Richard M. Bird
Recent experience with the administrative aspects of tax reform provides some useful lessons for would−be tax reformers in developing countries. To appreciate these lessons, one needs to examine the relation between tax structure reform and tax administration, along with some critical aspects of ''tax technology" in developing countries, as this chapter demonstrates. But it is also important to look at specific cases. Therefore the central part of the chapter is devoted to reviewing the administrative dimension of the recent Bolivian tax reform, with briefer references to experience in a few other Latin American countries.
Approaches to Tax Administration and Tax Reform
The most naive approach to the relation between tax administration and tax reform is simply to ignore administration.1 That is what happens in optimal tax theory, for instance, which gives administration little attention except in the choice of taxes. As Slemrod (1990) argues, if optimal tax theory is to become relevant to the formulation of tax policy, it must incorporate "tax technology" (administrability, for example, with respect to monitoring, measurement, and agency and enforcement costs) more explicitly. The size of the chasm now existing between optimal tax theorists and practitioners of the equally arcane, but quite different, art of tax policy is illustrated by the almost universal rejection by the latter of the clearest policy conclusion of the former, namely, that sales tax rates should be nonuniform.2
Tax reforms even in countries as sophisticated as Israel have at times fallen far short of expectations at least in part because of inadequate attention to the administrative dimension. Indeed, Israel may have made the same mistake twice. The fundamental administrative deficiencies of the 1975 Israeli reform are recounted in convincing detail in Radian and Sharkansky (1979), whose subtitle—"Partial Implementation of Ambitious Goals"—tells the story. Earlier, the principal economist authors of this reform had titled their own paper "The Political Economy of a Tax Reform" and had emphasized the pragmatic and compromised nature (from the point of view of pure theory) of what they felt they had been able to accomplish (Ben−Porath and Bruno 1977). Their outraged reaction (Ben−Porath and Bruno 1979) to a subsequent critical comment by Radian (1979) is thus perhaps understandable.
Having gone along with all kinds of politically necessary compromises, they were now being attacked for administrative unrealism! It is impossible to read this exchange, however, without feeling that this response completely misses Radian's central point, namely, that reforms that do not take administrative limits adequately into account are likely to turn out in practice to be quite different from the way they look on paper.
Given this history, it is rather ironic to read the following assessment by Jenkins (1989: 23) of Israel's latest tax reform exercise: "The 1987 tax reform commission in Israel is a classic example of people with good intentions proposing complex tax policies in order to ensure economic efficiency and fairness. Many of the proposals would be impossible to administer and were rightly resisted by the tax administration." Eco−
nomic expertise alone, it appears, cannot produce workable reforms even in a country like Israel, which for decades has been fortunate enough to possess an administration that, although not perfect, is well above the usual standard found in developing countries (Wilkenfeld 1973).
3— Tax Administration and Tax Reform: Reflections on Experience 76
Knowledgeable tax policy advisers are of course well aware of the importance of administration. The question is, what do they do about it? At one extreme, as already noted, they may either ignore it or, equally naively, assume that any administrative deficiencies can and will be quickly remedied once the right tax policy is put into place.3 At the other extreme, they may take the quite different view that administrative feasibility dominates tax structure decisions. This dominant view is perhaps most clearly epitomized in the "tax handles" literature (for example, Musgrave 1969), which in effect argues that the tax structure of developing countries is dominated by their limited ability to administer taxes. Ports and breweries are few in number and easily visible: import and excise duties are therefore prime sources of revenue in most developing countries. In contrast, capital gains and income from wealth are difficult to locate, measure, and tax; such taxes are therefore of minuscule importance.
