Robert P. Inman
Professor of Finance and Economics, Wharton School, University of Pennsylvania
Daniel L. Rubinfeld
Robert L. Bridges Professor of Law and Professor of Economics, University of California, Berkeley
Federalism is today a topic of intense intellectual debate in many countries throughout the world. In Europe, the former Soviet Union, South Africa and elsewhere the view that good government will involve a blending of local and central governmental decision-making is now well accepted. With a topic such as federalism that has been so widely debated from so many perspectives, it would be impossible to provide a comprehensive review of the literature. Rather, we provide a particular perspective, one that balances the twin goals of political participation and economic welfare. A carefully-specified theory of federalism has many useful applications: to government expenditure policy (Martinez-Vasquez 1994) , to tax policy (Inman and Rubinfeld 1996), to deficit policies (Inman 1990) , or to regulatory policy (Rose-Ackerman and Mashaw 1984) .
Confederations of small governments have been praised by many -- from Plato and Aristotle, through Rousseau, Montesquieu, Smith, and Mills, to contemporary federalist legal and economic scholars -- as that political institution most likely to encourage a trio of social virtues: political participation, protection of the sovereign rights of citizens, and economic efficiency. [On participation and the virtues of small governments see Pateman (1970) at 46, 110, 111, and Frug (1980) . On economic efficiency and the positive role of small governments see Smith (1776) and Tiebout (1956) .] Other scholars have been more skeptical. The Federalist Papers (particularly at No. 47), the intellectual blueprint for the U.S. Constitution, raises serious doubts about the virtues of loose confederations of small governments. Leading contemporary social scientists from political science (Robert Dahl and William Riker) and economics (Joseph Schumpeter and Paul Samuelson) have lent theoretical support to the Federalists' concerns. [On small governments and their consequences for meaningful political participation, Schumpeter (1943) , Dahl (1956) , and Dahl and Tufte (1973) . On the potential inability of small governments to protect group rights, see Riker (1964). For a discussion of why small governments might be economically inefficient, see Samuelson (1954) .] Like the Federalists, these scholars have argued that a strong, democratically elected central government is generally the more effective means to ensure political participation, protection of the rights of citizens, and more efficient and equitable resource allocations.
Which level of government is best suited to make public policies, given that we want those policies to be democratically decided, respectful of personal rights, and economically efficient? A considered answer -- whether for Europe's new Economic Union, the new South African Constitution, or to the regulation of anticompetitive firm behavior within the United States or the Economic Union -- requires a careful balancing of the benefits and costs of decentralized and centralized political structures. To fashion such a balance, it is essential that we first review the potential virtues of each federalist structure.
We begin by defining what is meant by decentralized and centralized federalist structures. In this essay, governments will be defined as "decentralized and state" or "centralized and national" according to the combination of two constitutional decisions. The first decision -- called the partition decision -- divides the single national citizenry into states. Following the partition decision, there will be a group of separate state governments and one central government to which each state will send (for simplicity) one representative to a central legislature, the single policy-making body of the central government. Policy decisions made by state governments will be decided by their separate legislatures. [Adding another legislative chamber or an executive with a veto to our analysis will only improve the ability of federalist systems to meet the participation, sovereignty, and efficiency objectives. For an analysis of the effects of bicameral legislatures on government performance, see Levmore (1992) . For an analysis of the effects of a president with veto powers on government performance, see Fitts and Inman (1992) .]
Courts and legislatures are concerned with the second constitutional decision -- the assignment decision -- which allocates final political responsibility for choices over a social or economic policy to either the legislature of the central government or to the legislatures of each of the separate states. If the assignment decision allocates ultimate responsibility for a social or economic policy to the states (or their localities), then we describe policy-making for this class of government decisions as "state" or "decentralized." In contrast, if the assignment decision gives final responsibility for a social or economic policy to the central government, then we describe policy-making for these assigned responsibilities as "national" or "centralized." Together, these two constitutional decisions define the essential federalist -- decentralized or centralized -- structure of the government. We realize, of course, that federalist structures may have decentralized and centralized components, as, for example, when the central government mandates the form and function of certain decentralized government activities.
The much debated virtues of these decentralized or centralized governmental structures include democratic political participation and enhanced economic efficiency. Which federalist structure or combination of structures favors which virtues given the realities of the political and economic marketplace? We review the arguments and the contemporary evidence in Section II (participation) and III (economic efficiency) below. Not surprisingly, neither a centralized nor a decentralized structure maximizes public participation and economic efficiency all the time. Which structure is preferred turns on a balancing of these values. In IV we discuss briefly how a society might set that balance.
Political participation is defined as "actions through which ordinary citizens of a political system influence or attempt to influence (political) outcomes." [ Nagel (1980) at 1.] Actions include voting, debating, marching, picketing, contributing, and passive and armed resistance. The aim of these actions is to influence the direction of government policy. Active citizens may be successful in moving public policies closer to their preferred outcomes because of their participation. In this case, citizens are said to have influence. Or active citizens may be unsuccessful, their political actions having no effect on policy outcomes. In this case, the citizens have attempted, but failed, to influence political outcomes.
The benefits to citizens of their own, and others' participation, can be divided into three categories: instrumental or utilitarian, developmental or educative, and intrinsic or consumptive. In the utilitarian view, perhaps argued most directly in the work of Bentham and James Mill, political participation serves a specific function; to ensure that government maximizes aggregated citizen utility or welfare. In democratic societies, political participation is the necessary check on abuses of citizen welfare by ill-informed or even greedy elected officials. Political participation guarantees that abusive elected officials will be exposed and defeated at the next election. For the utilitarians, political participation's value is fundamentally instrumental; it plays, in the words of Carole Pateman (1970) , a "protective function" for the public welfare.
For Jean-Jacques Rousseau and John Stuart Mill, political participation was instrumental to another, perhaps more important, end: it protected citizen liberties. While both Rousseau and Mill acknowledged the instrumental value of participation in protecting private economic interests and ensuring good (efficient) government, they extended the argument of the utilitarians to include participation's protective value for personal freedom as well. Participation ensures that no one individual or group is master over any other -- "The rights and interests of every or any person are only secure from being disregarded, when the person interested is himself able, and habitually disposed, to stand up for them." [ Mill (1960) at 186. See Pateman (1970) Chapter 2, particularly at 26-27, for a summary of the arguments of J. S. Mill and J. Rousseau in favor of political participation.]
For many scholars, however, participation does much more than just protect private economic interests and personal liberties. Aristotle, Rousseau, J. S. Mill, de Tocqueville, and contemporary commentators such as Dewey and Pateman view citizen political participation as serving an important educative function as well. By participating in the political process the individual learns that his private interests are intimately linked to the interests of others. This knowledge leads to a willingness to compromise, to put private interests aside and to call upon values of justice and common good when making public choices. There may be beneficial spillovers to private decision-making, too. Having recognized and accepted the validity of others' interests in the public arena, we may be more accepting of those interests in our private dealings. Simply put, we may be different people -- more tolerant, less selfish -- after political participation.
Finally, political participation may provide direct consumptive benefits to those who participate. From those who simply like to hear the sound of their own voice or revel in controlling others, to those who take genuine satisfaction in having contributed to a cause they feel is worthy, the act of political participation offers many a personal consumption benefit. How else might we explain widespread voting in national elections? [ Ordeshook and Riker (1973) Chapter 3.] While perhaps not the most compelling argument for increased political participation, it should not be ignored.
If political participation is an activity to be valued, how do we know if there is more or less of it within society? The instrumental, or influence, effects of individual political participation can be measured by a metric of political power . At one extreme, the individual fully controls the outcome, while at the other the individual has no influence over the outcome. If one person controls the outcome, then, of course, all others must be powerless. [See Nagel (1975) for a detailed discussion.] The participatory, or "actions," effects of political participation can be measured by a metric of political effort . One plausible effort metric would be the hours, or the hourly equivalent of dollar contributions, given by the individual to political activity.
