavior of others, laws should penalize not just the tax evaders but also those who collaborate either passively or actively in tax-evading activities. For example, in many countries, the tax evasion of professionals, such as doctors or independent contractors, is facilitated by their requests that customers pay them in cash or accept invoices that underestimate the payment. In addition, those who govern must obey the tax laws. When a country’s leadership engages in tax evasion or similar activities, it sends an unmistakable signal that noncompliance is acceptable.
Penalties
As can be anticipated from the preceding discussion, the severity of the penalties has some impact on the extent of tax evasion. Taxes may be paid in arrears without any intention to evade them, especially if the interest charges are low. Usually, interest charges and pecuniary penalties are applied to any tax in arrears that does not reflect tax-evading motivation. Tax evasion or fraud, however, is a more serious matter and, at least in the tax laws, carries much heavier sanctions.
Penalties for Tax Arrears
Usually, the amount of interest charged on taxes paid in arrears is calculated using either a fixed percentage point above some key central bank rate, or above the average of bank rates, or a specified percentage of the amount due in taxes for each month in arrears, up to a maximum amount. In some countries, additional surcharges are also applied.
Penalties on taxes paid in arrears vary depending on whether the cause is late filing of returns, failure to file returns, or filing incorrect returns. In the case of taxes withheld at source, penalties depend on the type of infraction. For example, penalties differ depending on whether the correct amount has been withheld or whether the amount withheld has been surrendered to the tax authorities. In all cases, repeated offenses or offenses not corrected or admitted within a specified time period are subject to higher penalties. Sanctions often take the form of a percentage of the tax due and range from 25 percent to 100 percent; several countries also charge penalties fixed in nominal terms.
Sanctions for Tax Evasion and Tax Fraud
Sanctions for tax evasion and tax fraud are much more severe (many times the amount of the defrauded amount) and include possible closure of an enterprise for a specified time and jail sentences, ranging from a few months to several years. Giving the tax administration the power to close establishments for a few days, without the possibility of appeal, has been an effective deterrent to tax evasion in Argentina and other Latin American countries.
III. Concluding Remarks
This paper has examined the factors that give rise to tax evasion. Tax evasion varies by sector (agriculture, industry, and commerce), organization of production (small trader or large company), and type of economic agent (salaried, self-employed, or capital owner). It is also affected by social ethics and the standards set by those that govern. Given these standards, it is further affected by the potential taxpayer’s attitude toward risk.
Tax evasion affects the horizontal and vertical equity of a tax system, as well as the efficiency of the market and the tax system. It certainly affects the revenue productivity of the tax system. Unchecked or poorly controlled tax evasion builds cynicism about the public sector. It tends to complicate the tax structure as legislators try to anticipate tax evasion through tax legislation. The use of effective and quickly applied penalties to counter tax evasion has an impact on its extent and spread. However, their application does not necessarily imply even a second-best solution for the correction of inequities or for the efficiency of the competitive mechanism if many tax evaders do not get caught or remain unaffected by penalties.22
The theoretical foundation for modeling tax evasion remains wanting. It is too simple to be of much practical use. The theory relates the taxpayer’s behavior toward tax compliance to his or her attitude toward risk, but ignores other factors that influence tax evasion. The theory assumes that taxpayers know precisely the probability of being caught and the consequences of such an event; however, tax administrators often keep this information confidential and the consequences of being caught may not be fully predictable.
Estimates of the evasion of income and consumption taxes have been selectively reported in the published literature for many countries. More information of a confidential nature exists as a result of exercises carried out by tax authorities or by technical assistance experts from international organizations. The methodologies utilized leave much to be desired because of the lack of data and, more important, because of what the data are not able to capture. The data may only partially capture the effects of tax evasion while including the effects of other leakages (such as legitimately used tax incentives or deductions whose total effect may be difficult to remove). Thus, it would not be prudent to base economic policy solely on the results that emerge from these estimations.