The result of this administration−dominated approach to tax policy may be a tax structure that burdens foreign trade and the organized modern industrial sector heavily and hardly impinges at all on most domestic
consumption or such hard−to−tax groups as the liberal professions, small traders, and farmers and the
nonmonetary sector in general. Traditional excise taxes, for example, score highly in administrative terms: the tax base is relatively easy to measure and monitor, and it is relatively simple to enforce and collect taxes on this base with little leakage into unauthorized pockets. Such taxes, levied as a rule on products for which demand is relatively inelastic and with which adverse external effects may be associated, may also look good in terms of the standard efficiency calculus.4
At the same time, heavy taxes levied on a narrow base may affect the organization and production of the taxed goods in unexpected and undesirable ways.5 Moreover, as a rule the distributional effects of the traditional excise taxes on alcohol and tobacco are considered undesirable (McLure and Thirsk 1978). This concern may be
overdone, however. In the context of many developing countries import and excise taxes may not be shifted forward, as is conventionally assumed.6 To the extent such taxes are paid out of rents, their incidence may even be progressive. Where there is a large nonmonetary sector, taxes impinging only on the monetary sector will also be progressive.7
Whatever the facts, if politically significant groups are dissatisfied with the outcome of the tax system for any reason and consider it unfair, the result may be decreased compliance where taxpayers can get away with it.
Decreased taxpayer morale owing to perceived unfairness may thus lead to increased shirking (evasion).8 At the extreme, a sort of Gresham's law of taxation may operate, with the combination of bad taxes imposed on
administrative grounds and the (in part consequent) evasion of good taxes leading to the virtual destruction of the tax system.
A related approach to the role of tax administration in tax reform may be illustrated by Tanzi (in Newbery and Stern 1987), who distinguishes between the statutory and the real (or effective) tax system. Tax reform thus conceived has two stages: changing what is said (the law) and changing what is done (the administration of the law). Policy change without administrative change is nothing. The converse is not true, however: in the
memorable words of Casanegra (1990: 179), "tax administration is tax policy." The implication is that changing tax administration dominates changing tax structure as a means to achieving effective tax reform. Or, as Tanzi (in Newbery and Stern 1987: 238) put it, "The basic truth to remember is that control over the statutory system (over the tax laws) may at times be accompanied by very little control over the effective system. If such is the case, changing the laws may mean far less than we believe."
Two quite different approaches to the administrative dimension of tax reform have been sketched—and perhaps caricatured—above. The first approach suggests that tax reformers should concentrate on getting the tax structure right and worry about fixing up administration, if at all, only after tax law has been tuned to the desired
specifications. There is no point in administering bad taxes better. In contrast, the second approach suggests that what really matters is administration. If policy outcomes are to be altered, administrative changes (which may also require legal changes) must come first. In reality, most authors who may appear to take one or the other of these views would likely say that this is a false choice: structure and administration are interdependent and tax 3— Tax Administration and Tax Reform: Reflections on Experience 77
reform must take both aspects into account.
In a recent study of tax reform in Jamaica, for example, Roy Bahl (in Gillis 1989: 169), argues that "a first principle for successful tax reform is to get the policy right and then deal with the administrative problems." This declaration suggests his tent is clearly pitched in the first camp noted above. The main substantive argument Bahl (in Gillis 1989) gives for this conclusion, however, is that the tax structure may be so complex and unfair that it simply cannot be administered effectively.9 This is an undeniably true statement, and not just for Jamaica.
Complaints about the compliance and enforcement problems caused by poor drafting abound in the literature in developing and developed countries alike.10
An especially dramatic instance occurred in the Philippines, where the nominally global and progressive income tax became so complex, hard to understand, and difficult to administer that a general sigh of relief went up when it was replaced a few years ago by a set of schedular taxes (Asher and Kintanar in Asher 1989b).11 This explicit reversion to schedular taxation is noteworthy both because it appears to have been motivated largely by
administrative considerations and because it is so contrary to the conventional wisdom on the virtues of global as compared to schedular taxation (see, for example, Shoup and others 1959). As noted later in the chapter, however, the Philippines may prove at least in this respect to be on the cutting edge of the new wave of tax reform in developing countries.12
Whatever one thinks of the Philippine reform, Bahl (in Gillis 1989) is quite right that it generally makes no sense to attempt to improve the administration of a badly designed tax. It does not follow, however, that attention should not be paid to improving tax administration as well as to tax design. Indeed, in his own work on Jamaica, Bahl (1988, 1991) has paid a great deal of attention to such critical aspects of tax administration as staffing, training, administrative procedures, and the collection and use of information. Like all good tax reformers, he is thus well aware of the interdependence of tax reform and tax administration.
