The Reason of Rules: Constitutional Political Economy
By Geoffrey Brennan and James M. Buchanan
- Ch. 1, The Constitutional Imperative
- Ch. 2, The Contractarian Vision
- Ch. 3, The Myth of Benevolence
- Ch. 4, Modeling the Individual for Constitutional Analysis
- Ch. 5, Time, Temptation, and the Constrained Future
- Ch. 6, Politics Without Rules, I
- Ch. 7, Rules and Justice
- Ch. 8, Politics Without Rules, II
- Ch. 9, Is Constitutional Revolution Possible in Democracy
Politics Without Rules, II:
Distributive Justice and Distributive Politics
In Chapter 7 we discussed the connection between rules and justice. We argued that a meaningful discussion of justice presupposes the existence of rules and that a just outcome emerges when the relevant rules have been followed.
Such a view of justice is not common, in either economic or broader circles. Orthodox conceptions of justice invoke the notion of distributive justice, or “equity.” In this view, the justice or injustice of an outcome is held to be an intrinsic property of the outcome itself and to be context independent. To determine whether one social outcome is more “just” than another, one need merely investigate the relative positions of individuals in each. Relative positions are typically measured by reference to individuals’ incomes or wealth, although sometimes consumption of, or access to, certain “crucial” consumption items (for example, health, housing, food, or education) can be regarded as a superior basis for comparison.
It is not our objective here to criticize this familiar conception of distributive justice; in what follows, we shall take the normative relevance of “equality,” or “less inequality,” somehow construed, as given. Our concern is, rather, to examine the concept of distributive justice from the constitutional perspective. Our task is to ask, given the normative judgment that equality is desirable, what set of institutional arrangements are likely to be most conducive to securing that “equality” (or “less inequality”). Are there any reasons for believing that the political order will generate a more nearly equal distribution of income than a free market would generate (as seems to be commonly assumed)? What “rules” of political order are likely to generate more “equitable” outcomes?
Before we address these issues, however, it is important to recognize that in posing them we adopt a perspective on distributive justice very different from the orthodox one. We shall therefore begin our discussion by attempting to characterize the orthodox approach and by showing how the constitutional view raises additional, crucial issues. We shall also indicate what is, in our judgment, analytically incoherent about the orthodox conception.
II. Distributive Justice: The Conventional View
Characterizing the normal conception of distributive justice involves posing and attempting to answer the question, How should total product be distributed? All possible distributions of some aggregate are considered, and some criteria are used to select the “best.” A common analogy is the division of a pie among contending children, each of whom is presumed to want more pie. Consider, for example, a simple two-person community, composed of persons J and K. In terms of simple geometry, the set of possible distributions is given by the line
JK in Figure 8.1. Niggardly nature presents the community with an aggregate
OJ (equal to
OK), which can be given all to J (at point
J) or all to K (at point
K), respectively, or distributed between J and K in various proportions. We represent the distribution as the ratio of J’s share to K’s share; at
E, for example, we can depict this ratio as the slope of the ray from the origin to
E. Alternative measures of the distribution might be the difference between
E and unity (that is, equality), or the variance of levels or the amount assigned to the poorer individual (at
E,OM) or something more complex. It is normally more or less presumed that, ceteris paribus, equality is best. In this abstract characterization it is perhaps difficult to justify anything other than equality (though it may not be a trivial matter to justify equality either).
Economists and others typically argue that this characterization is inadequate in two significant respects. First, the pie does not present itself as an aggregate. Rather, it comes already sliced, the size of the slices being determined by the relative productive capacities of individuals. That is, there is a point on the
JK line (let it be
Q in Figure 8.2:) that represents a sort of status quo point from which
redistribution via some means must occur. Second, as redistribution occurs, the size of the pie changes. The general presumption is that as we move away from
Q, the pie shrinks at an accelerating rate as “excess burden” arises on both the tax and the transfer side of the redistributive process. Accordingly, the relevant possibilities frontier becomes something like
UV in Figure 8.2: Only points on
UV are feasible; points on
JK other than
Q are not. Armed with this concession to reality, the normative analysis becomes one of “trading off” pie size against pie distribution, and the famous “equity-efficiency” conflict emerges. The “policy problem,” then, presents itself in two dimensions: first, that of organizing the operation of policy instruments so that the cost in terms of pie forgone in achieving the ideal distribution is minimized; and second, that of determining how much distributive justice should be sacrificed in order to keep the pie as large as possible.
III. The Constitutional Perspective and Institutional Incidence
There are several things wrong with this typical characterization of distributive justice. Consider first the conceptual philosophical issues. As a way of reflecting abstractly on one’s distributive objectives—as a way of giving content to the notion of distributive justice—the conjectural model is perhaps innocent enough. It seems natural to pose the question, “If I were assigned the task of dividing the pie, what should I do?” But this question presupposes that questions of distributive justice do, in fact, present themselves this way. It suggests that distributions are
chosen by some single agent, or by some decision rule.
