Journal Issue
Share
Article

Decentralizing Government

Author(s):
International Monetary Fund. External Relations Dept.
Published Date:
January 1997
Share
  • ShareShare
Show Summary Details

Decentralizing government operations can improve economic welfare. But decentralization requires close policy coordination among all levels of government.

OVER THE last few decades, a clear trend has emerged worldwide toward devolving spending and, to a lesser extent, revenue-raising responsibilities to subnational levels of government. This trend can be seen in countries with a long tradition of centralist government as well as in federalist systems, and in developing as well as industrial countries. The shift toward devolution is largely a reflection of the political evolution toward more democratic and participatory forms of government that seeks to improve the responsiveness and accountability of political leaders to their electorates, and to ensure a closer correspondence between the quantity, composition, and quality of publicly provided goods and services and the preferences of recipients.

Delegating spending

Decentralizing spending responsibilities can bring substantial welfare gains. Government resources can be allocated most efficiently if responsibility for each type of public expenditure is given to the level of government that most closely represents the beneficiaries of these outlays. But decentralization can also entail significant costs in terms of distributional equity and macroeconomic management. This can be especially important in large countries where the economic differences among regions are substantial and can lead to undesirable internal migrations, as well as social and political pressures. Decentralization can also be costly if it results in the substandard provision of certain public goods—such as primary education or basic health care—as this can affect productivity and the long-term growth prospects of the economy.

While these are valid points, they do not necessarily imply that the provision of public goods and services—other than those of a clearly national nature—should be centrally administered. The central government can influence how these goods and services are provided locally by setting policy guidelines for their delivery, by transferring resources to subnational governments to equalize their capacity to meet these guidelines, and by controlling ex post the level and quality of local services.

Decentralizing spending can have significant implications for macroeconomic management. Even if the overall level of expenditure of subnational governments is constrained by limits on their taxing and borrowing powers, changes in the composition of their expenditures can affect overall demand and the balance of payments in ways that may defeat national stabilization objectives. This may happen, for example, if subnational expenditure shifts toward items that have a relatively large impact on demand, such as transfers to consumers.

From a macroeconomic management perspective, therefore, central governments should retain responsibility for expenditures that have a particularly strong impact on demand or are sensitive to changes in the business cycle, such as unemployment benefits. In general, the greater the share of public expenditure assigned to subnational levels of government, the greater the need to involve them in the pursuit of any needed fiscal adjustment.

Weak institutions can throw the theoretical efficiency gains from decentralization out the window in practice, however. Overstaffing, poor technical skills and training of employees, and the inability to formulate and implement effective spending programs affect local (and central) governments worldwide. The incidence of corruption is not negligible. Moreover, sub-national governments often do not have modern, transparent public expenditure management systems, while some local jurisdictions are too small to fully realize the potential efficiency gains from decentralization.

Against the background of these considerations, it is not surprising that country experiences regarding expenditure assignments vary widely For example, the share of state (provincial) government expenditure in total general government expenditure varies from 18 percent in Spain to 45 percent in India (Table 1).

Table 1Distribution of expenditure varies widely
YearTotal general government expenditure1
As a

percentage

of GDP
As a percentage of the total
Central

government
State

government
Local

government
Industrial countries
Federal systems
Australia199546.559.036.05.0
Canada199360.141.741.217.1
Germany199557.259.224.116.7
Spain199355.970.418.211.3
United States199441.353.425.621.0
Unitary systems
Belgium199456.588.5n.a.11.5
France199556.582.3n.a.17.7
Netherlands199566.576.4n.a.23.6
Norway199460.268.4n.a.31.6
United Kingdom199554.177.3n.a.22.7
Developing and transition countries
Federal systems
India199330.854.745.3u
Argentina199221.855.144.9u
Brazil199356.665.724.89.5
Mexico199319.178.321.7u
Russian Federation199538.562.4u37.6
Unitary systems
Kenya199430.096.1n.a.3.9
Poland199551.883.8n.a.16.2
South Africa199450.166.325.48.3
Thailand199517.392.6n.a.7.4
Sources: IMF, 1996, Government Finance Statistics Yearbook; and IMF, 1997, International Financial Statistics (June). Note: n.a. means not applicable; u means data unavailable.

General government is defined to include the central government; social security system; and state, provincial, and local governments.

Sources: IMF, 1996, Government Finance Statistics Yearbook; and IMF, 1997, International Financial Statistics (June). Note: n.a. means not applicable; u means data unavailable.