At first glance, Bird (in Gillis 1989) may appear to be taking exactly the opposite line to Bahl's "first policy, then administration" argument My central point in that paper, however, was simply that the key to successful tax reform in developing countries is to design reforms that will work properly, assuming that the tax administration in place at the time of the reform will remain much as it is. The argument is not that administrative improvements along the traditional (and correct) lines set out by Surrey (1958) are not desirable and necessary. It is rather that experience in many countries in recent years has demonstrated that such improvements are so difficult and time−consuming to accomplish that would−be tax reformers must as a rule work with what now exists, since that is all there is likely to be for years to come.
But this is really the same message, with a different emphasis, as in Bahl (in Gillis 1989). There is no point in administering more efficiently an inherently bad tax structure (Bahl). Nor is there any point in designing a good tax structure that cannot, and will not, be administered effectively (Bird). In short, the key to successful tax reform is to design a tax structure that can be administered adequately with the available resources while at the same time making the best possible use of those resources from a long−term perspective. A tax system comprises both structure and administration, and structural and administrative reform must be considered together if tax reform is to accomplish the goals usually attributed to it.
Three Aspects of Tax Technology
In addition to simply recognizing the interdependence of the structural and administrative aspects of taxation, more research in the area of tax technology is needed to improve the success ratio of tax reform efforts.13 This section reviews three aspects of this technology and suggests some possible lines for future research.
Three Aspects of Tax Technology 78
The Importance of Administrative Incentives
Not one tax authority in the more than twenty developing countries in all parts of the world in which I have worked on tax reform has had available the sort of basic information on the characteristics of the taxpaying population that is needed to enforce taxes adequately and fairly. In many cases, the only current information available to the head of the tax administration has consisted of collection data, with no indication of who paid the tax, how it was paid, or whether it was paid on time. An obvious way to improve tax administration in most developing countries is thus simply to design and install better management information systems. Unfortunately, although much ingenuity has gone into designing and implementing better computer−based information systems (Lane 1990), little attention has been paid to the second, and critical, part of the problem: designing organizational and individual incentives to ensure that the information is used. An early paper by Haberstroh (1965) on
managing a tax department has not, it appears, generated any subsequent general analysis or any publicly available case studies. Nor has there been any systematic analysis of the optimal way to reward tax officials, although this subject cries out for study owing to the universality of the problems arising from the low salaries paid to tax officials.
In Indonesia, for example, revenue targets or quotas were for many years the principal way in which the income tax administration was managed and tax officials rewarded. This system was almost universally condemned by visiting experts as arbitrary, inequitable, and discouraging to good administrative practices.14 Analysis of the targeting system suggested, however, that the targets were set to support rather than discourage administrative effort. Even the apparent arbitrariness and inequity with which they were sometimes applied accorded with prevalent political and cultural norms. Rather than an archaic anachronism, targeting in the Indonesian context possessed the
important virtue of providing a predictable flow of revenue while permitting the central authorities to oversee the (as it were) "official tax farmers" who actually ran the system.
The designers of the Indonesian tax reform of 1985 were aware of this entrenched system and of the potential problems it posed. Gillis (1985), for example, explicitly noted that the success of the reform depended critically upon the readiness of tax officials to change their ways.15 In fact, as Gillis (1990) later noted, not only did the tax administration not support the income tax reform but even five years after the reform there was little evidence of any significant change for the better in administration. There have of course been some changes, and some improvements, in Indonesia's tax administration in recent years (Nasution in Asher 1989b). Nonetheless, the consensus appears to be that the important reforms in the income tax law have had less impact on taxpayers than expected largely because of the failure of the tax administration to mend its ways (Asher 1989a; Gillis 1989, 1990; Jenkins 1989). How any other outcome could have been expected, given that the same people were still responding to the same incentives in the same ways, is unclear.