But distributions are not chosen. Social outcomes, with their distributional characteristics “built in,”
emerge from a complex interaction of individual agents, each pursuing his own ends and each connected to others under a set of prevailing rules. Of course, it is conceivable that the prevailing rules could be such that outcomes (or specific distributions) would be chosen by a single agent. Totally despotic orders are not unknown. Is this the implicit model of politics that moral and social philosophers presume in dealing with questions of distributive justice? This model of politics does, indeed, seem to be what economists have in mind when they carry abstract philosophizing to the “relevant” policy level. But if the distribution is chosen by a despot, the model of Figure 8.1 is entirely inadequate. The relevant social setting will now involve
three persons, not two. In addition to J and K, who are claimants on the aggregate pie, there is H, who is to do the cutting and allocating. If we model H as we do J and K, treating all as having conflicting demands for pie, then H will appropriate the entire pie for himself. If, on the other hand, we presume H to have altruistic concern for J and K, then, on the assumption of behavioral symmetry, we must model J and K likewise. The set of feasible distributions in Figure 8.1 is reduced from
LL‘, since distributions outside those limits will involve voluntary transfer between the parties to secure
L‘. In the same way H can be presumed to transfer to J and K up to the limits of his charitable impulses, just as if all the output had accrued to H in the first place. In other words, any conceptualization of the distributive-justice problem in terms of the unilateral choice of a distributive outcome effectively presupposes that the status quo distribution is one in which the chooser owns everything. This manner of setting up the problem virtually guarantees that distributive justice will be infeasible. Moreover, any minimal-state, free-market distribution such as
Q in Figure 8.2 seems certain to be more equitable than that required to make redistribution possible. Considerations of distributive justice would, therefore, seem to argue persuasively in favor of rules that would deny the power to transfer to political institutions altogether.
The immediate response to this criticism is, of course, that the despot model is not presupposed. Assigning political agents the power to make transfers is not the same as assigning them an unqualified right to the use of the revenue from which transfers are to be made. Political agents cannot simply make transfers to themselves. But if this is so, we must ask why. We must, in other words, trace
how political agents are constrained in the distribution of the pie. And the relevant constraints here are not the familiar “economic” ones having to do with how the size of the pie changes when the level of public transfer activity increases. The relevant constraints are those embodied in the
rules of the political game that shape the magnitude and pattern of transfers. To put the point more generally, different institutional arrangements, ranging from those embodied in some Nozickean “minimal state” to those characterized by the modern welfare state, where explicit restrictions on the power to transfer are absent, will generate different patterns of distributive outcomes. The problem of practical politics is to choose a feasible institutional arrangement that will generate a distributional pattern most congruent with the requirements of distributive justice.
The central point here is a familiar one in the context of efficiency analysis. So-called market-failure problems (attributable to public goods, generalized externalities, and monopoly elements in private-goods supply), although sufficient to indicate the presence of unrealized gains from exchange, are now widely recognized as constituting an inadequate normative case for government intervention in market processes. There is no necessary presumption that simply because markets are imperfect, political processes will work better. On the contrary, as public-choice theory reminds us, there are very good reasons for doubting the capacity of political processes to achieve Pareto optimality. The normatively relevant comparison is between two imperfect institutions. The mere observation that one institution or the other is imperfect—that markets “fail”—is simply not sufficient to establish a case for government “intervention.” So much is accepted, at least in principle, although the point seems to require continual reiteration.
In the realization of distributive justice, however, there is a precisely analogous point, though it seems hardly to have been noted. The point is this: It is not enough to lament the distribution that emerges from—or is
presumed to emerge from
*54—a freely operating market. One must show that the effect of the political process on the distribution is in the direction of equality. Or, to put the question most starkly, is movement toward equality institutionally feasible? It is ironic that despite the extensive literature on distributional policy, this basic question is almost never posed, let alone answered.
*55 The virtue of the constitutional perspective is that it places this question firmly at center stage. It does so by shifting the domain of normative inquiry from the set of imaginable
income distributions to the set of feasible
institutional arrangements from which income distributions will emerge.
What seems to us necessary here is a set of “institutional incidence exercises.” Consider, for example, a familiar tax incidence exercise in orthodox public finance. Suppose that the “policy maker” is conceived to have access to several taxes: an excise on beer, a land tax, and a corporate profits tax. In the conventional setting, we might, for example, ask, What are the distributional consequences of imposing the beer tax rather than an equirevenue land tax? We trace the distributional implications of this tax substitution in terms of the “burdens borne” by various groups (consumers of beer, specific factors in the beer industry, etc.) and burdens reduced for others. Only then is it possible to apply the normative criteria of distributive justice to decide whether the tax substitution is desirable and what mixture of tax instruments is “best.” With the tax instruments available, only certain distributions are feasible and none of them may correspond to the ideal. As we have emphasized, this approach is inadequate because it ignores the fact that any such tax choice emerges out of a political process in which “distributive justice” will be secured only if it serves the interests of those who are decisive in the political game. The approach, however, does recognize that not all distributions are feasible and that the distributional implications of alternative policy instruments must be traced before any normative recommendations can be offered. There is, at least, the recognition that alternative taxes rather than alternative distributions are the objects of normative evaluation. Here, we extend that recognition to a more appropriate level of discourse. Rather than focusing on the distributional implications of alternative
taxes, we focus instead on the distributional implications of alternative
sets of rules. In that sense, our concern is with the “incidence” of alternative institutions.
