2 Assigning Expenditure Responsibilities

Teresa Ter-Minassian
Published Date:
September 1997
  • ShareShare
Show Summary Details
Ahmad Ehtisham, Hewitt Daniel and Ruggiero Edgardo 

Economic theory offers limited guidance for assigning expenditure responsibilities among different levels of government. At the micro level, the benefit principle suggests that a given service should be provided by the level of government that most closely represents the region that benefits from such service. For example, local public goods such as municipal services often only benefit persons who live in a given city or township; in these cases, decentralization to the municipal level is feasible and desirable for allocative and administrative efficiency. In contrast, national public goods, such as national defense, macroeconomic stability, and redistribution, require centralized administration and centralized policy.

Most goods that governments supply, however, do not fit neatly into either one of these two categories. For these mixed goods some degree of decentralization coupled with some centralized coordination of policy is both feasible and desirable, owing to unclear benefit regions, externalities, or national redistributional implications. For instance, there are certain efficiency advantages to local supply of primary education and preventive health care, such as possibly better quality through local supervision, and allowance for communities to express cultural and curative preferences. For tertiary education and hospitals, existence of economies of scale and externalities (their benefits accruing to more than one jurisdiction) imply that more centralized control may be warranted. However, the demand for minimum standards often requires that centralized decision making of policies be ensured for all these services. When conflicting goals arise, the analysis of which level should be assigned expenditure responsibility can become quite complicated and subject to normative bias.

The analysis of expenditure assignments is complicated by the fact that centralization-decentralization is a multidimensional choice. Expenditure assignments involve decisions on which level of government should be assigned the formulation, financing, and administration of policies. A number of combinations are possible. Intergovernmental grants allow policymakers to separate expenditure decisions, administration of programs, and their financing. Revenue sharing and grants serve as the coordinating mechanism that allows decentralized expenditure to coexist with centralized tax collection and with redistribution. Since the distributional implications of expenditure decentralization are often undesirable, tensions between expenditure assignments and financing arrangements often arise.

If economic theory provides little guidance on expenditure assignment—apart from purely local and national public goods—the examination of actual practices in this chapter confirms that a wide array of different solutions is possible. There is a great variety of intergovernmental fiscal arrangements and expenditure assignments among different countries, reflecting varying social preferences among the countries involved. Overlapping responsibilities in policy formulation, financing, and administration of services are also common. In general, industrial countries are more decentralized than developing countries, though the former tend to have centralized revenue collection to ensure efficient tax administration (Tanzi, 1994) and to finance redistribution from richer to poorer regions. Not only do intergovernmental fiscal arrangements vary widely across countries, often a specific country’s degree of decentralization changes over time. Among the more highly centralized industrial countries—France, Italy, and Spain—significant tendencies toward decentralization are now more evident.

At the macro level, an interesting question arising from expenditure assignments is the interaction between the level of decentralization and macroeconomic stability. This chapter examines the hypothesis that since control of government expenditure is a crucial element in obtaining macroeconomic stability, a more centralized intergovernmental arrangement would be compatible with greater macroeconomic controls. It concludes that macroeconomic control is not contingent on the actual administration of expenditure functions by the central government. A more important element is to have in place appropriate financing mechanisms that provide incentives for expenditure controls. Countries with centralized administration of expenditure functions are as prone to loss of macroeconomic control as those with decentralized administration. However, it appears that decentralized administration with poorly defined policy goals and lax financing mechanisms almost always have negative consequences for macroeconomic control. Thus, establishing a mechanism of expenditure control can be more complicated in a decentralized system that separates expenditure administration from financing responsibilities.

Overall, the findings in this chapter would weigh against major changes in expenditure assignments in most countries if these are sought as a means to achieving improved macroeconomic control. A country’s social preference, rather than economic theory, should be the major guide to defining the appropriate level of expenditure decentralization for mixed public goods. Once an administration system is in place, it is difficult and costly to change expenditure assignments. Policymakers having the objective of improving macroeconomic management should perhaps pay more attention to consistency between expenditure assignments and financing arrangements (revenue assignments and the system of intergovernmental transfers), rather than to the extent of centralization or decentralization of expenditure assignments. Increasing administrative and allocative efficiency, as well as matching public expenditure levels to preferences, is more important than the specific intergovernmental expenditure assignments.

This chapter first discusses allocative efficiency and the benefit principle. Then the relationship between expenditure assignments and financing arrangements is examined. Expenditure patterns in various countries for social protection, education, and health are considered next. The chapter concludes with a summary of findings.

Principles of Expenditure Assignment

Although traditional approaches assign allocative expenditure activities to local governments on a benefit principle, distributional and stabilization functions require primary responsibility at the central level.1 This principle does not lead to a sharp distinction between the expenditure responsibilities of different levels of government.

Shares of Local Expenditures

The systems of intergovernmental finance in the industrial federations have evolved gradually, much like their market and political institutions; the present systems result from historical circumstances and political events and reflect cultural traits. Among the six industrial federations (Australia, Austria, Canada, Germany, Switzerland, and the United States), the share of local (municipal) expenditures in total government expenditures varies from 7 percent in Australia to 24 percent in Switzerland; the state or provincial share of total expenditures varies from 14 percent in Austria to 43 percent in Australia.2 Transfers from higher-level governments vary from 16 percent of total local (municipal) revenue in Switzerland to 45 percent in Canada; the share of intergovernmental transfers in total receipts of states or provinces varies from 16 percent in Germany to 45 percent in Australia.