Similar problems are likely to arise in implementing any tax reform unless close attention is paid to the internal incentive structure of the tax administration. In Egypt, for example, tax officials currently receive much of their (legal) compensation in the form of(untaxed) incentive payments based on a sort of targeting system.16 In Senegal (and many other countries) tax officials receive much of their compensation from the additional tax and fines they collect—a system obviously fraught with the possibility of abuse. In Jamaica, at least prior to the recent tax reform, many tax officials relied heavily on (nontaxable) travel allowances to make a living wage—a system that encouraged unproductive travel around the island. Such aspects of the reward systems (compensation and performance appraisal) prevailing in different tax administrations are not minor matter's. As the Indonesian example illustrates, these details may determine the outcome of major reforms in tax structure.17
The Importance of Administrative Incentives 79
The Costs of Taxation and Other Quantitative Questions
The cost of levying taxes is obviously a relevant consideration in shaping the tax structure in any country.18 A recent study in the United Kingdom of the operating costs—public (administrative) and private (compliance)—of the tax system found that this cost came to about 4 percent of revenues, or 1.5 percent of GDP (Sandford and others 1989).19 The real resource cost of running the tax system is probably at least as large in developing countries and deserves more attention than it has received.20
Although there are obvious difficulties in carrying out cost studies (Bird 1982a), particularly in developing countries, a worthwhile research task would be to estimate the administrative and compliance costs of both existing taxes and, especially, of proposed tax reforms.21 Tax changes create such transitional costs as increased uncertainties, startưup costs, and learning costs. These costs, too, need closer study to ensure that the gains from change exceed its costs sufficiently to make the investment worthwhile.22 Frequent changes in tax structure (like large differentials in taxes that depend on minute differentiations in real circumstances) are likely to result in regressively distributed real social costs exceeding any benefits resulting from the change (or differential).
Economists may have neglected unduly the operating costs of tax systems. The same cannot be said of the
efficiency costs (excess burden or deadweight loss) that are the main focus of much economic analysis of taxes.23 Changes in economic behavior induced by taxation reduce the real level of wellưbeing every bit as much as does the using up of real resources in collecting the taxes in the first place. Although estimates of efficiency costs are considerably more speculative than those of operating costs, their conceptual appeal is such that many more such estimates may be found in the literature, at least for developed countries.
Although such estimates are subject to wide variation, most economists today would probably agree that efficiency costs are likely to be at least 2030 percent of revenues collected (and, some would say, perhaps much higher). Indeed, if one adds to such costs—as Usher (1986, also in Bird 1991) has persuasively argued we should—an allowance for the real cost of concealing income or sales from the tax authorities (the cost of tax evasion), the overall cost of collecting taxes would look even higher.24
Unfortunately, the amount of quantitative evidence available on such questions is risible. Malcolm Gillis (in Bird 1991), for example, reports two unpublished studies, in Colombia and Indonesia, which suggested that for every
$1 in bribes received by tax officials, governments lose $20 in tax revenues. The cost to taxpayers of evading direct taxes in those countries was thus probably as low as 5 percent of the tax evaded for those caught in the rather porous tax net—and much lower for those who slipped through unnoticed.25
Such estimates ignore the expected value of being caught and penalized. Despite the prominence of such calculations in the extensive economic literature on tax evasion (following on Allingham and Sandmo 1972), however, this omission is not significant. Tax evaders in most developing countries can realistically assign an
expected value of zero to the likelihood of being detected and penalized. The utility of extending the now conventional formal analysis of noncooperative games to developing countries thus seems dubious.26
Demonstrations of the dependence (or nondependence) of evasion on the height of tax rates and the precise design of the penalty structure are unlikely to provide much information of interest to tax authorities.27 Moreover, scattered evidence in different countries suggests that the more severe the penalty the less likely it is to be applied.