IV. The Incidence of Unrestricted Majoritarianism
In this and the ensuing sections, we shall attempt to derive the pattern of distributional outcomes that can be expected to emerge under majority rule, subject to a variety of institutional restrictions.
*56 Our object in doing this is to take political institutions more or less as we know them and ask, first, whether there is any theoretical presumption that political processes tend to generate greater “distributive justice” than, say, a predominantly market-oriented regime; and second, what restrictions on the operation of political institutions seem likely to be conducive to the pursuit of “distributive justice.” Throughout, we shall deal with an extremely simple, highly abstract model of majority rule.
Specifically, we shall consider a simple three-person community, composed of persons A, B, and C.
*57 All three vote, and all are assumed to cast their votes in accordance with their interests.
*58 There is no altruism; all aim to maximize their own real incomes.
*59 We define the length of a period to be the length of time between elections, so there is by definition one election each period. Finally, in order to focus solely on the distributional consequences of majority rule, we assume that the only thing government does is make transfers; that is, we abstract from any provision of public goods that the government may undertake.
*60 Because we wish to focus on the distributions that emerge, we shall depict all outcomes as a triplet of the form [
a is the total income of A,
b the total income of B, and
c the total income of C.
Consider, on this basis, the conceptual benchmark represented by the distribution under an entirely “free” market order. No transfers, either public or private, occur. This distribution will reflect individuals’ differing productive capacities, preparedness to take risks, propensities to accumulate, and differing histories. Therefore, we would not expect this outcome to reflect distributional equality, in general. We denote this “free-market outcome” by the vector
M = [
i is the
ith individual’s market income. Without loss of generality, let A be the richest and C the poorest, so that
c. The total of the
m‘s is national income more or less as conventionally measured.
Let us now introduce the prospect of governmental transfer, assuming that majority rule is the only operative constraint. As a point of departure, consider the case in which all transfers are “lump sum,” that is, total income can be redistributed at will without loss. Aggregate money income (and the aggregate wealth stock) are totally unaffected by the transfer process. On this basis, suppose initially that B and C form a decisive coalition. Then, clearly, they will rationally take all of A’s income (and wealth). To do otherwise would be to forgo income that the majority could costlessly appropriate and would prefer to have. The outcome will then be of the form
L = [0,
M. Note that no such outcome is stable. Outcome
L can be defeated at the next election by a coalition between, say, A and B involving an outcome of the form
J = [
b, 0], where
M; or alternatively by a coalition between A and C involving an outcome of the form
K = [
M. Some such coalition in which two of the three parties are made better off must always exist. The situation here is the familiar one in public-choice theory (and in game theory) in which every possible outcome can be defeated under majority rule by some other.
No outcome is inherently stable. Rather, we would expect continuous “cycling” as the composition of the decisive majority coalition changed. Nevertheless, whatever the precise identity of the majority coalition, all outcomes here will exhibit the same characteristic feature: the income of the minority member will always be driven to zero. Transfers will always be pushed to the confiscatory level.
We should note that parties to a prevailing majority coalition have an incentive to form binding agreements to maintain the coalition, for by entering such an agreement each majority member avoids a fifty-fifty chance of receiving a zero income in the
next period. Of course, it would not be rational to refuse unilaterally to break the coalition. One could then lose only if one’s current partner decided to break the agreement. Despite the continuous pressure for a coalition to break down, however, the possibility that a particular minority will be exploited systematically over a long sequence of electoral periods cannot be ruled out.
Three general points should be noted about this simple model. First, given the assumption that all transfers are lump sum, it is clear that the “benchmark” pretax, pretransfer income distribution is of no significance at all. Each individual has an expected average income over any sequence of electoral periods of 1/3
M. No one is intrinsically “richer” than anyone else. Differential labor productivities or differential propensities to save or take risks exercise no influence at all on the distributional outcome, not even as some relevant point of departure. No one can properly be said to “own” anything, except what he can extract from the political process itself. And what the individual does extract is his only for the duration of the current electoral period. Here, the simple pie-carving analogy is perfectly proper. There
is a fixed pie, and it
is up for grabs.
Second, because the precise composition of the decisive majority at each point is unpredictable, each individual naturally has the same expected income over any future electoral sequence, namely 1/3
M. In itself, this equalization of expected incomes can be interpreted as a major step toward distributive justice, as normally construed. The equality of
expected income, however, has to be evaluated in light of the fact that there will be, at any point in the electoral sequence, one party that has zero income. Moreover, there is the possibility that particular coalitions may turn out to be stable over many elections, in which event the same party will face totally confiscatory transfer policy for substantial periods. Given this fact, the failure of standard criteria of distributive justice to address distributions of
fluctuating parameters seems a major inadequacy. The impact of majoritarian political process on distributive justice in this simple model remains rather unclear.
