total administrative expenditures

An important issue on which economics might be expected to shed light is how much a government should spend for tax administration. By analogy with the economic theory of the firm, it may seem that the general principle governing the decision should be that expenditures should be carried to the point at which the last increment in costs just equals the additional revenue obtained—that is, to the point at which marginal cost equals marginal revenue. At that point, the excess of total revenue obtained over total administrative costs incurred will be at a maximum. That is true because expenditures will be incurred for all administrative activities that would yield revenue in excess of their cost, and no expenditures will be incurred for activities that would yield less than their cost.

Objections can be advanced against the analogy with the operations of a business firm. The tax department is not a profit-seeking enterprise, and taxpayers are not customers. Tax collection is a compulsory transfer of income rather than a voluntary payment for individually consumed goods and services. Equality of the value of resources devoted to the process and the revenue obtained at the margin does not have the same significance in regard to tax administration as it does in regard to the production and sale of marketable output. For marketable output, it may be assumed—subject to important, but not overwhelming, qualifications—that the use of resources is responsive to consumers’ desires and that a smaller or greater volume would satisfy them less well. For tax administration, no comparable assumption is warranted; citizens may value additional public services at more or less than the amount of revenue obtained from increased activities (Shoup (1969), p. 433). But that in itself does not justify dismissal of the marginal principle. If vigorous administration should push total revenue beyond a politically acceptable level, tax rates could be reduced. With lower tax rates, some curtailment of administrative expenditures might be indicated. The maximization of the excess of revenue over administrative costs, at the new tax rates, might still be regarded as proof that the real resources devoted to tax administration had been used effectively.

Difficulties would be encountered in applying the marginal principle, because information is not readily available on the relation between incremental cost and incremental revenue. In addition, arguments can be made for spending more or less than the principle indicates.

It could be argued that expenditures on tax enforcement should be pressed beyond the point of equality between marginal cost and marginal revenue, on the grounds that equal application of tax laws is an important goal of public policy that should not be evaluated on narrow commercial terms (Simons (1950), pp. 31-32). Law enforcement efforts in general are not governed by the objective of maximizing revenue from fines. It could also be argued that in the long run, additional costs incurred in rigorous tax administration will be economically justified by virtue of improved voluntary compliance. Although that contention appears plausible, it is difficult to produce reliable quantitative evidence on the relation between administrative costs and voluntary compliance.

In practice, most governments appear to spend less, perhaps substantially less, than the marginal principle would indicate. This impression is also hard to verify, because in only a few countries has information on marginal cost and marginal revenue been made public. When statistical comparisons are made, they usually relate total administrative expenditures to total revenue for all taxes combined or for all the taxes assigned to a particular administrative organization. Such data tell nothing about the relation between costs and revenue at the margin and are of limited value in appraising efficiency.

There are several reasons why governments might not wish to carry tax administrative expenditures as far as the marginal principle would indicate, even if the necessary information on marginal cost and marginal revenue were easily obtainable. First, legislators and officials may feel that the indicated level of administrative activity would antagonize taxpayers and provoke a withdrawal of cooperation and political support. This belief is the opposite of the attitude mentioned previously that would favor expenditures greater than those which would equalize marginal cost and marginal revenue in the short run and, like its opposite, is hard to evaluate. Second, it is politically difficult for governments to apply different budgetary standards to the tax department and to other departments or ministries that do not generate revenue (Farioletti (1973)). When budgetary stringency is the order of the day, as it frequently is, the tax department is likely to have its requests for funds cut back, more or less in line with other requests, even in those rare cases in which reliable estimates are available to show that the loss of revenue will exceed the saving in administrative costs. Salaries of tax officers are usually at the same level as the pay of other civil servants whose diligence and probity do not directly affect revenue. Third, taxpayers incur private costs of compliance that may substantially exceed the costs of administration. For example, a careful study found that, in England and Wales in 1970, the costs of compliance with direct personal taxes (including out-of-pocket costs and imputed costs) ranged between 1.8 and 6.0 times the administrative costs (Sandford (1973), p. 44). Administrative and compliance costs are partly substitutes; but at the margin, increased administrative activities may well entail increased compliance costs. Fourth, political decision makers may consider it prudent to avoid even a temporary rise of revenue above the expected level because they believe that citizens will react more adversely against a budget that is larger than they desire than against one that is smaller than they wish.

allocation of a fixed appropriation

The marginal principle, in my opinion, has greater value as a guide for the use of a fixed budgetary appropriation than as a standard for setting the amount of the total appropriation. Most tax departments receive such an appropriation as the outcome of a budgetary process that is influenced by a variety of factors. In deciding how the money shall be divided among the administration of various taxes or administrative processes, the relation between marginal costs and marginal revenue is a highly useful index. The maximum amount of revenue in relation to total administrative costs, and hence the most effective use of the appropriation, will be obtained by carrying each of the various activities to the point at which the difference between the marginal unit of revenue and the marginal unit of cost is equal to that of all the other activities. In this situation, no net gain in revenue could be secured by making a small reallocation of administrative resources.

