The Power to Tax: Analytical Foundations of a Fiscal Constitution

The Domain of Politics

What a government ought to do, is a mysterious and searching question, which those may answer who know what it means; but what other men ought to do, is a question of no mystery at all.... The question is not why governments are bound not to do this or that, but why other men should let them if they can help it. The point is not to determine why the lion should not eat sheep, but why men should eat their own mutton if they can.

—Perronet Thompson, "Greatest Happiness Principle,"
Westminster Review, 21 (July 1829);
reprinted in Utilitarian Logic and Politics, p. 141

Our concern in this book is with the fiscal constitution, with alternative means of constraining government's power to tax and to spend. We are not directly concerned with nonfiscal constraints on governmental powers, whether these be constitutionally or otherwise imposed. Our analysis would, however, be seriously incomplete if we did not recognize the relationship between fiscal and nonfiscal constraints. The potential substitutability between fiscal and nonfiscal constraints keeps us from claiming that the former are in all circumstances absolutely essential for keeping governments within appropriate limits. Conversely, the complementarity between fiscal and nonfiscal constraints keeps us from extending to wild extremes the argument that fiscal limits alone will accomplish the purpose of constraining Leviathan.

The construction of this book can perhaps best be interpreted as being predicated on the assumption that an appropriate set of nonfiscal rules is operative, setting the context within which alternative fiscal constraints can be analyzed and evaluated. Broadly speaking, the nonfiscal constraints in existence are simply assumed to be those that we generally observe in modern Western democracies. Beyond this, however, our analysis requires the assumption that the imposition of specifically fiscal constraints will not itself generate successful "escape" from fiscal limits through nonfiscal channels. Although we shall make no attempt at exhaustive analysis here, this chapter is devoted to an examination of the interdependencies between those means of limiting Leviathan which involve restrictions on the taxing power and those which do not. In particular, we include in the latter category restrictions on the domain of government spending, as well as specific types of procedural restrictions on the nature of political process.

8.1. Procedural Constraints on Political Decision Making

As we have noted earlier, our analysis departs from the conventional public-choice framework in which the government is modeled as a relatively passive responder to the demands placed upon it by voters-taxpayers-beneficiaries. We have argued in Chapter 2 that constitutional guarantees of electoral competition among politicians and parties are not sufficient to ensure that the Leviathan-like proclivities of governments are always kept in bounds. In that earlier discussion, we assumed a setting for electoral competition characterized by majority-voting rules both in elections among candidates for office and in choices among policy actions within legislative assemblies.

Once the inadequacies of electoral competition are acknowledged, there are two directions for reform. We might think of imposing constitutional constraints directly on the taxing-spending power, which is our basic subject matter in this book, or we might think of imposing constitutional changes in the procedures or rules through which political decisions are made. With some reforms of this latter type, resort to direct fiscal constraints may be neither necessary nor desirable.

By way of illustration, we can do no better than look to the seminal work of Knut Wicksell—work that has already been referred to several times in this book. Wicksell did not seek reform in the fiscal constitution, as such, but he did seek to achieve more responsive government through changes in the procedures through which taxing-spending decisions were made. In his idealized model, he proposed a rule of unanimity in legislatures. But he acknowledged the difficulties of such a rule in practice, and he seemed willing to settle for a qualified majority approval for all public outlay projects amounting to perhaps five-sixths of the total voting membership. Further, and at a more attainable level, Wicksell proposed that taxing and spending legislation be simultaneously considered. To Wicksell, legislatures should operate under an obligation to provide explicitly for the financing of each and every spending program authorized. In so doing, he recommended a form of earmarking somewhat similar to that indicated by the discussion in Chapter 7, but for rather different reasons. In the context of the modern discussion of constitutional policy, the proposal that government be required to balance its budget is perhaps a reasonably close analogue to the sort of procedural requirement that Wicksell suggested. We shall have more to say about "balanced-budget limitation" in Chapter 10; here, we simply note its status as a variety of "procedural constraint."

The characteristic feature of procedural reform lies in its explicit avoidance of imposing constraints or limits on the results or outcomes of collective decision making. The proposals listed above, and others that might be discussed, operate directly on the process of reaching or making collective decisions; they do not restrict directly the particular range of outcomes that might be attained. This is not, of course, to deny that there must exist some feedback relationship between the predicted effects of procedural changes and the direction of predicted results. Indeed, some such predictions must be used to inform any choice among alternative procedures. But an advantage of purely procedural reform lies in the outcome flexibility that is retained.