Third, the assumption of lump-sum transfers is highly unrealistic. The
level of income generated under the majoritarian system will necessarily be reduced by the revolving ownership rights. Consider, for example, the incentives for A, currently a party to the majority coalition, to accumulate capital assets when he recognizes that there is at least a one-in-three chance that he will be the exploited minority in the next election and have his assets confiscated. Incentives to acquire skills that add to labor income, to take risks, or to pursue entrepreneurial opportunities are substantially muted by the fact that the individual expects to reap only a fraction of the attendant benefits. Furthermore, securing membership in the majority coalition is a privately profitable activity. If, as seems likely, the time and energy individuals devote to becoming part of the decisive coalition have positive value when used in alternative ways, the expenditure of such time and energy is a net loss to the community. Such losses fall under the rubric of “rent-seeking” losses and are being increasingly recognized as major sources of inefficiency in contexts where institutional structures create opportunities for private gain that do not involve increased production.
All this implies that transfers can be “lump sum” only in a very narrow sense. It may, in any period, be profitable for the prevailing majority to confiscate all the minority’s income and assets, but in making plans for future electoral periods, all agents can be expected to bear in mind the prospect that they themselves will be minority members and that their assets will be confiscated. Therefore, there will be much less available for confiscation than there otherwise would be. In particular, the sum of posttax-transfer incomes will be significantly less than
M at all points in the electoral sequence.
In light of this fact, all may well be better off in the expected sense if restrictions are placed on taxes and transfers in order to limit redistribution from the minority to the majority at any point. Moreover, such restrictions would have a presumptively desirable distributional consequence in that the distribution would be less inequitable at each point in the electoral sequence. We shall therefore turn, in Section V, to a brief examination of various possible restrictions that might be imposed on the tax-transfer process.
V. Tax Rules and Distribution under Majority Rule
Retaining the basic model of majority rule set out in the previous section, we shall proceed to examine the nature of distributional outcomes that emerge from political process in the presence of fiscal restrictions of various types, assuming that tax-transfer operations generate “excess burdens” and generalized efficiency losses for the community. We shall consider, in turn, restrictions imposed on the taxing process and the transfer process, adding an additional restriction at each turn. First, we shall suppose that the constitution restricts the taxing authorities to a single-rate income tax, no other restrictions being applied. Then we shall add the requirement that the income tax be levied at a uniform proportional rate on
all individuals. We shall add the further restriction that transfers be paid in equal per capita amounts, in the form of a “demogrant.” Finally, we shall consider a direct
constitutional selection of the tax-rate-transfer level.
1. Flat-rate income tax
Suppose that the majority is restricted to the use of a flat-rate tax on money income in imposing taxes on the minority. That is, although only minority members pay the tax, there is a well-defined tax base, money income, and a single rate applied equally to all minority group members. The significance of this restriction is twofold. First, the single-rate restriction is important because it prevents the majority from levying a tax that discriminates over inframarginal units of the minority members’ money income. Just as a discriminating monopolist can charge higher prices on inframarginal units of product and thereby secure some of the value of the consumer surplus otherwise accruing to the purchaser, so a taxing authority can appropriate some (and in the limit all) of the citizens’ consumer surplus by means of an
appropriately regressive tax.
*62 The requirement of a uniform tax rate across all units of an individual’s income rules out any such prospects.
Second, the requirement that taxes be levied on money
income explicitly rules out the power of the majority to confiscate the minority’s asset holdings—something that would be feasible under a wealth tax, for example. Income from assets can be taxed at the same rate as income from other sources, but the assets themselves remain more or less inviolable.
Several things follow from these restrictions on the tax side. First, there will be, for each individual, a maximum amount of revenue,
i, that can be obtained from him through the flat-rate income tax. Second, there will be for each individual an associated minimal income level,
i, which he secures, posttax, when faced with the maximum revenue tax rate. Third, we know that the taxing process will inflict net losses on taxpayers, over and above the revenue collected, as individuals are induced by the tax to substitute tax-free leisure for income.
These claims can be readily justified analytically,
*64 but it is perhaps necessary here only to provide an analogy. We can think of the majority as owning a monopoly right in the sale of the minority’s factor services. The maximum tax revenue obtainable corresponds to the maximum monopoly profit from the sale of those services, and the minority’s minimal income can be thought of as the cost borne by the monopolist of providing those services and accruing to the basic productive factors (in this case, the minority). The excess burden is then analogous to the efficiency losses associated with monopoly.
On this basis, it is clear that the decisive majority will always apply the revenue-maximizing tax rate to the minority’s income. To do otherwise would be to forgo transfers that the majority coalition could obtain. This will not, however, drive the minority’s net money income to zero; tax rates will necessarily fall short of 100 percent. To apply a tax rate in excess of the revenue-maximizing rate will, by definition, reduce the revenues available for transfer to the majority. At the same time, the majority will never willingly impose taxes on itself, because taxes involve an excess burden. Even if the revenue were to be returned to the taxpayer group in lump-sum form, the group would still receive less than would be required to compensate for the imposition of the tax.