Whether an individual engages in political activity and whether those activities translate into political influence depends on the individual's personal desires to participate and to have power, and on how society's political environment translates individual participatory activity into influence. Important attributes of the political environment will include the political institutions that define voting rights, freedom of expression, and access to elected officials and the legislature. Also important, is the number of citizens who can participate in public decision-making.
As a theoretical matter, one would expect that participation would be higher in local politics simply because the benefits of influence are greater the fewer the number of disparate constituents. However, individual political effort is not so easily predicted. When participatory effort is more productive in producing political influence, as it is likely to be in small democratic states, individuals might actually participate for fewer hours, choosing to allocate some of their increased "political productivity" into other, non-political activities. Ultimately, how political institutions and the population of the nation and its states affect political influence and political effort is an empirical question. Certainly the early (Aristotle, Rousseau, J.S. Mill) and contemporary (Pateman, Frug) advocates of participatory politics see the small, democratic state as most conducive to high values of influence and effort for the average citizen.
How will the two constitutional decisions central to setting the federalist structure -- the partition decision and the assignment decision -- affect the level of political participation in a society? The answer is ultimately empirical: How do political influence and participatory effort change as political decision-making moves from large central governments to smaller, decentralized states and local polities? Participation theorists have not hesitated to make predictions, typically arguing that a federalist system with a large number of small states with substantial policy responsibilities will increase the political influence of each citizen and generate increased political effort as well. There is little doubt that this is true for political influence, especially in societies with strong guarantees of equal political rights. Less certain, however, is the relationship between political effort and the size of states with decisionmaking power; it is not difficult to present arguments that effort may rise, fall, or even remain unchanged as federalist systems become more decentralized.
The empirical evidence on how political influence and participatory effort change as the governmental structure becomes decentralized is sparse, but what is available generally supports the prediction that both influence and effort will rise as the size of government declines. Within-country comparisons of political influence and effort across small and large governments seem the most appropriate testing ground of the participation hypothesis. [See Dahl and Tufte (1973, pp. 44-53) for a review and critique of the cross-national evidence on size and participation.] First, citizens surveyed in five countries (United States, Britain, West Germany, Italy, and Mexico) all reported making greater effort to understand, and having more success in understanding, local rather than national political issues. In the surveyed countries, citizen efforts to influence government were also typically two to three times higher for local than for national governments. In supporting evidence, a detailed survey of political participation across Swedish cities found average citizens in small communities were more aware of local politics than their counterparts in middle size cities and both knew more basic local political facts than large city residents. Second, political effectiveness or influence -- at least as perceived by citizens -- also seems to increase as governments get smaller in size. Again, the five nation survey finds that citizens feel much more influential at the local than at the national level; typically, half of the citizens in the surveyed countries felt they had no influence over national policies while only 25 to 30 percent felt powerless at the local level. A more detailed analysis of the U.S. subsample showed a significant positive correlation between an index of political powerlessness and size of government. [ Finifter (1970) at 389-410. ] Indirect evidence from the Swedish study also implies that citizen influence declines as governments get larger; the gap between the percent of citizens and the percent of elected officials who would support a tax increase increases as communities get larger. Finally, evidence from the United States, the Netherlands, and Switzerland all show that politics in larger communities, states, and cantons are increasingly dominated by organized interest groups or political parties. While such groups make participation easier for some citizens, they may also it more difficult for others. If so, then Rousseau's concern that participation becomes more unequal as governments get larger is confirmed.
Any final conclusions regarding the effects of institutional decentralization on participatory effort and political influence must await more complete empirical analyses. Nonetheless, theoretical argument and the current empirical evidence lead us to the tentative conclusion that smaller governments, and thus more decentralized federalist structures, are likely to encourage the valued goal of greater political participation.
Using the economic efficiency criterion to choose the appropriate federalist structure is a complex endeavor involving a comparison of the economic benefits and costs of assigning responsibility to each level of government: in the U.S., to the national or central level, the state level, or the local government level. In a federal system there are two important dimensions to economic efficiency. First, inter jurisdictional efficiency involves the appropriate allocation of individuals and other resources, such as capital, among the different jurisdictions. Interjurisdictional efficiency is achieved when the public activities of these interacting governments satisfy the collective demands of individuals living in different jurisdictions at a minimum economic cost. [A full analysis would also include the economic costs of politically deciding what those policies should be and ensuring that the politically decided policies are what the government actually provides. These latter costs have been identified elsewhere as the transactions costs of government. They include the costs of selecting legislators, the decision-making costs of setting a policy, and the monitoring costs of ensuring that the policy is put in place; see Breton and Scott (1978) .] Second, intra jurisdictional efficiency requires a choice of public activities which satisfies the collective demand within a given jurisdiction, measured as the willingness to pay for those activities for all individuals within the jurisdiction, and again does so at minimum costs.
In our complex federal system both interjurisdictional and intrajurisdictional inefficiencies can arise, the former because of spillovers , the decisions and actions of individuals within one jurisdiction can have effects on individuals located in other jurisdictions that are not accurately reflected in the marketplace, and the latter because the political process may not generate outcomes within a jurisdiction which maximize the well-being of its residents, i.e., create political inefficiencies . In a broad sense, the choice of the appropriate jurisdiction to be responsible for a government activity involves a trade-off -- the larger the jurisdiction the less likely that there will be spillovers from one jurisdiction to the next, but the more likely that the political process will lead to a misallocation of resources within jurisdictions.
A decentralized system of governments is ideal from the point of view of intrajurisdictional efficiency, since assigning economic responsibility to the smaller and more homogeneous of the jurisdictions increases the likelihood that services provided and regulations promulgated will be consistent with the desires of each and every member of the population. In other words, in a decentralized system of governments, local tastes for public goods are likely to be less disparate than national tastes. The resulting reduction in the inefficiency of providing public goods (since citizens with similar tastes would prefer to consumer the same level of public goods) has been characterized as the decentralization theorem by Wallace Oates.
Unfortunately, decentralization has its disadvantages, due primarily to the spillovers that are likely to arise when jurisdiction size is small. A less decentralized system with larger jurisdictions can minimize the spillovers involved, while at the same time taking advantage of the lower cost of promulgating and enforcing certain regulations or dispensing the benefits of public programs, and the lower cost of obtaining and using information.
Our analysis of the political economy of federalism begins with a critique of the arguments for a strongly decentralized system. While we agree that a focus on the presence of substantial spillovers is appropriate, we stress the fact that decentralization will not guarantee an optimal public sector mix even when no spillovers are present. We believe that a complete theory of federalism cannot rely solely on the "Tiebout Model", an ideal model of a fiscal competition between many independent local governments. Appropriately specified, Tiebout's decentralized model of governmental competition is formally analogous to the purely competitive market model with complete information used to study market competition between private firms. In the Tiebout model, government policies are determined by market , or economic, exchange alone. Unfortunately, the assumptions needed to ensure that governmental competition is fully analogous to efficient private market competition are very strong, and may not hold in realistic public economies. [See, for example, Rubinfeld (1987) .]
A more appropriate analytic framework -- based upon political models of state government decisionmaking and interstate bargaining -- yields, we feel, more appropriate conclusions as to the likely economic performance of decentralized public environments. In the section that follows we study how such systems are likely to work. We conclude that when viewed as a political rather than market process, decentralized federalism may be economically inefficient. This conclusion suggests the need to consider a more centralized environment, using the central, or national, government to set policy. To be consistent, however, the analysis of central government performance must use the same political model that found inefficiencies at the decentralized level. This we do by explaining how central government regulatory policies also might be inefficient -- indeed, potentially worse, for political reasons, than the inefficient state and local policies which the central policies had hoped to correct.
According to one view, the world of decentralized, competitive governments will allow for a national regulatory marketplace, in which individuals move among local jurisdictions or states to select the regulatory combination that they most desire. Just as a private market place forces suppliers to produce their goods at the lowest cost, and to provide the goods most desired by consumers, so too will the pressure of taxpayers shopping, through exit and relocation, force governments to be efficient in their provision of public goods and public goods and services.