It must be emphasized that data on the total costs of administering various taxes and of carrying out separate activities, such as auditing or data processing, are useful for planning the allocation of resources only to the extent that it is realistic to contemplate the elimination of a tax or an activity. In that case, the total costs, if ascertainable, are indeed the marginal costs. In the more usual situation, however, the pertinent question is whether a little more or a little less expenditure shall be incurred in applying a particular tax or carrying out a particular process. Often these marginal variations will relate to the deployment of staff. A considerable part of total costs consists of overhead expenditures, which cannot be precisely allocated. The application of the marginal principle depends on the availability of information that may be difficult to obtain. Sample studies can provide a basis for estimating various points on the curves of marginal costs and marginal revenues (Welch (1954), pp. 216-17) but may be expensive. Even when good statistics are available, the application of the marginal principle may, in some cases, involve intricate calculations, the outcome of which is not intuitively obvious (Balachandran and Schaefer (1980)). There may be differences between short-run, medium-run, and long-run marginal costs, as staff and equipment cannot be instantaneously redeployed. It is easier to attain a uniform relation between marginal cost and marginal revenue for all taxes and all administrative processes if they are all assigned to one administrative organization than if they are distributed between two or more organizations. However, other considerations may outweigh this advantage of a unified revenue service.

Concentration of administrative resources on activities that will yield the most revenue may well result in the neglect of unremunerative taxes and of poor and remote regions. The requirements of administrative efficiency may thus conflict with the ideal of equal application of all tax laws. If so, compromises will have to be made. Similar compromises or trade-offs are common in the enforcement of other provisions of civil and criminal law. In regard to taxation, general considerations of administrative efficiency and equity may be qualified by awareness of special political sensitivities and by recognition of taxpayers’ costs of compliance and the nonfiscal purposes of certain taxes. On these matters, the director of taxation will be well advised to consult the minister of finance or other responsible political leaders, informing them of any intention not to enforce certain tax provisions because of inadequate resources, and they may decide to seek legislative changes to lessen conflicts between administrative efficiency and other desiderata. Despite difficulties and qualifications, the marginal principle remains highly informative as a guide to the allocation of administrative expenditures. Any important departure from the objective of equalizing the difference between marginal revenue and marginal cost in various activities should be accepted, in my opinion, only if there is explicit and persuasive justification for it.

views of tax administrators

Many tax administrators appear to disagree with the emphasis on marginal revenue as a measure of the “output” or “product” of a particular administrative activity. For example, a speaker at the 1977 general assembly of the Inter-American Center of Tax Administrators (Casanegra de Jantscher (1977), pp. 3-4) said:

There appears to be increasing agreement among tax administrators on the fundamental objective of tax audit. Stated simply, the objective is to promote a higher level of voluntary compliance among all taxpayers. . . . Obtaining additional revenue from audit assessments should be considered only a secondary objective of audit activities.

The stimulation of voluntary compliance does not necessarily conflict with the revenue objective. Improved voluntary compliance will be reflected in greater revenue or smaller administrative costs in relation to revenue over a period of years, if not immediately. Nevertheless, two important points are at issue. One is whether marginal revenue shall be defined restrictively as only the collections directly attributable to an activity or shall include an allowance for indirect effects. The other issue is whether only the current year’s revenue shall be counted or whether the estimated effects on future yields shall be included. The wide view and the long one have appeal. Common sense tells us that neither indirect effects nor future effects of administrative activities should be ignored. On the other hand, it is very difficult to obtain quantitative estimates of indirect effects or to find a basis for forecasting future effects. It is not unreasonable to suppose that direct, short-run revenue results may be correlated with indirect, long-run results. While it would be wrong-headed to rely solely on the data on costs and returns at the margin that can be quantified, it would be injudicious to apportion resources only on the basis of intuition or tradition.