In earlier chapters, we have found it to be convenient to model government as a monolithic entity that seeks to maximize net revenue or surplus. This model is useful in allowing us to discuss alternative tax constraints; the single maximand offers a device that allows predictions to be made about the behavior of an unconstrained Leviathan. It is clear that any discussion of procedural reform, as a possible substitute for fiscal constraints, must depart somewhat from such simplistic Leviathan construction. In the earlier analysis, the objective for constitutional change was one of limiting governmental behavior, of fencing government in, so to speak, without at the same time changing its "character." The procedural changes suggested by Wicksell, however, are directed at modifying the structure of governing itself and, hence, changing the behavioral model appropriate to describe governmental actions. The surplus-maximizing objective would not be appropriately assigned to a government that operated within the Wicksellian procedural constraints. Such a government would not evoke the Leviathan metaphor, and the whole analysis and discussion of possible constitutional reform would be quite different from that presented in this book.

We have written this book about potential change in the fiscal constitution on the presumption that major changes in governmental procedures for reaching decisions do not occur, and that government, as we observe it to operate, will continue essentially as it is now institutionally characterized. As our other works should have indicated, the stance taken here does not suggest that we personally prefer the strictly fiscal constraints over the procedural ones. Both means of controlling governments must be analyzed and examined. Our efforts here reflect nothing more than a division of labor.

8.2. The Rule of Law: General Rules

A different, if somewhat related, approach to that previously discussed as procedural may be examined under the rubric of the "rule of law." This approach, perhaps best identified by the arguments of F. A. Hayek in his The Constitution of Liberty (1960) and subsequent writings, aims to restrict the structure or pattern of allowable outcomes rather than either the procedures for reaching outcomes or the specific outcomes themselves. This objective is to be accomplished by requiring that outcomes of the fiscal process conform to the familiar and time-honored "rule of law." By this Hayek means that all rules involving taxes must be general. They must be universally applicable to all members of the political community, whether or not these persons are inside or outside the subset of persons that make the governmental decisions. This approach essentially reflects the specific application of the traditional legal norm of "equality before the law" to the taxing activities of government.

Historically, the constitutional requirement that taxes be uniform seems to stem directly from this legal norm. As our earlier analysis has indicated, tax uniformity may be generated as one means of restricting the revenue-gathering potential of government. In our analytic setting, such uniformity was taken to require nondifferentiation in tax rates among persons or among groups or among different tax bases. However, insofar as this sort of uniformity is to be found in modern fiscal systems, it probably reflects residues of the legal equality norm rather than any overt consciousness of the need to restrict Leviathan's fiscal appetite. The tenuousness of existing uniformity precepts is evident; taxes do and may vary among persons and among groups; different tax bases are subjected to differing levies. At best, existing legal norms, as currently interpreted, serve only to rule out totally arbitrary discrimination in taxation. For example, the United States government could not (or at least has not, to date, been able to) impose one rate of income tax on Catholics and another rate on Protestants, one rate on blacks and another rate on whites, one rate on women and another rate on men. If the government tried any such tax discrimination, it seems likely that it would be ruled out of court on constitutional grounds. Of course, were government able to discriminate beyond currently allowable limits, it could generate more revenues from its current sources, as our analysis in Chapter 4 indicated. The accepted legal norms do, therefore, limit Leviathan. The taxing power is not unrestricted.*99

This Hayekian interpretation of the rule of law as requiring generalized uniformity suggests two questions. How effective are existing legal limits? And how could a generalized extension of the legal norm of uniformity be used so as to constrain Leviathan beyond those limits already discussed?

These questions immediately raise issues of definition. How is uniformity or equality to be defined for tax purposes? Should "equality before the law" in taxation require equal payments by all persons in the polity? Or should such equality be interpreted to require that all persons in the jurisdiction confront equal rates of tax, hence allowing for proportional but not regressive or progressive tax structures? Hayek's argument to the effect that a proportional tax structure would meet the requirement for generality whereas a progressive rate structure would not do so seems to be dangerously arbitrary. To defend such a position requires considerably more analysis than Hayek has provided.

If we leave the questions of definition aside, however, does the generality requirement impose checks on the taxing proclivity of Leviathan? Would the stipulation that persons who are inside the ruling coalition face the same tax structure as those who remain outside the ruling coalition restrict the taxing power as such? With no accompanying constraints, would the mere fact that members of the ruling group themselves pay taxes like everyone else effectively result in less total tax gathering?