As Levi (1988), Lewis (1982), and others have stressed, good taxpayerưgovernment relations require attention not only to the control mechanisms (sanctions, audit probabilities) stressed in the economic literature but also to the important mechanisms of social cooperation that constitute part of the "institutional constraints" (Mansfield 1987) that enable tax systems to collect significant amounts of revenue.28
The Costs of Taxation and Other Quantitative Questions 80
Similar casual empiricism (for example, in Paraguay, where smuggling constitutes a major economic activity) suggests that, as a rule of thumb, the cost of smuggling (bribes, additional transport costs, and the like) is probably on average no more than 10 percent of the value of the smuggled goods. Although I am aware of no systematic study of this point, it is interesting to note that a uniform 10 percent tariff rate appears to be a relatively common recommendation in reform packages, such as that adopted in Bolivia recently (Thirsk 1989a).29
An even more casual observation (reported by Joel Slemrod in Bird, 1991) is that for every additional $1 spent on tax enforcement in Mexico, an additional $25 in revenues is collected.30 Gillis (1990) reports an even more startling figure from Indonesia, where a special "strike force" of auditors, focusing on a limited number of large firms reporting losses for 1985, produced a direct revenue return equal to 340 times the investment. The ripple or indirect effect—that is, the resulting improvement in "quasiưvoluntary compliance" (Levi 1988) as a result of the increased perceived probability of detection—would presumably have made these results even more impressive.
As Gillis (1990) says, ''This special team clearly demonstrated not only the presence of great 'slack' in tax administration, but also the tremendous potential returns available from investments in targeted audits.
Unaccountably, however, the group was disbanded at the end of 1987."
This interesting story suggests two lessons that merit further consideration. The first is that no government, no matter how authoritarian it may be, can with impunity enforce taxes too harshly on politically influential groups.31 More generally, as Shoup (1969: chap. 17) and Goode (1981) observed, no government anywhere pushes enforcement to the point at which marginal administrative cost is equated to the marginal revenue collected. Nor, as these authors properly stress, should any do so. Nonetheless, it would clearly be useful to understand this relationship better, both empirically and analytically.
The second lesson from the Indonesian story concerns the appropriate strategy for deploying auditing resources.
In terms of direct returns, there seems little question that targeting big taxpayers is likely to prove most rewarding.
As Muten (1981) has stressed, it often makes sense to concentrate scarce administrative resources on those few taxpayers, generally large firms, from whom taxes can actually be collected rather than dissipating them to little avail across the vast array of small and medium taxpayers. It may thus be sensible to create special administrative sections to exert tight control over large taxpayers, as has been done in Bolivia and Argentina. The large literature on how to deploy auditing resources optimally (see, for example, Balachandran and Schaefer 1980) usually similarly favors a selective, targeted, approach.