Given their limitations, methods of estimation include the matching of information from tax declarations with either national accounts data or survey (or sample) data blown up to population levels. Because of the unreliability of surveys (respondents may not reveal the truth regarding tax evasion) and because of their cost, the national accounts approach is more commonly used. If the objective is to estimate evasion of the value-added tax (VAT), however, a national input-output framework has to be used because of the VATs method of collection at different stages of production, some of which may be exempted from the VAT base (or taxed at different rates). An indirect way of measuring tax evasion has been to estimate the extent of the underground economy and, once that has been done, to estimate the taxes that should have been paid. It appears from the published literature that perhaps a third of potential tax revenue may be lost to evasion in selected Latin American and Mediterranean countries. Some estimates indicate even higher percentages, These estimates, however, must be taken with a grain of salt since they would, at times, imply very high tax burdens in the absence of tax evasion. The Appendix reviews a number of the estimation techniques discussed here.
If tax evasion is high, the role of tax administration becomes doubly important. The size of tax administration resources, the main target groups (large enterprises or all taxpayers), the efficiency with which the resources are utilized (collection costs), the ease with which taxpayers can pay taxes (compliance costs), the relation between the tax administration and the taxpayer (good public relations rather than the spreading of fear), and the methods of tax collection (withholding, presumptive taxes, minimum taxes, and cross controls) all play a role in determining the level of tax evasion.
Finally, one interesting aspect of evasion is its counterpart on the expenditure side of the budget, which has not received the attention that tax evasion has.23 While tax evasion is the nonpayment of taxes duly owed to the government, the equivalent phenomenon on the expenditure side is the abusive receipt of government payments. In a way, one finds a parallel in the comparison between indirect taxes and consumer subsidies, one being the reciprocal of the other. Activities connected with the illegal receipt of government expenditures may involve corruption. For example, the receipt of a percentage of government contracts, the receipt of pensions not deserved (for example, by claiming disability when one is not disabled), the payment of wages to so-called ghost workers (a phenomenon common in several developing countries), the taking of leave on the basis of fictitious illnesses, and so forth. This is the flip side of tax evasion: the government loses when taxes are not paid, but it also loses when it makes payments it should not.
APPENDIX
Estimating Tax Evasion
In recent years, many scholars and governments have attempted to measure the size of tax evasion in particular countries, either for specific taxes or for the whole tax system. The measurement of tax evasion is obviously fraught with difficulties. Many of these difficulties have to do with the fact that the information available is limited and often unreliable. However, a more philosophical difficulty is often not acknowledged—namely, the problem that the statutory tax system has been“contaminated” by the existence of tax evasion. In other words, it is not the system that would exist in the absence of tax evasion: statutory rates have often been increased to compensate for the revenue losses associated with tax evasion.24 But if this is true, then when one uses the current statutory rates to measure tax evasion, one exaggerates the size of the evasion, since the rates would have been lower if the evasion had not been there.
Various methods have been used to measure tax evasion. Some of these try to measure it directly, some indirectly. Among the direct methods one can identify are(l) the use of the national accounts, (2) the use of direct controls, (3) the use of household budget surveys, and (4) the use of direct surveys of taxpayer behavior. The indirect methods are largely related to estimates of the underground economy. Once the size of the underground economy has been measured, the extent to which the existence of the underground economy has implied tax revenue losses to the government must be assessed. In other words, undeclared income or some other unreported tax base must first be measured. Subsequently, an estimation of the unpaid tax must be made.
National Accounts Method
Perhaps the commonest and most often used method for assessing the size of tax evasion is comparing the estimate of a particular tax’s base made by the national accounts authorities with the base reported to the tax authorities after making appropriate adjustments. An early study that attempted this technique for several industrial countries was Tanzi (1969). A similar one for Argentine data was Herschel (1978). The Internal Revenue Service of the United States has been following this approach routinely for the income tax; various other authorities have used it for measuring the base of the VAT and other taxes. Given that the VAT is collected at various stages of production, careful use of information based on a sectoral input-output table would be necessary. This was initiated by Aguirre and Shome (1988) for Mexico. They developed a methodology for constructing the VAT base on a sectoral basis while allowing for the differential tax rates of the VAT. It was applied by Serra (1991) for Chile and clarified by Mackenzie (1992) methodologically. It has since been attempted in various unpublished technical assistance studies by IMF staff and is being used by technical units in the ministries of finance in many Latin American countries.
The difference between the base as reported to the tax authorities and the base as estimated by the national accounts authorities gives an indication of unreported income. If the tax is a fully proportional one, this unreported income automatically and directly provides an estimation of the unpaid tax. If the tax is progressive, as would be the case with income taxes, then the estimation of the unpaid tax becomes more complex since one would have to make assumptions about the effective tax rate at which the unreported tax base would have been taxed. In other words, the unreported income must be allocated among the tax brackets. It would also be necessary to reinstate, into the information based on income tax declarations, the various exemptions and deductions at the different tax brackets in order to make it comparable to the national accounts data.