Consider, for example, the case where A is in the minority, and B and C are in the majority. Then A will face the revenue-maximizing tax rate and receive a net total income of
a, whereas B and C will pay no taxes and will receive
a to divide between them in some proportion. That is, the resultant outcome will be
JA = [
a. Analogously, when B is the minority, the outcome will be
JB = [
b. The result would be equivalent if C were the minority.
As in the case of unrestricted majoritarianism, we cannot predict which majority coalition will prevail. No outcome exists that cannot be defeated under majority rule. “Cycling” can be expected, as the identity of the decisive coalition continuously shifts. Such cycling is here bounded, however, in the sense that no one’s income can be reduced to a level below
i. Obversely, the best that any group can conceivably do is to receive maximum revenue from
both of the other groups. This is, of course, unlikely, but a majority member may have to consent to be taxed in order to provide sufficient revenue to bribe another group to join the majority coalition. Suppose, for example, that
JA prevails and that the maximum revenue
a is split equally between B and C. Then A will seek to form a coalition that is best for himself. If there is more maximum revenue to be obtained from B than C (that is,
c), A will attempt to form a coalition with C. In order to do so, however, he may need to pay C more than
b < ½
a, for example), and A will therefore have to pay some positive tax himself.
Because of the revolving cycles and the general distributional indeterminacy involved here, it is difficult to specify exactly what the expected income will be for each group over the long electoral sequence. However, we can make certain general remarks. First, unlike the case of unrestricted majoritarianism, differential productivity, thrift, and so forth will exercise some influence on the expected distribution because the minimal income that each will receive when in the minority can be expected to reflect basic income-earning capability. There is, then, some substantive meaning here in the notion that some are richer than others. Second, and as in the unrestricted case, the effect of government transfer activity seems likely to be in the direction of equalizing
expected incomes. In general, we expect the maximum revenue obtainable from any group to be positively related to the tax-free income of that group. If so, the amount the poorest can expect to receive in transfers will exceed the amount the richest can expect to receive as transfers in absolute terms (and hence a fortiori as a proportion of pretax total income). The net effect is to condense the distribution of expected income over the long electoral sequence, but not to equalize it. In this sense, unrestricted majoritarianism generates
expected income that is closer to equality; on the other hand, at each point in the electoral sequence, the observed distribution of income will be more nearly equal in the presence of the tax restriction than in its absence. And, of course, the aggregate of incomes will be predictably higher.
2. Uniform proportional income tax
We now add to the previous tax restrictions the requirement that taxes be uniform across taxpayers. Under this restriction, the majority must face the same tax rate on money income that the minority faces. However, since there are no restrictions on the use of tax revenues, the majority will rationally appropriate all tax revenues for itself; no transfer at all will be paid to the minority. Accordingly, there will still be cycling in this case. No particular majority coalition will be intrinsically stable, for any outcome can always be confronted with another that is preferred by two of the three individuals.
If it were not for the presence of excess burden in the tax system, this case would not differ from unrestricted majoritarianism. Uniform nondistorting taxes would be imposed by the majority so as to appropriate
all the income of each individual. Revenues would then be divided in some proportions strictly among majority coalition members. In other words, the requirement of tax uniformity, in the absence of distortions within the tax system, does not constrain outcomes in any way. But because the tax system does distort—because it imposes an excess burden on majority members that increases at the margin with the tax rate—the majority will not rationally push the tax rate to its maximum revenue limits. Rather, the majority coalition will push taxes to the point where the cost to itself of the marginal dollar raised, including the excess burden imposed on majority members, is exactly one dollar. It is relatively simple to see that this will imply taxation at a rate lower than that required to generate maximum revenue from the minority. Specifically, let us consider tax-rate increases in the neighborhood of that maximum revenue level. An increase in the tax rate adds only a small amount to the net revenue collected from the minority (and hence to the net transfer received by the majority members) but adds substantially to the excess burden of taxation that majority members themselves face.
Specifically, the cost to individual
i of raising an additional dollar of revenue consists of two elements: the individual’s share of the tax revenue and the marginal excess burden he faces. That is,
i + (
i‘s share in the marginal dollar of tax revenue raised. Under a proportional money income tax,
i is simply the ratio of
i‘s money income to aggregate money income. It is clear that
dR tends to infinity as we approach the revenue maximum, because
dR becomes zero at the maximum revenue level. Therefore, the marginal cost to
i of an additional dollar of transfer also rises to infinity at maximum revenue. If each majority member shares equally in the tax proceeds (that is, each majority group receives half of each dollar of revenue received), each group will desire a tax rate at which
i takes the value ½.
i is larger for those with larger money incomes, and given that money income and total income are positively related, tax rates will be predictably lower when the rich are in the majority coalition than when the poor are. The poor therefore will lose relatively less when in the minority than the rich will lose when in the minority. Furthermore, although tax
rates will in general be lower than in the nonuniform case, aggregate revenue will tend to be higher than in the nonuniform case since all must pay taxes. This means in turn that gross transfers will be larger.
In summary, the poor will tend to do rather better under the uniformity restriction: They will receive larger transfers on average than in the nonuniform case when in the majority and pay lower taxes when in the minority. The requirement of tax uniformity works in favor of greater redistribution
on average and yet toward smaller variation over the long electoral sequence in the fortunes of any individual. The richest can do less well and the poorest less badly over the electoral cycle.
