FUNDACION DE ESTUDIOS DE

ECONOMIA APLICADA

THE PRINCIPLES OF FISCAL FEDERALISM* por Timothy J. Goodspeed**

Junio 1991

* Una versión en español de este trabajo figura en la colección "Cuadernos de Economía y Finanzas" de FEDEA. El autor está agradecido a FEDEA por ofrecer una atmósfera de trabajo estimulante y hospitalaria, y a FEDEA y al Instituto de Estudios Fiscales por el apoyo financiero recibido.

** Florida International University.

The Principles of Fiscal Federalism

I. Introduction

II. Stabilization

A. The Orthodox Approach

B. The Stabilizing Effect of Transfers Among Regions

III. Efficiency Issues

A. The Efficiency Benefits of Decentralization

1. Satisfaction of Different Demands for Public Services

2. The Explicit Link Between Benefits and Costs

3. The Decentralized Tax System and Migration Efficiency

4. Political Behavior and Efficiency

5. Competition Among Unrestricted Governments

B. Externalities: The Efficiency Problems with Decentralization

1. Non-Benefit Pricing in the Public Sector

2. Benefit Spillovers

3. The Production Function

4. Competition Among Restricted Governments

5. Grants as a Corrective Mechanism

IV. Redistributional Issues

A. The Appropriate Level of Government for Redistribution

B. The Redistributional Role of Grants

C. The Redistributional Role of Decentralized Income Taxes

D. Redistributional Expenditures

V. A Summary

I. Introduction

Most governmental structures are characterized by more than one level of government. For instance, Spain has a central government which governs the entire country, Autonomías which govern smaller geographic regions, and Municipios which govern even smaller geographic areas. In addition, Spain's entrance into the European Community adds an additional layer of government. Naturally, the simultaneous governance of regions by different governmental units raises questions concerning the appropriate level of government for the various expenditure and tax policies undertaken by the public sector. This paper reviews the insights that economic analysis provides in answering these questions.

Any review of fiscal federalism would be incomplete without reference to the work of Musgrave (1959) and Oates (1972). Musgrave suggested that the functions of government could be broken down into three parts: stabilization, allocation, and distribution. Stabilization refers to macroeconomic monetary and fiscal policies designed to stabilize the economy, allocation refers to the provision of public goods that are not efficiently produced by the private sector, and distribution refers to an alteration in the distribution of income dictated by the private market. Oates argues that the stabilization and distribution functions should be assigned to the federal level of government, but that there is some scope for the allocative function at the state and local levels.

Stabilization is assigned to the federal level since this level can rely on both monetary and fiscal policy while local governments can only rely on fiscal policy. Moreover, the openness of local governments will tend to diminish the force of the Keynesian multiplier.

Oates also argues that the federal level of government should be given primary responsibility with respect to distributive issues. He argues that a policy that attempts to redistribute income from rich to poor within a given region will result in the immigration of poor from other regions and the emigration of the rich from the region in question; this migration tends to negate the redistributive goals of the regional government. This argument relies primarily on the presumption that the mobility of residents at the local level will tend to thwart redistributive policies of lower levels of government.

Finally, it is argued that the allocative function should be shared; the federal government should provide those goods that are characterized by large economies of scale in population while local governments should provide goods that are are "congestible" so that lower per-person costs will not be realized as the population grows. The argument for local provision of these latter services is that local governments can more closely match the tastes of residents.

Many of these early ideas are still valid, though some have been modified and many new ideas have also become known. The remainder of this paper will explain more fully the rationale behind the assignment of different expenditure and tax functions and will develop the more recent ideas as well as explaining the classic ones.

II. Stabilization

A. The Orthodox Approach

As mentioned above, the orthodox position is that stabilization should be assigned to the federal level of government because that level has control over both monetary and fiscal policy. Moreover, the mobility of goods and factors of production will diminish any Keynesian multiplier effect. That is, the stimulation of demand in one area would lead to an immediate movement of goods to that area from others (i.e. imports will increase) to satisfy the increased demand; but this results in an outflow of financial assets (in response to the deficit in the balance of trade), so that income will not increase much. If goods are slow to move, labor will migrate in response to increases in labor demand so that wages will not rise much.

While macroeconomics has undergone radical change over the past two decades, the fiscal federalism literature has been virtually silent on stabilization policy. This has lead Gramlich (1985) to suggest that some of the orthodox stabilization arguments are outdated and may need to be rethought. While Gramlich holds Keynesian views and does not reject the orthodox view on its face, he presents evidence that labor is not very mobile across U.S. states. This leads him to suggest that states should in fact be undertaking more responsibility for stabilization policy.