Individual Income Tax
In the context of the individual income tax, an actual framework for nonwage earners may be described as follows. To declared personal income, adjustments should be made for those components of income that are included in the concept of income in the national accounts but are deductible for tax purposes.25 These include personal exemptions, deductions, investment allowances, and other deductible direct taxes that have been paid. The adjustments need to be made for individual tax brackets if the tax structure is progressive. The result of the exercise would be a series for gross taxable declared income (by income class). A comparison with gross taxable income from the national accounts would yield an estimate of undeclared nonwage income.
Tax evasion among wage earners is often limited because of withholding at the source and because wages are an important cost to enterprises. To claim this cost, they need to report the wages paid. However, it may also be difficult to estimate tax evasion by wage earners. Information on tax withheld by employers may not be readily available, since it is not the form in which wage income is usually declared for tax purposes. It may be even more difficult to obtain this kind of information by bracket or sector. Small and medium-sized firms that do not pay profit taxes would also tend to underreport tax withheld on wage income or may actually withhold less income than required by law. However, the overall revenue loss from this source should not be significant owing to the small firm size. In general, estimates of tax evasion from wage earnings would be attempted through sampling techniques.
Corporate Income Tax
Techniques similar to those used for the individual income tax may also be applied to the corporate income tax, adjusted for the kinds of deductions and incentives that apply specifically to the corporate sector. The task is not easy, however, since corporate sector tax incentives would have to be accounted for in addition to those surrounding the individual income tax. These incentives can be used legitimately or otherwise, making the task of adjustment difficult. One redeeming feature in the case of corporate income tax evasion is that the corporate form in developing countries is primarily confined to large and easily identifiable firms. It is common in developing countries for tax administrations to establish special units to control large taxpayers, mainly big—often foreign—corporations.
VAT
The VAT has emerged as the most important revenue earner in many countries, and attempts to estimate evasion of the VAT have become relatively commonplace over the past few years. The widest VAT base is all purchasable goods in the economy: GDP plus imports minus exports. Thus, the starting point is again the national accounts. However, the estimate can be made either from the expenditure side or from the production side.
The expenditure method can be summarized as follows. To total domestic expenditure (including imports) add net private expenditure from abroad and subtract nontaxed expenditure (typically, government expenditure on wages and salaries, fixed capital formation—except private expenditure on new houses—and changes in inventories) to obtain taxable expenditure. Adjust for taxes on expenditure to obtain adjusted taxable expenditure. Further, subtract exempted expenditures (typically, the financial sector, nonprofit and social organizations, small businesses below a legally defined threshold, and gross rents paid) but add back taxable inputs and the capital purchases of exempt sectors to obtain the potential VAT base.
The VAT base calculation from the production side is similar, except that zero-rated exports have to be subtracted and imports added.26 It is more convenient to use the production side method when the VAT contains more exemptions by economic sector than by product for final consumption. Sectoral data are more amenable to production side estimates, while exemptions for particular products are more amenable to expenditure side estimates. Further, given the nature of the VAT—that is, collection based on stages of production—sectoral data are again more amenable to base calculations.
Using the above-mentioned methodologies, IMF staff have made some interesting inferences regarding the relationship between the VAT rate and the ratio of VAT revenue to GDP. For example, Table A1 introduces the concept of a“revenue productivity ratio—the amount of revenue raised per point of the VAT rate.” The last column of Table A1 indicates that the average amount of revenue per point of the VAT rate is 0.37 percent of GDP for the 22 countries in the sample. A country whose ratio approaches 0.5 percent could be said to be performing at a high VAT effort.