According to this view, public activities should be decentralized, except when there are significant economic spillovers from the regulating jurisdiction to its neighbors. When spillovers result, decentralized governments may ignore important economic costs or benefits which will lead them to over-, or under-, provide the activity. In these instances central government review is appropriate.
We must be clear, however, about the assumptions built into the Tiebout model of decentralized governments on which this model relies. Each household and business is highly mobile and chooses to reside in the jurisdiction that offers the tax-expenditure-regulation package that it most prefers. As typically specified, five conditions define the competitive Tiebout public economy:
1. Publicly provided goods, services, and regulatory activities are provided at minimum average cost; [If assumption 1 holds, there is an efficient population size which minimizes the average cost per household of providing that government activity. This rules out using Tiebout competition to allocate "pure public" goods, those goods where additional users of the government's facility does not reduce the consumption benefits enjoyed by previous users. Typical examples of a pure public good include defense, television or radio signals, and new knowledge. Goods which can be allocated by the Tiebout competitive process are those goods where more users reduces, proportionally, the benefits enjoyed by current users. Rather than pure public goods, these goods are congested public goods. Education, highways, police protection, and parks are examples. The efficient scale of operation for such goods must be sufficiently small (e.g., 15,000 in population) so that each household can find a suitable jurisdiction in which to reside. Existing evidence suggests that this constraint can be met for many government services; see, for example, Ladd and Yinger (1989) at 83, 85.]
2. There is a perfectly elastic supply of political jurisdictions, each capable of replicating all attractive economic features of its competitors; [This assumption ensures that efficient alternatives exist; this is the task of public sector entrepreneur-politicians who run against, and defeat, inefficient incumbents; competitive political entrepreneurs play a role analogous to take-over "artists" in the private market. Or public sector entrepreneurs can appear as enterprising real estate developers who build new jurisdictions -- analogous to "new entrants" in a private market -- to draw away dissatisfied residents from inefficient governments. See Epple and Zelenitz (1981) .]
3. Mobility of households and businesses among jurisdictions is costless;
4. Households and businesses are fully informed about the fiscal and regulatory policies of each jurisdiction; [Assumptions 3 and 4 ensure that when possible inefficiencies arise, households and businesses can move to otherwise similar jurisdictions without those inefficiencies.] and,
5. There are no interjurisdictional externalities or spillovers. [Assumption 5 ensures that all public regulatory activities can be provided within these efficient jurisdictions -- larger governments or intergovernmental cooperation is not required. In the end, a regulated economy satisfying assumptions 1 to 5, and organized as a fully decentralized network of competing jurisdictions, will maximize economic efficiency. Citizens and businesses can consume their preferred levels of the public regulatory activity with a minimum expenditure of production and transactions costs.]
While a useful starting point, the Tiebout model does not provide a complete theory of federalism. The assumptions 1 through 5 may arguably apply to a local public activity such as rent control, which is imposed in California in only a relatively few cities, thereby leaving most households the opportunity to move to non-rent control jurisdictions as they see fit. However, the assumptions are less likely to hold for more centralized activities, where choices are more limited and moving is more costly.
If assumptions 2 through 5 do not hold, we must rethink the optimality of complete decentralization. The model of perfectly competitive private markets has long been an appealing paradigm for economists, in part because some real world markets do actually seem to operate in that manner, and in part because if they do, the efficiency consequences seem clear. Neither of these reasons applies without qualification to the Tiebout framework for competitive governments , however.
Consider assumption 2, that there is a perfectly elastic supply of political jurisdictions. The model of perfectly competitive markets yields an efficient allocation of resources in part because firms are free to make the efficient choices of inputs and outputs, and in part because entry into and exit from an industry are free. In the Tiebout model, competitive governments fill the role of firms. But governments may be constrained by internal politics when making their input and output choices, and free entry of new governments is by no means assured in actual public goods economies. [See generally, Epple and Zelenitz (1981) , arguing that limited entry permits competitive governments to maximize public "monopoly profits" from the provision of public goods and public regulations and the level and allocation of those profits will be determined politically within the jurisdiction.] For example, an entrepreneur cannot simply invest his capital to build a new city.
If assumption 2 does not hold, four potentially adverse consequences for economic efficiency are likely to arise. First, with a limited supply of jurisdictions each government is likely to contain citizens with different views of what constitutes the government's best policy. If so, local politics becomes important. If local decisions are made by majority rule, the outcome will not necessarily be economically efficient. [See Bergstrom, Rubinfeld, Roberts, and Shapiro (1988) .] Second, when the economy can no longer replicate attractive local jurisdictions, economic rents can be earned. The presence of these publicly created rents associated with the desirability of some locations over others can in turn create a tension between the fiscal entrepreneurs who wish to develop new communities and the owners of the desirable locations. As a result, there is not guarantee that the market for land and for other scarce resources will be economically efficient.
Third, when private sector job opportunities are not identical in all locations, productive private sector employment may be sacrificed as labor is attracted to less efficient locations for fiscal reasons. The consequence is a less efficient private economy. [See Buchanan and Goetz (1972) for the initial presentation of this argument and the summaries in Boadway and Flatters (1982) and Wildasin (1986). ]
Fourth, once the supply of jurisdictions is fixed, the analogy of the decentralized model of federalism to anonymous free market competition is no longer valid. Each local government may be aware of the potential effect of its action on other jurisdictions. It will then act strategically in making its decisions on the basis of its best guess as to how other jurisdictions will respond. In this framework the actions of each local jurisdiction, even if small, can have a substantial effect on the welfare of the rest of the decentralized world. Essentially, a self-interested decision by those in power in one jurisdiction can result in a series of distortions that arise throughout the effected network of other jurisdictions. For example, a decision by one jurisdiction to offer a tax break to attract new business could lead to a competition in which many jurisdictions offer a variety of tax business tax benefits.
While it is difficult to predict the exact consequence of the rather complex behavioral patterns that can arise when these strategic interactions are taken into account, it seems clear that an efficient outcome is unlikely to arise. States may, for example, underprovide welfare services, or more generally to underprovide public services in which they have particular production advantages and therefore should provide in abundance. [See Gramlich and Laren (1984) on welfare competition among states, and Pestieau (1977) .] This could occur because a state realizes that by providing a high level of welfare benefits it is likely to attract new low-income residents from nearby states, thus adding substantially to its cost of providing such benefits. States might also provide substantial tax benefits to attract business, or at least to avoid losing business to other competing states, and in the process undertax business capital and overtax the residential base. [See, for example, Break (1967) at 23-24 . See also Oates (1972) at 142-3 . For a review of the more recent literature, see Inman and Rubinfeld (1996) .
An equivalent phenomenon can occur with respect to regulatory policies. States might relax their environmental controls to encourage business migration, or simply to forestall the loss of business because other states had or were about to relax their environmental regulations. [This argument has been developed by Cumberland (1981) . See also Fischel at 119 (1975) , and more generally, Oates and Schwab (1986) .] Note that the competitive process which leads to a uniformity of regulation among states is not consistent with the advantages of the diversity associated with the Tiebout model, nor is it likely to be efficient. The result could be a "race to the bottom" resulting in an equilibrium in which environmental regulations are "undersupplied." [The uniformity of occupational regulation is described in Stigler (1971) at 130 . For a more general argument that the regulatory actions of individual states may not result in an optimal federal system, given the incentives that determine political choice, see Rose-Ackerman (1981) . See also, Revesz (1992) .]
