Individual Preferences and Public Policy

Robert T. Deacon, University of California, Santa Barbara
ABSTRACT - Attempts to construct behavioral models of the public sector have resulted in the characterization of government as an exchange process, in which tax payments are the quid pro quo for public services received. The most widely discussed political-economic model follows from a characterization of citizen-voters as well informed, economically rational agents, embedded in a competitive political environment. Challenges to this theory question not only its positive implications, but also its normative content.
[ to cite ]:
Robert T. Deacon (1981) ,"Individual Preferences and Public Policy", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 517-522.

Advances in Consumer Research Volume 8, 1981      Pages 517-522


Robert T. Deacon, University of California, Santa Barbara


Attempts to construct behavioral models of the public sector have resulted in the characterization of government as an exchange process, in which tax payments are the quid pro quo for public services received. The most widely discussed political-economic model follows from a characterization of citizen-voters as well informed, economically rational agents, embedded in a competitive political environment. Challenges to this theory question not only its positive implications, but also its normative content.


The range of issues currently being raised in the field of state and local government finance is both diverse and volatile. The popular movement to limit taxes and public expenditures is probably the most visible of these, though court imposed changes in public school finance, and state and federal funding for local government programs are of continuing interest. Had a phenomenon such as tax limitation arisen a mere twenty years ago, it seem safe to speculate that the major topics of concern to the economics profession would have been with the distributional effects of tax cuts and the adequacy of alternative revenue sources to make up shortfalls. In contrast, the current literature is filled with studies of voting behavior on tax limit referenda, attempts to discover whether such phenomena are indicative of "political disequilibrium," and if so, what features a new equilibrium is likely to embody. This dramatic shift in emphasis reflects a rapid evolution in economic analysis of the public sector, one which has resulted from an attempt to conceptually bridge revenue and expenditure sides of the public budget with a behavioral paradigm (Buchanan 1975). In this emerging framework, government is characterized as an exchange process, in which taxes are the quid pro quo for services received. Individuals are endowed with essentially the same motivations, preference structures, and informational requirements faced in the private sector. But in the role of public sector consumer, the individual's opportunities and constraints are critically shaped by the nature of political institutions.

Viewing the citizen-taxpayer as consumer raises a variety of questions, What is the nature of his public sector demands, and how are they revealed? How accurately are these demands articulated, and how do public supply institutions respond to them? Can a concept of equilibrium, similar to that used to study private markets, be employed to characterize public sector outcomes? This general class of questions is addressed in the following pages. The next section briefly discusses the institution of voting, and its role as a device for revealing preferences. Following this, the concept of equilibrium in majority rule system is explored, and competing views of the nature of the political process are examined. Concluding comments are presented in the final section. At each step, existing empirical evidence, as it relates to key conceptual questions, is briefly summarized. [More extensive discussions of several of the topics treated here appear in Deacon (1977).]


The notion that individuals act in self- interest when choosing among ballot alternatives has been implicit in public choice theory from its beginning (Bowen 1943, Downs 1957, pp. 27-45). Extension of the utility maximization hypothesis from the realm of the market to the voting booth has apparently been so natural that it is axiomatic in the theoretical literature. The view that individuals compare ballot alternatives and vote for those that offer preferred outcomes suggests the use of voting choices as information on revealed preferences for public goods.

To illustrate this approach, consider the following simplified situation (adapted from Rubinfeld 1977) involving a referendum to change the level at which a single public service is provided. Such choices are fairly common in public school districts in the U.S., where ballots offer alternative property tax levies. The preferences of individual i are represented by a conventionally shaped utility function in which the public service in question (q), and consumption of a composite private good (xi) enter as arguments,

Ui = Ui(q,xi) .   (1)

Normalizing so that the price of the private good is unity the individual's budget constraint is expressed

Ii = xi + tiE   (2)

Where Ii is income, E is total public service expenditure by the jurisdiction in question, and ti is the individual's tax share, the amount by which his tax will rise if public spending increases by one dollar. To complete the problem, spending is related to services received by a (convex) cost function for the public service

E = E(q)   (3)

The consumer's preferred public sector outcome is characterized by maximizing (1) subject to (2) and (3). The solution to this problem is the consumer's desired public spending level,

E*i = Di(ti,Ii) .   (4)

The functional label Di is used to indicate that the relationship is a conventional demand curve, where ti is interpreted as a "tax price. Consequently, E*i is expected to be decreasing in ti and increasing in Ii (if the public good is normal).