B. The Stabilizing Effect of Transfers Among Regions

In addition to Gramlich's criticism, an argument has recently emerged from discussions of the developments in the European Community. This argument, suggested in papers by Eichengreen (1990) and Sachs and Sala-i-Martin (1991), is concerned with the extent to which transfers by the central government from one region to another are necessary for a common currency area. While this literature is not directed at fiscal federalism per se, it has some interesting implications for the conduct of stabilization policy in a federal system of government.

This literature suggests that intergovernmental transfers are a form of insurance. Such transfers will mitigate depressions or booms in any one area of the common currency area; since individuals are risk-averse, everyone will gain from such insurance. Moreover, it is argued that such transfers are necessary for the survival of a common currency area. The implications for the EC are clear: the viability of a common currency in Europe may depend on transfers that mitigate large income fluctuations in any one country.

The fiscal federalism literature has rarely considered this insurance argument. Yet, it has some interesting implications for stabilization policy. First, it supports the view that stabilization policies should be carried out by the central level of government. Insurance is provided to guard against unpredictable fluctuations in income across regions. If fluctuations in income across regions are random and labor is immobile at least in the short run, then all will benefit from provision of the insurance, but a region will be unable to provide the insurance because its revenues will fall in conjunction with the incomes of its residents; therefore, the federal government should provide the insurance.

A second implication of the insurance argument is that it provides an additional rationale for lump-sum grants to relatively poor regions, or at least to regions that are currently depressed. As will be discussed later, the justification for such grants has traditionally been concerned with equity; the insurance argument suggests an efficiency rationale since everyone will be made better off by provision of the insurance.

III. Allocative Issues

The decentralization of certain public services provides benefits to the economy because resources may be allocated more efficiently, and so more and better services can be produced by middle and lower level governments for the same cost as would be incurred had those services been centrally provided. However, decentralization also may present certain problems in the efficient allocation of resources. We first consider the beneficial effects of decentralization and then discuss certain efficiency problems that may arise.

A. The Efficiency Benefits of Decentralization

1. Satisfaction of Different Demands for Public Services

Probably the most widely discussed benefit of decentralization is that it allows different regions to tailor the services that are provided to the needs of the community. For instance, a large urban area may need a very different transportation system than a smaller rural area. A central government may be unaware of the different needs of the areas, or may have a very crude mechanism to allocate funds for transportation. Since local governments are much closer to their citizens, they may be better able to discern these different needs and provide the appropriate type of transportation system. In addition, the citizens of a particular region may be more willing to pay for a transportation system that is more closely tailored to their needs. Another example might be education. A community that contains many children per household may be willing to spend much more on education than a community with few children per household.

Moreover, diversity in demand for services is not limited to the physical characteristics of a particular community. Just as the demand for products in the private market depends on prices, income, and tastes, so too does the demand for public services. Thus, in a society that is diverse in tastes or income, it can be expected that certain groups will have a higher demand for certain services, will value these services more than other groups, and will therefore be willing to spend more on such services.

Just as efficiency in the private market is characterized by the satisfaction of diverse demands, resources in the public sector will be allocated more efficiently if diverse demands can be satisfied. Suppose, for instance, that a consumer in the private market wishes to buy blue shirts while his friend wants to buy white shirts. Clearly, the two consumers will be happiest if the first consumer is allowed to buy blue shirts while the second buys white shirts. If we require both consumers to buy white shirts, the consumer who wants blue shirts will be unhappy. If the public sector produces shirts, the decentralized solution in which one region provides blue shirts and another white shirts will better satisfy consumers than the centralized solution in which only white shirts are provided to both regions.

For the public sector to satisfy diverse demands for services, two components are necessary. The first is decentralized services, which allows a variety of options from which consumers can choose. The second ingredient is the ability to freely choose a region in which to live. Thus, a consumer who prefers a large amount of public services and is willing to pay for these services, but who happens to live in a region that provides very little in the way of public services, can move to a region that better satisfies his preferences. The ability of consumers to choose the region which provides the level of services most closely tailored to their needs has been termed "voting with the feet" and is associated with the work of Tiebout (1956).