Assumption 3, that citizens and businesses are costlessly mobile between political jurisdictions is also problematic. The assumption applies best to suburban jurisdictions among which individuals and firms have a good deal of mobility. When the number of cities and towns within a metropolitan area is substantial, families and firms have significant choice in their provision of public goods and public regulations. Empirical evidence supporting the Tiebout hypothesis has focused entirely on this local public sector. [For studies of household relocation, see Gramlich and Rubinfeld (1982) , who used survey data for Michigan and concluded that their results were "reasonably consistent with a Tiebout interpretation: in large metropolitan areas there is quite extensive grouping; in smaller areas, there is some grouping." (at 554-55). Oates (1972) at 162-179 studied the metropolitan New Jersey region, using tests of the capitalization of public goods differentials into property values, and concluded that public goods preferences could explain migration among neighboring cities and towns. For evidence that businesses are sensitive to local taxes and services when relocating within a metropolitan area, see Bartik (1988) and McConnell and Schwab (1990) , suggesting that environmental regulations are not an important factor affecting business location decisions.] Activities of local governments in metropolitan areas -- save those with significant spillovers -- are likely to meet the competitive test. [See Courant and Rubinfeld (1981) .] Yet much economic activity occurs at the state level, where assumption 3 is unlikely to hold. Studies of migration show that most household moves are within metropolitan areas, not between states. Those who do move between states are typically job motivated, and are not primarily seeking a more attractive public service/public regulation environment. [ Easterbrook (1983) at 35 n. 28 . See also the detailed evidence on interstate and interregional location of households reviewed in Greenwood (1986) .] If assumption 3 does not apply, then the model cannot guarantee overall (i.e., market plus public sector) efficiency.
Assumption 4, that households and businesses are fully informed, presents another problem for the Tiebout argument. Households and firms must have reasonably accurate information about the tax, spending, and regulatory policies of the communities in which they will reside. Yet, it is often difficult for individuals and firms to ascertain the implications of regulations within their own jurisdiction, let alone make a reasonable judgment about the public activities that will exist in other jurisdictions at a future point in time after which they have relocated. [ Stigler (1971) at 130 , has argued, "(t)he voter's expenditure to learn the merits of individual policy proposals and to express his preferences ... are determined by expected costs and returns" and that, in the case of regulatory policy, this cost-benefit calculus dissuades voters from discovering very much.] When we view the location decision as involving a choice of public sector packages, and we take into account the fact that other jurisdictions are likely to be changing those packages over time, the informational demands that must be satisfied for the competitive system of governments to work effectively seem very great.
When assumption 4 no longer holds, i.e., when there is incomplete information about the costs and levels of regulatory activity, an important principal-agent problem arises. Households and firms have two options: go it alone and risk being exploited by better informed fiscal entrepreneurs and local factors of production, or hire an agent to protect household and firm interests. Complications arise when information about agent performance is costly and some deceitful agents survive, or when the expertise provided by agents to households and firms cannot itself be kept private. [To the extent the market for "agents" works perfectly -- all inefficient or exploitive providers are exposed, and only truthful agents survive -- the final outcome will still be efficient, but some fiscal surplus will be allocated to the agents as payment for services performed. Such payments can be made as a contingent fee conditional on the performance of the agent in managing the purchase of the public good, or as a competitively determined lump-sum payment to a previously identified -- by reputation or certification -- "truthful" agent.] In both instances there will emerge externalities in the market for agents that can cause an inefficient level of information to be provided about the fiscal performance of the Tiebout communities. [When information about agent performance or abilities is available at no cost, then agent efforts can be fully monitored (avoiding all problems of moral hazard) and agent abilities will be fully known (avoiding all problems of adverse selection). The agent market can then work perfectly. When information about agents is costly, however, there is no guarantee that the efficient agent market will arise; see Holmstrom (1985) . Similarly, when the information provided by the agents to households cannot be kept private, then there will be externality-caused inefficiency in the information market.]
The final assumption, requiring no spillovers, is an important real world violation of the Tiebout framework that may require central government intervention. A spillover or externality arises when the activities of a jurisdiction affect directly the well-being of an individual or firm in another jurisdiction and the decisionmaking of the first jurisdiction does not take these effects on others into account. [For a discussion of public goods externalities see Williams (1966) and Brainard and Dolbear (1967) among others. This literature is critically reviewed and extended by Pauly (1970) . For a discussion of tax externalities see the survey in Inman and Rubinfeld (J. Pub. Econ. 1996) .] One occurs when a spillover offers a positive benefit to a neighboring community, for example, through the regulation of socially damaging air pollution emitted by a locally valuable firm. When such positive spillovers exist, state and local governments may provide too little of the activity. The converse case provides a third example. Here a socially productive firm produces only local pollution, for example, a garbage or nuclear waste site. The incentive is for the local community to impose stringent regulations on such a firm so as to force its relocation onto a neighboring community. A negative regulatory externality results. If the neighboring communities respond with similar regulations, the firm may have no place to go, even though it is in the interests of all jurisdictions to have the firm locate in at least one locality. With negative spillovers, there may be too much state and local activity. [For an illustration of this point with respect to the regulation of solid wastes, see David and Lester (1989) . In these circumstances, a federal or state set of uniform standards may be preferred by all. In fact, uniform national regulation not only can take advantage of economies of scale in administration, it can also produce scale economies of production and distribution for firms, while at the same time reducing their location costs; see Rose-Ackerman (1994) . Augmentation of national standards by states or localities can be allowed, but only if the additional regulations do not generate spillovers by benefiting local households and businesses at the expenses of nonresidents.] The general point is clear: when assumption 5 is violated, there is no guarantee that decentralized competition will produce an efficient allocation of public and private resources.
Our review of the assumptions of the Tiebout framework lead us to conclude that the analogy between the model of competitive governments and the model of competitive markets is not sufficiently strong to serve as the basis of a theory of federalism. If we are to have a compelling efficiency basis for drawing the lines between decentralized and centralized governmental policy it must be found in an analysis based upon paradigms other than Tiebout's model of public sector competition. Our discussion in the next two subsections seeks to provide that paradigm.
Economic efficiency demands that state public activities be efficient both with respect to the preferences of residents of the state ( intra jurisdictional efficiency) and with respect to the preferences of residents outside the state ( inter jurisdictional efficiency). Separate political institutions exist through which each dimension of efficiency might be achieved. State legislatures and their appointed state agencies are the means through which residents within a state might hope to obtain intrajurisdictional efficiency. Interstate cooperative agreements struck between residents, or their elected agents, of neighboring states are the means to interjurisdictional efficiency. We consider the potential of each, beginning with an analysis of state legislative processes.
When elected representatives join together in a state legislature, they are free to fashion coalitions and set regulatory policies as they wish, subject only to the constraint that decisions receive the approval of a majority of legislators. Majority-rule legislatures making decisions over multiple public policies must resolve a fundamental structural defect: the propensity of the majority-rule process to cycle from one policy outcome to another. [This fundamental observation regarding majority rule processes has been well recognized at least since the work of M. de Condorcet in the "Essai sur l'Application de L'Analyse à la Probabilité des Decisions Rendues à la Pluraliste des Voix" (1785) and is known popularly as the paradox of voting: even though majority rule leads a legislature to vote for policy A over policy B, and for policy B over policy C, it is certainly possible, and often likely, that policy C will be defeat policy A. Kenneth Arrow (1963) proves that this paradox of voting is a symptom of a more fundamental problem with all democratic processes: there is no guarantee that such processes can find a best outcome or are immune to manipulation. Thus, the design of political institutions will almost inevitably be a search among the second-best. For a survey of this literature in the context of the new political economy, see Inman (1987) .] When no winning coalition is capable of holding its majority against small policy variations offered by a losing minority, then either no decision will be made or final policy outcomes will be random and uncertain.