Each country has unique features. In Australia, local (municipal) government is of modest importance, and the state governments depend heavily on federal transfers. Switzerland is characterized by strong local governments that meet most expenditures from their own revenues and depend relatively little on transfers.

Expenditure devolution in India can be traced to the Government of India Act of 1935, which provided a degree of political autonomy to the provinces (states). With independence, developmental and infrastructure concerns have been addressed by the extra-constitutional Planning Commission, while the finance commissions have addressed issues of revenue sharing and the financing needs of the states. Expenditure control, given India’s federal structure, is a major issue.

In countries in transition to a market economy, such as China and Russia, overall expenditure control and the devolution of expenditure responsibilities are critical to economic stabilization and are of prime policy concern. Existing institutions and methods provide a starting point to determining desirable directions of reform.

Centralized Expenditures

Centralization, that is, vesting the central government or administration with power over policy, facilitates uniformity in the provision of public services.3 In a number of circumstances, this is necessary. First, certain goods have nonrival consumption within an entire country. Therefore, by definition the level of service, such as through defense and macroeconomic policies, is uniform. Decentralization would lead to an inefficient allocation of resources. Second, when significant economies of scale are present, decentralized administration can be inefficient, depending on the function being considered and the size of the country. Third, undesirable population and capital movements can result from variations in policy and the level of provision between jurisdictions.

In addition to the pursuit of macroeconomic stability, the following activities of government require centralization.

Provision of National Defense and Other National Public Goods

For certain goods, the whole population is affected by a given expenditure decision. Nonrival consumption exists throughout the country, and by definition everyone has access to the same level of service. If decentralization of policy decisions is attempted, each jurisdiction would consider only the direct benefits to its residents and ignore the benefits to residents of other jurisdictions—leading to the free-rider problem. For instance, while localities on the borders of a nation are greatly concerned about national defense, interior communities do not have the same incentives and might simply depend on the frontier regions to protect them. Such behavior, encouraged by an inappropriate system of expenditure assignments, leads to an undersupply and constitutes a misallocation of resources.


The central government is more efficient than local governments in ensuring income redistribution and in establishing minimum standards of public services across regions, because the ability of localities to support low-income groups is severely limited by the localities’ revenue bases, as well as the interjurisdictional mobility of the poor, the rich, and businesses. Redistribution by a given locality in isolation can attract the poor from neighboring localities. At the same time, imposing taxes on local factors of production to finance the redistribution can lead to flight of capital and of the wealthy and a consequent shrinkage of the resource base. Fixed factors, such as land and natural resources, vary considerably across regions, and the resulting revenue bases may afford some localities little scope for attaining minimum standards of public services in the absence of transfers from higher levels of government. Though the general government may be more efficient in ensuring income redistribution, the local governments are better able to identify target groups and to mobilize support for them (see the section on Social Protection). In general, the greater the mobility of the populations, the more difficult and costly it is to redistribute income within a locality.

The establishment of minimum access to education, health, and other human services across regions is a socially desirable goal of central government. Such redistributive policy implies substantial expenditure activities—or at least financing responsibility—for the central government.

Provision of Local Public Goods

Government services that benefit a clearly limited geographic region, or local public goods, could be provided by that level. Decentralized provision facilitates regional variety in the mix and the level of local public goods, which can greatly enhance social welfare. For instance, in a high crime area where most of the housing is constructed of stone, the local population will prefer high police expenditures over spending on fire prevention. Alternatively, in a given locality, the road network might need repair, while in another an upgrade of water and sanitary services might be preferable. Once it is accepted that regional variety is desirable, local provision can solve two of the more difficult problems confronting government decision making: the demand revelation problem and efficiency in government production.

The demand revelation problem concerns how much of a given public service to supply. While the market determines the exact quantity of each private good by equating supply and demand, no parallel mechanism exists for public goods. Instead, policymakers need to assess desired levels of supply of each public good, given costs of supply and public preferences. Although democratic elections provide some guidance and polls can also be useful, politicians have their own agendas that may not be in line with the public interest.

Ensuring efficiency in government production relates to public expenditures at all levels of government. For market-supplied goods, competition among producers, the profit motive among managers, and the desire to earn high incomes among workers encourage efficiency and work effort. Although this does not ensure that market production is always efficient, it is clear that competition tends to push the market in that direction. There is no parallel mechanism or motivation for the central government. Instead, a government has to depend on bureaucratic forms of control to induce efficiency.4 Furthermore, government officials and politicians can benefit by shirking responsibilities and by diverting government resources for personal use. Because this is a serious problem in many countries, more efficient government provision is needed, with streamlined procedures and public surveillance of government bureaucracies.

Decentralization of government promises a partial solution. The Tiebout and tax competition models point out that competition among local governments induces localities to provide an efficient configuration of local public goods. The key to the Tiebout Model is mobility of residents (see Tiebout, 1956; and Gordon, 1983). The framework examines a situation in which numerous localities exist in a metropolitan area. Since residents are mobile between localities, families will tend to choose the area that offers them the best combination of public services and taxes to pay for them. Therefore, local governments must be efficient in the provision of services to retain residents and attract new entrants.