2. The Explicit Link Between Benefits and Costs

The mechanism of voting with the feet suggested by Tiebout implies that decentralization coupled with the ability to freely choose among communities provides a way for the nice efficiency properties of the private market to be duplicated by the public sector. However, for the Tiebout mechanism to work perfectly, several conditions need to be met. The first requirement for decentralization to be perfectly efficient is that public services are priced properly.

The proper pricing of public services, whether provided locally or nationally, begins with the basic proposition that an individual's tax payments and the services provided by the government be as closely linked as possible, with the individual's tax payment corresponding to the share of public benefits that he receives, and that this link be as clear as possible to the citizens. By closely linking the benefits that an individual receives from a public service to the cost of providing the service, the citizen will be best able to judge whether or not he would like to pay more in taxes to finance additional services.

This link between benefits and costs of government programs has an important implication for the financing of expenditures provided by lower levels of government. In particular, it suggests that lower levels of government should finance their spending from their own revenue sources rather than from outside revenue sources. Without such own-financing, the link between benefits and costs is broken. For instance, if a local government relies on the central government for financing, it will ask for as much money as possible and it will spend all the money it receives since its citizens bear only a small share of the costs of the program. The citizens have no incentive to ask their elected representatives to control spending since there is no clear link between tax payments and local service provision. Thus, a first principle for financing of lower levels of government is that, to the extent possible, lower levels of government should rely on their own revenue sources.

3. The Decentralized Tax System and Migration Efficiency

A second question concerning the proper pricing of decentralized public service relates to the type of tax system that is most appropriate for decentralized governments. That is, given that lower levels of government finance their own expenditures, what type of tax system is most appropriate? To answer this question, one must first understand something about the production function for public goods.

A "pure" public good in the sense of Samuelson (1954) is characterized by declining per-person costs of production as the population that consumes the good rises. For instance, a missile may provide a given level of protection to 10, 100, or 1000 people for a given total cost. Thus, the per-person cost of protection is much cheaper for 1000 people than for 10. Such a good can be most efficiently provided by the central level of government since this level of government serves the most people and thus provides the lowest per-person cost.

Other goods that are provided by the government are subject to "congestion." This term is used to describe goods whose total cost of production rises with the number of consumers. For instance, a small public park may be able to serve 10 people at the same total cost, but beyond this some crowding or congestion takes place so that the park may have to be twice as large to accommodate 20 people.

In fact, many empirical studies of goods that are provided by local governments in the United States suggest that the production function of these goods exhibits constant marginal congestion costs. That is, if it costs $1 to provide the good to 1 person, it will cost $2 to provide the good to 2 people, $3 to provide the good to three people, and so on. Such goods provided by the government are referred to as "publicly provided private goods." Since there is no cost advantage to the provision of such goods by the central government, and diverse tastes can be satisfied if provided by local governments, these goods are more efficiently provided by lower level governments.

What is the most efficient tax system that can be used as a pricing mechanism for these local public goods? Traditionally, efficiency in a Tiebout model has been concerned with migration. For migration efficiency, each person should pay the additional cost of providing the service to him. Since each person can freely move to a community that provides the level of public services that he is willing to pay for, mobility insures that each person is satisfied with his tax payment and the level of services that are provided, and the tax system insures that each person pays the additional cost of providing service to him.

To illustrate, suppose that it costs a jurisdiction $100 per person to produce one unit of public service and that each of two jurisdictions have 10,000 people. If jurisdiction 1 provides 1 unit of the public service and jurisdiction 2 provides 2 units, it will cost $1 million to produce the public service in jurisdiction 1 and $2 million in jurisdiction 2. A tax system that charges each person the cost of providing the public service to him will charge each person $100 per unit of public service.

To see why this is efficient, remember that we have assumed the additional cost of public service to be $100 per-unit per-person, and consider an individual in jurisdiction 1 who decides that he wants to consume more public services and so contemplates a move to jurisdiction 2. His move will impose costs of $200 on that jurisdiction since jurisdiction 2 must provide 2 additional units of the public service. But notice that his tax payments will be just equal to the additional costs he imposes. He will only move if he is willing to pay the additional cost of consuming additional public services.

The first studies were done by Bergstrom and Goodman (1972) and Borcherding and Deacon (1971). See Inman (1979) for a review of studies thereafter.

One might note that the tax system described above is the infamous "poll tax" that helped bring down the Thatcher government. Thus, it seems to be an unpopular tax both with citizens and with politicians. We will consider the effect of other tax systems later in this paper.