The demand curve is drawn in Figure 1. Suppose that the existing level of public expenditure is E0, and that the issue before the voter proposes a marginal increase in spending. Since voter i prefers a level of public outlays larger than E0, a yes vote on the proposed increase is expected. The general rule for voting on marginal increases such as this can be written:

vote "yea" if E*i > E0,

vote "no" if E*i < E0,

abstain if E*i  = E0

Intuitively, if one had prior information on the shape of demand functions and on individual tax prices, then voting responses could be predicted. For example, among the class of voters who have demand curves similar to Di in Figure l, those who face tax prices above t0 would be expected to vote against the measure, while those with tax prices below t0 would support it. Consequently, "the demand function (for public education) ought to be the most effective discriminant function for distinguishing 'yes' from 'no' votes that households cast" (Peterson 1975, p. 104).



A number of public education demand functions have been estimated from school referendum data, though "probit" or "logit" techniques ere usually employed in estimation rather than discriminant analysis. In some cases, individual attributes and voting choices have been available from survey data (Peterson 1975, Rubinfeld 1977). More often, however, only aggregate voting returns (e.g., by precinct) are reported, a condition that necessitates interpreting demand responses as pertaining to an average or representative voter in each group (Barkume 1978). In general, these studies have obtained significant price and income effects that are of expected sign. Moreover, since it is often reasonable to postulate that demands will depend upon other observable attributes such as age, education, family status, etc., a variety of additional hypotheses have been examined. For example, Rubinfeld (1977) found school employees much more likely to vote for tax increases than otherwise similar voters, Others have examined the effects of numbers-of school age children in a family upon voting responses. [In general, voting studies of the type described here, can only measure the relative magnitude of price and income effects. In a series of papers, Eli Noam has developed and applied a technique that permits estimation of the absolute effects of price, income, and other variables upon demands, and applied the method to the determination of optimal budgets. See Noam (1979) and references cited therein.]

If voting data could only be analyzed within the confines of the example developed earlier, the application of this approach would be severely limited. With noneducational referenda, diverse arrays of services may be involved in a single ballot, and payment may be exacted by changing incomes or market prices, rather than simply altering the level of a single tax. One technique that has been developed to analyze such complex alternatives employs the concept of a mixed indirect utility function (Deacon and Shapiro 1975). Within this framework, ballot choices are characterized in terms of utility comparisons, rather than public expenditure demand functions, and utilities under alternative ballot outcomes are expressed as functions of income, prices, tax rates, and public goods levels. Application of this technique to referenda involving such issues as scenic resource conservation, rapid transit provision, and public hospital financing has yielded results that are generally in close accord with the underlying theory.

Prior to casting a ballot, the voter must decide whether or not to vote at all, i.e., whether the benefits of showing up at the polls exceed the costs (Downs 1957). Potential benefits depend upon the likelihood that a single vote will decide the outcome (as, for example, measured by the closeness of the election) and upon the expected difference in outcomes under the two alternatives on the ballot. Costs will vary with the imputed value of the individual's time, and the resources required to obtain and evaluate information regarding the alternatives. This economic model of the "turnout" decision has been successfully employed in a number of empirical studies (Barzel and Silberberg 1973, Chapman and Palda 1980). Moreover, it indicates that those who vote may not be a random sample of the populace. This conjecture was borne out in an analysis of the Headlee tax limitation amendment in Michigan by Gramlich et al. (1980), who found that nonvoters tended to be relatively uninformed or uncertain regarding the likely impacts of the measure, and generally inclined to be against it. The notion that voting costs may significantly influence electoral outcomes, by "selecting" the kinds of preferences revealed in elections, remains to be fully integrated into existing models of the collective choice process.

