The Online Library of Liberty

A project of Liberty Fund, Inc.

• Foreword
• Geoffrey Brennan
• Preface
• Acknowledgments
• The Power to Tax
• 1.: Taxation In Constitutional Perspective
• 1.1.: The Notion of a “constitution”
• 1.2.: The Logic of a Constitution
• 1.3.: The Means of Constitutional Constraint
• 1.4.: The Wicksellian Ideal and Majoritarian Reality
• 1.5.: The Power to Tax
• 1.6.: The Enforceability of Constitutional Contract
• 1.7.: Normative Implications
• 2.: Natural Government a Model of Leviathan
• 2.1.: Leviathan As Actuality and As Contingency
• 2.2.: Monopoly Government and Popular Sovereignty
• 2.3.: The Model of “leviathan”: Revenue Maximization
• 2.4.: The Model of Leviathan As Monolith
• 2.5.: The Constitutional Criteria
• 3.: Constraints On Base and Rate Structure *
• 3.1.: Government As Revenue Maximizer Subject to Constitutional Tax Constraints
• 3.2.: Tax-base and Tax-rate Constraints In a Simple Model
• 3.3.: One Among Many
• 3.4.: Tax Limits and Tax Reform
• Appendix: Progression In the Multiperson Setting
• 4.: The Taxation of Commodities
• 4.1.: The Conventional Wisdom
• 4.2.: Constitutional Tax Choice
• 4.3.: Alternative Forms of Commodity Tax: the Choice of Base
• 4.4.: Uniformity of Rates Over Commodities
• 4.5.: Uniformity of Rates Over Individuals
• 4.6.: Discrimination By Means of the Rate Structure
• 4.7.: Summary
• Appendix
• 5.: Taxation Through Time Income Taxes, Capital Taxes, and Public Debt
• 5.1.: Income Taxes, Capital Taxes, and Public Debt In Orthodox Public Finance
• 5.2.: The Timing of Rate Announcement
• 5.3.: Income and Capital Taxes Under Perpetual Leviathan
• 5.4: Leviathan’s Time Preference
• 5.5.: The Time Preference of the Taxpayer-citizen With Respect to Public Spending
• 5.6.: The Power to Borrow
• 5.7.: Conclusions
• 6.: Money Creation and Taxation
• 6.1.: The Power to Create Money
• 6.2.: Inflation and the Taxation of Money Balances: a “land” Analogy
• 6.3.: Inflation and the Taxation of Money Balances
• 6.4.: Inflationary Expectations Under Leviathan
• 6.5.: Inflation, Wealth Taxation, and the Durability of Money
• 6.6.: The Orthodox Discussion of Inflation As a Tax
• 6.7.: The Monetary Constitution
• 6.8.: Inflation and Income Tax Revenue
• 6.9.: Monetary Rules and Tax Rules
• 7.: The Disposition of Public Revenues *
• 7.1.: The Model
• 7.2.: Public-goods Supply Under a Pure Surplus Maximizer: Geometric Analysis
• 7.3.: The Surplus Maximizer: Algebraic Treatment
• 7.4.: The Nonsurplus Maximizer
• 7.5.: Toward a Tax Policy
• 8.: The Domain of Politics
• 8.1.: Procedural Constraints On Political Decision Making
• 8.2.: The Rule of Law: General Rules
• 8.3.: The Domain of Public Expenditures
• 8.4.: Government By Coercion
• 9.: Open Economy, Federalism, and Taxing Authority
• 9.1.: Toward a Tax Constitution For Leviathan In an Open Economy With Trade But Without Migration
• 9.2.: Tax Rules In an Open Economy With Trade and Migration
• 9.3.: Federalism As a Component of a Fiscal Constitution
• 9.4.: An Alternative Theory of Government Grants
• 9.5.: A Tax Constitution For a Federal State
• 9.6.: Conclusions
• 10.: Toward Authentic Tax Reform Prospects and Prescriptions
• 10.1.: Taxation In Constitutional Perspective
• 10.2.: Tax Reform As Tax Limits
• 10.3.: Tax-rate Limits: the Logic of Proposition 13
• 10.4.: Tax-base Constraints
• 10.5.: Aggregate Revenue and Outlay Limits: Ratio-type Proposals For Constitutional Constraint
• 10.6.: Procedural Limits: Qualified Majorities and Budget Balance
• 10.7.: Toward Authentic Tax Reform
• Epilogue

2.2.