Some administrators accept the marginal framework of analysis but believe that the function of tax administration is not just to prevent underpayments but also to avert overpayments. In an interesting paper on tax auditing by the State of California that was published in 1954, a senior tax official argued that the correction of overassessments was as important as the detection of underassessments. Hence, he suggested that audit performance be measured by adding together additional assessments and refunds, without regard to sign, and showed that this approach called for relating marginal cost to marginal return, defined as the marginal amount of “misplaced tax” (Welch (1954)). While I agree that both ethics and an expedient regard for public relations dictate that a tax department should not confine itself to discovering underpayments and should make prompt refunds when overpayments are found, I doubt that it is realistic to give equal weight to the correction of both kinds of error. Special attention to underpayments may be justified on the grounds that taxpayers are likely to be more prone to understatements. A director of a tax department who is especially attracted to quantitative evaluation of performance might develop an activity measure that combines positive and negative adjustments of assessments with a greater numerical weight for the former. Conceivably, an imaginative official could go even further and differentiate among additional assessments by assigning greater weight to the discovery of understatements of tax liability associated with behavior that is regarded as especially harmful on economic grounds than to understatements associated with other behavior. For example, greater weight might be given to an additional assessment obtained by preventing the successful manipulation of international transfer prices, which would result in the evasion of both taxation and foreign exchange regulations, than to an assessment of unreported income that had been hoarded in the form of national currency (Morag (1957)). The marginal principle is versatile enough to cover various quantifiable objectives (Wertz (1979)). But it is advisable to refrain from developing overly elaborate schemes that may prove to be impractical and tend to discredit the quantitative approach.

staff training

In many countries, the expenditures of the tax department include outlays for formal programs of staff training. Experts’ reports often recommend that such programs be strengthened and expanded. Economic theory corroborates the productivity of training but points to a caution about the allocation of its costs. To an increasing extent in recent years, economists have recognized the contribution of training to productivity and have employed the concept of human capital as an aid to thinking on the subject. Human capital represents the present value of the increased productive capacity resulting from education, training, and experience. Human capital, however, differs from physical capital in one important respect. Human capital belongs to the individual worker and is taken with him when he changes jobs, whereas physical capital usually belongs to the organization that purchased it. For that reason and because of inadequate appreciation of the importance of training, there is a risk that economic incentives will be too weak to elicit the socially optimum amount of training. Although a tax department should not take an overly narrow view of the subject, it could justifiably draw an important distinction between general training to impart a skill such as bookkeeping, which is useful in a variety of jobs, and specialized training, such as instruction in departmental procedures, which is useful mainly for the purposes of tax administration. On the one hand, staff members engaged in general training could justifiably be paid at reduced rates during this period in recognition that their future earning capacity will be enhanced and that there is no assurance that they will remain with the tax administration organization. Alternatively, some general training might be scheduled after regular working hours or might be subsidized by the ministry of education. On the other hand, staff members undergoing highly specialized training should be paid at the regular rate, as it may be assumed that the tax department will obtain the benefit of the increased productivity due to the specialized training. Admittedly, it is not always possible to distinguish clearly between general training and specialized training, and some forms of specialized training may enhance a tax officer’s earning opportunities in the private sector.

tax burdens and administration

In discussions of tax administration, it is often asserted that additional administrative expenditures are justified because they will produce additional revenue in a manner that will be less burdensome to taxpayers than raising tax rates or imposing new taxes. There is some truth in this; well-administered taxes, being more uniformly applied than poorly administered ones, are more equitable and cause smaller economic distortions. But the argument should not be pushed too far. Revenue obtained by better tax administration, like other revenue, represents a diversion of resources from private consumption and investment to utilization in the public sector or a transfer of income from taxpayers to recipients of government payments. This is not the place to discuss the factors that bear on the real costs of taxation, and which establish economic limits to taxation, but in general it would seem that these factors relate to revenue obtained by better administration as well as to other revenue (Goode (1980)). Incomplete tax enforcement may be a safeguard against the harmful effects of excessive tax rates or poorly designed taxes.

audit criteria, best-judgment assessments, and forfaits

Tax assessments rely primarily on accounting data, but economic analysis and statistics can serve as useful supplements. Economic studies can be undertaken to elaborate and update procedures used in assessing the income tax and the value-added tax. Three applications of such studies may be distinguished: (1) to establish criteria for identifying cases for detailed examination or field audit; (2) to help officials to make best-judgment or administrative assessments when the information supplied by taxpayers is believed to be incorrect or incomplete; and (3) to provide, as an alternative to conventional assessments, methods for determining estimated income or value added of certain classes of taxpayers who are not expected to maintain full accounts. The first application is a strictly administrative matter; it is an elaboration of the rules of thumb used by all experienced tax officers and the judgmental guidelines that have been developed by many tax departments as aids in checking on the plausibility of information submitted by taxpayers. The second application may also be possible without specific statutory authorization, but the third application depends on legislation that allows it. Despite fundamental legal differences, the second and third applications can be similar in practice (Lapidoth (1977)).