4. Political Behavior and Efficiency

Efficiency depends not only on an efficient migration pattern, but also on the provision of an efficient level of public services within a region, which in turn depends on the voting procedure and the internal allocation methods of government. While the actual workings of a government can become quite complex, economists have suggested some simple models that try to capture the important political aspects.

One model of political behavior that is popular among economists is called the "median voter model." This model suggests that politicians will act to satisfy the preferences of the median voter. The idea behind this is that the politician will be able to obtain the 50 percent of the vote necessary for election by satisfying the preferences of the median voter. Although the model depends on many assumptions, it has an important implication: democracy works. That is, politicians will submit to the will of the people (or of the average person) in making decisions.

What does the median voter model imply about the economic efficiency of political decisions? Political decisions will not necessarily be economically efficient under majority rule, and will depend on the tax-price faced by the decisive voter. One interesting result that has emerged is that, if preferences are symmetric, then the satisfaction of the median voter's preferences produces a level of public services within a jurisdiction that is economically efficient.

See Bergstrom (1979).

5. Competition Among Unrestricted Governments

An additional issue that has arisen recently is the effect of competition among governments at lower levels. Since competition among firms in the private market is normally thought to improve the allocation of resources, so too competition in the public sector among lower levels of government should lead to an efficient solution.

Two issues have arisen recently in this regard. The first is whether competition will lead benevolent governments to choose an efficient tax system and provide public goods efficiently. The second concerns the control of the growth of government.

The argument that competition among lower levels of government will lead to efficient tax-expenditure decisions has been suggested recently by Oates and Schwab (1988). They argue that competition will lead governments to impose taxes that just reflect the benefits of the services that are provided. If taxes are set too low, the desired level of services cannot be funded; if taxes are set too high, resources will leave the jurisdiction. Therefore, taxes will be chosen that reflect the benefits of public service provision.

The result that competition leads to an efficient outcome depends to some extent on the behavior of politicians. The Oates/Schwab argument that competition leads to efficient tax and expenditure decisions depends on an efficient voting model.

A quite different argument concerning the benefits of competition depends on a model of political behavior that suggests that government is a "Leviathan" that has growth rather than the preferences of the electorate as its objective. This Leviathan model of government behavior suggests that governments are inefficient overspenders, and is associated with the work of Brennan and Buchanan (1980).

In this context, it is argued that competition will limit the inefficient growth of government. The empirical evidence on the Leviathan theory is somewhat mixed and does not directly address efficiency questions; rather, the empirical evidence is used as a test of the different theories of political behavior.

B. Externalities: The Efficiency Problems with Decentralization

The previous section has argued that decentralization of certain services will result in a better allocation of resources in the economy. It was also noted that certain restrictions are necessary for the efficiency properties of a decentralized public sector to work perfectly. In this section, we explore the problems that may arise from decentralization.

These problems all stem from what economists term "externalities." Externalities arise when one person or firm's actions affect another without the effect being reflected in a market price. For instance, a firm may lower its costs by producing in a way that creates a lot of air pollution. While the air pollution is a real cost to society, the firm does not have to pay for this cost in the private market. The private market thus produces too much pollution.

1. Non-Benefit Pricing in the Public Sector

One externality can arise from a tax system that does not conform to the efficient "poll tax" system mentioned in the previous section. Suppose, for instance, that an income tax system is used. The income tax provides an incentive for individuals to move to a wealthier jurisdiction. To see why, suppose that it costs a jurisdiction $100 per person to produce public services. If each of two jurisdictions have 10,000 people, it will cost $1 million to produce the public service in each jurisdiction. Suppose one jurisdiction has total income of $10 million while the other has total income of $100 million. To pay for the public service, the first jurisdiction must impose a 10% tax while the second must impose a 1% tax. An individual in the poorer jurisdiction will therefore have an incentive to move to the richer jurisdiction since he can consume the same level of public service but pay only a 1% tax.

To see why this is inefficient, consider again a production function that exhibits constant marginal congestion costs. Since we have assumed the additional cost to be $100 per person, the individual that moves to the richer jurisdiction will impose costs of $100 on that jurisdiction. Unless his income happens to be $10,000 (the average income of the richer community), he will impose costs that differ from his tax payments. Since the average income of the poorer jurisdiction is only \$1000, it is likely that his tax payments will fall far short of the additional costs he imposes, and the richer community will have to make up the difference in higher taxes.