Although problems with these models lie in the imposition of distributional considerations and assumptions regarding mobility, particularly of households, which may not be warranted in practice, they do point to important potential benefits of decentralized provision. On a more intuitive level, the advantages of decentralization stem from possible efficiency gains from smaller operations and the potential gains for consumer choice from variety. Decision making is normally easier in smaller organizations, and a small local government is more likely to cater to local needs and desires. However, large diseconomies of scale are possible in small local government units, as seen, for instance, in Austria. Furthermore, although local bureaucracies may be more efficient than national bureaucracies, the net effects of decentralization on the overall size of the bureaucracy are indeterminate.

With increased decentralization comes the possibility of loss of macroeconomic control as local bureaucracies multiply. Such a proliferation makes monitoring and evaluation more problematic, and the quality of administration may deteriorate, given a shortage of skilled administrators. Tendencies to exaggerate expenditures and to participate in bargaining games with higher levels of government may increase, particularly if financing arrangements are shared.

Assessment of Expenditure Patterns

A difficulty in assigning expenditure responsibilities in some countries lies in unclear distinctions between levels of administration. This is in addition to the difficulties associated with the separation between regulation, financing, and administration of expenditures at a given level. As pointed out in a comprehensive review by Levin (1990), regulation is difficult to measure, and whereas finance and administration can be identified, neither fully represents the complete range of activities carried out by the government. The difference between financing and administration lies in intergovernmental transfers, where the grantor government often has no administrative role in providing goods or services. The financing (administration) of expenditures at a particular level is determined inclusive (exclusive) of grants given and net (inclusive) of grants received. Centralization ratios based on the administration of expenditures are likely to underestimate the role of the central government, particularly with respect to the redistributional transfers generally made by the central authorities. The appropriate classification of extrabudgetary funds, particularly for social security, would also affect the reported degree of centralization.

Some of the resources, such as shared revenues, conditional grants, and borrowing, may provide higher levels of government a degree of control. In some cases, the central government mandates certain expenditures, thus exerting some control over local functions, even when these are administered and fully financed by the lower level of administration. At the opposite extreme, expenditure functions may be mandated and financed by the central government, but administered by the subnational governments, including minimum educational or health standards. Here, the effective control over the level of expenditures rests with the administering lower-level authority (for example, health care in Italy), even though officials of the central administration may be represented at the lower level (for example, social expenditures in Pakistan). The loss of central control in the latter case arises from the effective underwriting by the central government of deficits or expenditure overruns at lower levels of government. Thus, in most cases, the form of transfers from higher levels of government and the associated incentives for limiting expenditure growth appear to be more important in controlling overall expenditures than simplistic categorizations of centralization or decentralization (see Chapter 4).

A wide variety of expenditure patterns and associated financing mechanisms can be observed across the world. Unconditional grants and own resources accord the greatest level of autonomy to local governments in providing local goods and services.

As Table 1 indicates, unconditional grants form a predominant share of transfers in the United Kingdom, France, and Australia—giving local governments the capacity to provide local goods, but not requiring them to do so. On the other hand, only 11 percent of grants in the United States and around 20 percent in the Scandinavian countries and Canada can be classified as free of regulations. Among developing countries, very few distinguish between different types of grants, although a substantial proportion of total expenditures is financed through grants in India (27.8 percent) or Indonesia (17.6 percent), excluding transfers for local expenditures in both cases.

Table 1.Percentages of General Government Expenditure Financed by Intergovernmental Grants(Average of latest three years available)
Ending YearTotal Expenditures1EducationHealthSocial Security and WelfarePercent of Intergovernmental Grants Unclassified
New Zealand19812.1
South Africa198614.9
United Kingdom198713.
United States198714-942.818.713.210.9
Source: Levin (1990) based on Government Finance Statistics Yearbook, Vol. 13 (Washington: International Monetary Fund, 1989).

Includes supranational authorities’ share of general government expenditures in Belgium (2.2 percent), Denmark (2.2 percent), France (1.4 percent), Germany (1.8 percent), Luxembourg (2.7 percent), and the United Kingdom (1.9 percent).

Data for general government do not include local government.

Source: Levin (1990) based on Government Finance Statistics Yearbook, Vol. 13 (Washington: International Monetary Fund, 1989).

Includes supranational authorities’ share of general government expenditures in Belgium (2.2 percent), Denmark (2.2 percent), France (1.4 percent), Germany (1.8 percent), Luxembourg (2.7 percent), and the United Kingdom (1.9 percent).

Data for general government do not include local government.

Where identifiable, conditional transfers indicate the extent to which higher levels of government wish to promote a particular activity at the subnational or local level. Conditional grants finance a substantive proportion of expenditures on education in Zimbabwe, the United States, and Canada, but very little in the United Kingdom and Germany. However, in the latter group of countries unconditional grants are more important than in the former—thus indicating that even in a highly “centralized” system, as in the United Kingdom, local authorities have some freedom for maneuver in the allocation of resources for education, despite the establishment of nationwide norms and curriculums.