The discriminant function system of the U.S. Internal Revenue Service is an example of a formal set of standards for selecting income tax returns for audit. However, it is derived from statistical studies of the results of special audits of randomly selected tax returns and does not appear to draw on economic analysis or statistics from outside sources (United States, Comptroller General (1976)).

The Israeli tachshiv (plural tachshivim) is an example of the use of economic analysis and data to assist officials in making best-judgment assessments in cases where comprehensive and reliable information is not available from books of account. Tachshivim are standard assessment guides that are prepared for various trades and professions by economists of the tax department and constantly brought up to date. In 1977, they were said to cover about 100 economic sectors (Lapidoth (1977), pp. 124—47; Wilkenfeld (1973), pp. 144-50).

France, in its forfait system, appears to have gone the furthest in legally establishing alternative bases of assessment, using indicators to determine estimated income and value added rather than as a check on assessments ostensibly based on conventional records. Forfaits are used in France for assessing the income tax of farmers, unincorporated business enterprises, and professional persons whose gross receipts fall below stipulated levels and also for assessing the value-added tax of small enterprises. For agriculture, forfaits reflect average rates of return per hectare for different types of farming and different regions, as estimated by the agriculture department. For other forfaits, French tax inspectors have at their disposal a monograph for each important trade or economic activity that includes descriptive material and information on gross profit margins for various activities or products. The monographs are prepared on a regional or national basis by the research division of the tax department, and representatives of business or professional organizations are invited to comment on them. In addition, the French tax administration has statutory power to assess income tax by reference to certain external indicators of the taxpayer’s style of life, each of which is assigned a specific value. The external indicators are derived primarily from household consumption surveys of the national statistical institute. They include the rental value of the person’s principal and secondary residences; ownership of cars, motorcycles, boats, aircraft, racehorses, saddle horses; and employment of servants.

Whether a forfait system is needed and is politically acceptable, or even constitutionally permissible, in a country depends on a number of factors that will not be discussed here. However, my impression is that there is considerable scope in many countries for productive research to extend and improve indirect or external indicators of net income and value added, regardless of whether the results are used merely as internal guidelines for the tax department or serve as alternative bases of assessment. Where possible, the studies should use samples of firms or individuals for whom trustworthy conventional accounts showing net income and value added, and also information on suitable indirect indicators, are available. For business enterprises, farmers, and professional persons, the method is to correlate the indirect indicators with gross receipts and the deductions that are appropriate for arriving at net income or value added. The indicators may include input factors such as purchases, number of workers, amount and quality of equipment, size and location of premises, imports, use of raw materials or components, and consumption of fuel and electricity. They may also include output factors, particularly gross sales or receipts but also physical units in some cases. Separate indicators are likely to be needed for various trades or industries and for different geographic areas.

The other class of indicator, comprising identifiable consumption items, is intended as an aid in approximating total personal income on the basis of the taxpayer’s living style. Although this expedient has a long history, in most cases its economic basis appears tenuous. A tax department cannot be expected to undertake the extensive and costly survey that would be required to furnish better economic support for the presumptive indicators. It may be possible, however, to follow the French example and improve conventional indicators by referring to family budget data that have been compiled by the central statistical office or another agency for the preparation of a consumer price index or other purposes.

Of course, reliance on indirect indicators is less satisfactory than assessment of the income tax or value-added tax by reference to reliable accounting statements or other direct information. A forfait system or a rigidly applied set of guidelines tends to convert the levy into a tax on the indicators rather than a tax on the nominal base, and the de facto tax may have economic effects quite different from those normally attributed to an income tax or a value-added tax. But there may be no good alternative during the lengthy period required to promote reliable bookkeeping and voluntary compliance with sophisticated tax laws. Exemption of the smallest enterprises is a common feature of value-added taxes, but large exemptions are unfair and offer encouragement to forms of enterprise that are often inefficient and slow to adopt modern methods. A formal qforfait system is likely to appear more equitable and to be politically more acceptable than the outright application of different forms of taxation to small and large enterprises. Forfait systems usually allow taxpayers to “graduate” to the normal regime by establishing satisfactory accounts. There is, of course, a risk that firms whose assessments would be higher under the normal regime will be deterred from improving their accounts. As a safeguard, it may be advisable gradually to restrict eligibility for the forfait system and to keep forfaits rather high.