It seems clear that it would, at least in some plausible settings. Consider a simple example in which Leviathan takes the form of an exploitative majority coalition. Suppose further that the uniformity requirement on the tax side involves the requirement that revenues be raised by a completely general income tax. To avoid definitional questions of the type indicated above, suppose that all citizens-taxpayers have identical tastes and pretax incomes, so that the choice between progression and proportionality involves no question of discrimination between individuals. Given here that the majority and minority are of virtually identical size, the contribution that members of the majority make to each dollar of revenue collected is virtually 50 cents. It may therefore seem as if the majority would rationally vote for maximum revenue: it makes a net gain of almost 50 cents for every dollar of revenue collected. This conclusion would be valid if taxes were lump-sum. Uniformity in and of itself—specifically in the absence of any restrictions on the power to tax—would not limit Leviathan at all: the exploitative majority would end up with all the income. In the presence of base limits of the sort we have analyzed earlier, however, the uniformity requirement does serve as an additional means of constraint. For where the tax to be levied involves limited revenue, it also involves an excess burden for all taxpayers. This excess burden is present because each of the taxpayers, including those in the majority coalition, will rationally attempt to minimize his own tax payment by substituting away from the taxed good. At some level, the marginal excess burden sustained by members of the majority coalition will be equal to the marginal transfer received from members of the minority. The majority coalition will not seek to push the extent of redistribution beyond this point, because they lose more in additional excess burden from the tax system per dollar of additional transfer received than they gain in transfers.

Figure 8.1

Consider a simple example. The tax base is taken to be money income, X, leaving leisure exempt. The "demand" curve for X for the economy as a whole is depicted as the D curve in Figure 8.1. Given the assumption of identical tastes, this "demand" for X can be divided into two (virtually) identical parts, with D_m being the aggregate demand for X over all members of the majority coalition. With uniform taxes, the most revenue that can be obtained from the taxation of X is shown by the shaded areas, R*, in Figure 8.1: one-half of this revenue (or one-half of any other quantity of revenue, for that matter) will be paid by members of the majority coalition. The welfare loss sustained by taxpayers at this maximal level of revenue will, in the linear case, be precisely one-half the maximum revenue yield (as demonstrated in Chapters 3 and 4).

Figure 8.2

We can transfer the information in Figure 8.1 into a more usable form by depicting in Figure 8.2 the costs to majority members of any transfers they receive. Such costs are of two parts: first, since taxes are uniform and the majority is a bare 50 percent of the population, majority members pay in taxes 50 cents out of every dollar of transfer revenue. This is shown as the horizontal line at 50 cents in Figure 8.2. Beyond this, however, there is the excess burden of the tax system. One-half of this excess burden, also, will be borne by the decisive majority. To determine total marginal costs to the majority, including excess burden, of higher transfer levels, we need to add to the 50-cent line in Figure 8.2 the marginal excess burden per dollar of revenue raised. Clearly, the addition to welfare cost or excess burden associated with an extra dollar of revenue approaches infinity as revenue approaches its maximum. For as the tax rate approaches the revenue-maximizing rate t*, the excess burden is increasing at an increasing rate while the revenue level is increasing at a decreasing rate. For tax-rate increases in the neighborhood of t*, excess burden rises while the revenue increase is zero: the extra welfare loss associated with an additional dollar of transfers (or an additional dollar of revenue) is quite literally infinite.

On this basis, we can draw from the 50-cent line upward a curve labeled TMC in Figure 8.2 that depicts the total marginal cost to majority members of various levels of transfer, including that part of the excess burden of the tax system that majority members bear. At R*, the TMC curve will approach infinity, indicating that the excess burden per dollar of extra revenue is approaching infinity. The area between TMC and the 50-cent line up to R* measures the total excess burden of R* borne by majority members: since the majority is one-half of the electorate, this will be one-half of aggregate excess burden, or ½ W* in Figure 8.1.

Given TMC, we are in a position to predict the level of transfers that the decisive majority will rationally vote for. Since the value of a dollar transfer to the majority is always $1, the level of transfers emergent under majority rule will occur where the TMC curve intersects the $1 line. At this point, the cost of an extra dollar's transfer, including the excess burden of taxes that the majority pays, is exactly $1. We depict this point by R_m in Figure 8.2. Clearly, R_m will be to the left of R*: the majority will not push taxes to the revenue-maximizing limit. The revenue level R_m depicts the total revenue raised for transfers in the presence of a generality requirement on the tax side.