Financing patterns are relevant to local expenditures, given the empirical phenomenon of fiscal illusion, whereby transfers appear to stimulate expenditures of recipient governments beyond expectations. This phenomenon is observed in several countries for which a reasonably long time series is available (see Winer (1983) for Canada, and Gramlich (1979) for a review of U.S. studies). The evidence is consistent with the hypothesis that the separation of decision making caused by substituting own resources with grants increases local expenditure beyond theoretically expected levels, although the degree of illusion tends to fade over time.

A number of studies point out that the counterpart of fiscal illusion at the local level is the tendency of grants to lead to lower expenditures by higher level governments (see Hewitt, 1986; and Logan, 1986). Therefore, the impact of fiscal illusion on general government expenditures is uncertain. As discussed in Chapter 4, the type of grant and the mechanism whereby it is implemented affect the overall level of local expenditures.

The public choice literature (for example, Brennan and Buchanan, 1980) suggests that decentralization per se might curb the growth of government expenditures—the Leviathan hypothesis, although the evidence is mixed. Cross-sectional analyses based on international data5 tend to reject the hypothesis. It appears that the form of financing of expenditures is more important in determining expenditure levels than the extent of decentralization. For instance, substitution of own sources with transfers per se can have a stimulative effect on expenditure and could lead to a higher than optimal level of expenditure.6 The abolition of subnational taxes in 1973 in Italy to centralize tax policy and improve tax administration, coupled with the effective removal of financing responsibility at the subnational level, led to higher expenditure. De facto decentralized revenue and expenditure responsibilities in China in the period up to 1993 have also exacerbated the growth of expenditures and of associated overall budgetary and macroeconomic imbalances. Thus, reduced overall expenditures cannot be taken as an inherent characteristic of decentralization. In fact, the fiscal illusion thesis suggests the opposite.

Brosio, Hyman, and Santagata (1980) found that public per capita expenditure in Italy was indeed positively affected by the separation of the subnational governments’ ability to raise own revenue from their expenditure assignments—which had resulted from the fiscal reform in the 1970s. This increase was immediate, although it tended to decline with time owing to the financial problems soon encountered by the subnational governments, as the central government limited the amount of transfers granted to them. Rizzo (1985) also found a positive impact of decentralization on government size in Italy, when coupled with a decline in fiscal responsibility of local administrators. A study of decentralization and government size in Latin America (which controls for a positive correlation between rising income and the size of government (Wagner’s hypothesis)) found that decentralization financed by central government transfers, rather than through own revenues, was likely to increase government expenditures (Reid and Winkler, 1992).

Fiscal illusion implies that grants powerfully influence allocation decisions of local government. From the Gramlich (1979) study, it appears that the expenditure elasticity of revenue-sharing grants is approximately unity. This suggests that revenue-sharing grants do not crowd out local revenue collection but instead are put entirely into increased local expenditures. The elasticity associated with matching grants appears to be in the order of two: matching grants approximately double expenditures. Although these empirical results are somewhat surprising, one explanation offered in Hewitt and Heffley (1989) suggests that the higher expenditures result from an interaction between mobility and grants.

To avoid suboptimal supply, grants are clearly needed when local goods have interregional externalities. In these cases, matching grants can induce localities to take account of the positive benefits afforded to nonresidents. In the resulting arrangement, there are decentralized administration and a sharing of both financing and policy control.

The issue of expenditure assignments and the associated pattern of grants should be viewed in the changing context of a country’s macroeconomic conditions. The Italian case reflects the consequence of revenue centralization since 1973, coupled with the assignment to subnational governments of the responsibility for the administration of major expenditure programs, such as health and social assistance, while responsibility for their financing was mostly with the central government (Table 2). There is no local incentive to control costs, and the lack of accountability at the local level has led to an uncontrolled growth of total public expenditure. The grant system developed during the 1970s was meant to be a temporary bridge between the fiscal reform of 1973 and the reform of local taxation and financing—which was started in 1993. In the meantime, ad hoc grants have multiplied as the need arose.

Table 2.Italy: Arrangements for Financing, Policymaking, and Administration of Major Expenditure Items, 1991
Social welfare1
HealthFor own earmarked revenues
Social welfare
EducationSupport personnel, buildings and furniture, for specific schools in secondary educationFor functions specified under financing arrangements
Social welfare
EducationSupport personnel, buildings, and furniture in elementary educationFor functions specified under financing arrangements
Social welfareCash and in-kind benefits to people on the municipal poor listFor functions specified under financing arrangement

Pensions and family allowances.

Pensions and family allowances.

Social Protection, Education, and Health

There is considerable diversity of experience relating to the share of expenditures on social protection, education, and health administered at each level of government, as seen in Table 3. Among industrial countries, education is primarily provided at the central government level, although in the United Kingdom and the United States, it is provided by local governments, and by state governments in Australia7 and Germany. Among developing countries, education and health care are primarily provided by the central government in some cases, as in Indonesia, but by the state (or provincial) governments in others, such as in India (data on local governments are not available in either country). In other countries, such as Brazil, detailed information is not available for the breakdown of functional expenditures at subnational Levels of government, despite the importance of such a classification for overall policymaking.