depreciation allowances

Even in the assessment of the income of firms that maintain accounts meeting established standards, some recourse to economic analysis and statistics may be needed. An important element in the measurement of business income is depreciation allowances, or capital consumption allowances. Accountants have tended to view these allowances as a conventional and systematic method of spreading the cost of capital assets over their useful life. Economists tend to regard true depreciation or economic depreciation as the decrease in the market value of a capital asset during any time period owing to wear and tear and obsolescence and to think that tax assessments ought to reflect economic depreciation. The economic concept, however, is difficult to apply owing to the absence of active markets for secondhand equipment, particularly equipment of a specialized nature.

The measurement of either accounting depreciation or economic depreciation is greatly complicated by inflation, and in recent years this aspect of the problem has overshadowed other issues. With rapidly rising prices of capital goods, conventional depreciation allowances based on the original cost of assets lose their validity as measures of capital consumption and as elements in the computation of net income. Businessmen complain that the allowances are inadequate to finance the replacement of plant and equipment—a different point but also an important one. Many countries have permitted or required one-time revaluations of depreciable assets after periods of inflation, and a few countries have introduced continuing arrangements for revaluation (Lent (1975); Casanegra de Jantscher (1976)). The advisability of such adjustments and the merits of different techniques are being actively discussed in other countries.

The two principal techniques may be called current purchasing power (CPP) accounting for depreciation and current cost accounting (CCA). The CPP approach adjusts depreciation allowances for changes in the purchasing power of money as measured by a broad index, whereas the CCA approach adjusts the allowances for changes in the prices of the particular assets owned by the firm. Clearly, the CPP approach is administratively much simpler and, in my judgment, should be preferred for this reason. On this point, my judgment may be influenced by my opinion that CPP is also fairer than CCA, because the latter omits from taxable income the real gains enjoyed by owners of assets whose value rises faster than other prices.

An alternative proposal, advanced by Nicholas Kaldor in another context in his well-known report on Indian Tax Reform ((1956), pp. 72-79), merits further consideration as a simple means of dealing with depreciation allowances. Kaldor’s proposal was that instead of taking annual allowances over the useful life of the asset, the taxpayer be permitted to take a lumpsum allowance, when the asset is put in service, equal to the discounted value of all the annual allowances that would be available under normal accounting. The immediate lump-sum allowance, unlike normal allowances, would not be subject to erosion in real value because of increases in the price level occurring after the investment was made. A lump-sum, first-year allowance has recently been proposed in the United States to obviate the need for inflation adjustments and to simplify depreciation accounting (Auerbach and Jorgenson (1980)). Discounting to compute the present value of future allowances under normal depreciation would recognize the economic advantage of obtaining the tax benefit of depreciation allowances earlier than under the normal method. This is an essential feature of the proposal that distinguishes it from the practice of allowing immediate write-off of the full cost of capital investments that exists in some countries. The latter practice is very generous and, on certain plausible assumptions, is virtually equivalent to full tax exemption for the profits from the investments (Shoup (1969), pp. 301-302).

Two critically important features of the proposal for lump-sum allowances would be the determination of the normal useful life of assets and the choice of the appropriate discount rate. The length of life and the depreciation profile (straight line, declining balance, sum of the years’ digits, or other) would be set as under normal depreciation. The choice of the discount rate is debatable. With stable prices, it might appropriately be set equal to the interest rate on loans to creditworthy business borrowers. Under inflationary conditions, however, interest rates reflect both pure time discount and an allowance for the expected loss in the purchasing power of money. Use of the market interest rate under these conditions would mean that the discounted lump-sum allowance would not serve as a method of neutralizing the effect of inflation on depreciation allowances. If, however, the lump-sum allowances reflected discounting only for “pure” time discount, granting them would be an administratively simple method of alleviating the effects of inflation. To illustrate, if pure time discount (that is, the interest rate that would prevail if the price level were expected to be stable) is 4 per cent and prices are expected to rise by 10 per cent a year, the market interest rate should be approximately 14 per cent, consisting of 4 per cent for pure time discount and 10 per cent for the expected decrease in the purchasing power of money. The discounted lump-sum allowance for an asset that would be eligible for 5-year, straight-line depreciation under normal accounting (and would have no salvage value) would be 89.0 per cent of its cost with a 4 per cent discount rate, but only 68.7 per cent of cost with a 14 per cent discount rate. For assets with longer lives, the difference would be more dramatic; for a 20-year asset, the lump-sum equivalents would be 68.0 per cent at a 4 per cent discount rate and 33.1 per cent at a 14 per cent discount rate.