General government is defined to include the central government; social security system; and state, provincial, and local governments.

Assigning revenue

There is a degree of consensus in the public finance literature on desirable criteria to guide the assignment of revenue-raising responsibilities across the various levels of government. Nevertheless, country experiences do not always conform to those criteria and vary widely (Table 2).

Table 2Distribution of tax revenue among different levels of government
YearTotal general government tax revenue1
As a

percentage

of GDP
As a percentage of the total
Central

government
State

government
Local

government
Industrial countries
Federal systems
Australia199528.976.619.93.5
Canada199338.753.536.510.0
Germany199541.173.021.06.0
Spain199333.286.64.68.8
United States199427.065.720.613.7
Unitary systems
Belgium199445.794.8n.a.5.2
France199542.489.8n.a.10.2
Netherlands199544.796.3n.a.3.7
Norway199440.378.6n.a.21.4
United Kingdom199534.896.4n.a.3.6
Developing and transition countries
Federal systems
India199314.961.838.2u
Argentina199219.857.242.8u
Brazil199325.771.426.02.6
Mexico199316.384.615.4u
Russian Federation199529.060.0u40.0
Unitary systems
Kenya199421.197.8n.a.2.2
Poland199540.092.1n.a.7.9
South Africa199427.691.43.15.5
Thailand199518.294.9n.a.5.1
Sources: IMF, 1996, Government Finance Statistics Yearbook; and IMF, 1997, International Financial Statistics (June).Note: n.a. means not applicable; u means data unavailable.

General government is defined to include the central government; social security system; and state, provincial, and local governments.

Sources: IMF, 1996, Government Finance Statistics Yearbook; and IMF, 1997, International Financial Statistics (June).Note: n.a. means not applicable; u means data unavailable.

General government is defined to include the central government; social security system; and state, provincial, and local governments.

It is generally recognized that both distributional and, especially, macroeconomic management considerations argue against arrangements (such as existed in the former Yugoslavia) that would assign all or most taxing powers to subnational governments, with upward revenue sharing, especially if the sharing formula is renegotiated frequently. For one thing, such arrangements do not facilitate income redistribution through the tax system. More important, they deprive the central government of any tax tool for macroeconomic management. Therefore, upward revenue-sharing arrangements can be viable only in countries (such as Germany) with a long-established tradition of close policy coordination among different government levels or in loose confederations or common economic areas in which the responsibility for stabilization policies continues to rest primarily with the member states. Moreover, since substantial regional variations in the bases and/or rates of certain taxes could lead to distortions in the flow of goods and factors of production, efforts have to be made to harmonize tax policies even in common economic areas.

Arrangements that assign all or most taxing powers to the central government are undesirable as well. By separating spending authority from revenue-raising responsibilities, these arrangements obscure the link between the benefits derived from public expenditures and their price, namely, the taxes levied to finance them. Therefore, the alternative favored in the literature and most frequently observed in countries around the world is one that provides for the assignment to each level of government of its own sources of revenue, in combination with various types of intergovernmental transfers to bridge any resulting gap between revenue and expenditure assignments.

In this area, country practices differ widely Some countries completely separate the tax bases for different levels of government, while others allow different levels of government to tap the same tax base (tax overlapping). Examples of tax separation can be found in Australia and India, and, of nonshared taxes, in Germany. By contrast, there is a considerable degree of tax overlapping in Canada and the United States. In general, the central government should be assigned taxes that are levied on the more mobile tax bases; are more sensitive to changes in income—that is, having higher income elasticity; and are levied on tax bases that are distributed very unevenly across regions.

Intergovernmental transfers

Reflecting the fact that most major taxes are typically assigned to the central government, while substantial and growing expenditure responsibilities are devolved to regional and local governments, sizable vertical imbalances (pretransfer fiscal deficits) frequently emerge at the subnational government level. There are also horizontal imbalances, since the revenue-raising capacity of subnational governments varies and different regions may face different cost and demand pressures as they attempt to meet their assigned expenditure responsibilities.

The gap between revenue and spending is met through intergovernmental transfers (grants and revenue sharing), borrowing by deficit jurisdictions, or a combination of the two. Designing a solid system of intergovernmental transfers is crucial not only to redistribute resources within a country but also to ensure that limits on borrowing by subnational governments can be set and enforced effectively.