Monopoly Government and Popular Sovereignty

The essence of the problem at the constitutional level is how to constrain the natural proclivities of government so as to achieve results that are consonant with those which are desired by the potential taxpayer-voter-beneficiary, as he views his own role in postconstitutional periods from the initial constitutional perspective. As we have noted earlier, current public-choice theory, as well as the prevailing political ethos or public philosophy, has concentrated its attention largely on voting rules and arrangements as the primary means of constraining governmental behavior. Our analysis shifts emphasis to nonelectoral means of achieving these ends. Our object is, of course, not to deny that electoral processes may constrain in some cases over some range, any more than we would seek to deny that dictatorial governments may exercise their discretionary power benevolently in some cases over some range. Rather, we are seeking to show that the assumption that electoral processes are sufficient to constrain self-seeking government is extremely vulnerable. We do so by appeal to a sequence of observations which, taken together, seem to us to support this position fairly thoroughly.

These observations are of three basic types. First, there is the observation that for certain types of decisions in relation to the use of resources, electoral processes are inappropriate even where those processes do generate outcomes congruent with electoral demands. Second, there is a set of purely analytic questions about majoritarian political processes which raise doubts as to whether such processes can be predicted to constrain governments effectively, in all or most cases. Third, there is a set of observations about historical experience—about the growth of government in Western democracies, and attempts by the electorate at large to limit that growth—which suggest that, in fact, democratic electoral processes may not have been constrained. We examine each of these basic types of argument in turn.

Analytical justifications for nonelectoral decision processes. Our point of departure here is the observation that electoral processes could conceivably be used to constrain governments much more severely than is typically the case in Western democracies. Decisions could be taken by popular referenda much more commonly. Nor, as James C. Miller3 has recently argued, are the reasons for avoiding the referendum option entirely technical: computer and telephone voting techniques seem to provide inexpensive means of popular voting (or large sample voting) on an extensive range of issues over which politicians and bureaucrats currently exercise discretion. One might argue—and indeed the whole theory of “representative” democracy must now be dependent on such an argument—that the reason for not relying more extensively on popular government is that it is simply undesirable to do so. Accordingly, one set of arguments that merit investigation here relates to the question as to why one might avoid reliance on majoritarian political process, even where this is taken to provide effective constraint. We offer three arguments along such lines.

1.The constitutional domain. There are aspects of in-period activity that should not be and cannot be, within the terms of constitutional construction, subject to determination by in-period political process. For example, to the extent that the constitution lays down private property rights and laws against interference with those rights, these laws must be interpreted and enforced essentially independently of in-period politics. (Otherwise, a decisive majority might conceivably simply suspend the property rights of minority members—murder of a Republican may become legitimate under Democratic rule.) Once this point is admitted, however, and once it is recognized that the enforcement of such laws requires resources, possibly of some magnitude, it becomes clear that questions of the extent of resources so allocated, and how they are to be used, must over some range be independent of simple majoritarian consent. They must therefore be made independent of in-period political process.

This is an important point and has far-reaching implications. The machinery by which laws are enforced, including not only the judiciary but also the police force, cannot be made obedient to simple majority will. Yet they must clearly be constrained in some way. Their powers must be set forth, and means of enforcement must be found. The “separation of powers,” the presence of courts of appeal, and no doubt purely internal moral constraints all have to be relied on here—in addition to the fiscal constraints we set out in the ensuing chapters. Our point at this stage is simply that electoral constraints, as such, cannot be relied on.