Enterprises that were able to make full use of the lump-sum allowances would be protected from the effect of inflation on the adequacy of depreciation allowances, regardless of the actual rate of inflation over the useful life of their assets. In practice, it is impossible to ascertain exactly what the pure time discount rate should be. Economists have devised methods for estimating the expected rate of inflation—and of deriving the pure time discount rate from market interest rates—but these may have more appeal to econometricians than to legislators or tax administrators. In past periods of price stability, long-term interest rates in the range of about 3 per cent to 5 per cent prevailed in countries with developed financial structures, and a figure in that range might be an acceptable rate of discount for use in a lump-sum depreciation system. An objection to such a system is that it would tend to favor established firms, with large incomes against which the discounted lump-sum allowances could be immediately written off, as against new or less profitable firms, which might lack sufficient income to absorb the full allowance. That discrimination could be mitigated by carrying forward unused allowances and increasing them by an interest adjustment (which should approximate the market interest rate rather than a pure time discount rate, in order to provide some protection against erosion, caused by inflation, of the real value of amounts carried forward).

inventory valuation

The valuation of inventories (stocks) is another item affecting the assessment of taxable profits that is distorted by inflation. The effect of inflation can be reduced by reversing the traditional assumption that goods are sold in the order of acquisition (the first-in-first-out, or FIFO, method) and assuming that those most recently acquired are sold first (the last-in-first-out, or LIFO, method). The LIFO method is fairly simple when a few homogeneous items are involved, but it becomes complex when a variety of goods are held in the inventory. In the latter cases, adjustment of the inventory value by reference to one or more price indices is feasible (Rottenberg (1980)). As is true of depreciation allowances, the use of a single broad index measuring change in the purchasing power of money is simpler than using indices of the prices of the particular kinds of goods held in the inventory, and in my judgment the simpler method would also be more equitable. Both methods use economic statistics to supplement accounting data. The tax department should inform itself about the coverage and reliability of the price index or indices.

comprehensive adjustments for inflation

Adjustment of depreciation allowances and inventory values would be, at best, an incomplete correction for inflation. A comprehensive system would have to take account of gains and losses on liabilities and assets fixed in money terms and of some other items (Casanegra de Jantscher (1976); Aaron (1976)). It would have important administrative implications, but I shall not discuss them, as I see little inclination to adopt such a system outside countries that experience extremely high and prolonged inflations.

time discount

The time discount and interest rate factors, to which reference has been made in the comments on depreciation allowances, are economic concepts that merit close attention in tax administration. Even with a stable price level, it is always advantageous to a taxpayer to postpone payment because he can use the money meanwhile to gain interest or profit income by lending or investing, to avoid paying interest on amounts that he would otherwise borrow, or just to remain more liquid and thus to be in a position to meet unforeseen needs or to take advantage of attractive opportunities. Time discount may seem less relevant for a government than for an individual or private enterprise; but as most governments are debtors, early receipt of revenue usually allows savings on interest payments. When revenue is to be used to finance productive investments in the public sector, the advantage of early receipt is similar for the state and for a profitable private enterprise. Under inflationary conditions, the time of payment of taxes becomes much more significant, because postponement of payment reduces its real value (Hirao and Aguirre (1970); Tanzi (1977) and (1980)). As most countries are currently experiencing inflation, it is especially important for tax departments to be conscious of the costs of delay in tax collection and to take all reasonable and feasible steps to reduce the lag between accrual of tax liability and collection. It may be worthwhile to give up some refinements of assessment in order to speed up collection. Delinquent accounts should be subject to realistic interest rates or should be indexed by reference to a suitable broad price index. Collection at source and other current payment arrangements for direct taxes should be strengthened and extended. Frequent remittances of withheld taxes and excises and sales taxes should be required.

interlocking systems

Economists have been intrigued by the possibility of devising interlocking systems that would rely on the self-interest of taxpayers to induce them to divulge information to the tax department. Usually their suggestions go no further than the familiar idea of matching invoices with value-added tax or sales tax declarations and cross-checking between these declarations and income tax returns, and economists are rarely cognizant of the practical difficulties of these operations. Kaldor, in his report on India (1956, p. 2), advanced a more elaborate proposal that he described as follows:

The five taxes—income tax, capital gains tax, annual wealth tax, personal expenditure tax and the general gift tax—would all be assessed simultaneously, on the basis of a single comprehensive return; and they are “self checking” in character, both in the sense that concealment or understatement of items in order to minimise liability to some of the taxes may involve an added liability with regard to others, and in the sense that the information furnished by a tax-payer in the interest of preventing over-assessment with regard to his own liabilities automatically brings to light the receipts and gains made by other tax-payers.