Some economists have sharply criticized the national accounts approach on the grounds that if tax evasion is significant, the national accounts are likely to be underestimated. Therefore, the calculation described above may be meaningless in those cases. However, these economists fail to realize that the information that the national accounting offices receive from the tax authorities often contributes very little to the estimation of the national accounts since the national accounts authorities often rely on other methods for measuring production. For example, the agricultural sector’s income is often underreported to the tax authorities because of tax evasion, but the estimations for the national accounts are made on the basis of sampling or surveys of directly observed average productivity per acre and the average prices at which the crops are sold.27
Value-Added Tax, 1988
(Countries with single-rate VATs)
Value-Added Tax, 1988
(Countries with single-rate VATs)
| Actual VAT | |||
|---|---|---|---|
| revenue: | Revenue | ||
| Country | GDP ratio | VAT rate | productivity ratio |
| Bolivia | 2.6 | 10.0 | 0.26 |
| Chile | 8.8 | 16.0 | 0.55 |
| Costa Rica | 3.8 | 10.0 | 0.38 |
| Denmark | 9.5 | 22.0 | 0.43 |
| Dominican Republic | 1.6 | 6.0 | 0.27 |
| Ecuador | 2.5 | 10.0 | 0.25 |
| Finland | 8.4 | 19.1 | 0.44 |
| Grenada | 4.5 | 6.0 | 0.75 |
| Guatemala | 2.4 | 7.0 | 0.34 |
| Guinea | 0.6 | 13.6 | 0.05 |
| Haiti | 1.8 | 10.0 | 0.18 |
| Indonesia | 4.5 | 10.0 | 0.45 |
| Israel | 9.8 | 15.0 | 0.65 |
| Korea | 3.3 | 10.0 | 0.33 |
| Madagascar | 1.5 | 15.0 | 0.10 |
| Mauritius | 2.1 | 5.0 | 0.41 |
| New Zealand | 6.7 | 12.5 | 0.54 |
| Norway | 9.4 | 20.0 | 0.47 |
| Panama | 1.1 | 5.0 | 0.22 |
| Peru | 2.0 | 13.0 | 0.15 |
| Taiwan | 2.6 | 5.0 | 0.52 |
| United Kingdom | 6.0 | 15.0 | 0.40 |
| Average | 4.3 | 11.6 | 0.37 |
Value-Added Tax, 1988
(Countries with single-rate VATs)
| Actual VAT | |||
|---|---|---|---|
| revenue: | Revenue | ||
| Country | GDP ratio | VAT rate | productivity ratio |
| Bolivia | 2.6 | 10.0 | 0.26 |
| Chile | 8.8 | 16.0 | 0.55 |
| Costa Rica | 3.8 | 10.0 | 0.38 |
| Denmark | 9.5 | 22.0 | 0.43 |
| Dominican Republic | 1.6 | 6.0 | 0.27 |
| Ecuador | 2.5 | 10.0 | 0.25 |
| Finland | 8.4 | 19.1 | 0.44 |
| Grenada | 4.5 | 6.0 | 0.75 |
| Guatemala | 2.4 | 7.0 | 0.34 |
| Guinea | 0.6 | 13.6 | 0.05 |
| Haiti | 1.8 | 10.0 | 0.18 |
| Indonesia | 4.5 | 10.0 | 0.45 |
| Israel | 9.8 | 15.0 | 0.65 |
| Korea | 3.3 | 10.0 | 0.33 |
| Madagascar | 1.5 | 15.0 | 0.10 |
| Mauritius | 2.1 | 5.0 | 0.41 |
| New Zealand | 6.7 | 12.5 | 0.54 |
| Norway | 9.4 | 20.0 | 0.47 |
| Panama | 1.1 | 5.0 | 0.22 |
| Peru | 2.0 | 13.0 | 0.15 |
| Taiwan | 2.6 | 5.0 | 0.52 |
| United Kingdom | 6.0 | 15.0 | 0.40 |
| Average | 4.3 | 11.6 | 0.37 |
Sampling Method
The second method of estimation, the tax compliance measurement method, has been used largely by the United States. In this method, a random sample of about 55,000 taxpayers is selected from data available to the Internal Revenue Service and to the social security administration. This sample is closely examined for possible tax evasion by the taxpayers chosen.28 The average tax evasion for the sample is then used to obtain results for the whole population. The results, called the gross gap, or the tax gap, represent the unpaid income taxes on legally earned individual and corporate income. For 1987, the last year for which this information was published, the tax gap amounted to $85 billion, of which $63.5 billion was tax evasion by individuals, $21.4 billion was tax evasion by corporations, and $1.1 billion was tax evasion by nonfilers. Those who generate the data have expressed skepticism that this money could actually be collected.