The analysis of voting behavior is of central importance to any model of the public economy, and mist precede judgements concerning the ability of democratic systems to respond to citizens' preferences. To the extent that voting behavior is consistent with economic rationality, it becomes possible to predict outcomes and to formulate refutable hypotheses regarding the electoral process. In turn, it is the possibility of prediction that allows one to construct positive political models, and to make normative statements concerning actual political system.


To address questions regarding the determinants of political outcomes and the nature of political equilibrium, an explicit model of the political process under majority rule is necessary. The first economist to suggest such a theory was Harold Hotelling (1929). The paradigm he offered was extended in a remarkable article by Howard Bowen (1943), and further refined by Downs (1957).

To motivate the developer of this theory, consider the following problem. The range of possible public sector outcomes (policies) is represented as a set (a1',a2', .... an'), and elements of this set will be placed before-the voters in a pairwise fashion. Each citizen-voter has preferences for the various alternatives, and in principle, these preferences can be represented in term of individual utilities. Assuming that that preferences of voters are known, the problem is to predict which alternative will be chosen under simple majority rule. In general, the existence and characteristics of a solution to this problem depend upon the satisfaction of certain conditions regarding the shape of preferences and the way in which alternatives are offered.

It has been known, at least since the 18th century, that majority voting on pairs of alternatives need not exhibit transitivity. Cycles can develop in which a1' defeats a2', a2' defeats a3', and a3' defeats a1', etc., even if the preferences of individual voters are fully transitive.

Without transitivity, equilibrium would not be possible unless some sort of stopping rule were invoked; but use of such a stopping rule makes the final outcome "path dependent,'' or sensitive to the order in which alternatives are offered. If, however, individual preferences satisfy a certain regularity property then the cycling problem disappears. For simplicity, suppose that indifference among pairs of alternatives is ruled out, and that the number of voters is odd. Then majority rule will be transitive if the list of alternatives (a1',a2', .... an') can be arranged in a sequence (a1, a2,... an) such that the following condition is satisfied for each voter: If U(ai) > U(ai+j), then U(ai+j) > U(ai+j+k), for all i, j, k > 0. This statement of the so-called "single peaked" condition guarantees transitivity, though considerably weaker conditions will also suffice (Sen 1968). Intuitively, this condition is satisfied if each voter's utility, when plotted against the sequence alternatives, forms a single peaked curve, i.e., each voter has one uniquely preferred alternative, and utility declines monotonically as one moves away from this alternative in either direction.

If preferences are single peaked then the median of individually preferred outcomes is dominant in that it will be preferred by a majority of voters over any other alternative that may be paired against it. To see why this is true, let am be the median preferred alternative and suppose it is paired against some other alternative (am+h) that is higher in the sequence. Since am is the median preferred point, a simple majority of voters have preferred alternatives that are at or below am in the sequence; since their preferences are single peaked, each voter in this majority prefers am to am+h. Of course, the same argument applies to alternatives below am. [Hotelling (1929) speculated that political candidates and their platforms could be arrayed along a scale from liberal to conservative, and that this political spectrum formed a natural sequence over which individual preferences are likely to be single peaked. From this it follows that vote maximizing candidates in a two party system will be led to adopt similar platforms toward the middle (median) of the distribution of political opinion.]