Kaldor conceded that the system would not prevent evasion in cases where both the seller and buyer of goods or investment property had an interest in understatement.

Higgins (1959, ch. 23, pp. 524-44) carried Kaldor’s idea to its logical conclusion and proposed a fully integrated “self-enforcing incentive tax system for underdeveloped countries.” It would include: (1) a personal income tax applying to capital gains as well as ordinary income; (2) a corporation income tax; (3) a general sales or turnover tax; (4) a wealth tax with graduated rates; (5) a tax on excess inventories defined as inventories in excess of those “normal” in relation to sales in the trade or industry; and (6) a personal expenditure tax. Liability for all these taxes would be computed by the statistical office, leaving the tax department “only the simpler task of collecting the tax after each taxpayer’s liability had been clearly established.”

In addition to the usual collation and cross-checking in respect of income tax and sales tax or value-added tax, the Kaldor-Higgins schemes would take advantage of the interrelations between expenditure tax, wealth tax, and capital gains tax. Personal expenditure would be defined as the excess of income over saving, and saving would equal the increase in net wealth. Hence, a person who understated his personal expenditure by overstating his saving would incur additional liability for wealth tax. A seller of investment property who understated his capital gain by either omission or underreporting would deprive the buyer of evidence of an investment made and could thus expose him to an overstatement of his expenditure tax liability (provided the buyer fully reported his own income). The excess inventory tax would be intended to discourage underreporting of sales and thus to aid in enforcing the sales or value-added tax and the income tax of the seller.

These proposals appear so unrealistic that a detailed critique is not worthwhile. In my opinion, their authors exaggerated the proclivity of taxpayers to refined calculations, the capacity of tax departments to use the great mass of data that would be generated, and the receptivity of governments to fiscal innovations. I suspect that most tax administrators will regard the idea of a self-enforcing tax system as fantastical. Even if put into operation, the proposed systems would not prevent evasion in cases in which both parties to a transaction omit it from their records or understate its amount. Both parties could evade the related taxes, and as no conflict of interest would arise between them, neither would have an economic incentive to report correctly. For example, a merchant who failed to report part of his purchases and sales could evade income tax, sales or value-added tax, wealth tax, excess inventories tax, and personal expenditure tax. His trading partners could do likewise. More subtly, the classification of consumption expenditures as business costs, or the consumption by a merchant of part of his stock in trade, could allow evasion of income tax, sales or value-added tax, and personal expenditure tax.

Shoup’s conclusion (1969, p. 436) on the self-checking proposals is judicious; the separate taxes would reinforce each other in the sense that the incremental cost of adding, for example, either a personal expenditure tax or a wealth tax to a revenue system that included the other tax would be smaller than the cost of administering either tax alone. But a more efficient use of administrative resources in relation to revenue might be obtained by concentrating on the income tax and the sales tax or the value-added tax.

incentives for supplying information

A less ambitious method of using economic incentives to ensure accurate reporting of taxable transactions is to offer a reward to the nontaxable party for insisting on proper documentation. That method has been used in some jurisdictions in connection with retail sales taxes or value-added taxes at the retail level. Retailers may be required to prepay the tax by buying coupons or stamps to be given to customers (who usually are not legally taxable parties, though they may bear the economic burden of the tax). Alternatively, retailers may be required to give customers copies of prenumbered invoices supplied by the tax department. Customers are offered an inducement to demand the coupons, stamps, or invoices by the government’s willingness to redeem them for a small fraction of their face value or to enter them in a lottery drawing (Hart (1967)). As the redemption value or lottery prizes must be small in relation to the tax for the scheme to be attractive to the government, they will not be strong incentives to forgo collusive understatement of large sales and division of the evaded tax between seller and buyer. This kind of collusion is less feasible in small transactions; but for these, the reward to consumers may be too small to motivate many of them. On the whole, the device seems likely to be weak and capricious in operation.