While this is clearly a problem in theory, evidence in Goodspeed (1989) suggests that it may be much less of a problem in practice. While the income tax provides some incentive to move to the richer community, housing prices are likely to be higher in that community as well. Thus the individual must weigh the benefit from his tax gain against the increased housing costs he will face. Since housing is such a large part of an individual's budget, he is much less likely to move, and the efficiency problem associated with the use of income taxes by lower levels of government is likely to be quite small.

2. Benefit Spillovers

One can think of a public service as having benefits over a given geographic area or number of people. A pure private good benefits one person while a pure public good benefits everyone. Trash collection by one truck benefits the area that can be covered by the truck during a given time period.

Olson (1969) has suggested that the optimal level of government to provide a good is that level that exactly matches the geographic area of benefits. If the benefits from a public good provided in one jurisdiction "spill over" into another jurisdiction, an externality is created because the government does not take into account the benefit provided to people who are outside of its area of influence.

For instance, one might argue that the benefits of education extend to an entire country. To the extent that lower levels of government provide public services that benefit larger geographic areas, public services will be underprovided in a decentralized public sector since the benefits that accrue outside the region will not be taken into account when decisions concerning the level of service are made.

3. The Production Function

The production function provides a third area in which externalities may be present so that a decentralized public sector provides inefficient levels of public services. As we have mentioned, to the extent that public goods enjoy cost savings as population increases, decentralization becomes less desirable. Thus, the shape of the cost function with respect to population is important.

We have mentioned that many public services provided at the local level in the U.S. have been found to be characterized by constant marginal congestion costs. Presumably, production is governed more by physical laws, and is unlikely to change drastically from country to country. Still, to the extent that the production process is different, country differences will be important. In addition, previous work in this area has come under some scrutiny of late, but recent evidence has not been able to reject the basic findings of the previous work.

An important aspect of the production function for education has been discussed in the United States. The idea is that a child's education depends on the characteristics of his peers. That is, it has been suggested that, given all other inputs into education, a child will obtain a better education the more educated are his peers. Thus, a child from a poorly educated family will obtain a better education if his schoolmates come from highly educated rather than poorly educated families. Conversely, a child that comes from a highly educated family will be worse off if his peers are poorly educated.

Since the decentralized solution would most likely result in relatively homogeneous schools, this aspect of the production function, if true, has profound and difficult implications for society. It suggests that a mixing of children from poorly and highly educated families will make the children from poorly educated families better off and those from highly educated families worse off. Society will be better off with this mixing if it values the gain to children coming from poorly educated families more than the loss of the children coming from highly educated families.

See Oates (1988) and Edwards (1990).

See Bradford, Malt, and Oates (1969), de Bartelome (1990), and Oates and Schwab (1991).

4. Competition Among Restricted Governments

We discussed above the argument that competition among governments will lead to a more efficient allocation of resources. There are certain exceptions to this efficiency argument, however.

First, if there are restrictions on the types of taxes that governments can impose, an efficient outcome may not be possible. For instance, if only taxes on mobile resources are available, a government may impose inefficiently low taxes so that the mobile resource does not flee. The consequence of low taxes is, of course, low service levels.

Second, just as in the private market, the presence of market power by a single jurisdiction can thwart the efficiency properties of decentralization. That is, monopolistic or oligopolistic behavior by a jurisdiction will normally result in inefficient output levels.

5. Grants as a Corrective Mechanism

We have noted several problems that may result from decentralized decision-making. Some, such as the externality caused by local income taxes, are not likely to result in large efficiency losses. For those externalities resulting from decentralized decision-making that are perceived to cause problems, a literature on grants from the central to lower levels of government has been developed. This literature suggests that "matching grants" can be designed to correct for the inefficiency.

To see how a matching grant would work, suppose that education is decentralized and provides benefits to society as a whole that are not taken into account by lower levels of government. The central government can design a grant to encourage spending on education. For each dollar of local government spending on education, the grant would designate additional money from the central government that could be spent on education. This is called a "matching" grant as opposed to a "lump-sum" grant because the central government matches funds that are allocated for education by the local government.

See Wildasin (1989) and Wilson (1987), for instance.

See Mintz and Tulkens (1986).

Thus, matching grants can be used to correct for any externality-related inefficiency resulting from decentralization. A matching grant is preferred to a lump-sum grant for this purpose because the matching grant has both income and price effects and therefore encourages more spending by the local government for a given grant level.