Table 3.Magnitude of General Government Expenditures and Portion Administered by Each Level of Government1(Average of latest three years available)
Total ExpendituresEducationHealthSocial Security and Welfare
Country and Ending YearTotalEducationHealthSocial Security and WelfareCentral governmentState governmentLocal governmentCentral governmentState governmentLocal governmentCentral governmentState governmentLocal governmentCentral governmentState governmentLocal government
(In percent of GDP)(In percent of general government)
Argentina2 (1987)
Australia (1987)
Austria (1987)51.870.413.716.9
Belgium (1987)56.785.911.9
Bolivia (1986)11.185.910.63.4
Brazil (1987)34.165.824.59.6
Chile (1987)
Colombia (1984)
Denmark (1986)
Finland (1987)43.054.745.3
France (1985)
Germany (1983)
Hungary (1988)
India2 (1986)
Indonesia2 (1988)
Ireland (1987)55.872.527.5
Luxembourg (1987)
Malawi (1984)
Mexico (1984)
Netherlands (1988)
New Zealand (1981)43.286.913.1
Norway (1986)47.266.433.6
Pakistan (1979)
Paraguay (1984)11.395.14.9
Poland (1988)
Romania (1985)
South Africa (1986)33.374.812.512.7
Sweden (1987)61.659.840.2
Switzerland (1984)
Thailand (1982)
Tunisia (1982)
United Kingdom (1987)
United States (1987)
Yugoslavia (1987)
Zimbabwe (1986)
Source: Levin (1990) based on International Monetary Fund, Government Finance Statistics Yearbook. Vol. 13 (Washington: International Monetary Fund, 1989).

Excluding intergovernmental grants.

Data for general government do not include local government.

Includes supranational authorities’ share of general government expenditures in Belgium (2.2 percent), Denmark (2.2 percent), France (1.4 percent), Germany (1.8 percent), Luxembourg (2.7 percent), and the United Kingdom (1.9 percent).

Source: Levin (1990) based on International Monetary Fund, Government Finance Statistics Yearbook. Vol. 13 (Washington: International Monetary Fund, 1989).

Excluding intergovernmental grants.

Data for general government do not include local government.

Includes supranational authorities’ share of general government expenditures in Belgium (2.2 percent), Denmark (2.2 percent), France (1.4 percent), Germany (1.8 percent), Luxembourg (2.7 percent), and the United Kingdom (1.9 percent).

Social Protection

Social protection mechanisms embody elements of insurance, together with redistribution. This is true in the industrial societies, where formal social insurance instruments cater for old age and loss of income, including unemployment, together with family allowances that limit the risk of poverty due to childbearing, and measures to provide for social assistance. Informal arrangements that mimic the above arrangements are also occasionally found in traditional societies in developing countries—but these suffer from a number of difficulties that mirror the level at which the service is provided (see Ahmad, Dreze, Hills, and Sen, 1991).

A major element in the provision of coverage for life-cycle and employment contingencies is the pooling of risk. Thus enterprise-level provision for the aged or the disabled, as was the case in former centrally planned economies, such as China and the countries of the former Soviet Union, imposed on older enterprises major costs that were inconsistent with the efficient functioning of a market economy. Similarly, in several countries (for example, Bangladesh, India, Jordan, Pakistan, and most countries of the former Soviet Union), subsidies to keep inefficient public sector enterprises afloat have proven difficult to remove owing to an absence of effective unemployment insurance. And, while community-based provision for the disabled and aged tends to be effective in identifying the appropriate recipients, because of the lower costs of local information, such methods are not appropriate when the whole community is affected, say, by drought. Thus, by their nature, social insurance mechanisms should not be based on local pooling of risks, and the safety net should be spread as far as administratively possible.

It is clear, however, that the informational advantages of local provision predominate when there is a need for fine targeting, as with most forms of social assistance. The link with local levels of resources (at the margin) is also clear, with the roots of assistance based on charitable provisions by churches, and the more formalized local provisions based on individual wealth in many Middle Eastern and African countries (see Ahmad, 1991). With financing from higher levels of government at the margin, there is a lower incentive to pay, together with a reduced incentive to target the neediest.

The devolution of certain social expenditure responsibilities, for example, for consumer and enterprise subsidies, as in some of the former centrally planned economies in Eastern Europe and Asia, has contributed to an intractable problem of expenditure and revenue bargaining at lower levels of government. This has degenerated into a loss of control over such expenditures, together with an increase in the associated overall deficit. An attempt to introduce a degree of control, by moving to matching earmarked grants, has had a regressive impact, because the system of price and enterprise subsidies is more heavily concentrated in the cities. Moreover, the matching ratios for central grants can be as high as two, and poorer provinces are often unable to comply with the high cofinancing requirements.


Education represents a classic case of conflicting goals and different levels of assignment. Significant cross-country differences can be found between the provision of preschool, primary, secondary, tertiary, and higher levels of education, although standardized information is not available.

In most countries, primary education is supplied free. More developed countries also provide free secondary education and heavily subsidize higher education. The question of why governments choose to subsidize education is not simple. In the first instance, education does not possess the characteristics of a pure public good. Educational benefits mostly accrue to the recipients by increasing their human capital and therefore increasing their expected lifetime income. No strong evidence suggests that the private rewards to education are less than the productivity gain to society and, indeed, private education is to be found at all levels in most countries. The difficulty is that the poor usually have limited access to private education, because of imperfect capital markets and information. Thus, ensuring that low-income persons have at least a minimal level of education is an excellent way of redistributing income. Moreover, the poor in general have a higher discount rate than that of society, and the present value of the future income through education is greater for society than for the poor.