Dividing revenues. Revenue-sharing arrangements of various types are quite common and tend to be geared mainly to correcting, to varying degrees, the vertical imbalances generated by revenue and expenditure assignments. Tax revenue can be shared on a tax-by-tax basis or on the entire pool of central government tax revenues. Tax-by-tax sharing is the norm in many countries, including Argentina, Brazil, Germany, India, and Russia. The problem with this type of arrangement is that it gives the central government an incentive to concentrate its collection and enforcement efforts on the taxes that are either not shared or shared to a lesser degree. Moreover, the central government also has an incentive to concentrate rate increases (for example, for stabilization purposes) on those taxes, and this can distort the tax system. For these reasons, revenue sharing based on the entire pool of central government revenues is preferable.

Revenue-sharing arrangements with coefficients set in law or in the constitution (as in Brazil or Colombia) give some predictability to revenues, which is important for budget planning. However, they impart considerable rigidity to the central government budget. In particular, they can substantially dilute the impact of fiscal tightening. For example, if the central government tries to tighten fiscal policy by raising shared taxes, this move also gives subnational governments more spending money. This outcome could be avoided if the portion of revenue going to the subnational governments was levied at a constant rate on the shared tax base. Fixed revenue-sharing arrangements can also have procyclical effects as tax revenue automatically rises during a boom, thus increasing the spending capacity of the subnational governments, while declining revenue during economic downturns forces them to cut back spending abruptly. To address this problem, some element of flexibility could be introduced in these sharing arrangements—for example, by relating the transfers to a moving average of central government revenues or by requiring sub-national governments to build up revenue stabilization funds to even out cyclical fluctuations in shared taxes.

The distribution of shared revenues among subnational jurisdictions is often made on a derivation basis, with each jurisdiction getting the same share of the revenue collected in its territory. But this does not address the problem of horizontal imbalances. To even out income among jurisdictions, some countries utilize formulas based on redistributive criteria. For example, in Germany, shared revenue from the value-added tax is apportioned on a per capita basis, which entails a degree of redistribution to the less affluent states. India utilizes formulas that combine population, income per capita, indicators of backwardness, and the state’s own tax effort.

Grants. Besides revenue sharing, the main mechanism for intergovernmental transfers are grants from higher (federal or state) to lower (state or local, respectively) levels. There are general-purpose grants—unconditional transfers, generally aimed at addressing horizontal imbalances—and specific-purpose grants that carry conditions regarding the use of the funds and/or the performance to be achieved in the programs financed through them. Block grants fall between these two categories. They are earmarked to finance broad areas of expenditure, such as education, rather than specific programs.

The choice between conditional and unconditional transfers should reflect a number of considerations. On the one hand, imposing conditions limits the autonomy of subnational governments, partly negating the arguments for decentralization. However, imposing conditions may be justified by distributional considerations, such as the need to ensure minimum nationwide standards for the provision of services of national concern, such as primary education, health care, or pollution control. Whether or not they are desirable in theory, the design and enforcement of appropriate conditions for grants are not easy in practice, and controls on the use of grants often end up being more formal than substantive. It is even more difficult to specify and enforce conditionality on the performance of the programs supported by the grants.

Within conditional grants, the choice of whether or not to impose matching requirements also has to take into account various considerations. Matching requirements may induce a redirection of resources of subnational governments to areas of spending considered priorities by the central government, but obviously at a cost for the local provision of other services. Also, matching requirements may place poorer, resource-constrained regions at a disadvantage vis-à-vis richer ones in the utilization of federal grants. Finally, budgetary and, more broadly, macroeconomic management considerations argue against open-ended grants.

Borrowing

Countries generally adopt one of four approaches to controlling subnational borrowing: (1) sole or primary reliance on market discipline, (2) cooperation by different levels of government in the design and implementation of debt controls, (3) rules-based controls, and (4) administrative controls. Examples of the first model can be found in Brazil (until recently) and in Canada. Notable examples of the second model are Australia, Germany, and the Scandinavian countries, although Germany’s subnational governments must also follow the “golden rule,” which prohibits borrowing to finance current expenditure. Other types of rules, prevailing in different countries, stipulate limits to the absolute level of subnational governments’ indebtedness; allow new borrowing up to a level of debt consistent with a maximum debt-service ratio; or ban or restrict certain types of borrowing that involve greater macroeconomic risks, such as borrowing from the central bank or from abroad. Examples of a rules-based approach can be found in, among other countries, Spain, Switzerland, and the United States.