IV. Redistributional Issues

A. The Appropriate Level of Government for Redistribution

As mentioned previously, the orthodox position is that redistributional activities should be undertaken by the central level of government. Two arguments are used to support this. One is that redistribution from rich to poor on the local level will result in an outflow of the rich and inflow of the poor that will largely nullify the desired redistributional change. A second is that redistribution provides benefits to society as a whole; thus, spillover benefits will result in inefficiently low levels of decentralized redistribution.

This view has not gone unchallenged, however. In particular, Pauly (1973) suggests that redistribution can be viewed as a local public good for which different regions have different demands. He suggests that people care more about the poor within their own region than the poor outside their region. Consequently, the same efficiency argument suggested previously, that decentralization allows the satisfaction of diverse demands, can be applied to redistribution.

In addition, Tresch (1981) suggests that the central and lower levels of government may have different redistributional goals. He argues that the appropriate level for redistribution depends on the relative superiority given to the redistributional preferences of the central versus lower levels of government.

B. The Redistributional Role of Grants

In addition to the role of grants as a corrector of externalities, grants from the federal government to lower levels have also been discussed as a redistributional tool. The discussion of this issue has taken place in the U.S. judicial system as well as in economic journals. Such grants are often termed "equalizing grants" because the stated purpose is to equalize tax bases across jurisdictions.

See Boadway and Wildasin (1984) and Brown and Oates (1987), for instance.

There may be an economic justification for these grants if, for instance, poor communities face higher costs of providing public services. For example, in an earlier section, the effect of peers on the level of education a child obtains was discussed; these peer effects may lead to higher costs in poorer jurisdictions to obtain a given level of education. If equal access to a good education is a societal goal, grants that equalize the cost per pupil may be justified. This efficiency rationale for redistributional equalizing grants has recently been explored in Oates and Schwab (1991).

However, in some cases, grants that redistribute income from rich to poor regions can have a perverse effect. Suppose, for instance, that the goal is to redistribute from rich to poor people. Further suppose that some rich individuals live in a poor region and vice-versa; while redistributive grants for this case would be progressive on average, such grants would actually be taking some income from the poor who happen to live in a rich region and redistributing this to the rich that live in the poor region.

Recent work by Wildasin (1990) suggests a redistributional role for matching grants. Wildasin suggests that redistribution by one jurisdiction will affect other jurisdictions because migration of poor laborers to the redistributing jurisdiction will depress wages there and increase wages in the region from which emigration takes place. This externality leads to too little redistribution, and matching grants should be used to encourage redistribution by local governments.

C. The Redistributional Role of Income Taxes

The orthodox prescription for redistribution to be a function of the central government suggests that taxes based on ability-to-pay, such as an income tax, should be used primarily by higher levels of government. Some doubt is cast on the orthodox efficiency argument because it has not held up to recent empirical scrutiny; recent work by Goodspeed (1989) suggests that the efficiency loss that would result from migration induced by local ability to pay taxation is likely to be quite small for the United States. This result implies that local governments in the United States could use income taxes if they so desired without large efficiency losses, and it should be feasible (from an efficiency standpoint) for other countries to use ability to pay taxes at the local level for purposes of fiscal redistribution.

In addition, certain OECD countries rely primarily on ability to pay taxes at the state and local level. For instance, according to the OECD (1989), nearly 100 percent of the state and local tax revenues of Sweden and Finland were in the form of income and profits taxes in 1987. Other OECD countries follow the orthodox prescription. For instance, Australia, Greece, Ireland, the Netherlands, New Zealand, and the U.K. all had zero revenue from income taxes at the state and local levels in 1987.

The apparent inability of the traditional arguments to explain the observed pattern of local income taxes in the OECD and the recent empirical evidence that contradicts the orthodox efficiency argument led Goodspeed (1991) to suggest that countries are able to choose from a range of redistributive mechanisms. More specifically, he found evidence to support his hypothesis that countries will differ in the extent to which income taxes are used at lower levels of government depending on the relative importance of redistribution within and between regions, the demand for redistribution in a country, and the cost of redistribution.

D. Redistributional Expenditures

Of course, in addition to redistribution through the tax system, redistribution can occur through expenditures that are targeted to the poor. The orthodox argument, recently codified by Brown and Oates (1987), suggests that migration induced by such expenditures can cause increased costs of redistribution for lower levels of government. This represents an externality that will result in lower transfer payments if redistributional expenditures are decentralized. The conclusion again follows that the central government should take responsibility for redistributional expenditure programs.