If legislatures are to reach decisions, additional legislative institutions must be discovered for overcoming this inherent instability of the majority-rule process. Two institutional approaches are commonly used by legislatures. The first assigns agenda-setting powers to a small subset of members, say the speaker of the house or a key legislative committee. Other members in the legislature then simply vote -- up or down -- on the items in the approved agenda. Most likely, policies will be approved by a bare majority -- a minimal winning coalition -- in this strong agenda-setter legislature. [For articulation of the general theory of how this process overcomes cycling, see Baron and Ferejohn (1989) . For a careful description of what legislative institutions are used to assign agenda control, see Shepsle (1979) . There is an extensive literature which tests for the direct influence of agenda-setters -- typically congressional committees -- on policy outcomes; see Krehbiel (1992) for a critical but balanced review.] A second strategy shares agenda-setting powers among all members, giving each legislator a right to select his most preferred policy in that policy area most germane to the legislator's constituents. This second approach to legislative decision-making involves each legislator deferring to the preferred policies of all other legislators, provided the other legislators defer to the legislator's own policy requests. The guiding principle here is a norm of deference -- "You scratch my back, I'll scratch yours" -- and it typically results in legislative proposals which are approved unanimously. For this reason such legislatures are often called "universalistic." [The guiding theory can be found in Weingast (1979) and Niou and Ordeshook (1985) . The theory has been developed for, and its applications must be limited to, understanding how legislative bodies allocate government projects and government regulations -- what has been called distributive politics; see Lowi (1969) . A norm of deference is not an appropriate model for predicting how legislatures will decide re distributive policies such as welfare or progressive taxation, politics where one group constituents must lose to another. Details regarding the legislative institutions needed to enforce the norm of deference can be found in Weingast and Marshall (1988) for the details of how such a legislative norm of deference might be enforced.]
A single legislator uncertain as to whether she will be one of the lucky winners in the bare majority legislature will typically favor the more open rules of the universalistic legislature. Intuitively, the universalistic legislature is preferred because under universalism activities that benefit the legislator's district are nearly certain to be chosen, and a substantial portion of the costs will be borne by other districts. [If benefits exceed average project costs in a majority of districts -- a likely case -- then a legislature run by a norm of deference will be the preferred legislative institution. Barry Weingast (1979) and Emerson Niou and Peter Ordeshook (1985) prove exactly when this outcome holds. See also Schwartz (1989) or Inman and Fitts (1990) , both of whom extend the analysis to legislative projects of variable size. The conclusion that legislators prefer universalism to bare-majority rule extends to this more general case.]
Universalistic state legislatures operating under a norm of deference run a significant risk that their public activities will be economically inefficient, however. Consider, for example, the typical case of a state regulation which raises industry prices within a state, effectively reducing output below competitive levels. Those who benefit from such a regulation are state residents who own capital and other "specialized" factors of production employed in the regulated industry. Those who lose are the state's consumers who buy the output of the regulated industry and the state's taxpayers who pay for the regulatory bureaucracy.
Assume further that the industry would be competitive, pricing at marginal cost, if unregulated, but would act as a profit-maximizing cartel if regulated. The cartel would choose to maximize its profits by raising its price above the competitive level, and by reducing its output. Three conclusions follow from this framework. First, industry producers within the state prefer to be at a regulated output near the cartel profit-maximizing level, because as regulation nears this level industry profits, the "marginal benefits" of regulation to the interest groups that own the specialized inputs that gain from regulation, are maximized. Second, state consumers of industry output prefer the competitive output, for at zero industry regulation, the tax costs and lost consumer benefits from regulation -- the "marginal costs" of regulation -- are minimized. Finally, any regulation of a competitive industry is inefficient from the perspective of all state residents, because the additional cost to state taxpayers and consumers everywhere greater than the additional benefit to state producers.
How then will a universalistic state legislature set government policy for the state? Under the norm of deference each elected state representative is allowed to propose a state policy which maximizes the returns to his constituents. Consider how one legislator (perhaps with a group of other legislators representing similar districts) might view an activity when all of the benefits, but only a small fraction of the state's costs, are felt within the district(s). In this case, constituents would internalize all of the marginal benefits, while constituents will take into account only a small fraction of the marginal costs. For the affected legislator(s), the preferred activity would be a higher than efficient output. But not all legislators will propose inefficient activities. Only those legislators from districts whose local economy has concentrated benefits will find the activity to be in their interests. When they do, state politics will generate an intra jurisdictional inefficiency.
States with strong political parties capable of disciplining party members or states with strong governors capable of mobilizing broad-based citizen support may be able to control these intrajurisdictional inefficiencies of universalistic legislatures, but not perfectly. Consider first the ability of strong political parties to control policy. By definition strong parties form a stable legislative majority. Members who might be tempted to deviate from the party's platform can be disciplined so that deviations will not occur. Government policy will therefore be the policy which is in the interests of the stable, majority party. In this stylized framework, only those producers in legislative districts belonging to the majority party will be favored. Minority party districts will receive no benefits.
When proposing an activity for their constituents, representatives in the majority party will be required by the party to consider the effects of their proposed activity on constituents in all districts controlled by the party. From the perspective of the majority party, therefore, an activity in one district will still provide benefits to party constituents, but will now impose only a fraction (perhaps slightly greater than 50%) of the total cost of the activity on party members.
From the majority party's perspective, the preferred activity level in a party-controlled district will be an efficient level, where party marginal benefits equal party marginal costs. Because the party preferred level internalizes the costs to party members, the inefficiency must be less than the inefficiency created by the universalistic legislature. The loss of intrajurisdictional efficiency declines accordingly. Put simply, by internalizing more of the costs of government activity, strong political parties can improve the efficiency performance of state regulations. [For empirical evidence that strong political parties may have controlled legislative inefficiencies, see Inman and Fitts (1990) at 111-115 . Political parties may even promote full economic efficiency. This limiting case occurs if the political party controls all representatives in the legislature. Here there is an incentive for all districts to elect a representative to the stable majority party, since this is their only way to join the controlling coalition and to receive constituent benefits. With a single stable majority party, therefore, there is strong pressure towards full economic efficiency. Donald Wittman makes this argument in "Why Democracies Produce Efficient Result" (1989).]
So too can strong governors. Michael Fitts and Robert Inman present an analysis of executive leadership in economic policy where the executive who has access to needed political resources -- appointment powers, control over appropriated budgets, and an executive veto -- and with the appropriate political incentive -- running in a national or state-wide election -- will seek efficient economic reform. Using appointment powers and discretionary executive authority over policies and the budget, the governor can fashion a reform coalition in which each member district agrees to sacrifice its inefficient policy for a compensating executive-controlled substitute policy. Facing a universalistic legislature, executive-led reform will improve economic efficiency for all state residents. A successful reform package must share the benefits of this reduced inefficiency with residents in the original district and the rest of the state's taxpayers and consumers. On balance, reform makes all residents in the state better off, thereby increasing the governor's chances for re-election. But only strong governors, those with sufficient executive-controlled resources to pay compensation to those districts made worse off, can hope to achieve reform. Even then reform will typically fall short of full efficiency.
We conclude that state politics will often encourage inefficient governmental activities for state residents, and that these intra jurisdictional inefficiencies are likely to be greatest in those states with weak political parties and weak executives. Importantly, these intrajurisdictional inefficiencies occur even when there are no spillovers to citizens outside the state. Matters are even worse, however, when there are spillovers, for here even strong political parties and strong governors will find it politically useful to support the (now nationally) inefficient activity. The case of activities with out-of-state spillovers involves inter jurisdictional inefficiency.
As with the case of intrajurisdictional inefficiency, the marginal benefits continue to reflect the benefits to residents within the state. The marginal costs, however, now encompass the national marginal costs to the residents of the state (state marginal costs plus the marginal costs of lost consumer benefits to out-of-state residents). These national marginal costs are greater than the statewide marginal costs discussed previously because they reflect the additional lost consumer surplus suffered by out-of-state residents. Government policy remains a political decision within the state, however, set by the balancing of state constituents' marginal benefits and costs.
Large interjurisdictional inefficiencies can remain even when strong parties and strong governors successfully control their own state's intrajurisdictional inefficiencies. Even though the inefficient regulation declines with a strong party (or strong governor), significant interjurisdictional inefficiencies may remain. Even a fully efficient state government setting policy efficiently within the state could impose large interjurisdictional inefficiencies on out-of-state citizens; the "state" of OPEC is a prominent example.