Let us now suppose that the generality requirement does not apply. The majority will now apply taxes to the minority alone, and will do so up to the revenue-maximizing limit. Since the minority is one-half the population, revenue will be ½ R*; and the majority will receive all that revenue in transfers. The excess burden sustained by minority members will be one-half of this maximum revenue, or ¼ R*, given linearity assumptions.

The implications of the generality requirement for a given tax base can now be gauged by appeal to a direct comparison of the two equilibria. In the absence of generality: aggregate tax revenue is ½ R* dollars; the net benefit of transfers to the majority coalition is ½ R* dollars; and total excess burden is ¼ R*. In the presence of the tax generality requirement: aggregate tax revenue is R_m; the net benefit of transfers to the majority coalition is the area between TMC and the $1 line up to R_m in Figure 8.2; and total excess burden is twice the area between TMC and the 50-cent line up to R_m in Figure 8.2. Since R_m is less than R*, and the benefits to majority members in the generality case are less than ½ R_m, those benefits are also less than ½ R*. Majority members lose by the generality constraint. In the case of general taxation, however, both majority and minority members sustain excess burdens due to taxation, and the total excess burden in this case may exceed that in which only minority members pay taxes. In sum, the generality requirement clearly reduces the fiscal exploitation of the minority, but it will increase total tax revenue and it may increase the aggregate excess burden attributable to taxation. The net effects of the generality rule on the tax side are therefore not entirely unambiguous.

The model outlined here in some sense makes the case for the effectiveness of uniformity constraints in its most generous form. As the size of the decisive coalition falls from one-half of the electorate to an even smaller proportion of the citizenry—to a ruling class, a bureaucratic elite, a president-premier-king—the bite of any uniformity constraint falls: the contribution of the decision makers to the cost of transfers to themselves becomes smaller and smaller. In terms of Figure 8.2, a reduction in the size of the dominant or ruling coalition would have the effect of displacing the TMC curve downward throughout its range, and hence shifting the solution toward R*.

In the discussion of this section, we have set aside the possibility that the Leviathan surplus should itself be subject to tax. Although this is not inconceivable in the case of the exploitative majority, it becomes rather strained where the transfers take the form of public expenditures of particular benefit to the ruling class or the bureaucratic elite.

8.3. The Domain of Public Expenditures

The ambiguity of the uniformity norm when applied only to the tax side of fiscal operations should not be taken to prejudice the question of its extension to the expenditure side. As it happens, requirements for Hayekian equality of treatment or generality have historically never been applied to the spending side: on the contrary, quite arbitrary discrimination in the distribution of the benefits of public spending among persons and groups seems to be characteristic of modern fiscal systems.*100 But this fact of fiscal experience, in itself, offers no logical basis for rejecting legal requirements for uniformity on the spending side of the fiscal account as a possibility to be considered. But other types of constraint on the spending side deserve some attention here, and are historically more common. In this section, we seek to explore briefly some of these restrictions on expenditures, along with the implications of generality requirements.

If we imagine, in the foregoing example, that restrictions on generality were extended to the spending side of the account, it is clear that the potential for redistribution in the majority's favor would be entirely removed: the requirement of an identical share in revenue for all would obliterate the possibility of any individual obtaining more than he paid in taxes. This would not, to be sure, hold in a slightly more general setting in which pretax incomes differ, provided that uniformity on the tax and expenditure sides is defined asymmetrically. For example, if proportional taxation is taken to satisfy the uniformity rubric on the tax side, but equal per capita shares are required on the expenditure side (a structure the so-called "linear negative income tax" simulates), some redistribution would still occur. There would clearly be less redistribution than where no uniformity restrictions were imposed on the transfer pattern, since richer majorities could never transfer income in their own direction; but where uniformity is defined symmetrically on both sides of the fiscal account, exploitation of a minority by a decisive majority is removed.

Somewhat similar restrictions on the power of an exploitative majority are achieved by the requirement that the spending activities of government be restricted to the provision of genuinely "public goods" of the pure Samuelsonian type. Such goods are by definition equally (and totally) consumed by all citizens; consequently, if there is uniformity on the tax side, the possibility of a majority redistributing resources in its own favor is substantially removed. It could indeed be argued that the restrictions on the domain of government activity which are extant in most written constitutions were drafted with such considerations in mind—they can certainly be rationalized along such lines.