This is an important result and merits some emphasis. The requirement of tax uniformity in the presence of majority rule (without any corresponding uniformity restrictions on the transfer side) leads to a more equitable outcome, both in terms of expected income and in terms of the outcome at any point in the electoral cycle, than emerges in the absence of the uniformity requirement.
3. Uniform proportional income tax plus demogrant
Consider now the case in which uniformity restrictions are imposed on both the tax and the transfer side of the budget. The transfer uniformity is taken to require that revenues be divided equally among all citizens, in the form of an individual payment to all. Each group therefore receives thirty-three cents of each dollar of tax revenue raised. Uniformity on the tax side is defined, as in the preceding section, to involve a proportional income tax levied at the same flat rate on all individuals.
Given these restrictions, there will be some desired tax rate for each individual. This rate will be lower for any individual the higher his income, because he will be contributing a higher pro rata share of any dollar of tax revenue raised. In fact, any individual with income above the average will desire a zero tax rate—and hence a zero demogrant. Such an individual will contribute to revenue raised in proportion to his share of national income and will receive in return a per capita share enduring a net loss. Obversely, an individual with less than average income will desire a positive tax rate, one at which the cost to him of raising an additional dollar of revenue, including the marginal excess burden, will be equal to his per capita share of one-third. Clearly, the lower the individual’s income, the higher will be the desired tax rate (and concomitantly tax revenue and demogrant). Here we have
no cycles. Preferences over the single parameter
t can be arrayed along a single spectrum, and majority rule will lead to the median income earner, in this case B, being decisive in determining the tax rate to be applied. There is no scope for a decisive coalition between A and C, because given the restrictions on taxes and transfers there is no arrangement that will make both A and C better off: A wants lower tax rates, C higher ones. Hence, the tax rate levied will be the tax rate desired by the median-income individual. As we have indicated, only if income of the median-income earner is less than average will he desire positive transfers; and he will desire more transfers the lower his income is. The extent of redistribution will depend, therefore, on the extent to which the income distribution is skewed toward the lower end.
In contrast with the earlier cases, then, the pattern of transfers in this case is entirely determinate and emerges identically at every point in the electoral sequence. The extent of such redistribution depends solely on the difference between mean and median income. Whereas the focus of normative analysis is on the
variance of the income distribution, or some equivalent measure of inequality, the extent of redistribution under majority rule, so restricted, depends solely on the degree of
skewness. In other words, majority rule, with uniformity restrictions on both tax and transfer sides, serves to connect redistribution to the wrong parameter. Changes in the distribution of income that are essentially irrelevant to redistributive ethics can generate substantial changes in the pattern of political transfers, whereas changes in the income distribution that are of considerable normative significance may either leave the extent of transfers from rich to poor unchanged or alter them in a “perverse” direction. Some simple comparative static results from our three-person analysis suffice to support these claims. Suppose, for example, that the income of the poorest falls, ceteris paribus. Then average income will by definition also fall. Median income will remain unchanged, however, so the extent of transfers will
fall. Obversely, if the income of the poorest rises, ceteris paribus, transfers will rise. Consider, on the other hand, an increase in median income, ceteris paribus. In this case, transfers will fall, even though the relative share of the poorest in the total pie has declined. Of course,
some changes in the income distribution will generate changes in the transfer pattern that are consistent with the dictates of distributive justice as conventionally interpreted. For example, as the income of the richest increases or decreases, ceteris paribus, the level of transfers will rise or fall. But as we have shown, the simple median voter model does not generate patterns of political redistribution that are an appropriate expression of what distributive justice requires.
VI. Direct Constitutionalism and Distributive Justice
The central message of the institutional incidence analysis undertaken in the previous two sections is that ordinary majoritarian politics is a highly imperfect mechanism for securing distributive justice—at least as distributive justice is normally conceived. Two types of problems seem to be associated with majority rule. The first is one of distributional indeterminacy. The implementation of political transfers will always be such that the direction of transfer is away from the minority and toward the decisive majority, and the poorest cannot be expected to be in the decisive majority any more often than anyone else. Therefore, although the poorest can expect to be net beneficiaries of the political roulette wheel in the long run, the fact that political transfers will regularly flow in a perverse direction is one to be reckoned with. Exactly how one should evaluate a sequence of systematically variable distributional outcomes is not a question to which social philosophers or normative policy analysts have given much attention. Nonetheless, it should be clear that without some answer to this question majoritarian institutions cannot generally be evaluated in terms of criteria for distributive justice. Is an income stream that varies randomly over time as “just” as a stream that yields income equally in each period? If not—and specifically, if the constant stream is to be preferred—there seems a good prima facie case for restricting majority rule in some way to produce, if possible, more stable distributional outcomes.