For the median voter theorem to have any empirical content, the condition of single peaked preferences is clearly pivotal. In practice, there are a number of empirically important contexts in which the condition seems quite reasonable, and most theoretical and empirical applications have been restricted to such situations. [See, however, Barzel and Deacon (1975), and Sonstelie (1979), for analysis of voting outcomes in situations where preferences are not single peaked.] The simple referendum example described in the preceding section is a case in point. The individual whose public expenditure demand is shown in Figure 1 prefers a public outlay of E*i. If actual spending is increased or decreased from this level, utility declines monotonically, as shown by the curve labeled Ui in Figure 2. Of course, the same argument applies to other individuals as well; correspondingly, Figure 2 also illustrates public spending preferences for two other voters, h and j. In this example, i is the median voter, and E*i is his preferred expenditure level. Notice that any proposal to increase (decrease) spending from E*i would be opposed by a majority of the electorate, h and i (i and j). (The other points labeled in Figure 2 are explained below.)


If the median voter's optimum appears on the ballot it will defeat any alternative motion. But whet guarantees that it will ever be placed before the voters? If motions, or political alternatives, are supplied competitively by political entrepreneurs who seek to maximize their chances of being elected, then the median outcome seems very likely. Nonmedian strategies simply invite defeat in this context. In noncompetitive environments, however, other outcomes are clearly possible. Since the competitive case has received primary attention in the literature, it is discussed first.



The median voter characterization of public sector outcomes has been used extensively in the analysis of state and local government spending patterns. From Figures 1 and 2, the competitive majority rule outcome (E*i) represents one point on the median citizen's demand curve. By relating public service expenditures in a sample of jurisdictions to the median voter's tax price, income, and other attributes expected to influence demands, the possibility of estimating an entire demand function emerges. Early attempts to apply this notion were limited to resting qualitative implications of the theory, e.g., that increases in the size of the non-residential component of the tax base reduce tax prices to all voters, and hence lead to larger expenditures (Bart and Davis 1966). Although the empirical models were somewhat loosely specified, the results obtained were sufficiently promising to encourage further research. The empirical methodology was refined considerably in a group of studies designed specifically to estate parameters of demands for public services (Borcherding and Deacon 1972, Bergstrom and Goodman 1973). In each case, the median citizen's demand was assumed to be a log-linear (linear in logarithms) function of median income and median tax price in a jurisdiction. Specification of tax price usually presents the most difficult empirical task in such studies. An individual's tax price for a public service depends directly upon the cost of producing public output, and upon the share of total taxes paid by the citizen in question (see equations (2) and (3)). Empirically, variations in public service costs are usually related to prices of inputs (Borcherding and Deacon 1972, Deacon 1978). If services are financed by a property tax, then median tax shares can be estimated from information on median housing values and total assessed valuation in each jurisdiction (assuming that the median citizen owns and occupies a house of median value). Application of this methodology to state and local government expenditures has yielded estimates of price and income elasticities that are typically significant and of expected sign. Moreover, a degree of consensus has emerged regarding the actual magnitudes of these effects. Demands are almost invariably found to be price inelastic, with elasticities usually ranging from .2 to .6. Income elasticities are typically positive, but less than unity, e.g., between .4 and 1.0. [There are, however, notable exceptions. Public parks and recreation services often exhibit income elasticities greater than 1.0; estimated income elasticities for public welfare expenditures are sometimes negative.]

Identifying the relevant demand attributes for the median citizen in a jurisdiction is particularly difficult. If, for example, public service demands vary with age, the level of education, or the number of children in a household, the assumption that median demands can be uniquely related to median income and tax price alone, may well be inappropriate. Bergstrom and Goodman (1973) have developed a very useful theorem in this regard. Within the context of log-linear demands and certain regularity properties for income distributions, it provides conditions that are sufficient to specify the form of the median demand in a jurisdiction. [For further discussion of this theorem, and an empirical examination of the assumptions underlying it, see Inman (1978).]