Might we imagine a political response from neighboring states to correct these interjurisdictional spillovers? Economists have designed systems of grants (matching and non-matching, specific and general) as well as tax and subsidy arrangements which can greatly increase the effiency of a decentralized system. However, a more distinctive approach, which we pursue in depth here, is the facilitation of interstate agreements, or Coasian bargains, to remove interjurisdictional inefficiencies. The bargain would require the out-of-state residents to "compensate" the producers from the regulating state for the profits they lose when the inefficient regulation is removed. As Coase and others have pointed out, there are sufficient benefits to in-state and out-of-state consumers and taxpayers harmed by the activity that compensation can be paid to the state-created cartel, with the out-of-state residents remaining better off.
In practice, such compensation would be paid through an interstate agreement with neighboring states paying compensation to the regulating state, with the resulting compensation then passed on by that state's governor or legislature to within state producers. But we expect such a Coasian bargain between out-of-state consumers and within-state producers will be hard to effect. Successful Coasian bargains require five assumptions to be met:
1. There are no, or very small, resource costs associated with the bargaining process;
2. Preferences over bargaining outcomes and the resources of households are common knowledge;
3. Bargaining agents perfectly represent the economic interests of their constituents;
4. All bargaining agreements are costlessly enforceable; and
5. The parties can agree to a division of the bargaining surplus.
Only when assumptions 1 through 5 hold will interjurisdictional efficiency be achieved. How likely are these assumptions to be valid?
On its face, assumption 1 seems defensible. The institutions for bargaining -- elected state officials -- are already in place, and the marginal costs of applying these institutions to one additional political agreement are likely to be very small, but only to a point. Legislatures do not solve all problems in every year. The legislative agenda is limited, and at some point it simply does not pay to bring additional policies to the floor for consideration. Those policies which will be excluded are likely to be the marginal problems, those which have only a small effect on the economic well-being of voters.
Even if 1 holds, assumptions 2 to 5 may fail. If the preferences of the participants to the bargain are not common knowledge (2 no longer holds), then there may be a strategic advantage to concealing costs and benefits, demanding greater compensation from the agreement if you represent the "winning" state or offering less compensation if you represent one of the "losing" states. If both parties adopt the "conceal" position, however, no agreement may be forthcoming.
In choosing their bargaining strategy, jurisdictions must assess the extent to which other parties are likely to make concessions. In game theoretic terms, the parties try to assess the threat points of the other parties -- the minimum offer that they are willing to accept in order to accept a negotiating offer. In addition, they try to estimate the extent to which the other party will be willing to make concessions prior to reaching their threat point. If the parties make poor estimates of each other's threat point, or miscalculate the chances that the other party will accept a compromise offer, they may take a hard line in the bargaining process. Clearly, therefore, Coasian bargainers will not be immune to possibilities for strategic behavior when they negotiate, and to the extent that they succumb, Coasian bargaining may break down. [See Polinsky (1980) ; Crawford (1982) ; and Kreps and Wilson (1982) .] Such attempts at free-riding behavior are likely to undo a Coasian bargain when preferences are not common knowledge and the number of states with affected consumers is large. [For the first treatment of the effects of group size on voluntary agreements see Olsen (1965) . The effect of group size and other variables on bargaining has been studied extensively through the use of experiments. See, for example, Hoffman and Spitzer (1982) . See also, Bagnoli and McKee (1991) .]
Even if consumer preferences and industry profits are common knowledge, there is no guarantee that state elected officials will choose to make the constituents of the Coasian bargain better off (3 no longer holds). While the favored industry is well represented in its state (why else would there be regulation?), consumers of the product may not be so fortunate in their states. They may be in the minority in a majority rule state. In contrast to private market agreements, here the affected parties are heard only if generally elected representatives choose to make their case.
Assumption 4 is perhaps the least problematic of the five Coase assumptions. If states do reach an agreement and sign an interstate compact, typically the agreement is legally enforceable. [See Ellickson (1979) . Ellickson discusses the idea of creating a series of public property rights and duties to encourage bargaining solutions to environmental spillovers. Note that monitoring costs, a serious problem in many contracts, are not likely to problem here. States which have agreed to curtail their monopoly overcharge regulation will know if they have been paid compensation. Consumer states which pay compensation will be able to observe if the regulating state has re-introduced regulation. For a discussion of importance of monitoring costs to contract enforcement, see Williamson (1985) .] Enforcement difficulties arise only when important contingencies which might affect the agreement cannot be foreseen in advance. For example, if a future event occurs which alters the benefits and costs of a particular policy, then the state choosing the activity may decide to break the previous interstate agreement.
Arguably the most important and perhaps least appreciated source of bargaining failure is the loss of assumption 5. Even if 1 through 4 hold, the parties may not be able to agree as to how the economic surplus generated by the bargaining process can be divided. Each Coasian bargain does two things: it establishes an efficient exchange thereby creating economic surplus, and it distributes that surplus among the bargaining parties. Proponents of Coase emphasize the first outcome and have largely ignored the important difficulties of the second. Yet it is well known that the division of any economic pie is a bargaining problem which may have no solution. [When redistribution games might have equilibrium outcomes is an important topic of recent economic research; see Rubinstein (1982) and more recently Binmore, Rubinstein and Wolinsky (1986) .] The often observed result in "divide-the-pie" games that bargaining breaks down because the proposed division is not "fair" illustrates the point. In such situations, one state rejects an economically beneficial offer from another state because the proposed offer violates the first state's exogenous, and no doubt politically motivated, sense of economic fairness.
What is seen as an exogenous norm of fairness in a one-time game may be a rational endogenously chosen strategy in a repeated political game. For example, a governor seeking re-election cannot appear "weak" when bargaining with another state. If the money involved in a regulatory agreement is modest relative to the overall state budget -- as it is likely to be -- then the political symbolism of the "share" from the divide-the-pie game may be far more important for re-election prospects than the actual dollars involved. If so, and if both governors must demand more than 50%, then agreement will not be possible. [For evidence from the experimental literature that a sense of fairness is crucial to the ability of parties to reach agreement, see Roth (1985) .] Strategic interplay becomes even more complicated, and agreement less likely, as the number of bargaining states increases beyond two. [This is true even if preferences are fully known and the free-rider problem is not at issue; see Haller (1986) .]
We conclude that economically inefficient policies are likely at the state and local level. Our analysis of state and local political institutions reveals strong political pressures towards favoring narrow, geographically concentrated interests at the expense of more diffused interests. While strong political parties and powerful governors have a political incentive to internalize the costs, even in these states some inefficiencies will remain for residents. [For evidence that state regulations are politically motivated, and in particular, are more likely for geographically concentrated industries and occupations, see Stigler (1971) .] These intrajurisdictional regulatory inefficiencies are compounded when the effects of government activity spill over onto consumers in other states. In these instances, even strong political parties and governors have a political incentive to choose activities. The hope that voluntary interstate compacts might control these interjurisdictional inefficiencies seems to us to be very optimistic. The assumptions required for such Coasian agreements are very demanding and unlikely to hold in practice. This seems particularly so when the efficiency costs on any one individual are small, costs are not common knowledge and are diffused over groups in many states, the affected individuals are themselves a minority within a state, and state politicians treat the division of proceeds from the bargain as a signal of their political "toughness." [For evidence of the difficulties states have in fashioning Coasian agreements, even when the number of states is few, see Kolstad and Wolak (1983) , and Coates and Munger (1995) .] When any one of the assumptions 1 to 5 fail to hold, Coasian agreements will not arise and state created interjurisdictional regulatory inefficiencies will remain.
Within federalist public economies, whenever state and local governments are economically inefficient, it is appropriate to ask whether the central government might do a better job. [See, for example, Oates (1972) particularly at chapters 1 and 2.] But the comparison must be one a consistent one, allowing for potential political inefficiencies at both the state and federal levels of government. Deciding whether to centralize regulatory policy at the national level first requires an analysis of how the national legislature will set such policies.
The effects of state regulatory policies on industries and consumers outlined previously generalizes to central government regulatory policy-making as well. In this instance industries are concentrated in particular congressional districts or states: tobacco in North Carolina, defense production in Washington and California, automobiles in Michigan. Like their counterparts in state government, congressional legislators have an incentive to look for central government activities which will benefit households and firms located in their districts. However, such national activities will have significant costs for U.S. taxpayers, costs which may exceed the benefits.