One way of restricting majority rule to achieve a determinate distribution is to require uniform proportional taxes and disbursement of revenues on an equal per capita basis. If such restrictions are imposed, however, a second problem with majority rule emerges: The transfer pattern generated will be insensitive to ethically relevant considerations. Specifically, the level of transfers will be perversely responsive to changes in the level of income of the poorest. Indeed, there is no guarantee that transfers will occur. Any transfers that do occur, of course, will be such as to benefit the poorest most of all. This is a direct concomitant of the restrictions imposed—that taxes be levied in proportion to income and that revenues be paid on an equal per capita basis. Nevertheless, majority rule does not seem capable of determining a
level of transfers that is ethically appropriate or sensitive to the relevant parameters of the distribution.
It may, of course, be argued that the highly austere model of political process used in the foregoing sections is so remote from “real-world democracy” that nothing of any relevance can be concluded. Though fully conceding the starkness of the simple majoritarian model used, we reject the criticism. The object is to investigate the nature of the pressures on distributional outcomes that arise from electoral processes. Those pressures necessarily reflect the nature of the institutional setting, of which electoral competition under simple majority rule is a major (and many would argue predominant) feature. The danger of less formal treatment of democratic institutions is that one can become captivated by the prevailing political religion. There is, of course, much
talk of distributive justice in the political rhetoric. The substantive issue is whether there is any reason to believe that distributive justice will be secured.
Given the simple model of majority rule we have developed, the answer is at best equivocal. Majority rule may be able to secure a more equitable outcome in the income distribution and may do so at tolerable cost. Majority rule, however, seems likely to work much better in this respect if constrained by the requirement of uniform proportional taxation. Such a requirement promotes more equitable outcomes in that it restricts the advantages to a decisive coalition of richer individuals in organizing large-scale transfers toward themselves, while restricting to a lesser extent the analogous advantages for the poorer. In all such cases, however, because transfers flow often enough in the “wrong” direction, the cost of securing any net redistribution in an expected sense is rather high—and seemingly unnecessarily so. For example, in cases 1 and 2 of Section V, it seems likely that all parties would receive larger returns if there were an agreement to secure the average distributional outcome at every point in the electoral sequence. This could also be construed to have purely distributional advantages. In a Rawlsian maximin setting, where no individual knows to which group he will belong, an arrangement that generates the expected distribution at every point has the advantage of entirely precluding the outcome in which the poorest are the exploited minority. Given the maximin criterion, this characteristic would seem to be decisive.
Might, then, some directly
constitutional transfer arrangement be distributionally superior? If, in some quasi-Rawlsian setting behind the veil of ignorance at the constitutional level, individuals can be presumed to have well-defined preferences in relation to future income distributions and to transfer policies that might be implemented to constrain those distributions, why not seek to institute those policies at the
constitutional level? Suppose, for example, that a uniform proportional tax rate is to be levied and that the resultant revenues are to be expended in equal per capita grants. Also suppose, however, that the tax rate (and hence grant level) were not determined by the median voter under majority rule but, instead, the tax rate (or the grant level) were selected
constitutionally. Transfer policy would cease to be a matter for in-period political determination; the pattern of government grants would, instead, be part of the rules of the game.
There are four points to be made about such an arrangement. First, and somewhat surprisingly, the demogrant system is not necessarily as efficient a means of achieving a given
expected distributional outcome as the uniform tax alternative examined in Section V.2. That is, reductions in the variance of the poorest’s income stream over the electoral cycle are achieved at some loss in aggregate income. The reason for this seems to be that requiring transfers to be paid to all individuals increases the total revenue requirement of the transfer level that the poorest insists on when in the decisive coalition, and does so sufficiently to offset the disadvantages to the poor when in the minority. Recall that in the model discussed in Section V.2, tax rates are lower when there is a coalition of the richest than when there is a coalition of the poorest, so the disadvantages to the poor from being in the minority are already somewhat moderated. In any event, to achieve the same expected income for the poorest via a demogrant system does require a higher tax rate and hence larger excess burdens than in the case where transfers are restricted to majority members, even though the poorest will be in the exploited minority some of the time.
Second, and on the other side of the ledger, the constitutional demogrant system obliterates incentives to expend resources in an attempt to ensure that one is part of the decisive coalition, incentives that exist whenever majoritarian cycles prevail. If political rent seeking of this type is at all large (which, as we have already argued, it could very well be), then the lower excess burdens on the tax side in a majoritarian system restricted only by the requirement of uniform proportional taxes may well be more than offset. In this event, efficiency considerations favor the explicit constitutional transfer package.
Third, although it seems natural to refer to such an arrangement as “constitutionally determined
transfer,” there is something rather misleading about such terminology. There is, after all, no sense in which these policies would “redistribute” resources or income among persons. Rather, the constitutional order would involve a set of property rights in which each individual genuinely “owned” only a portion of the product that the labor embodied in his person generated. On the other hand, each would own an equal share of some portion of the product that the labor embodied in other persons generated. The government could not be taken to be
transferring income from the rich to the poor—by virtue of gratuitous altruism, hatred of the rich, or any other aspect of “Robin Hoodism.” The arrangement would simply be part of the institutional structure under which individuals’ basic property rights were defined and enforced. In fact, this is perhaps not so different from current institutional arrangements in Western democracies. Citizens do, by virtue of the prevailing powers of the state, make claims on the personal productivity of other individuals—claims the state legitimizes and effects through the tax-transfer system. The crucial difference between what is here suggested and what currently prevails is that under the constitutional alternative, the claims individuals make on one another under state aegis would be
uncontingent, not varying according to fluctuating electoral fortune. The rights that citizens possessed by virtue of such constitutional arrangement of taxes and transfers would then be no different from other things they could properly construe as
owning. At this point, notions of distributive justice and formal justice (as examined in Chapter 7) overlap. Because distributive justice considerations are applied to the rules rather than merely to outcomes, and because these considerations are endorsed and applied under constitutional consensus, the outcomes emergent under rules satisfy the requirements of “distributive justice,” whatever the characteristics of the distribution of income those outcomes may exhibit at any given time.