2.Differences between “constitutional” and “in-period” preferences. The rational individual may have preferences at the constitutional level which at the in-period level will have changed in a systematic way. For this reason, he may wish to set aside some in-period electoral outcomes that would emerge, in favor of his constitutional judgments. One of us has explored two examples of this possibility4 in detail elsewhere, but a simple case here may be useful. Suppose that the citizenry feel benign toward the “poor” and, at either the constitutional or postconstitutional level, would desire that substantial welfare payments be financed from tax revenues. Quite apart from the tax costs, these welfare payments would have a net cost in terms of disincentive effects, and at the margin the total of these costs would be equated to the net utility gains attributable to the altruistic inclinations of the citizenry at large. At the constitutional stage of choice, the costs and benefits over the whole sequence of future periods could be taken into account in arriving at some optimal level of welfare. Suppose, however, that no constitutional determination of welfare spending is made and decisions are left to period-by-period politics. In this case, there will be a higher level of welfare or transfer payments than in the case where the determination is made constitutionally. The reason for this difference is that the benefits of the welfare transfers to the current and directly observed generation of recipients (those out of work last period, for example) will be reckoned without adjustments for the disincentive effects on welfare recipients themselves. Potential recipients are observably “poor”; transfers will not make them “poor.” At the constitutional level, however, the possible effects of a transfer program in creating potential recipients can be fully taken into account. The program can be examined ab initio and abstractly.5

The interesting feature of such cases is that individuals may be led to make period-by-period choices that produce a situation that no one of them wants. Further, this result arises not from free-riding prisoners’ dilemma elements in social interactions, but rather from genuine differences between the constitutional and in-period perspectives. These cases are collective analogues of the purely private inclination to create binding “moral” rules by which the individual may constrain his own future actions. The characteristic feature of all such examples is the conflict between the individual’s current and future preferences, as revealed by his actions taken at different times in relation to choices at a specific point in time.6

One additional example of such differences relates to redistribution as “justified” by Rawlsian-type preferences exhibited behind the veil of ignorance. Armed with information about his position in society, the individual may (and generally will) adopt profoundly different attitudes to redistribution from those he might have adopted behind such a veil. In such a setting, constitutional preferences over the income distribution may be embodied in specific fiscal rules that will generate results quite different from those that would be produced under unconstrained majority voting. In general, transfer policies may need to be set constitutionally: in-period electoral process will simply not generate the constitutionally desired results, and this failure will be predicted at the constitutional level of cognition.7

3.Rational ignorance. As emphasized by Downs,8 and elaborated by many other scholars, information about politically provided goods and services is a “public good” in the technical sense. Because the acquisition of such information is costly at the margin, and because the benefits are not fully appropriable to the informed voter, all individuals will rationally remain underinformed9 about the issues involved in elections and about public policies more generally. This fact has two possible implications. First, recognizing the likelihood of such ignorance on the part of the electorate, the constitutional decision maker may prefer a set of institutions in which effective political power is more narrowly held than under genuine majority rule. Rather than choose to constrain politicians to behave in accordance with grossly underinformed electoral preferences, he may prefer to assign discretionary power to a smaller number of (representative) politicians-bureaucrats over some range and to depend on nonelectoral constraints to ensure that this power is used in the interests of the electorate.10 Second, the necessary asymmetry between information held by the electorate as distinct from the politician-bureaucrat offers scope for misleading the electorate, and a differential power which can within limits be exercised by politicians-bureaucrats in whatever way they choose.

Theoretical analysis of majority rule and its inadequacy to constrain. Each of the foregoing arguments was designed to illustrate the possibility that individuals might, at the constitutional level, explicitly choose nonelectoral means of constraining government in preference to relying on electoral processes, given that those electoral processes generate outcomes in line with electoral preferences. There is, however, no guarantee that majoritarian democracy will operate in this way. Even where electoral processes are intended to constrain the exercise of government power, will they necessarily do so? How do majoritarian political processes actually work? Do they in fact constrain the behavior of those who hold the power of government, and if so, to what extent? There are at least two matters here that merit discussion: one relates to the way in which it has become customary to model majority rule, the other to the role of bureaucracy in determining political outcomes.