At the same time, there are at least three reasons for spreading auditing talent more thinly. The first is to encourage greater compliance by small taxpayers by making it clear that they are not immune to enforcement efforts (Casanegra 1990). A second reason, equally obvious, is in the name of those eternal verities of
developmental tax policy, equity and economic growth. Enforcing taxes most rigorously on the largest companies may maximize revenue, but it is both unfair and may tend to encourage the expansion of less efficient, and less taxed, sectors of the economy.32 Finally, as Cowell and Gordon (1989) have demonstrated, random audits may be most efficient even in terms of the conventional economic analysis of evasion when the possibility of total
noncompliance is taken into account.33
The Link between Expenditures and Tax Administration
The efficiency and effectiveness of any tax administration reflects many factors. The importance of an adequate legal framework both for the tax structure and such procedural aspects of administration as assessment, collection, review, appeal, and penalties is clear (Yudkin 1973). So is the need for adequate internal management and
effective deployment of resources by the administration itself. In addition, however, it is critical both that the political authorities support effective tax administration and that taxpayers themselves are neither implacably antagonistic to the administration nor excessively tolerant of evasion. When a tax administration lacks political support and is badly run, it will lose credibility with the public, and thus further erode its political (and probably resource) base. Argentina's recent experience affords a clear demonstration of this vicious circle at work.34
The Link between Expenditures and Tax Administration 81
How people feel about taxes and tax administration at least to some extent reflects how they feel about
expenditures.35 When people are antagonistic to government, when they feel that it is wasting their money and not acting in their best interests, taxes and tax officials are likely to be even more unpopular than usual.36 This line of argument should not be misunderstood. I am not arguing that a good tax system is one in which there is a high degree of voluntary compliance. Such voluntary compliance is largely a myth. Even if every individual citizen fully supports all government actions and willingly accepts a resulting high tax level in principle, it is still in his individual interest to reduce his share of the total tax burden. Tax systems depend more upon taking money away from people before it gets into their hands—and scaring them into paying the balance—than they do on goodwill. Nonetheless, the boundaries of what is considered reasonable tax evasion in any society are elastic, and how far these boundaries are stretched depends at least in part upon how the government is perceived by its citizens. People seem to be more willing to pay taxes if they think the funds are well used—particularly for their own benefit—and if they feel fairly treated relative to the individuals and groups with whom they compare themselves. As Bates (in Gillis 1989) has stressed, one can understand the politics of tax reform only if one considers both the costs of taxes and the benefits of expenditures for specific political actors.
The principal functions of government are to deliver services and to manage conflict. Traditionally, economists have considered only the first of these objectives. Even the best−intentioned government can live up to its good intentions only if it is in power, however, and it can stay in power only if it can muster sufficient support, including tax revenues. Some critical support generally comes from groups to whom, so to speak, government delivers the goods, whether as format or informal tax concessions or as public expenditures favoring their interests. Governments reward those whose support and cooperation is needed to accomplish particular policy goals.37
In some instances, one key to tax reform may be to link expenditures and revenues more explicitly than has usually been done in practice or advocated in the tax reform literature.38 The principle of benefit taxation—of levying taxes in accordance with the benefits received from the expenditures they finance—is an old one, with a rationale well−grounded in both equity and efficiency (Bird 1978). Few seem to realize, however, the extent to which this approach can usefully be applied in developing countries that find it difficult to raise taxes and to spend the proceeds efficiently.
The precise outcome of benefit taxes and user charges depends on the details of the particular expenditures and financing methods in question. Although generalizations are as suspect in this as in most areas of economic policy, the more effective use that can be made of expenditure−revenue links, the more acceptable any given tax system (or proposed reform) may be. It may also be easier to administer, although this is far from certain since the cost of administering benefitrelated systems can be high and it is by no means clear that greater acceptance implies greater compliance. As in the case of administrative incentives and the costs of taxation, much more careful study of particular cases is needed before any firm conclusions can be drawn with respect to the costs and benefits of particular types of expenditure−revenue links in different contexts.
Tax Administration and Tax Reform In Latin America
Two types of tax reforms have taken place recently in Latin America.39 The first type, of which Bolivia affords the most dramatic example, is a drastic overhaul of the tax system in a short time. The second, of which Colombia is perhaps the best example, is a more gradual evolution through a series of reforms over a longer period of time.
The results of both processes are similar, however—and very different from the direction of reform favored in such well−known reports as those by Musgrave (1981) and Musgrave and Gillis (1971) and Shoup (1965) and Shoup and his colleagues (1959). This section considers briefly a few of the apparent lessons in tax administration and tax reform suggested by this recent experience, particularly in the muchpublicized Bolivian case.
Tax Administration and Tax Reform In Latin America 82