The attempt to model demands in a public choice framework has shed light on a variety of related issues. For example, results obtained by Peterson (1975), Bergstrom and Goodman (1973) and others, are consistent with the proposition that renters perceive themselves as bearing only a fraction of the tax liability on rental property they occupy. Others have noted relationships between spending and the composition of the tax base that indicate the degree to which voters expect taxes levied on non-residential property to be borne by outsiders (Ladd 1975). These results are particularly important for public finance theory because they imply that questions regarding tax incidence and other distributive aspects of public policy cannot really be divorced from resource allocation issues, once a collective choice process is introduced. In this context, policies designed simply to redistribute tax burdens in the population, e.g., "circuit breaker" tax relief, or equalization formulas for school finance, may alter public service levels in ways that are, perhaps, predictable (Inman 1978).

In summary, a large body of theoretical and empirical research has been conditioned upon the median voter hypothesis. The fact that a measure of agreement in empirical results has emerged from diverse studies lends some support to the underlying paradigm. [Moreover, Peterson (1975) and Rubinfeld (1977) summarize evidence to indicate that demand estimates from median voter expenditure studies are in close agreement with results obtained from voting data.] However, the model has by no means escaped criticism (Romer and Rosenthal 1980b). To many, the level of abstraction it employs is stratospheric; moreover, the basic premise that communities actually attain Hotelling-Bowen-Downs equilibrium has largely eluded direct tests. One source of concern is the informational requirements that the model imposes upon voters. Several authors have raised the possibility that complexity in taxing institutions may cause voters to incorrectly perceive tax liabilities. If such a misperception takes the form of "fiscal illusion," a systematic underestimation of tax bills, the effect would be to increase expenditure beyond the median voter's "fully informed" optima. To date, however, attempts to shed empirical light upon this proposition have been rather inconclusive (Wagner 1976, Clotfelter 1978). [As Bergstrom and Goodman (1973) point out, a powerful law of large numbers applies to medians as well as to means.]

As a consequence, purely random errors in perception of tax liabilities or service levels tend to cancel out. A methodologically more telling criticism concerns the fact that the median voter model minimizes the role of interest groups, and assigns politicians and bureaucrats the passive role of seeking out and supplying the demands of the voting middle class. This is essentially an assumption of competition in the political process, and it has been challenged by a number of writers who take a much more monopolistic view of the public economy. Niskanen (1971) was the first to develop a complete theory along these lines, and more recent attempts to formulate models of noncompetitive equilibrium in the public sector have adopted his basic approach.

According to Niskanen, government bureaus enjoy considerable monopoly power as sole suppliers of public services. Consequently, they can control the range of alternatives offered to the citizenry, or their elected representatives. Comparing market relationships in public versus private sectors, "the primary difference . . . is that the [government] bureau offers a total output in exchange for a budget, whereas a market organization offers units of output at a price" (Niskanen 1971, p. 25). In effect, bureaus have the power to confront their constituents with "all or nothing choices." In the context of a majority rule system, this amounts to a limited control over the agenda, or range of alternatives facing voters.

In Figure 2, individual i would receive utility of Ui(0) if no public output were provided. If faced with an all or none offer, both i and j (a simple majority) would prefer expenditure levels up to EN to the prospect of receiving no public service at all. By restricting the menu of available alternatives to a single pair, one of which involves no service, the monopoly bureau can effectively select any expenditure level between 0 and EN. By way of characterizing the bureau's motives, Niskanen offers salary, perquisites of office, prestige and power as plausible maximands, but argues that in general these will be positively related to the size of the total budget. With budget maximization as an implicit goal, the Niskanen equilibrium expenditure level is EN in Figure 2. [It seems clear that Niskanen intended his model to apply primarily to federal governments. In any case, the degree of oversupply certainly depends upon the elasticity of demand and, hence, upon the availability of substitute services. At the local government level, the possibility of migration among jurisdictions presents substitution possibilities that are exceedingly costly at the federal level.]