1.The operation of majority rule. As is well known, unless preferences are single-peaked11 or parties have to announce policies simultaneously,12 majority rule generates cyclical “social preferences.” In other words, there are at least three positions, A, B, and C, such that a majority exists that prefers position A to position B, a majority exists that prefers B to C, and a majority exists that prefers C to A. Much of public-choice theory has indeed been preoccupied with this simple majority cycle. For example, in the simplest of the Downsian two-party models, one party, party I, has to announce its policy before the other, party II, and the announced platforms are binding.13 Downs shows that party II will always defeat party I. Suppose that we consider a simple three-voter world, with no utility interdependence, and party I announces a payoff to the three voters, a,b, and c, of $100 each. This platform can then be depicted as the payoff vector (100, 100, 100), where the first term indicates the payoff to a, the second term the payoff to b, and the third term the payoff to c. Party II can then announce a policy involving the payoff vector (101, 101, m), where m is less than 98 (or any permutation of these payoffs) and win the election. This possibility leads Downs to the conclusion that interparty competition does not in general lead to Pareto optimality, except perhaps by accident: a party can always ensure victory by announcing its policies last and by organizing appropriate transfers from some minority to the corresponding majority.

One important feature of this model, however, is an implicit inconsistency in the behavior of politicians. Politicians are postulated to maximize their expected returns, which depend on their election—and hence are modeled as maximizing the probability of being elected. It is therefore somewhat bizarre that they are not postulated to maximize the advantages of election when they are successful. Since party II knows that it can win the election if it has the right to announce its policy platform after party I, then II will respond to party I’s platform involving payoff vector (100, 100, 100) with a platform that both assures II of being elected and maximizes the “surplus” that it has at its disposal. In this case, party II’s “best” policy is one involving the payoff vector (101, 101, 0) (or any permutation of these payoffs), with the party itself appropriating the excess of total product over that product which is required to be left in the hands of voters in order to secure election. The implications of this simple model are then twofold: first, the rational party will exploit the maximal minority to the maximal extent feasible; and second, it will expend in payments to the majority the minimal sum necessary to secure election. As a result, a significant proportion of total resources is available for discretionary use by the successful party. In the non-single-peaked case, therefore, with sequential announcement of policy platforms, important dimensions of “monopoly government” emerge out of simple majority rule.

A number of questions might be posed about the foregoing model. One that arises out of the original Downsian discussion is whether the problems of majority rule in this setting might be avoided by requiring parties to announce their policies simultaneously. In the setting in which parties seek to maximize the probability of election, it does seem as if simultaneous announcement would serve to constrain parties fully. If, for example, we retain the three-voter case in which aggregate income is $300 and in which income can be costlessly transferred among voters, it seems clear that each party will aim for a platform in which two of the three voters share the $300 between themselves and the third receives nothing (the precise identity of the majority being determined at random). A party choosing this strategy would certainly expect to defeat one that chooses to distribute the $300 randomly among the three voters. What is significant in this particular case, however, is that if the party chooses to retain some of the $300 for its own use, the probability of being elected does not fall to zero. If we postulate that parties aim to maximize the expected returns from election, R, then
R = P_E · S,
where P_E is the probability of being elected and S is surplus obtained from election, and it can be shown that neither party would ever select a strategy that involved a zero value for S.14 Each party would rationally appropriate some of the $300, even where the other party did not—and as each party rationally increases its surplus, so the other party will increase its own. A sort of independent adjustment equilibrium may emerge in this setting, in which both parties (being essentially identical, by hypothesis) have identical surplus and identical probability of victory. At this equilibrium, it will be true for both parties that the elasticity of probability of victory with respect to surplus changes is −1; that is,
implies that
Thus, simultaneous announcement of policies in this setting is not fully constraining, as Downs claims.15 Rather, scope is left for genuinely monopolistic behavior on the part of noncooperating parties.

Beyond this, of course, one can hardly rule out the possibility of explicitly cooperative behavior. Politicians of all persuasions do have interests in common, and the potential for them to exploit these interests at the expense of the electorate is very considerable. As standard duopoly theory suggests, increasing the number of “firms” from one to two is not usually sufficient to ensure results analogous to pure competition: the costs of cooperation in the two-party case are surely too low to rule out collusive behavior.