In fact, the requirement that government spending be restricted to pure public goods is unnecessarily strict. The domain of government activity could be extended to those goods which are nonexcludable, even if jointness is incomplete: the crucial characteristic for these purposes is that each individual's consumption—or access to consumption—be identical. Indeed, government could be allowed to finance and provide fully partitionable "private" goods and services embodying no jointness efficiencies at all provided that equal quantities be made available to all. In all such cases, the distributional consequences are essentially the same as when transfers are made to all on an equal-share basis. To the extent that uniformity on the tax side is interpreted to require anything other than equal absolute amounts of tax per taxpayer, possibilities of redistribution remain but are substantially restricted by the equal-share requirement on the expenditure side.

Two somewhat different and more general points should be made in relation to this discussion. First, we have interpreted the power to redistribute negatively, in the sense that such power offers Leviathan the opportunity to redistribute in its own direction. This interpretation is consistent with the basic thrust of our model of constitutional choice: in this model, there are no preferences for redistribution as such at the constitutional level, although of course some redistribution might emerge to the extent that taxpayers-donors wish to make transfers to recipients in-period.*101 It is, however, clear that if the individual exhibits preferences over distributional matters at the constitutional level—whether they be Rawlsian, or otherwise—there is no guarantee that assigning the power to redistribute to government will be desirable: the transfer patterns that emerge from the assignment of such power may not at all satisfy any set of moral norms.

Second, restrictions on the domain of public spending of the sort we have been discussing may be implemented under an umbrella of rules against bureaucratic and political "corruption." That such rules exist and that institutions for their enforcement exist under most constitutions can hardly be questioned. These rules clearly restrict the domain of public activity in one sense. Their significance lies in the inhibitions they place on the ability of those exercising discretionary power to appropriate resources directly. Neither such rules nor the restrictions on spending discussed earlier serve to prevent Leviathan's appropriation of revenue surplus by indirect means—perquisites of office, overexpanded bureaucracies, and so on. For this reason, even if such restrictions could be expected to be totally effective—which seems unlikely—there would remain a role for tax limits of the type we have explored in preceding chapters.

8.4. Government by Coercion

To this point, we have examined nonfiscal constraints that might serve as possible substitutes for fiscal constraints on the activities of government. We have been interested in determining the extent to which the presence or potential introduction of such nonfiscal limits might reduce the need for any imposition of constitutional controls over the power to tax. A wholly different set of interdependencies emerges when we look at nonfiscal constraints as necessary complements to the fiscal controls. Here the issues to be analyzed concern the potential effectiveness of overtly fiscal constraints in view of their critical dependence on the maintenance and enforceability of nonfiscal instruments that will serve to prevent the former from being successfully avoided by government with Leviathan proclivities.

If we remain strictly within the revenue-seeking model of government, the problems to be discussed here do not formally arise since, by our somewhat artificial assumption, the single maximand is tax revenues. Hence, any constraint on the power or authority to tax must be effective by definition. Once we depart from this artificial construction, however, to grant that Leviathan's instrumental desire for tax revenues is for the purpose of ultimately acquiring command over real goods and services, the possible avoidance of any explicit tax limit or constraint must be reckoned with. If, when confronted with a legal limit on its taxing power, government should find it relatively easy to secure real goods and services by nontax means, there would be little purpose in the whole tax-limit exercise.

As we have noted, there exist legal-constitutional restrictions on the government's power to take, and, indeed, without some such restrictions there would presumably be no raison d'être for the institution of taxation itself. The government could, in the absence of legal constraints on the taking of power, simply coerce persons into relinquishing possession of the goods and services it wants. In the established legal traditions of Western nations, governmental coercion of this variety has been and continues to be considered beyond the legitimate exercise of state authority. Certain exceptions prove the rule—conscription for military services and eminent domain are two. In the latter case in particular, governmental taking is accompanied by the legal requirement for "just compensation," which becomes a part of the more general legal requirement for "due process."

The existence of legal limits to the taking power does not, of course, guarantee against an extension of such power. And it must be acknowledged that any imposition of constraining tax limits on government will create additional incentives for the direct "taking" of goods and services from persons, independently of the fiscal channel that involves first, taxation, and subsequently, governmental purchase of such goods and services in the marketplace. If tax limits are expected to be effective at all, legal restrictions on governmental taking power must be maintained in the face of increased incentives.