Budget Survey Method
The third direct method relies on household budget surveys. These surveys show the relationship between the spending of families and their declared income. A family that earns its declared income and spends much more than that income can be suspected of tax evasion unless other factors, such as accumulated wealth or borrowing against future income, account for these differences. The results from this method are not very reliable and they can only provide a gross order of magnitude. The Italian authorities have used the rationale behind this method in the development of the redditometro, an index that establishes a minimum taxable income for taxpayers on the basis of external indices of wealth, such as expensive cars and second houses.
Direct Taxpayer Survey
A few countries, especially Nordic countries such as Sweden and Norway, have used direct surveys of taxpayers. A random sample of taxpayers is chosen and they are asked, among other questions, to describe their tax-reporting behavior.29 This approach has received several criticisms, which range from whether individuals remember their past tax behavior to whether an individual would be willing to convey accurate information about an activity that may be considered antisocial, even when he or she is assured anonymity. The common belief is that tax evasion is often underestimated by these surveys even when they guarantee anonymity for the taxpayers.
Indirect Methods
Indirect methods essentially relate to the quantification of the so-called underground economy, which has been attempted for various countries. The connections between this quantification and the size of the tax evasion are often ambiguous and difficult to establish, especially when taxes are progressive. For example, if those who participate in the underground economy are mostly people with very low incomes, who would have paid very few taxes, then the existence of an underground economy may not imply the existence of significant tax evasion.
There is often a lot of confusion in how people define the underground economy. In some cases, people refer to taxes not paid; in other cases, they refer to the alleged underestimation of the national accounts. Often they do not specify which of these two definitions they have in mind. The problem is that, in many cases, one could have tax evasion with no underestimation of the national accounts or, alternatively, little or no tax evasion with underestimation of the national accounts. A further confusion comes from the fact that the attempt to evade taxes is not the only impetus to an underground economy, since corruption, regulations, and various forms of prohibitions are also important factors. Despite these questions, the underground economy is often taken as a proxy for tax evasion.
Discussion of the various methods used to measure the underground economy requires extensive elaboration. Perhaps it would suffice just to mention the methods used.30 The first method is the so-called expenditure and income discrepancy method, which assumes that the hidden incomes will show up as expenditures, so that the difference between the national accounts measured from income flows and the national accounts measured from consumption flows can indicate the size of the underground economy. This method was first applied to the United Kingdom. A second method is the employment census method, which tries to compare measured unemployment with the probable participation rate for the population in certain age classes. This method has been applied to Italy, Spain, and some other countries. Third is the physical input method, which is based on the idea that there is a predictable relationship between the use of some inputs, such as electricity, and the value of the output. Finally, there are various versions of the so-called monetary approach, an approach that associates evasion with currency or money holding. This monetary approach, developed in various forms by Guttman (1977), Feige (1979), and Tanzi (1980), has been widely used in a large number of countries to estimate the size of the underground economy.31 All of the above approaches have problems. It would be prudent, therefore, not to base economic policy, or even estimates of tax evasion, solely on the results that emerge from these estimations.
To conclude, there are many methods that have been used to estimate tax evasion. In recent years, IMF technical assistance missions have routinely calculated the potential yield of selected taxes using some variant of the national accounts method. These results cannot be provided because of the confidentiality of the reports, but they appear to be quite useful in calculating the extent of tax evasion and in policymaking undertaken by the authorities.
REFERENCES
Aguirre, Carlos A., and Parthasarathi Shome, “The Mexican Value-Added Tax: Methodology for Calculating the Base,” National Tax Journal, Vol. 41 (1988), pp. 543–54.
Altingham, M.G., and A. Sandmo, “Income Tax Evasion: A Theoretical Analysis,” Journal of Public Economics, Vol. 1, No. 3/4 (November 1972), pp. 323–38.
Ballard, Charles L., John B. Shoven, and John Whalley (1985a), “General Equilibrium Computations of the Marginal Welfare Costs of Taxes in the United States,” American Economic Review, Vol. 75 (1985), pp. 128–38.
Ballard, Charles L., John B. Shoven, and John Whalley (1985b), “The Total Welfare Cost of the United States Tax System: A General Equilibrium Approach,” National Tax Journal, Vol. 38 (1985), pp. 125–40.
Banerji, Arup, “Tax Evasion, Enforcement, and Intertemporal Choice,” Proceedings of 1991 National Tax Association Conference (Williamsburg, Virginia: National Tax Association, 1991), pp. 90–100.