Niskanen's budget maximization hypothesis has been recently extended by Romer and Rosenthal (1980a). In their model, the budget setting power of bureaucrats is somewhat attenuated, but significant nonetheless. The distinctive features of their approach hinge upon the existence of a "reversion level" to which public expenditures will return if the government's budget proposal is not approved. The reversion level is determined outside of their model, presumably at the constitutional stage. Returning to Figure 2, suppose the reversion level of spending is E0; in effect, this is the bureau's minimum budget offer. Correspondingly, ER(E0) is the maximum expenditure level that could gain a majority over E0, and thus represents equilibrium in the Romer-Rosenthal characterization. Introduction of a reversion level into the budget maximization framework yields a novel implication; in the absence of uncertainty, equilibrium expenditure is inversely related to the reversion level. "By facing the voters with a 'take-it-or-leave-it' choice, the setter [bureaucrat] exercises a threat over the voters. The worse the status quo [reversion level], the greater this threat, and consequently, the greater the gain to the setter [bureaucrat] from being able to propose the alternative" (Romer and Rosenthal 1978, pp. 35-36). Of course, this relationship only holds for reversion levels that are below the median optimum, E*i; beyond that point, the equilibrium budget is trivially equal to the reversion level.

Designing tests to distinguish between competitive and non-competitive political models is difficult at best. As is the case in the private sector, qualitative implications from competitive models are often fully consistent with non-competitive behavior as well. For example, in either Niskanen or Romer-Rosenthal versions of budget maximization, equilibrium public service levels are inversely related to service costs and directly related to incomes. The introduction of a reversion level in the Romer-Rosenthal framework, and its hypothesized effect upon budgets does, however, present an opportunity to separate the two models, since reversion levels should have little or no effect upon budgets in a competitive environment. Apparently the first attempt to examine this question empirically was carried out by Holcombe (1980). In Michigan, school boards have authority to levy property taxes up to a certain rate (which varies across districts) without approval of the electorate. Tax rates in excess of this statutory limit must be approved by majority vote. This system bears a very close resemblance to the Romer-Rosenthal non-competitive model with a reversion level. Holcombe (1980) presents anecdotal evidence to indicate that some districts attempted to exercise monopoly power by making non-incremental offers to the voters. [In one district, the reversion level was a tax rate of 6.0 mills, and the school board offered an alternative of 16.75 mills. The measure passed by a scant 1.1% (Holcombe 1980, pp. 267-268).] On average, however, he found no evidence that budgets exceeded competitive levels. Intuitively, Holcombe (1980) found that excess levies were generally approved by substantial majorities (65.3 percent voting in favor, on average). Thus, school boards could evidently have obtained much larger school budgets (with smeller majorities) than were observed. More formally, Holcombe (1980) made certain simplifying assumptions regarding the general shape of individual preferences for alternative budgets, and assumed that the distribution of preferred budgets was normal within each district. Using these assumptions, together with information on election returns and the actual alternatives offered in school levy referenda, he was able to estimate the median preferred expenditure level in each school district. Comparing his estimates to actual budgets in over 200 districts, he found that observed budgets deviated from median preferred levels by a mere 2.4 percent on average. Ironically, actual levies were below his estimated median points, but the difference was not statistically significant.

The institutions of school finance referenda in Oregon are quite similar to those in Michigan. Romer and Rosenthal (1980) conducted a test of their model using data on educational expenditures and reversion levels in Oregon school districts. A negative relationship between the two variables was expected. After controlling for factors such as income, tax price, and state subventions to local districts they found that reversion levels were significantly related to school outlays, but the direction of the effect was not monotonic. Except for school districts with extremely low reversion levels, actual budgets were positively related to reversion levels, contrary to expectations.