There is thus a prima facie case for public provision of education, although this often coexists with private provision. Education may have to be compulsory since there are private costs involved, for example, the loss of income from child labor, or the reduction in unpaid family services performed by children. Such private costs apply in all countries.

Primary and secondary education is normally administered by the local (that is, municipal) governments. University education is often more centralized to take advantage of economies of scale. There is little question that local administration of primary schools is desirable, although in larger countries there may be diseconomies associated with small-scale operations. First, by definition education services have to be geographically spread out. Second, smaller schools generally provide higher-quality education, although this may be achieved at a higher cost. Third, the direct involvement of parents in schools is observed to be a beneficial determinant of school quality. This benefit can only be achieved through decentralization of administration, as well as local monitoring. Jimenez and others (1988) found that the quality of primary education in the Philippines was higher in localities with higher local participation and monitoring. Despite the desirability of local administration, cost differentials across regions may be substantial. If these differences are due to unavoidable spatial or demographic factors, additional financial support from the central government may be justified.

At the same time, the distributional implications of the quality of education are substantial. Decentralization of financing, as observed in the United States, leads to large differences in educational institutions. Per capita education expenditures in the United States diverge widely in certain geographical regions, leading to a perpetuation of income differences across generations. Most societies consider inequality of educational services unfair. Thus, many societies find it desirable to decentralize control of state-supported schools, while at the same time taking steps to ensure that some minimal quality standards are obtained. In addition, a highly educated population is widely viewed as an important means of promoting development. Therefore, for allocation as well as distributional reasons, the entire nation should be interested in ensuring a minimum educational standard.

The methods used to promote minimum educational quality vary widely among countries. In some, the education program is centralized, with school costs paid entirely by the central government, such as France and Italy. However, at the same time, the staff of schools are required to react to the needs and desires of students and parents. In other countries, schools are effectively supported almost entirely by grants, for example, the United Kingdom. In this case, minimum standards are assured along with local administration, but financed by a block grant. In other circumstances, local financing is supplemented with grants based on needs, although (as with health care in Italy) the respective levels of need and associated responsibility may open up possibilities of bargaining between different levels of government.


Health care affords an interesting example of the interrelationships between public and private provision and the different levels of government at which the public provision would be most effective. Public health facilities, including preventive and diagnostic care, tend to be provided primarily at the local (or municipal) level. Public health facilities and the prevention of communicable diseases have a pure public good characteristic and are best provided by higher levels of government (provinces in some countries or the central government). Curative facilities, particularly at the hospital and specialized levels, have associated economies of scale and are most appropriately provided at higher levels—regions, provinces, or the center, depending on institutional arrangements. Many of these facilities can be provided by private agents, although overlapping insurance and provision can greatly add to overall costs, and to ensure equity, a system of referrals for the poor needs to be devised. As with other forms of mixed provision, the types and forms of health care provision vary widely.

There is a trend toward decentralizing much of health care provision to lower levels of government, although there have been serious problems both with implementation and financing. Managerial ability at the local level is often in short supply (Mills, 1990). In many cases, financing and supervisory mechanisms have had to be reevaluated following decentralization. For instance, Nigeria implemented a National Health Policy in 1988, emphasizing primary health care and secondary facilities to be administered and financed by state and local governments, with the federal government retaining responsibility for teaching and specialty hospitals. Given that shared revenues and grants formed the basis of much of state and local revenues, increasing costs forced a revision in the amounts allocated primarily to local authorities.

The interactions of financing and local provision are seen most clearly in China. Until the late 1970s, preventive and primary health care (for example, with the “barefoot doctors”) was organized and financed by localities and communes with remarkable success. The nature of mortality and morbidity in China changed from the typical developing country pattern (with parasitic and infectious diseases predominating) to one associated with middle- and higher-income countries and a virtual eradication of parasitic diseases (see Ahmad and Hussain, 1991). The economic reforms led to a substantial degeneration of the primary health care system and a reliance on payments for services, even for the inoculation of children. Rising incomes, soaring fees, and medicine costs have led to a spiraling cost of health care. At the same time, many of the poor who cannot afford the costs of health care are now effectively excluded from coverage. With the neglect of preventive care, morbidity rates have recently risen for some infectious diseases, such as typhoid and hepatitis, particularly in poorer regions (see Yu Dezhi, 1992). Attempts to introduce private insurance mechanisms in rural areas and in cities suggest that costs per head remain high, and that these mechanisms are applicable mainly in richer areas. Health care services in poorer regions depend crucially on central government transfers, given limitations on own resources of local governments.

Difficulties with financing health care can also be found in industrial countries, such as the United States, where the provision of private care is costly, and large groups of the population are excluded from basic coverage. At the other extreme, the Italian case reflects difficulties with a universal system of multilevel provision, almost entirely financed by transfers from the center until 1993. The central government determines allocations across regions, as well as the establishment of minimum service levels that lower tiers of government are obliged to provide. However, the expenditure norms of the central government’s “guaranteed uniform level” are fraught with considerable ambiguity, leading to ad hoc provisions and an erosion of control. Attempts to restrict funding have either led to a lowering of the quality of service or to considerable expenditure overruns.