Becker, Gary, “Crime and Punishment: An Economic Approach,” Journal of Political Economy, Vol. 76, (1968), pp. 169–217.
Bird, Richard, and Milka Casanegra de Jantscher, eds., Improving Tax Administration in Developing Countries(Washington: International Monetary Fund, 1992).
Cowell, Frank A., Cheating the Government: The Economics of Evasion (Cambridge: MIT Press, 1990).
Cremer, Helmuth, and Firouz Gahvari, “Tax Evasion and Optimal Commodity Taxation,” Journal of Public Economics, Vol. 50 (1993), pp. 261–75.
Dunn, D., “Tax Compliance with Untaxed Fringe Benefits: Evidence from Mexico” (unpublished; Washington: International Monetary Fund, 1992).
Feige, Edgar L., “How Big Is the Irregular Economy,” Challenge, Vol. 22 (1979), pp. 5–13.
Goode, Richard, The Individual Income Tax (Washington: The Brookings Institution, 1964).
Goode, Richard, “Some Economic Aspects of Tax Administration,” Staff Papers, International Monetary Fund, Vol. 28 (1981), pp. 249–74.
Guttman, Peter M., “Subterranean Economy,” Financial Analysts Journal, Vol. 33 (1977), pp. 26–27, 34.
Hansson, Ingemar, and Charles Stuart, “The Welfare Costs of Deficit Finance,” Economic Inquiry, Vol. 25 (1987), pp. 479–96.
Herschel, Federico J., “Tax Evasion and Its Measurement in Developing Countries,” Public Finance, Vol. 33 (1978), pp. 232–68.
Internal Revenue Service, Estimates of Income Unreported on Individual Income Tax Returns (Washington: Government Printing Office, 1979).
Mackenzie, G.A., “Estimating the Base of the Value-Added Tax (VAT) in Developing Countries: The Problem of Exemptions,” Public Finance, Vol. 47 (1992), pp. 257–70.
Musgrave, Richard, The Theory of Public Finance: A Study in Public Economy (New York: McGraw-Hill, 1959).
Pechman, Joseph A., Federal Tax Policy (Washington: The Brookings Institution, 1966).
Richupan, S., “Determinants of Income Tax Evasion: Role of Tax Rates, Shape of Tax Schedule, and Other Factors,” in Supply-Side Tax Policy: Its Relevance to Developing Countries, ed. by Ved P. Gandhi (Washington: International Monetary Fund, 1987).
Sandford, Cedric T., M. Godwin, and P. Hardwick, Administrative and Compliance Costs of Taxation (Bath, England: Fiscal Publishers, 1989).
Serra, Pablo, “Estimación de la Evasión Tributaria en el Impuesto al Valor Agregado,” Servicio de Impuestos Internos, Publicación No. 91 /05 /C (Universidad de Chile, Agosto 1991).
Silvani, Carlos, “Value-Added Tax Data Series” (unpublished; Washington: International Monetary Fund, 1992).
Sisson, C.A., “Tax Evasion: A Survey of Major Determinants and Policy Instruments of Control,” IMF Departmental Memorandum DM /81/95 (Washington: International Monetary Fund, 1981).
Slemrod, Joel, “Optimal Taxation and Optimal Tax Systems,” Journal of Economic Perspectives, Vol. 80 (1990), pp. 157–78.
Smith, Stephen, Britain’s Shadow Economy (Oxford: Clarendon Press, 1986).
Stella, Peter, “An Economic Analysis of Tax Amnesties,” IMF Working Paper 89/42 (Washington: International Monetary Fund, 1989).
Tanzi, Vito, The Individual Income Tax and Economic Growth(Baltimore: Johns Hopkins University Press, 1969; A Japanese edition is also available).
Tanzi, Vito, “The Underground Economy in the United States: Estimates and Implications,” Banca Nazionale del Lavoro, Quarterly Review, No. 135 (December 1980), pp. 427–53.
Tanzi, Vito, ed., The Underground Economy in the United States and Abroad (Lexington Books, 1982).
Tanzi, Vito, “The Underground Economy in the United States: Annual Estimates, 1930–80,” Staff Papers, International Monetary Fund, Vol. 30 (1983), pp. 283–305,
Uchitelle, Elliott, “The Effectiveness of Tax Amnesty Programs in Selected Countries,” Quarterly Review, Federal Reserve Bank of New York, Vol, 14, (1989), pp. 48–53.