It seems, therefore, that the prevailing views of majoritarian processes and electoral competition under majoritarian rule may be unduly sanguine about the capacity of electoral rules to constrain the exercise of political power. The reasons for this may be largely analytical—theorists have perhaps been overly preoccupied with the equilibrium properties of various electoral arrangements, while ignoring the role of the politicians as players in the political game. As a result, politicians have been mere cardboard cutouts, and the notion of constraining their behavior has been of less real interest than the puzzling conundrums of voting theory. To be sure, public choice has not been particularly optimistic about the possibility of the political mechanism generating outcomes with desirable welfare properties: majority rule has been recognized to generate outcomes that may be nonoptimal or inefficient by ordinary Paretian standards. This notwithstanding, it is clear that we may not have been pessimistic enough.

2.The role of the bureaucracy. Just as politicians have the power, through selection of policy platforms, to secure results distinct from those desired by the electorate (or even, as we have seen, those desired by the decisive majority), so do bureaucrats exercise genuinely discretionary power in the selection and implementation of policy proposals. Moreover, whereas the actions of politicians may be somewhat constrained by the threat of electoral defeat, the actions of bureaucrats are not.16 Indeed, by their very nature, bureaucrats act as monopolistic suppliers. Whether their role is to supply politicians with information about alternative policies, or to design the specifics of policies to be implemented—by appeal to their special skills and information—or to implement the policies (i.e., directly produce public goods) themselves, they do so in a setting in which competitive provision of such expert advice, or alternative sources of supply of the relevant public goods, are unavailable.

One could, of course, imagine a setting in which politicians contracted out the provision of public goods to the lowest-cost provider under a system of competitive tenders. Such a setting does, in fact, apply in some cases—for example, in highway construction or public works. Tasks might be contracted out to firms which do not work solely or even predominantly for the government, and which certainly have no right of tenure in public employment. But such a setting is the exception rather than the rule. Most of the human resources that are used up by governments are in permanent employ by government and do have right of tenure. They are in an inherently monopolistic position in the provision of public goods and services.

The precise way in which this monopoly power is exercised may, of course, vary. In one version of the theory of bureaucracy,17 bureaucrats are modeled as seeking to maximize the size of their budgets—a model that has some analytic features in common with those used in the book and developed subsequently in this chapter. In others, direct income maximization is taken to be the motivating force behind bureaucratic behavior. For our purposes, it hardly matters. What does matter is that the power that bureaucrats possess by virtue of their position as monopoly suppliers of public goods and services and of their role as “agenda setters” in the political arena18 is to be viewed as lying predominantly outside the coverage of electoral constraints: the potential for nonelectoral constraints to “improve” outcomes in line with the expected desires of the typical citizen-taxpayer at the constitutional level is considerable.

The broad brush of history. All of the foregoing analytics would be hardly to the point if there were clear empirical evidence to the effect that electoral processes have in practice worked pretty well in reflecting citizen preferences. But the facts, to the extent that they can be determined, hardly seem to support such a conclusion. On the contrary, the basic story seems to be an unambiguously sorry one for majoritarian political institutions. Over the last century or so, government has grown enormously in almost every country for which information is available, and probably to a magnitude completely unimagined by the Founding Fathers of the United States or any others of that period in history. In order to give a feeling for rough orders of magnitude, in 1902 the United States expended 7 percent of GNP on government activities at all levels—by 1970, the proportion was well over 30 percent. Of course, the precise interpretation of these figures is fraught with hazards. To the extent that the relative cost of publicly supplied goods has been rising, the increases in expenditure shares may reflect this cost increase alone. On the other hand, to the extent that much public-sector output is of the nature of Samuelsonian public goods, where total output is consumed equally by all consumers (or, less restrictively, where there are economies of scale in consumption), the expenditure share figures might disguise a larger share of public output in total output, due to the growth in GNP alone. Nor does the growth of government per se imply that government is too large: there is no presumption that government was the “right size” in 1902, or that increases in size do not accurately reflect electoral wishes. Moreover, it needs to be emphasized that the dimensions of our argument are essentially static. Even if it could be shown that electoral constraints had failed, our argument could not in itself explain the high growth of government, because to do so it would need to explain why electoral constraints have become less constraining over time.