Finally, once it is recognized that criteria of distributive justice are applicable to the choice among rules rather than some imagined choice among imaginable distributional outcomes, one cannot simply presume that the set of rules most conducive to distributive justice will be ones that “transfer” income. It is certain that economists generally approach the issue of redistributive policy with a strong predilection for cash transfers as the most efficient form of securing desired distributions. But this proposition may not be true once the domain of discourse shifts from selection of the desired
distribution toward the institutional structure from which desirable distributions may emerge.
There could be all sorts of rule changes that promoted greater equity in the pattern of emergent outcomes. In basketball, for example, we could diminish the natural advantages of tall players either by raising the height of the basket by five feet or by attempting to handicap tall players explicitly in some way (say, by weighting goals scored by the inverse of the height of the scorer). It is not obvious that the latter scheme would be more just, in any acceptable meaning. In the same way, rule changes in social affairs need not involve explicit redistribution through political process to secure more equitable outcomes. In light of our analysis of transfer patterns under simple majority rule, it would hardly be surprising if more efficient routes to distributive justice could be found.
The central object of this chapter has been to examine the question of distributive justice through constitutional eyes. Our point of departure has been the observation that the “benevolent despot” model of political processes is totally inadequate for the purposes of discussing distributive justice, just as it is in other pieces of normative social analysis. We cannot adequately characterize governments as “choosing” among alternative distributions. Rather, distributions emerge from the complex interaction of autonomous agents, all making individual choices under a given set of rules. The evaluation of alternative distributions is to be seen as a preliminary step only, as a piece of abstract normative theorizing, which of itself is incapable of revealing anything at all about policy-relevant matters. What is required is an examination of the way changes in the rules under which individuals interact change the pattern of distributive outcomes. Such examination we call “institutional incidence analysis.”
The analytic core of this chapter contains just such an institutional incidence analysis for an extremely simple model of majority rule. We make alternative assumptions about the excess burdens generated by taxes and transfers and examine the distributional effects of imposing various restrictions on the ways taxes can be levied and transfers paid. The general conclusion to be derived from this analysis is that majority rule is at best a highly imperfect means of pursuing distributive justice. Majority rule either generates cycles (that is, is basically distributionally indeterminate) or, when appropriately restricted, generates a specific pattern of transfers that is perversely responsive to normatively relevant changes (say, to changes in the income of the poorest).
All this suggests the possibility of lifting the determination of the redistributive or transfer budget out of the jurisdiction of in-period majoritarian politics and making it a matter of explicit constitutional compact. This can, of course, be done. But a constitutional decision to embed some transfer operations in the general rules of the game cannot be presumed. Expected effects on the distribution of income will presumably be relevant in the choice among alternative sets of rules, and rules generating more nearly equal patterns of distribution may, ceteris paribus, be preferred. Chosen rules, however, need not involve “handicapping” the more successful players. There is available, at the constitutional level, a wider set of options, and the means chosen for achieving more equitable outcomes need not necessarily involve interpersonal transfers.
The analysis here, to be sure, falls short of being definitive. It does, we believe, nevertheless point in the right direction. Traditional discussions of distributive justice are hopelessly remote from the real world of distributive politics. Indeed, crucial political constraints are typically ignored altogether, and the institutional feasibility of distributive justice is brushed aside as an irritating minor technicality. This may be an acceptable procedure for the moral philosopher, but for the social analyst it is totally inadequate. Our object here has been to expose that inadequacy and to indicate the sort of inquiry required to fill the void.
For present purposes, it seems clear that faith in the capacity of ordinary majoritarian processes to generate equitable patterns of distribution is naïve indeed. In the absence of rules designed to restrict the operation of majoritarianism in particular ways, almost anything goes. Equality in an
expected sense emerges only because of the intrinsic unpredictability of the future pattern of political coalitions, and what remains to be distributed in the world of “politics without rules” cannot be anything but small.
The Economic Prerequisite to Democracy (Oxford: Basil Blackwell, 1981).
Capitalism and Freedom: Problems and Prospects, ed. Richard T. Selden (Charlottesville: University Press of Virginia, 1975), pp. 52-77.
Toward a Theory of the Rent-Seeking Society (College Station: Texas A&M University Press, 1980).
The Power to Tax (Cambridge University Press, 1980), ch. 3, for a detailed exposition of this point.
Power to Tax, chs. 5 and 6.