Without an objective definition of “minimum standards,” together with a hard-budget constraint on the local authorities and effective auditing of local and regional expenditures, health care deficits in Italy are likely to continue into the future. Complete centralization, as in the United Kingdom and France, may not provide a full answer. Supervision and associated informational requirements may remain costly, and any attempt to provide a basic minimum is likely to result in implicit rationing and queues. The United Kingdom has experimented with the introduction of competition with respect to supply. The Dutch reforms of their health care system addressed the issue of market structure more than decentralization: maintaining free diagnostic and primary care at the local level, but introducing a greater reliance on market forces for curative care. A system of referrals has been instituted to ensure equity, and government regulations are maintained with respect to quality and cost controls.

The choice of the appropriate level of administration for health care thus depends crucially on the type of service: preventive and public health clearly being provided best at the local level. Others depend crucially on the administrative capabilities of the country, as well as the financing arrangements.

Concluding Remarks

Control of overall expenditure levels is important in a successful macroeconomic strategy. This chapter has attempted to show that this control is not contingent on the actual administration of expenditure functions by the central government, but rather on the adoption, within a system of intergovernmental fiscal relations, of financing mechanisms that provide appropriate incentives for control. Indeed, countries with centralized administration of certain expenditure functions are as prone to loss of macroeconomic control as those in which the administration is decentralized. However, it is also the case that decentralized administration, with poorly defined policy goals and lax financing mechanisms, almost invariably leads to a loss of macroeconomic control. Overall expenditure control, rather than the allocation of responsibilities, would appear to be of most concern from the macroeconomic perspective.

The other main conclusion of this chapter is that the observed wide variety of expenditure assignment patterns reflects varying social preferences among countries—coupled with little theoretical guidance regarding the costs and benefits of decentralization—particularly for mixed goods such as social protection, education, and health. Smaller, decentralized governments have the advantage of being more flexible, administratively simpler, and often able to more closely react to what local residents want. A major disadvantage of decentralization is that its redistributive implications at times lead to unacceptable regional disparities of income. Furthermore, differing social policies can unleash undesirable and inefficient mobility incentives that have contributed at times to an impoverishment of central cities (for example, Washington, D.C.), as persons in need of economic assistance are attracted to relatively more generous localities while high-income families and many businesses are repelled by the steep tax rates needed to support expensive social benefits. Using intergovernmental grants to allow centralized financing along with local expenditure administration is not an entirely satisfactory solution. Some of the advantages of decentralization are lost by such arrangements, as intergovernmental systems can become excessively complicated and nontransparent and can generate their own allocative inefficiencies through fiscal illusion.

These findings imply that substantial reshuffling of expenditure responsibilities, as a means to improving macroeconomic management, is seldom warranted. Existing intergovernmental institutions could generally be retained and reforms should concentrate on ensuring compatibility of expenditures and financing. This will then allow public reforms in the area of public expenditure policy and management to focus attention on more fundamental issues: improving the efficiency of the overall expenditures, increasing administration efficiency, and designing programs that generate little negative economic incentives, including corruption.

There are striking gaps in the information base on expenditure responsibilities and outcomes, even for heavily researched countries, such as Brazil and India. In order to evaluate expenditure programs at each level of government, data should be available on state and local expenditures classified into meaningful functional and economic categories, and on different types of grants and borrowing arrangements at each level of administration. Such information is seldom readily available. Thus, a major effort should be initiated to generate the necessary data at lower levels of government. Such information would help to establish expenditure priorities, to evaluate trade-offs at different levels of government, and to pare off unproductive expenditures—making for a more efficient, equitable, and sustainable public sector.

Musgrave (1959), pp. 181–82.

Figures in this section are the authors’ calculations from data in International Monetary Fund, Government Finance Statistics Yearbook, 1991, and refer to the following years: Australia (1990), Austria (1990), Canada (1989), Germany (1990). and Switzerland (1984).

This does not have to be the case, however. If local or regional variety is sought, it is normally desirable to allow subnational levels of government to make the allocation decisions.

Private businesses, particularly large corporations, can face similar problems and in fact many large corporate bureaucracies are not necessarily more efficient than government bureaucracies.

Subject to the major caveats concerning the measurement of decentralization, as discussed in the next section.

This is demonstrated theoretically by using a full-information median voter model (Brosio, 1985).

University education in Australia is assigned to the states under the Constitution. However, most universities are administered by independent syndicates, and almost all the financing comes from the central government through the states. There is thus considerable overlapping of responsibility even where the assignment might seem to be fairly clearly defined.


    AhmadEhtisham1991“Social Security and the Poor: Choices for Developing Countries,”World Bank Research ObserverVol. 6 (January) pp. 10527.

    AhmadEhtishamJeanDrezeJohnHills andAmartyaSeneds.1991Social Security in Developing Countries (Oxford: Clarendon Press).

    AhmadEhtisham andAtharHussain1991“Social Security in China: A Historical Perspective,”Social Security in Developing Countriesed. by EhtishamAhmadJeanDrezeJohnHills andAmartyaSen (Oxford: Clarendon Press).

    BirdRichard M.1986Federal Finance in Comparative Perspective (Toronto: Canadian Tax Foundation).