However, as the size of government increases (for whatever reason) the question of whether standard electoral constraints are adequate to keep government power within acceptable bounds does naturally arise. Could the trend be reversed even if the citizenry wanted to reverse it? And if so, what means would be required?

It is in the light of these questions that another important facet of recent history suggests itself for consideration. This is the so-called “taxpayers’ revolt” which attracted widespread attention in 1978 in the wake of California’s Proposition 13. Regardless of the long-term significance or insignificance of this movement, several aspects of it are noteworthy for our purposes. First, the revolt emerged not from within normal “parliamentary” process and interparty competition but from outside this sytem. The enormous success of Proposition 13 in California in the face of indifference and even opposition from most of the political establishment must surely raise some doubts about the extent to which normal political processes reflect the popular will. And these are doubts that are not entirely allayed by the spectacular policy reversals of a considerable number of the movement’s original antagonists. Second, the revolt took the form not of once-and-for-all tax and expenditure cuts but of explicit constitutional constraints designed to be operative over an indefinite future. The avowed intention was to constrain the size of government below the level that would prevail under normal electoral processes. The obvious implication is that a significant body of citizens—and the overwhelming majority in some places—do not trust the in-period political process to produce results in accord with the electoral will, for whatever reason.

On balance, the appeal to the experience of history, and not least to recent events in the United States, does suggest that government, in its current institutional setting, is close to being out of the control of the electorate. Governments today seem to be substantially closer to the revenue-maximizing margins of behavior than they were in preceding decades and centuries. From this we might infer that a specifically designated and more explicitly restrictive tax constitution may now be appropriate, whereas it was not in prior periods. Whereas little or no attention might have been given to such matters in the nineteenth and early years of the twentieth century, a relative neglect that may well have been more or less rational, the forces of history may now have moved such problems into places of importance and relevance. With reference to the United States in particular, the absence of a more complete set of tax rules in the initial constitutional document may be at least partially rationalized in this fashion. The writers of that document simply could not bring themselves to imagine governments with the authority and appetites that the modern Leviathan is observed to possess.

The various arguments that we have put together in the foregoing discussion represent, we believe, sufficient grounds for taking seriously the role of nonelectoral constraints in the analysis of constitutional choice. The line of reasoning makes a persuasive case, we think, for a general model of the political mechanism in which majoritarian electoral processes are not effective in constraining the power of government.

We should emphasize, however, that the model of government which we set out in the next section and on which our analysis is based is not offered primarily as a description of political reality. Our basic aim is not to effect a revolution in the study of public choice—to overturn the newly emerged orthodoxy about how majoritarian democracies actually work. Even if such a revolution were feasible and desirable, it could only represent a digression from our main argument here. We do believe that there is explanatory potential in our model, and in other settings we should not be averse to exploiting that. Our point here is simply that we do not need this stronger positive or empirical case to establish that the model is both interesting and directly relevant to the issues of constitutional choice.

[3. ] See James C. Miller III, “A Program for Direct and Proxy Voting in the Legislative Process,” Public Choice, 7 (Fall 1969), 107-13.

[4. ] See the discussion of the “punishment dilemma,” in Buchanan’s The Limits of Liberty (Chicago: University of Chicago Press, 1975), chap. 8, and the discussion in Buchanan’s paper, “The Samaritan’s Dilemma,” in Altruism, Morality and Economic Theory, ed. E. S. Phelps (New York: Russell Sage Foundation, 1975), pp. 71-85.

[5. ] In a constitutional choice, the desired level of welfare payments will be that level which maximizes the expression: whereas period-by-period decision making will produce the result where B_i is aggregate dollar benefit in period i and C_i is aggregate dollar cost in period i.

[6. ] For an attempt to model this problem for the individual formally, see A. M. Shefrin and Richard Thaler, “An Economic Theory of Self-control,” Working Paper No. 208, Center for Economic Analysis of Human Behavior and Social Institutions (Stanford, Calif.: National Bureau of Economic Research, October 1977).