If the agents of Leviathan, the rulers, seek to use the fiscal system or the taking power for the purpose of acquiring command over goods and services for their own direct consumption or use (over and above that share which they might legally be required to return to members of the community as public-goods benefits), the potential dangers embodied in dramatic extensions of the taking power may not loom as significant. Established legal traditions are important, and overt coercion on the part of governmental agents could presumably be held within reasonably narrow limits, despite enhanced incentives offered to such agents. But a much more severe, and possibly intractable, problem arises when we allow the agents of Leviathan to incorporate what we may call "nonpersonal" arguments in their utility functions, when we allow these agents to promote or to seek to further a set of "[jectives" that can be plausibly "legitimized" on what may be called "public interest" or "general welfare" grounds. In such a setting, tax limits per se may be ineffective in containing government. This seems to be a fact that must be squarely faced.

Consider a familiar example. Prior to the mid-1960s, individual citizens in the United States were, for the most part, unconstrained in their access to and their usage of the air- and waterways except to the extent that limits were inherent in the laws of nuisance as carried down from the English Common Law. In a reconstructed scenario different from the one actually followed, the government could have, in recognition of the pollution-environmental problems, declared "clean air" and "clean water" to be "public goods." It could have then levied taxes for the financing of such "goods," which in this case would have involved the purchase of individuals' agreements to reduce polluting activities. As we know, such a scenario was not followed; government did not utilize the fiscal route to the accomplishment of its avowed and newly found environmental objective. Instead, the government simply enacted laws that embodied direct prohibitions on specific types of activity, or in lieu of this, enacted laws that authorized administration agents (bureaucrats) to define the scope of prohibited activities. Individuals were simply prevented from being allowed to do things that had previously been available to them. In a real sense, government used the taking power; it took valued rights from persons, and without questions of due process being raised, and without compensation. The alleged "public good" was secured without resort to taxing and spending.

It seems evident that the presence or the absence of tax limits would have had, and could have, relatively little effect on extensions of governmental activity of the sort exemplified in the wave of environmental regulation of the late 1960s and the 1970s in the United States—air and water pollution, automotive and occupational safety, and consumer protection. Indeed, it may be persuasively argued that the interferences with personal freedoms reflected in regulatory laws of this nature present more serious issues than the more indirect extensions of governmental power by means of the fiscal process and reflected in explicit taxation.*102

There are, of course, relationships between the extension of direct governmental regulation and the size of the budget. Regulatory action implies regulatory agency, and agency in turn implies a regulatory bureaucracy, which, in its own turn, implies bureaucrats who work for money rather than peanuts. And as they impinge on agency budgets, tax limits can have a constraining influence. As the pollution control examples reveal, however, differing means of accomplishing differing governmental objectives have widely differing budgetary implications. The environmental regulatory bureaucracy requires financing, and from tax revenues, but the fact that it is empowered to regulate directly rather than through the fiscal process of spending on compliance very substantially reduces the bureaucracy's demands on the government treasury. It is relatively easy to envisage a federal budget making up no more than 20 percent of GNP that would reflect more interference with personal liberties than an alternative budget of 40 percent of GNP, but with substantially less direct regulation.

As economists here, we might call upon ceteris paribus and suggest that, under a given legal environment concerning direct regulatory action by government, the imposition of fiscal constraints must remain potentially effective. The government that commands 20 percent of GNP is less intrusive in the economy than the government that commands 50 percent, provided that the legal setting for direct regulation in the two cases is at all comparable. But this sort of argument would ignore the very real feedback that exists between the government's proclivity to regulate directly and to tax. A government subjected to tax-limit pressures will surely be predicted to exert more efforts through the legal process toward opening up direct regulatory channels.

There is little that we can do here other than to acknowledge the "limits of tax limits" in this respect. Tax limits, or fiscal constraints generally, can be expected to curb government's appetites to the extent that the utility function of governmental decision makers contains arguments for privately enjoyable "creature comforts," for final end items of consumption. Such constraints become much less effective, and may well be evaded, if the motive force behind governmental action is "do-goodism." The licentious sinners we can control; the saintly ascetics may destroy us.

Acknowledging the limits of tax limits amounts to saying that there are other elements of the political-legal constitution that warrant attention, over and beyond those that we analyze in this book. In particular, we should note that some of the procedural constraints discussed earlier in this chapter can serve to constrain direct regulatory behavior as well as the taxing power. As such, these procedural constraints are more general in their impact since they modify the decision-making structure itself. If such more general procedural changes are not within the realm of the possible, fiscal constraints will require the accompaniment of legal limits on the exercise of direct regulation. The argument for the imposition of fiscal limits, derived from the choice calculus of the individual who places himself behind the veil of ignorance at some constitutional stage of decision, takes on meaning only to the extent that it lends support, at the same time, to the companion imposition, or enforcement, of severe restrictions on the range of direct regulation.