By direct extension of the incentive argument, advantages could be attributed to pecuniary rewards for informers who bring evasion to the attention of the tax department and to bonuses for tax officers who uncover additional tax liabilities. The former practice is widespread but can be socially divisive if too freely used. The risk that incentive compensation payments to tax officers may stimulate excessive zeal and misdirected efforts and the difficulty of identifying the contributions of individual officers are generally recognized and have inhibited the use of such payments. Incentive compensation functions best where the output of individuals can be clearly identified, measured, and subjected to quality controls, which is rarely possible in tax administration.

market-enforced self-assessment of property

Another proposal for inducing private citizens to do some of the work of tax administration calls for a market-enforced system of self-assessment of taxable real estate. It would be an extension of a provision found in several countries according to which taxpayers are required to declare the value of their property for taxation, subject to the threat of being compelled to sell it to the state at the declared value, usually increased by a margin of tolerance. That provision has not been very effective in ensuring full valuations because the state is generally not interested in acquiring property and lacks knowledge of the values of particular pieces of property. The suggested modification would rely on private citizens who are informed about market conditions to act as the enforcing agents. Self-assessed values would be made public, and an owner would be compelled to sell his property to anyone for the declared value plus a premium of, say, 10 per cent to 20 per cent (Harberger (1965); Strasma (1965)). Knowledgeable speculators motivated by a lively sense of acquisitiveness would be counted on to prevent undervaluations. Harberger (1965, pp. 119-20) characterized the scheme as “simple and essentially foolproof” allowing “no scope for corruption” and having “negligible costs of administration.” He conceded, however, that “the beauty of this scheme, so evident to economists, is not . . . appreciated by lawyers. …” Lawyers and other skeptics have less confidence in the impersonal operation of market forces and more sympathy for property owners who might ignorantly misvalue their property or cautiously overvalue it in order to ensure themselves undisturbed possession. Possibilities would exist for abuse by purchasers acting on inside information about developments that would increase property values. Another version of the proposal would dispense with the forced-sale provision and would impose a penalty on the owner who declined to sell for the declared value plus the stipulated margin, with the penalty to be shared by the frustrated bidder. This would meet some of the objections to the proposal by converting it into a suggestion for a special form of reward for informers but could expose property owners to blackmail. Both versions of the proposal would be less threatening to owners who are well informed, ready to take risks, mobile, and not sentimentally attached to their property than to other owners and would offer opportunities to alert and well-financed bidders. The real costs of valuing property would not be avoided (though they might be reduced), as bidders would need either to make expert appraisals or to engage others to do them (Holland and Vaughn (1969); Bird (1974)).

penalties

By application of reasoning that is used for the valuation of investment risk, economists have attempted to evaluate the efficiency of penalties for tax evasion. The probable money value of a penalty is the product of the probability of its application and its amount—pA, where p is the probability and A is the amount. For example, a $1,000 penalty that is expected to be applied in only 1 per cent of the cases in which it might legally be assessed is actuarially equivalent to a $100 penalty applicable with a 10 per cent probability or a $10 penalty that is regarded as certain to be applied. Some theorists have suggested that enforcement efforts be guided by this principle. They assume that administrative costs and the probability of detecting evasion are directly related and infer that economies in costs can be obtained without sacrifice of revenue if the amount of penalties is increased so that pA is held constant (Allingham and Sandmo (1972); Kolm (1973)).

A theoretical objection to this suggestion is that it assumes that taxpayers are risk neutral—that is, that they are equally deterred by, or equally indifferent to, actuarially equivalent penalties. Observation indicates, however, that some people are risk averters and are concerned to avoid risks that are actuarially small. They would be likely to be more deterred by large penalties than the formula indicates and would suffer a burden of risk-bearing that, in principle, should be recognized as a social cost and deducted from the savings of administrative costs obtained by reducing p while increasing A (Polinsky and Shavell (1979)). Other people appear to be risk seekers and are ready to gamble against highly adverse odds if the possible prize is large; buyers of lottery tickets come to mind. They might well pay little attention to penalties that were uncertain. There is no way of knowing how taxpayers are distributed among the risk neutral, the risk averters, and the risk seekers. Hence a tax department could not confidently select values of p and A that would allow it to reach its revenue objective at minimum cost, even if it had finely calibrated estimates of the cost of attaining various levels of p. Taxpayers, for their part, would not have reliable information about p and might behave erratically.

Most important, the purely economic approach is inconsistent with widely accepted beliefs about justice and equitable taxation. A system that substitutes seldom applied but severe penalties for active tax enforcement is liable to abuse and is likely to be regarded as arbitrary and unjust and to result in the progressive deterioration of voluntary compliance. Penalties have not only the function of deterring evasion but also, as a legal writer has pointed out, the objective of “keeping complying taxpayers satisfied with their compliance.” And “this satisfaction may at least to some extent be a matter of retribution, a familiar aspect of criminal law” (Oldman (1965), pp. 317-18).