    BrennanGeoffrey andJames M.Buchanan1980The Power to Tax: Analytical Foundations of a Fiscal Constitution (Cambridge, England: Cambridge University Press).

    BrosioGiorgio1985“Fiscal Autonomy of Non-Central Government and the Problem of Public-Spending Growth,” in Public Expenditure and Government Growthed. by FrancescoForte andAlanPeacock (Oxford: Blackwell).

    BrosioGiorgioD.N.Hyman andW.Santagata1980“Revenue Sharing and Local Public Spending: The Italian Experience,”Public ChoiceVol. 35No. 1 pp. 315.

    BuchananJames M.1975The Limits of Liberty (Chicago: University of Chicago Press).

    ChelliahR.J.M.G.Rao andT.K.Sen1992“Issues Before the Tenth Finance Commission,”Economic and Political WeeklyVol. 27 (November21) pp. 253950.

    GerberRobert I. andDaniel P.Hewitt1987“Decentralized Tax Competition for Business Capital and National Economic Efficiency,”Journal of Regional ScienceVol. 27No. 3 pp. 45160.

    GordonRoger H.1983“An Optimal Taxation Approach to Fiscal Federalism,”Quarterly Journal of EconomicsVol. 98 (November) pp. 56786.

    GramlichEdward M.1979“Intergovernmental Grants: A Review of the Empirical Literature,” in The Political Economy of Fiscal Federalismed. by Wallace E.Oates (Lexington, Massachussets: Lexington Books).

    HewittDaniel1986“Fiscal Illusion from Grants and the Level of State and Federal Expenditures,”National Tax JournalVol. 39 (December) pp. 47183.

    HewittDaniel1991“Transfers to Local Governments,” in Public Expenditure Handbooked. by Ke-youngChu andRichardHemming (Washington: International Monetary Fund).

    HewittDaniel andDennisHeffley1989“Unfolding the Flypaper: The Effects of Inter-governmental Grants in an Open Local Economy,”IMF Working Paper 89/58 (Washington: International Monetary Fund).

    HewittDaniel andDubravkoMihaljek1992“Fiscal Federalism,” in Fiscal Policies in Economies in Transitioned. by VitoTanzi (Washington: International Monetary Fund).

    JimenezEmmanuelVicentePaqueo andMa. Lourdes deVera1988“Does Local Financing Make Primary Schools More Efficient? The Philippine Case,”PPR Working Paper WPS 69 (Washington: World Bank).

    LevinJonathan1990“Measuring the Role of Subnational Governments,” in Public Finance with Several Levels of Governmented. by RémyPrud’homme (The Hague: Foundation Journal Public).

    LoganRobert R.1986“Fiscal Illusion and the Grantor Government,”Journal of Political EconomyVol. 94 (December) pp. 130418.

    MieszkowskiPeter andGeorgeZodrow1989“Taxation and the Tiebout Model: The Differential Effects of Head Taxes, Taxes on Land Rents, and Property Taxes,”Journal of Economic LiteratureVol. 27 (September) pp. 10981146.

    MillsAnneed.1990Health System Decentralization: Concepts Issues and Country Experience (Geneva: World Health Organization).

    MuellerDennis C.1989Public Choice II (New York: Cambridge University Press).

    MusgraveRichard A.1959The Theory of Public Finance: A Study in Public Economy (New York: McGraw-Hill).

    OatesWallace E.1985“Searching for Leviathan: An Empirical Study,”American Economic ReviewVol. 75 (September) pp. 74857.

    OatesWallace E. andRobert M.Schwab1988“Economic Competition Among Jurisdictions: Efficiency Enhancing or Distortion Inducing?”Journal of Public EconomicsVol. 35 (April) pp. 33354.

    OkunArthur M.1975Equality-Efficiency: The Big Trade-Off (Washington: Brookings Institution).

    Pisany-FerryJ.1991“Maintaining a Coherent Macro-Economic Policy in a Highly Decentralized Federal State: The Experience of the EEC,”paper prepared at the OECD Conference on Fiscal Federalism in European Economies in Transition (ParisApril).

    ReidGary andD.R.Winkler“Decentralization and Government Size in Latin America” (unpublished;Washington: World Bank1992).

    RizzoIlde1985“Regional Disparities and Decentralization as Determinants of Public Sector Expenditure Growth in Italy, 1960–1981,” in Public Expenditure and Government Growthed. by FrancescoForte andAlanPeacock (Oxford: Blackwell).

    TanziVito“Tax Assignments” (unpublished;Washington: International Monetary Fund1994).

    TieboutCharles M.1956“A Pure Theory of Local Government Expenditure,”Journal of Political EconomyVol. 64 (October) pp. 41624.

    WinerStanley L.1983“Some Evidence on the Effects of the Separation of Spending and Tax Decisions,”Journal of Political EconomyVol. 91 (February) pp. 12640.

    YuDezhi1992“Changes in Health Care Financing and Health Status: The Case of China in the 1980s,”Innocenti Occasional Papers EPS 34 (Florence).

    ZaxJeffrey S.1989“Is There a Leviathan in Your Neighborhood?”American Economic ReviewVol. 79 (June) pp. 56067.

    Other Resources Citing This Publication