Usher, Dan, “Tax Evasion and the Marginal Cost of Public Funds,” Economic Inquiry, Vol. 24 (1986), pp. 563–86.
U.S. Department of the Treasury, Internal Revenue Service, Report on the Application and Administration of Section 482 (Washington: Internal Revenue Service, 1992).
Webley, P., and others, Tax Evasion: An Experimental Approach (Cambridge: Cambridge University Press, 1991).
Vito Tanzi is Director of the Fiscal Affairs Department. He holds a doctorate from Harvard University. Parthasarathi Shome is Chief, Tax Policy Division. He holds a doctorate from Southern Methodist University. The opinions expressed in the paper are those of the authors and are not necessarily those of the International Monetary Fund.
The tax gap emerges from comparing taxable income declared to tax authorities with taxable income calculated from other, presumably more accurate, sources.
Scholars often make a distinction between tax evasion and tax avoidance. In theory, tax evasion implies violation of the law whereas tax avoidance implies taking advantage of ambiguities in the law to reduce the tax burden. This distinction, however, is not always easy and in some countries, such as India, the courts have considered tax avoidance with the intention of evading taxation as tax evasion.
Several Latin American countries have requested IMF technical assistance for measuring tax evasion.
In recent years, new developments in industrial organization and in technology have introduced new ways of evading taxes.
During the electoral campaign, now-President Bill Clinton argued that the reduction of tax evasion by multinationals could generate a good deal of revenue. Recent work by the U.S. Internal Revenue Service has given some support to this view (see U.S. Treasury (1992)).
There is actually a close relationship between Allingham and Sandmo’s theory of tax evasion and Becker’s (1968) theory of crime.
Becker’s (1968) theory assumes that individuals evaluate the expected benefits and costs of various activities, including criminal activities, and choose those that provide the highest income, taking account of the associated net costs.
The theory assumes a close positive relationship between the costs of administration and the probability of catching tax evaders. The importance of this assumption should be kept in mind.
See especially Cowell (1990) for a review of this literature.
In some cases, penalties may be so delayed that they lose their deterrence effect.
Sometimes the taxpayer benefits from the delay owing to the low interest rates charged on the taxes that were due. Some countries require an advance payment of the assessed tax even when the taxpayer contests the assessment.
However, appeals are not costless in terms of time, worry, and legal or other fees.
But, see Goode (1981).
For a recent important contribution to the literature on tax administration, see Bird and Casanegra (1992).
However, see Slemrod (1990).
This literature concludes that in the presence of tax evasion some of the standard conclusions regarding optimal taxation do not hold.
Thus, the compliance cost per dollar paid can be defined as the excess cost to the taxpayer in terms of lost time, payments to lawyers and accountants, and so forth of $1 of tax payment.
In some cases, they have to spend considerable time just getting the right forms.
Anecdotal reports have referred to countries where some key posts in the tax administration have been in high demand or even“sold” to the highest bidders. Obviously these posts provide possibilities of high“incomes.”
Italy has perhaps been the most imaginative in the use of presumptive taxes in recent years.
In fact, the theoretically advocated and practically followed procedure of detecting tax evasion through selected audits raises serious questions of equity when many other tax evaders remain undetected and unpunished.
For some discussion of this issue, see chapter 8 of Smith (1986).
This point was clearly recognized by Luigi Einaudi, the prominent public finance scholar who became president of Italy. He once remarked that if all the Italian tax laws on the books were fully enforced the Italian level of taxation would be 120 percent of national income.
The reverse is also true for capital gains, which may be in the concept of taxable income but not in the national accounts.
In the expenditure side method, exports are already excluded from the domestic expenditure base.
For a discussion of this point, see the paper by Reuter in Tanzi (1982).
See Internal Revenue Service (1979). This method is different from that outlined in the previous section. The sampling method that is being described here is based on a sample selected for special scrutiny on a continuing basis and is used in lieu of the national accounts method.
See the paper on Norway by Isachsen, Klovland, and Strom, and the paper on Sweden by Hansson in Tanzi (1982).
These methods are discussed in Tanzi (1982).
Tanzi (1980, 1983) used this method to estimate first the size of the underground economy in the United States and then tax evasion.