Some additional indirect evidence on the budget maximization hypothesis has been obtained from comparisons of spending levels in jurisdictions where expenditures are set by elected representatives rather than by direct democracy. As Romer and Rosenthal point out, "the possibilities of both monopoly power acquired by elected officials and coalition of minority politics suggest that expenditures will be greater in representative democracies [than in direct democracies]" (1979b, p. 161). Both McEachern (1978) and Holcombe (1980) have examined this hypothesis empirically. In the former study, debt levels in states that require a simple majority referendum to increase debt limits were compared to debt in states with no such requirements. After controlling for certain variables that might independently affect debt levels, no significant difference in per capita debt in the two groups of states was found. In a similar fashion, Holcombe (1980) compared per student educational expenditures in states that require spending referenda, to states where expenditure levels are chosen by elected officials; again, no significant differences were noted.

Finally, both Citrin (1979) and Courant, et al. (1980) conducted surveys of voting age citizens in states considering tax limitation referenda (California and Michigan). When asked about their preferences for increasing or decreasing government expenditures and taxes by various amounts, it was found that the median respondent desired no change; this was true not only for total spending, but for virtually all of the individual services surveyed. However, the observation that responses were consistent with competitive public sector equilibrium is rather surprising in this case, since both measures were adopted. This apparent anomaly may simply indicate the potential pitfalls in survey research, or imply that closer examination of the specific referenda in question is warranted.


The attempt to distinguish competitive from non-competitive outcomes in the public sector is clearly important. This is true for essentially the same reasons that support the study of market power in the private economy; competitive markets attain a degree of efficiency not present in monopolistic regimes. When considering the question of efficiency in the public sector, at least two concepts are important. The first, sometimes called "process efficiency," concerns the degree to which inputs are combined in a cost minimizing fashion. This topic has not been addressed in the present paper, though numerous studies have concluded that, due to the nature of the rewards system, public enterprises are not efficient in this sense. The second concept concerns whether, given the structure of costs, the correct level of public output is produced; it is this notion of efficiency that is of primary interest here. In a general sense, the level of output is "correct," if the marginal benefits of additional output just offset marginal costs. Under certain, not terribly unrealistic conditions regarding the distribution of demands and tax shares in the populace, the competitive Bowen equilibrium is efficient, or at least approximately so, in this sense (Bowen 1943).

From the empirical analysis presented to date, no strong indication of non-competitive budget maximization has emerged. However, given the limited number of attempts to test this hypothesis, final conclusions are clearly premature. In any case, the current degree of polarization in views of the public sector, with some analysts apparently believing that all outcomes are competitive and others that none are, seems inappropriate and counterproductive. It seems reasonable to speculate that the evidence on this question will eventually show some elements of monopoly power in public organizations, at least under certain conditions, as is the case in private markets. If so, the relevant questions will concern the degree of monopoly power observed and the conditions under which it is likely to occur (e.g., at which levels of government, for what kinds of services, under which types of institutional structure). With a bit of imagination, it is even possible to envision development of a public sector analog to the traditional study of industrial organization in private markets.


Barkume, Anthony J. (1978), "Tax Prices and Voting Behavior: The Case of Local Educational Financing," Economic Inquiry, 16.

Barr, J. L. and Davis, O. A. (1966), "An Elementary Political and Economic Theory of Local Governments," Southern Economic Journal, 33, 149-65.

Barzel, Y. and Silberberg, Eugene (1973), "Is the Act of Voting Rational?" Public Choice, 19, 51-58.

Barzel, Y. and Deacon, Robert (1975), "Voting Behavior, Efficiency, and Equity," Public Choice, 21, 1-11.

Bergstrom, Theodore C. and Goodman, R. P. (1973), "Private Demands for Public Goods," American Economic Review, 63, 280-96.

Borcherding, Thomas E. and Deacon, Robert T. (1972), "The Demand for the Services of Non-Federal Governments," American Economic Review, 62, 891-901.

Bowen, H. R. (1943), "The Interpretation of Voting in the Allocation of Economic Resources," Quarterly Journal of Economics, 58, 27-48.

Buchanan, J. M. and Tullock, G. (1962), The Calculus of .Consent, Ann Arbor.