As with redistributional taxes, the case against decentralized redistributional expenditure policies depends on the mobility of residents or on the possible consequences that are perceived by local public officials. Empirically, this is a very difficult matter to sort out because it is difficult to determine the degree to which observed welfare payments that are decentralized would differ from a hypothetical centralized solution.

Of the evidence that exists, Brown and Oates suggest that mobility is strong enough to support centralization of redistributional expenditures. However, Gramlich (1985) notes that migration between U.S. states is much less than within states, and suggests that states should perhaps be doing more in the way of redistribution.

As a final note, it should be added that, along with much of the public finance literature, we have discussed tax and expenditure policies separately. However, it is important to note that it is the combination of tax and expenditure policies that is of importance, and the empirical literature on the effect of the combined tax-expenditure package is lacking. The reason is that it is quite difficult to assign the benefits received by individuals from many public expenditures. Yet, to properly assess both allocative and redistributive aspects of the impact of decentralization, the effect of both the expenditure and tax system should be simultaneously considered.

V. A Summary

A federal system of government in which tax and expenditure responsibilities are shared by different levels of government can provide an efficient and equitable framework for government. To conclude, we will review the basic propositions suggested by the literature on fiscal federalism.

The primary benefit of decentralized government is that, for many public services, it provides a more efficient system of government than a centralized system. More specifically, decentralization provides a link between the benefits and costs of public services. This link allows citizens to evaluate the relative costs of different programs, and demand more or less government services by either voting or moving to a community that provides better services. Moreover, the ability of lower level governments to determine their own tax-expenditure package allows different preferences to be satisfied. People with a strong desire to consume public services can locate in jurisdictions that provide a lot of services if they are willing to pay higher taxes for such services. By providing competition between communities, decentralization may also lead to more efficient service provision and competition for votes may lead politicians to be more responsive to citizen demands.

The design of a decentralized system should not constrain the types of taxes that can be used by different levels of government, except by providing a framework in which beneficiaries of public services (i.e. residents) are taxed. Lower level governments will then have an incentive to choose taxes that are best suited to their needs. Some economies in the administration of lower level taxes may be realized by a central collection agency, although local authorities, by being "closer to the people," may be better able to detect tax evasion.

Some efficiency problems may arise in a decentralized system because of externalities. In some cases, such as the use of income taxes by lower levels of government, these externalities are not likely to be large so that the national government need do nothing. In other cases, matching grants should be used to correct for externalities; for instance, such grants could be used to encourage services such as education that provide positive spillover benefits. Of course, matching grants will be effective only if lower levels of government raise their own revenue and hence have some expenditures to match.

Some equity problems may arise in a decentralized system of government so that redistributive policies should be undertaken primarily through the national tax and welfare system. However, to the extent that the distribution of income within a region is unequal, there is some scope for redistribution by lower level governments. For instance, the use of income taxes by lower levels of government may be appropriate for redistributive reasons if the distribution of income within the region is unequal. Equalizing grants may be a component of redistributive policy. However, such grants should be used primarily to guarantee certain societal norms such as equal access to educational services, and not as a primary tool for equalizing incomes.

Finally, specific purpose lump-sum grants should be used to establish a floor for services such as education and health that are deemed to be necessary for basic well-being. The lump-sum grants may have an equalizing component for services that are more expensive to provide in poorer or more urban areas. For efficiency purposes, it is desirable for the floor to be low enough so that own-source revenue is used at the margin; this will establish the important link between benefits and costs of services. However, the actual level of the floor depends on societal values.

Thus, a decentralized system of government provides many advantages in terms of efficiency. Some externalities may also be generated; to the extent that these are large in magnitude, the central government can use matching grants to correct the resulting misallocation of resources. The equity properties of a decentralized system are harder to evaluate. While a good case can be made for centralized redistributive policies, there is some scope for lower level governments to engage in redistributive policies as well.

References

1. Bergstrom, Theodore. 1979. "When Does Majority Rule Supply Public Goods Efficiently?" Scandinavian Journal of Economics. 81, 215-226.

References

1. Bergstrom, Theodore and R.P. Goodman. 1973. "Private Demands for Public Goods." American Economic Review. 63, 280-296.

References

1. Broadway, Robin and David Wildasin. 1984. Public Sector Economics, Second Edition. Boston, Mass.: Little, Brown and Company.