[7. ] In a setting where individual sovereignty is accepted, the precise normative authority of constitutional as opposed to in-period preferences is by no means clear. One cannot say whether the one should take precedence over the other without appeal to additional value judgments. One can, however, at a purely positive level, observe that decisions about in-period outcomes may be taken at the constitutional level that would never emerge out of ordinary political process, whatever the electoral rules, because in-period preferences would not endorse them. By their nature, such constitutionally preferred policies (including potentially much of redistributional policy) must emerge independently of current electoral processes.

[8. ] Anthony Downs, An Economic Theory of Democracy (New York: Harper & Brothers, 1957), chaps. 11, 13.

[9. ] That is, they will possess less than Pareto-optimal levels of information.

[10. ] This is, in essence, the argument typically put in favor of governmental decision making, rather than decision making by means of continual referenda. The absence of continual referenda does, however, undoubtedly lead to greater discretionary power held by bureaucrats and politicians.

[11. ] To secure single peakedness in preferences, substantial restrictions on both the domain of public spending and the permanence of tax institutions would be required, restrictions that need not characterize the “fiscal constitution.”

[12. ] See Downs, An Economic Theory of Democracy, chap. 10.

[13. ] Ibid.

[14. ] Suppose, to take a simple case, that party I distributes $280 randomly between two voters (let them be A, B), the identity of whom is unknown to party II, and retains the remaining $20 for itself. To establish the proposition that choice of this strategy does not reduce to zero party I’s chances of winning, all we need to ask is whether party II can be certain of victory if it distributes the full $300 to the electorate. The answer is clearly no. If II distributes $150 to either B or C and nothing to A, I will win if he pays more than $150 to either B or C and the remainder to A. Party I’s offer could be (120, 160, 0), and II’s (0, 150, 150); and I would win the election. There is clearly a whole range of possible arrangements in which I defeats II, even though the sum of the payoffs to electors is smaller for party I.

[15. ] Downs, An Economic Theory of Democracy, chap. 10.

[16. ] Political appointees may populate the upper echelons of the bureaucracy (as they do in American institutions), in which case they may be constrained in much the same way as politicians are (or are not) by electoral proceedings, depending on their tenure. But in some parts of the bureaucracy (the military establishment, for example) and in all parts of the bureaucracy where the executive is officially apolitical (as in British institutions), bureaucrats are not at all subject to electoral constraints—whether those constraints are effective or not.

[17. ] William Niskanen, Bureaucracy and Representative Government (Chicago: Aldine-Atherton, 1971).

[18. ] The importance of this role is emphasized and an interpretation of Niskanen’s theory along such lines offered in ongoing work by our colleagues Robert Mackay and Carolyn Weaver. See, for example, their “Monopoly Bureaus and Fiscal Outcomes: Deductive Models and Implications for Reform,” in Gordon Tullock and Richard E. Wagner, eds., Deductive Reasoning in the Analysis of Public Policy (Lexington, Mass.: Lexington Books, 1978); “Agenda Control by Budget Maximizers in a Multi-bureau Setting,” Public Choice (Summer 1981), forthcoming; “Commodity Bundling and Agenda Control in the Public Sector: A Mathematical Analysis,” Virginia Polytechnic Institute and State University Working Paper CE 79-6-1; and “On the Mutuality of Interests between Bureaus and High Demand Review Committees,” Public Choice, 34 (1979), 481-91. See also Arthur Denzau and Robert Mackay, “Benefit and Tax Share Discrimination by a Monopoly Bureau,” Journal of Public Economics (1980). For an analysis of constitutional limits as a control on the monopoly power of agenda setters, see Arthur Denzau, Robert Mackay, and Carolyn Weaver, “Spending Limitations, Agenda Control and Voters’ Expectations,” National Tax Journal, 32 (June 1979), 189-200; and “On the Initiative-Referendum Option and the Control of Monopoly Government,” Papers of the Committee on Urban Public Economics, 5, 1980.