Buchanan, J. M. (1975), "Public Finance and Public Choice," National Tax Journal, 28, 383-94.

Chapman, Randall G. and Palda, Kristian S. (1980), '"Voting Participation in a Public Consumption Perspective," Advances in Consumer Research, Vol. VIII.

Citrin, Jack (1979), "Do People want Something for Nothing: Public Opinion on Taxes and Government Spending," National Tax Journal, 32, 113-130.

Clotfelter, Charles T. (1977), "Public Spending for Higher Education: An Empirical Test of Two Hypotheses," Public Finance, 8.

Courant, Paul N., Gramlich, Edward M., and Rubinfeld, Daniel L. (1980), "Why Voters Support Tax Limitation Amendments: The Michigan Case," National Tax Journal, 33, 1-20.

Deacon, Robert T. (1977), "Private Choice and Collective Outcomes: Evidence from Public Sector Demand Analysis," National Tax Journal, 30, 271-387.

Deacon, Robert T. and Shapiro, P. (1975), "Private Preference for Collective Goods Revealed Through Voting on Referenda, "American Economic Review, 65, 943-55.

Deacon, Robert T. (1978), "A Demand Model for the Local Public Sector," Review of Economics and Statistics, 50, 185-93.

Downs, A. (1957), An Economic Theory of Democracy, New York.

Gramlich, Edward M., Rubinfeld, Daniel L., and Swift, Deborah (1980), "Voter and Non-Voter Preferences for Public Spending and Spending limits," Institute of Public Policy Studies, University of Michigan.

Holcombe, Randall G. (1980), "An Empirical Test of the Median Voter Model," Economic Inquiry, 18, 260-74, and preliminary mimeo with same title, Texas A and M Univ., 1976.

Hotelling, Harold (1929), "Stability in competition," Economic Journal, 29, 41-57.

Inman, Robert P. (1978a), "Optimal Fiscal Reform of Metropolitan Schools: Some Simulation Results," American Economic Review, 68, 107-21.

Inman, Robert P. (1978b), "Testing Political Economy's 'As If' Proposition: Is the Median Income Voter Really Decisive?'', Public Choice, 33, 15-66.

Ladd, Helen F. (1975), "Local Education Expenditures, Fiscal Capacity, and the Composition of the Property Tax Base," National Tax Journal, 28, 145-58.

McEachern, William A. (1978), "Collective Decision Rules and Local Debt Choice: A Test of the Median Voter Hypothesis," National Tax Journal, 31, 129-36.

Niskanen, William (1971), Bureaucracy and Representative Government, Chicago: Aldine-Ahterton.

Nose, Eli M. (1979), "Tax Limitations and Fiscal Federalism," National Tax Journal, 32, 181-88.

Peterson, George M. (1975), "Voter Demand for Public School Expenditures," in Public Needs and Private Behavior in Metropolitan Areas, John E. Jackson (ed.), Cambridge, 99-115.

Romer, T. and Rosenthal, H. (1978), "Political Resource Allocation, Controlled Agendas, and the Status Quo," Public Choice, 33, 27-45.

Romer, T. and Rosenthal, H. (1979e), "Setters, Reversions, and Median Voters: Evidence from School Expenditure Referenda," Carnegie-Mellon University, Pittsburgh.

Romer, T. and Rosenthal, H. (1979b), "The Elusive Median Voter," Journal of Public Economics, 12, 1-27.

Rubinfeld, Daniel L. (1977), "Voting in a Local School Election: A Micro Analysis," Review of Economics and Statistics, 59, 30-42.

Sen, A. K. (1968), Collective Choice and Social Welfare, San Francisco: Holden-Day.

Sonstelie, Jon (1979), "Public School Quality and Private School Enrollments," National Tax Journal, 33, 343-54.

Wagner, Richard E. (1976), "Revenue Structure, Fiscal Illusion, and Budgetary Choice," Public Choice, 31.