References

1. Borcherding T. E. and R. P. Deacon. 1972. "The Demand for the Services of Non-Federal Governments." American Economic Review. 62, 891-901.

References

1. Bradford, David, R. A. Malt, and W. E. Oates. 1969. "The Rising Cost of Local Public Services: Some Evidence and Reflections." National Tax Journal. 22, 185-202.

References

1. Brennan G. and J. M. Buchanan. 1980. The Power to Tax: Analytical Foundations of a Fiscal Constitution. Cambridge: Cambridge University Press.

References

1. Brown, Charles and Wallace E. Oates. 1987. "Assistance to the Poor in a Federal System." Journal of Public Economics. 32: 307 - 330.

References

1. de Bartelome, Charles. 1990. "Equilibrium and Inefficiency in a Community Model with Peer Group Effects." Journal of Political Economy. 98, 110-133.

References

1. Edwards, John H. Y. 1990. "Congestion Function Specification and the Publicness of Local Public Goods." Journal of Urban Economics. 27: 80-96.

References

1. Eichengreen, Barry. 1990. "One Money for Europe? Lessons from the U.S. Currency Union." Economic Policy. 10, 118-187.

References

1. Goodspeed, Timothy J. 1989. "A Re-Examination of the Use of Ability to Pay Taxes by Local Governments." Journal of Public Economics. 38: 319-342.

References

1. Goodspeed, Timothy J. 1991. "The Assignment of Ability to Pay Taxes in the OECD Countries." in R. Prud'homme, ed., Public Finance with Several Levels of Government.

References

1. Gramlich, Edward M. 1985. "Reforming U.S. Federal Fiscal Arrangements." in John Quigley and Daniel Rubinfeld, eds., American Domestic Priorities: An Economic Appraisal. Berkeley, CA: University of California Press.

References

1. Inman, R. P. 1979. "The Fiscal Performance of Local Governments: An Interpretive Review." in P. Mieszkowski and M. Straszheim, eds. Current Issues in Urban Economics. Baltimore: Johns Hopkins University Press.

References

1. Mintz J. and H. Tulkens. 1986. "Commodity Tax Competition Between Member States of a Federation: Equilibrium and Efficiency." Journal of Public Economics. 29.

References

1. Musgrave, Richard. 1959. The Theory of Public Finance. New York: McGraw-Hill.

References

1. Oates, Wallace E. 1972. Fiscal Federalism. Princeton: Harcourt Brace Jovanovich, Inc.

References

1. Oates, Wallace E. 1988. "On the Measurement of Congestion in the Provision of Local Public Goods." Journal of Urban Economics. 24: 85-94.

References

1. Oates, Wallace E. and Robert M. Schwab. 1988. "The Allocative and Distributive Implications of Local Fiscal Competition." University of Maryland Working Paper, June.

References

1. Oates W. E. and R. M. Schwab. 1991. "Community Composition and the Provision of Local Public Goods: A Normative Analysis." Journal of Public Economics. 44, 217-237.

References

1. OECD. 1989. Revenue Statistics of OECD Member Countries, 1965-1988. Paris: OECD.

References

1. Olson, Mancur. 1969. "The Principle of 'Fiscal Equivalence': The Division of Responsibilities Among Different Levels of Government." American Economic Review. May, 479-487.

References

1. Pauly, Mark V. 1973. "Income Redistribution as a Local Public Good." Journal of Public Economics. 2: 35 - 58.

References

1. Sachs, Jeffrey and Xavier Sala-i-Martin. 1991. "Fiscal Federalism and Optimum Currency Areas." Unpublished paper.

References

1. Samuelson, Paul. 1954. "The Pure Theory of Public Expenditures." Review of Economics and Statistics. 36, 386-389.

References

1. Tiebout, Charles. 1956. "A Pure Theory of Local Government Expenditures." Journal of Political Economy. 64, 416-424.

References

1. Tresch, R. W. 1981. Public Finance: A Normative Theory. Plano, Texas: Business Publications.

References

1. Wildasin, David. 1989. "Interjurisdictional Capital Mobility: Fiscal Externality and a Corrective Subsidy." Journal of Urban Economics. 25: 193-212.

References

1. Wildasin, David. 1990. "Income Redistribution in a Common Labor Market." Indiana University Working Papers in Economics Number 90-023.

References

1. Wilson, John D. 1987. "Trade, Capital Mobility, and Tax Competition." Journal of Political Economy. 95: 835-856.