Whether such economically inefficient activities will be legislated in the central authority depends upon the institutions of legislative decision-making. If the legislature is "universalistic," then we can expect it to favor inefficient activities. [For evidence that U.S. congressional institutions have supported a universalistic legislative process, see Weingast and Marshall (1988) . See also Silberman and Durden's study of the labor market regulations ( Silberman and Durden (1976) ) and Kalt and Zupan's study of environmental regulations (Kalt and Zupan (1984) ).] As with state legislatures, reform of policy-making will require strong presidential leadership and/or political parties capable of controlling the inefficient policy demands of members.
Given that state and central legislatures face the same political pressures to choose inefficient activities, and neither seems, at the moment at least, institutionally well-equipped to resist those pressures, we conclude that neither states nor the central government can be preferred unambiguously as the appropriate domain for choosing economic activities. The current evidence shows economic inefficiencies arising from choices at both levels of government.
All this said, there remains a final argument, a variant of Justice Brandeis' famous plea for "states as laboratories," which we feel tries to tip the balance in favor of decentralizing policy for economic efficiency. Here the laboratory is the state legislature and the experiment is the political ability of a state's governor or a strong political party to control the legislature's excesses. Under our analysis, those states with strong governors or strong political parties will typically have more efficient policies. If so, they become models of "good government" which other states might copy. Recent research by Timothy Besley and Anne Case(1995) suggests that residents do use the economic performance of neighboring states as a "yardstick" against which to measure the performance of their own political process, typically the governor. Governors who fail to control legislative inefficiencies face a significantly lower chance of re-election. Only the decentralization of regulatory policy to the states will give this process of political "yardstick" competition a chance to promote efficiency.
We have concluded to this point that political arguments and the current evidence on participation appear to support the use of small decentralized governments to maximize citizen political participation. However, a more centralized alternative can be supported by economic arguments and evidence concerning the economic efficiency of governmental activity. As a result, trade-offs between the potentially competing values of political participation and economic efficiency may be required when evaluating the optimality of a federalist structure.
How are such choices to be made? The analyses in sections A and B can be used to clarify the potential trade-off between the federalist values of political participation and economic efficiency. Consider how economic efficiency, measured in the dollars of economic output, is likely to change as the public sector becomes more decentralized through the assignment of responsibility to more and more, and thus smaller and smaller, governments. Economic benefits rise as responsibility for policy moves from a single central government to the assignment of responsibility to lower level governments, but only to a point. Efficiency improves because decentralization limits the reach of inefficient government activities; rather than the country as a whole bearing the burden of inefficiency, only residents within a state are so burdened. Further, interstate political competition may point the way to more intra jurisdictionally efficient policies. But as we decentralize, we also increase the likelihood of inter jurisdictional inefficiencies -- that is, spillovers. Moving initially from a centralized structure, the gains from intrajurisdictional efficiencies exceed any losses from interjurisdictional inefficiencies. Eventually, however, inefficient spillovers come to dominate, and the overall performance of the economy declines.
There is a similar relationship between decentralization and the extent of political participation. Consistent with our theoretical and empirical analysis of section II, greater decentralization is likely to increase political participation, though perhaps at a decreasing rate.
Combining our efficiency and participation analysis, we conclude that political participation rises with increased decentralization, but economic efficiency first rises and then falls. It is when efficiency declines that we face a trade-off. Increasing decentralization initially allows efficiency and participation to both increase, but beyond a point more political participation through decentralization must mean less economic efficiency.
The choice of an "optimal" level of decentralization ultimately depends upon the relative importance one places upon the competing federalist values of economic efficiency and political participation; alternative weightings for the values of efficiency and participation will dictate different federalist structures. For example, if efficiency is weighted more highly than participation, the optimal federalist structure will move away from a fully decentralized system so as to increase efficiency and reduce participation. We look forward, with anticipation, to seeing how these federalism issues will be resolved in the emerging constitutions throughout the world.
Arrow, A. Kenneth (1963) , Social Choice and Individual Values .
Bagnoli, Mark and Michael McKee (1991) , "Voluntary Contribution Games: Efficient Provision of Public Goods", 29 Econ. Inquiry, 351-pp.
Baron, David and John Ferejohn (1989) , "Bargaining in Legislatures", 83 Amer. Pol. Sci. Rev., 1182-pp.
Bartik, Timothy (1988) , "The Effects of Environmental Regulation on Business Location in the United States", 19 Growth & Change, 22-pp.
Besley, Timothy and Anne Case (1995) , "Incumbent Behavior: Vote-Seeking, Tax-setting, and Yardstick Competition", 85 Amer. Econ. Rev., 25-pp.
Bergstrom, Theodore, Daniel Rubinfeld, Judith Roberts, and Perry Shapiro (1988) , " A Test for Efficiency in the Supply of Public Education", 35 J. Public Econ., 289-pp.
Binmore, Kenneth, Ariel Rubinstein and Asher Wolinsky (1986) , "The Nash Bargaining Solution in Economic Modelling", 17 Rand J. of Econ., 176-pp .
Boadway, Robin and Frank Flatters (1982) , "Efficiency and Equalization Payments in a Federal System of Government", 15 Canadian J. of Econ. , 613-pp.
Brainard, William C. and F. Trenery Dolbear (1967) , "The Possibility of Oversupply of Local Public Goods", 75 J. Polit. Econ., 86-pp .
Break, George (1972) , Intergovernmental Fiscal Relations in the United States (1967) at 23-24.
Breton, Albert and Anthony Scott (1978) , The Economic Constitution of Federal States .
Buchanan, James M. and Charles Goetz (1972) , "Efficiency Limits of Fiscal Mobility: An Assessment of the Tiebout Model", 1 J. Pub. Econ., 25-pp.
Coase, Ronald (1960) , "The Problem of Social Costs," J. Law and Econ., 1-pp.
Coates, Dennis and Michael Munger (1995) , "Strategizing in Small Group Decision-Making: Host State Identification for Radioactive Waste Disposal Among Eight Southern States", 82 Public Choice, 1-pp.
Cooter, Robert (1982) , "The Cost of Coase" , 11 J. Leg. Studies, 1-pp.
Courant, Paul and Daniel Rubinfeld (1981) , "On the Welfare Effects of Tax Limitation", 16 J. Pub. Econ., 289.
Crawford, Vincent (1982) , "A Theory of Disagreement in Bargaining", 50 Econometrica, 607.
Cumberland, John (1981) , "Efficiency and Equity in Interregional Environmental Management", 2 Review of Regional Studies , 1-pp.
Dahl, Robert (1956) , A Preface to Democratic Theory , (1956).
Dahl, Robert and Edward Tufte (1973) , Size and Democracy .
David, Charles E. and James P. Lester (1989) , "Federalism and Environmental Policy" in James P. Lester, ed., Environmental Politics and Policy: Theories and Evidence, 57, 59, Duke University.
M. de Condorcet (1785) , "Essai sur l'Application de L'Analyse à la Probabilité des Decisions Rendues à la Pluraliste des Voix," Paris.
Easterbrook, Frank (1983) , "Antitrust and the Economics of Federalism", 26 J. L. & Econ., 23.
Ellickson, Robert (1979) , "Public Property Rights: Vicarious Intergovernmental Rights and Liabilities as a Technique for Correcting Intergovernmental Spillovers", in Daniel Rubinfeld (ed.), Essays on the Law and Economics of Local Governments.
Epple, Dennis and Alan Zelenitz (1981) , "The Implications of Competition Among Jurisdictions: Does Tiebout Need Politics?", 89 J. Public Econ ., 1197.
Finifter, Ada W. (1970) , "Dimensions of Political Alienation", 30 American Political Science Review .
Fischel, William (1975) , "Fiscal and Environmental Considerations in the Location of Firms in Suburban Communities" at 119 in Edwin Mills and Wallace Oates, eds., Fiscal Zoning and Land Use Controls .
