Chapter

3 Tax Assignment

Editor(s):
Teresa Ter-Minassian
Published Date:
September 1997
Share
  • ShareShare
Show Summary Details
Author(s)
John Norregaard

A key issue in the literature on fiscal federalism is the question of how subnational authorities might best be financed. This complex issue has no easy solutions, given the wide variety of systems actually applied in different countries and at different times in specific countries. Although there is no ideal system of financing state or regional and local governments, because every country faces different problems and different perspectives, some basic objectives may provide broad guidelines on how tax assignment can best be carried out. Tax assignment can hardly be looked at in isolation. It is an issue intimately related to the question of expenditure assignments across different levels of government, which was discussed in detail in Chapter 2. Even a carefully designed system of intergovernmental expenditure allocation will not work satisfactorily unless it is supported by an equally well-thought-out financing system, and vice versa.

This chapter focuses on the questions to be addressed when decisions are being made on tax assignment among different levels of government. The term “tax assignment” here describes the level of government responsible for determining the level and rate structure of various taxes, whether their revenue is to be collected or received by that level, or shared with others.

Tax Assignment and Tax Sharing

The general principles of decentralization must guide the assignment of taxes to different levels of government. According to these principles, as laid out in the traditional local finance literature, regional and local governments should ideally fulfill mainly allocational functions by providing services that accrue primarily to the local population, services whose costs the local constituency bears as far as possible. In the same vein, because of the degree of openness of local economies, the literature on fiscal federalism argues in favor of limiting regional and local government roles in economic stabilization, as well as in distributional policies.

In very broad terms, the assignment of funds to local jurisdictions may in principle follow one of three options. The first, and probably least attractive, option assigns all tax bases to local jurisdictions and then requires them to transfer upward part of the revenue to allow the national government to meet its spending responsibilities. As this option may hinder effective income redistribution across the national territory, as well as the effectiveness of fiscal stabilization, it may not represent the most efficient way of raising public resources and may provide inadequate incentives for the local jurisdictions to participate in the financing of the national economy. This system resembles that previously in force in the former Yugoslavia and is somewhat similar to the system of negotiated tax-sharing previously practiced in Russia. The system previously in force in China, generally recognized to have inhibited the government’s ability to pursue stabilization policies, had analogous features to this extreme model.

A second option, on the other extreme, is to assign all taxing powers to the center, and then finance subcentral governments by grants or other transfers, either by sharing total revenue or by sharing specific taxes. The main disadvantage of this option is that it completely breaks the nexus between the level of tax revenue collected and the decisions to spend that revenue, which constitute the basic prerequisite for a multilevel governmental system that enhances efficiency. Without this connection, the risk is that fiscal illusion will lead to overprovision of local government services. Also, because of the risk of frequent, discretionary cuts in transfers to local levels of governments, this system could also make it difficult to establish a stable system of service provision at the local level. This kind of system bears some resemblance to that once applied in the former Soviet Union and in Hungary. Substantial grant financing of local governments is still practiced in a number of industrial countries, such as France, Italy, and the Netherlands.

The third broad option is the more normal one of assigning some taxing power to the local jurisdictions, if necessary (that is, if vertical imbalances persist) complementing the revenue raised locally with tax-sharing arrangements or other transfers from the central government. This option leads directly to the question of which taxes should be assigned to local jurisdictions and which taxes should remain the responsibility of the national government (the tax assignment problem). By assigning taxes and thus letting the local jurisdictions bear the tax burden at the margin associated with expenditure decisions, the budgetary actions of local governments will be guided by tax-benefit considerations and will in this way improve economic efficiency.1

The tax assignment problem is typically not an either/or problem with a specific tax placed clearly and solely under the responsibility of either the local, the state, or the central government: rather, in reality (for most taxes) a spectrum of different designs exists ranging from full and complete local autonomy to systems with some local discretion and to others with no local autonomy whatsoever. In other words, even if a specific tax, such as an income tax, has been assigned to the local level because it is found to satisfy the criteria for a “good local tax” (see below), it is possible to design the income tax with varying levels of local revenue autonomy. Table 1 illustrates this important point in a very general way by providing a ranking of different tax designs with respect to the degree of autonomy that they leave with the local governments. For the sake of illustration, the table also includes the main nontax sources of revenues for local governments, although obviously a ranking of this nature can only be broadly indicative.