In a number of countries, the central government has direct administrative control over borrowing by subnational governments. This control may take a variety of forms, including setting annual (or more frequent) limits on the overall debt of individual subnational jurisdictions (or on some of its components, such as external borrowing); reviewing and authorizing individual borrowing operations (including their terms and conditions); and/or centralizing all government borrowing with on-lending to subnational governments for approved purposes (generally investment projects). Control generally encompasses not only the ex ante authorization of proposed borrowing but also ex post monitoring. Direct central government controls are, of course, more common in unitary states (such as France, Japan, Korea, and the United Kingdom) than in federations. One example of the latter is India, where federal government approval is required for borrowing by the states if they have outstanding debt to the federal government—as is currently the case for virtually all the states.

Each of these systems has advantages and disadvantages, the balance of which makes it more or less suitable to a particular country’s circumstances. Moreover, as these circumstances evolve—as fiscal and macroeconomic imbalances improve or worsen—this balance may change. Although appealing in principle, sole reliance on market discipline for government borrowing is unlikely to work in many circumstances. This is because one or more of the conditions for its effectiveness frequently are not realized. However, market discipline can be a useful complement to other forms of borrowing controls—it can provide a “reality check” for subnational governments that are trying to evade those controls. In this respect, greater transparency and dissemination of information on the finances of subnational governments are highly desirable, and governments should be encouraged to make any necessary changes in the relevant legal and institutional framework to promote these objectives.

Furthermore, the current decentralization trend seems likely to be in conflict with systems of administrative controls imposed by the central government on subnational borrowing. Rules-based approaches to debt control would appear preferable to administrative controls in terms of transparency and certainty, and are also preferable to statutory limits defined in the context of the annual budget process, which may be unduly influenced by short-term political bargaining. There is, in any case, a clear macroeconomic rationale for barring all levels of government from borrowing from the central bank (or, at a minimum, severely restricting their ability to do this). Borrowing abroad by subnational governments should also be strictly limited, in accordance not only with their debt-servicing capacity but also with macroeconomic (especially monetary and balance of payments) considerations.

“Weak institutions can throw the theoretical efficiency gains from decentralization out the window in practice”

In principle, a good case can be made for limiting all borrowing to investment purposes. However, the so-called golden rule may not be sufficiently restrictive in countries that need to generate government savings to finance at least a part of public investment. Moreover, it may not be desirable to allow government borrowing to finance investments that do not have adequate rates of return. In practice, it may be difficult to prevent governments from getting around the golden rule by labeling certain current expenditures as investments.

These considerations argue for setting the global limits on the debt of individual subnational jurisdictions on the basis of criteria that mimic market discipline. Even under rules-based approaches, there is scope for increased cooperation for all levels of government in containing (or reversing, if needed) the growth of the public debt.

Conclusion

The design of intergovernmental fiscal relations is influenced by political, social, and cultural factors, as well as by economic considerations. Within the narrower economic context, this design reflects a balance among different (and not always easily reconcilable) objectives, namely the efficient allocation of government resources, income redistribution, and macroeconomic management. The balance of these objectives and its evolution over time also tend to reflect a country’s social and political history, current conditions, and the presence or absence of serious macroeconomic imbalances.

It must be recognized that a high degree of decentralization may come into conflict with distributional objectives. This is especially the case, of course, in countries characterized by large regional income disparities. In these cases, a system of equalization-oriented vertical transfers from the center (as in Australia), or a horizontal redistribution mechanism (as in Germany), are likely to be necessary to preserve economic and social cohesion. It is important that such mechanisms be designed, however, so as not to discourage tax effort and cost effectiveness by the sub-national governments.

Substantial decentralization is also likely to make it more difficult for the central government to carry out macroeconomic stabilization through budgetary policies. Therefore, decentralization should progress more slowly in countries experiencing acute fiscal or macroeconomic imbalances. In these countries, it is especially important that a hard budget constraint be imposed on the subnational governments, through a design of intergovernmental fiscal relations that ensures for the subnational jurisdictions an adequate ex ante balance between expenditure responsibilities and their own revenues plus clearly defined transfers from the center, and that bars them from borrowing.

In countries that do not face serious macroeconomic or fiscal imbalances, it should be recognized that substantial decentralization of revenues and expenditures requires the central government to involve more actively the subnational governments, especially states and large municipalities, in macroeconomic management and makes them co-responsible for the achievement of shared economic objectives.

This article draws on material to be published in Fiscal Federalism in Theory and Practice by Teresa Ter-Minassian (IMF, forthcoming).

Teresa Ter-Minassian, an Italian national, is Deputy Director of the IMF’s Western Hemisphere Department.

Other Resources Citing This Publication