Fitts, Michael and Robert P. Inman (1992) , Controlling Congress: Presidential Influence in Domestic Fiscal Policy", 80 Georgetown Law Journal, 1737.
Frug, Gerald (1980) , "The City as a Legal Concept", 93 Harvard Law Review, 1059.
Gramlich, Edward M. and Deborah S. Laren (1984) , "Migration and Income Redistribution Responsibilities", 19 J. Human Res., 489.
Gramlich, Edward M. and Daniel L. Rubinfeld (1982) , "Micro Estimates of Public Spending Demand Functions and Tests of the Tiebout and Median Voter Hypotheses", 90 J. Polit. Econ., 536.
Greenwood, Michael J. (1986) , "The Factor Market Consequences of U.S. Immigration", 24 J. Econ. Literature, 1738.
Haller, Hans (1986) , "Noncooperative Bargaining of N 3 Players", 22 Econ. Letters, 11.
Hoffman, Elizabeth and Matthew L. Spitzer (1982) , "The Coase Theorem: Some Experimental Tests", 25 J. Law & Econ., 73.
Holmstrom, Bengt (1985) , "The Provision of Services in a Market Economy" in Robert Inman, ed., Managing the Service Economy: Prospects and Problems, 183.
Inman, Robert (1987) , "Markets, Government, and the New Political Economy," Alan Auerbach and Martin Feldstein (eds.), Handbook of Public Economics , Vol. II.
Inman, Robert P. (1990) , "Public Debts and Fiscal Politics: How to Decide", 80 American Economic Review, 81.
Inman, Robert and Michael Fitts (1990) , "Political Institutions and Fiscal Policy: Evidence from the U.S. Historical Record", 6 (Special Issue) J. Law, Econ. and Org. , 79.
Inman, Robert P. and Daniel L. Rubinfeld (1996) , "Designing Tax Policies in Federalist Economies: An Overview", 60 Journal of Public Economics, 307.
Kalt, Joseph and Mark Zupan (1984) , "Capture and Ideology in the Economic Theory of Politics", 74 Amer. Econ. Rev. , 279.
Kolstad, Charles and Frank Wolak (1983) , "Competition in Interregional Taxation: The Case of Western Coal", 91 J. Pol. Econ., 443.
Krehbiel, Keith (1992) , Information and Legislative Organization , Ann Arbor, The University of Michigan Press, 1992.
Kreps, David and Robert Wilson (1982) , "Reputation and Imperfect Information", J. Econ. Theory , 253.
Ladd, Helen F. and John Yinger (1989) , America's Ailing Cities .
Levmore, Saul (1992) , "Bicameralism: When Are Two Decisions Better than One", 12 Int. Rev. Law & Econ., 145.
Lowi, Theodore J. (1969) , The End of Liberalism .
Martinez-Vasquez, J. (1994) , "The Challenge of Expenditure-Assignment Reform in Russia", 12 Env. and Plan., 277.
McConnell, Kenneth and Robert Schwab (1990) , "The Impact of Environmental Regulation on Industry Location Decisions: The Motor Vehicle Industry", 66 Land Econ., 67.
Mill, John Stuart (1960) , "Representative Government," Three Essays.
Nagel, Jack (1975) , The Descriptive Analysis of Power .
Nagel, Jack (1980) , Participation.
Niou, Emerson and Peter Ordeshook (1985) , "Universalism in Congress", 29 Amer. J. Pol. Sci., 246.
Oates, Wallace E. and Robert M. Schwab (1986) , "Economic Competition Among Jurisdictions: Efficiency Enhancing or Distortion Inducing?", 35 J. Pub. Econ., 333.
Olsen, Mancur (1965) , The Logic of Collective Action.
Ordeshook, Peter and William Riker (1973) , An Introduction to Positive Political Theory .
Pateman, Carole (1970) , Participation and Democratic Theory.
Pauly, Mark (1970) , "Optimality, `Public' Goods, and Local Governments: A General Theoretical Analysis", 78 J. Polit. Econ., 572.
Pestieau, Pierre (1977) , "The Optimality Limits of the Tiebout Model", in Wallace Oates, ed., The Politics of Fiscal Federalism .
Polinsky, A. Mitchell (1980) , "Resolving Nuisance Disputes: The Simple Economics of Injunctive and Damage Remedies", 32 Stanford L. Rev., 1075.
Revesz, Richard (1992) , "Rehabilitating Interstate Competition: Rethinking the 'Race-to-the-Bottom' Rationale for Federal Environmental Regulation", 67 N.Y.U. L. Rev., 1210.
Riker, William (1964) , Federalism: Origins, Operation, Significance , Boston, Little, Brown, and Co..
Rose-Ackerman, Susan (1981) , "Does Federalism Matter? Political Choice in a Federal Republic" , 89 J. Pol. Econ., 152.
Rose-Ackerman, Susan (1994) , "Environmental Policy and Federal Structure: A Comparison of the United States and Germany", 47 Vanderbilt Law Rev., 1587.
Rose-Ackerman, Susan and Jerry Mashaw (1984) , "Federalism and Regulation", in George C. Eads and Michael Fix, eds. The Reagan Regulatory Strategy: An Assessment , 111 (Urban Inst.).
Roth, Alvin (1985) , "Towards a Focal Point Theory of Bargaining", in A. E. Roth, ed. Game Theoretic Models of Bargaining .
Rubinfeld, Daniel L. (1987) , "The Economics of the Local Public Sector", in Alan Auerbach and Martin Feldstein, Handbook of Public Economics, Volume II , (1987).
Rubinstein, Ariel (1982) , "Perfect Equilibrium in a Bargaining Model", 50 Econometrica, 97.
Samuelson, Paul (1954) , "The Pure Theory of Public Expenditure", 36 Review of Economics and Statistics, 387.
Schumpeter, Joseph (1943) , Capitalism, Socialism, and Democracy .
Schwartz, Thomas (1989) , "Representation as Agency and Pork Barrel Paradox", Mimeo, UCLA.
Silberman, Jonathan and Garey Durden (1976) , "Determining Legislative Preferences on Minimum Wage: An Economic Approach", 84 J. Pol. Econ . , 317.
Shepsle, Kenneth (1979) , "Institutional Arrangements and Equilibrium in Multidimensional Voting Models", 28 Amer. J. Pol. Sci., 27.
Smith, Adam (1776) , An Inquiry into the Nature and Causes of the Wealth of Nations.
Stigler, George (1971) , "The Theory of Economic Regulation", 2 Bell J. Econ., 3.
Tiebout, Charles (1956) , "A Pure Theory of Local Expenditures", 64 Journal of Political Economy , 416.
Weingast, Barry (1979) , "A Rational Choice Perspective on Congressional Norms", 23 Amer. J. Pol. Sci., 245.
Weingast, Barry R. and William Marshall (1988) , "The Industrial Organization of Congress" 96 J. Pol. Econ. 132.
Wildasin, David E. (1986) , Urban Public Finance , Chicago, Adademic Publishers, Hartcourt Press.
Williams, Alan (1966) , "The Optimal Provision of Public Goods in a System of Local Governments", 74 J. Polit. Econ. , 18.
Wittman, Donald (1989) , "Why Democracies Produce Efficient Result", 97 J. Pol. Econ., 1395.
Prof. Rubinfeld, Prof. Inman, please complete the following citations:
Arrow, A. Kenneth (1963), Social Choice and Individual Values .
Break, George (1972), Intergovernmental Fiscal Relations in the United States (1967) at 23-24.
Breton, Albert and Anthony Scott (1978), The Economic Constitution of Federal States .
Dahl, Robert and Edward Tufte (1973), Size and Democracy .
Ladd, Helen F. and John Yinger (1989), America's Ailing Cities .
Lowi, Theodore J. (1969), The End of Liberalism .
Nagel, Jack (1975), The Descriptive Analysis of Power .
Nagel, Jack (1980), Participation.
Schumpeter, Joseph (1943), Capitalism, Socialism, and Democracy .
Williamson, The Economic Institutions of Capitalism (1985).