Table 1.Fiscal Autonomy in Subcentral Governments
Own taxesBase and rate under local control.
Overlapping taxesNationwide tax base, but rates under local control.
Nontax revenuesFees and charges. Generally, the central government specifies where such charges can be levied and the provisions that govern their calculation.
Shared taxesNationwide base and rates, but with a fixed proportion of the tax revenue (on a tax-by-tax basis or on the basis of a “pool” of different tax sources) being allocated to the subcentral government in question, based on (1) the revenue accruing within each jurisdiction (also called the derivation principle) or (2) other criteria, typically population, expenditure needs, and/or tax capacity.
General purpose grantSubcentral government share is fixed by central government (usually with a redistributive element), but the former is free to determine how the grant should be spent; the amounts received by individual authorities may depend on their tax efforts.
Specific grantsThe absolute amount of the grant may be determined by central government or it may be “open-ended” (that is, depend on the expenditure levels decided by lower levels of government), but in either case central government specifies the expenditure programs for which the funds should be spent.

Complete local fiscal autonomy over revenues requires in principle that local governments can change tax rates and set the tax bases. In many countries, however, the central government either defines local tax bases or sets relatively narrow limits to the capacity of local governments to influence the tax base. In some countries (for example, Norway), the central government also sets out limits to the possible variation of local government tax rates.

Taxes assigned to lower levels of government may take the form of own taxes (sometimes referred to as tax separation systems), defined as taxes accruing solely to lower levels of governments, which can determine the rate and, in some cases, also have some autonomy to influence the tax base. An alternative system is represented by overlapping taxes (sometimes called piggybacking systems of local taxation) with the same (or almost the same) tax base for the different levels of government, but with the right of each level of government to set its own tax rate on that common base. This is the system of personal income taxes applied in, for example, the Nordic countries. In Canada, the income tax system used by the provinces involves levying the tax as a percentage of the federal tax revenue accruing within each province. As opposed to tax separation systems, a system of overlapping taxes may involve administrative advantages with regard to assessing the base and to tax collection. This, however, may be at a potential cost of reduced transparency as the tax levied at each level of government may be less easily identifiable for the taxpayers.

Some, in particular federal, countries prescribe in their constitution the system of tax assignment to be applied. Thus, in India, the Constitution prevents overlapping tax powers so that one type of tax can be levied by only one level of government. Likewise, the modalities of local government taxing powers are specified in the Constitutions of Nigeria and Brazil. In Switzerland, the federal government is prohibited by law from imposing indirect taxes, whereas in Australia a similar rule applies for the states.

The question of local fiscal autonomy may be considered almost completely independent from the question of who actually administers and collects the tax. The allocation of these tasks should be determined on the basis of where they can be carried out most efficiently, although one consideration may be that local accountability may be encouraged if the tax is assessed and collected locally.

Probably the single most critical issue in the discussion of subcentral fiscal autonomy, when looked at from the tax side, is whether the authorities concerned can determine their own tax rates. It could be argued that the case for local discretion, as far as the tax base is concerned, should be limited, because changing tax base definitions (for example, by allowing local governments to set individually the amount of a basic allowance, to introduce special tax reliefs, or to exempt specific sources of income or groups of taxpayers) could lead to distortions in the allocation of resources across localities, and also could have important redistributional consequences—an area in which local autonomy is generally believed to be unwarranted. If local governments cannot alter their tax rates, they cannot alter the level of their services in accordance with local preferences. In some countries, subcentral authorities rely mainly on taxes whose rates are fixed by the central government (for example, the countries with extensive tax-sharing arrangements, such as Portugal and Germany) or whose rate is subject to a ceiling. (Norway is a special case in this regard in that all local governments apply the ceiling rate of the local income tax.)

The importance attached to a lack of discretion in local tax policy depends mainly on the role subcentral authorities are supposed to play. To the extent that they are seen mainly as agents, implementing the policies laid down by other tiers of government, their limited autonomy with respect to tax policy would not appear to be serious. In contrast, if they are meant to implement their own expenditure programs and independently set their service levels in accordance with local preferences, their inability to determine tax rates and thus the level of their own revenues is a serious problem owing to the potential conflict between expectations, needs, and wishes of the local population, and the actual revenue potential available to local governments.

The main arguments against providing subcentral authorities with extensive fiscal autonomy center on the risk of increasing economic disparities between areas or localities and alleged restraints on central government macroeconomic control. Administrative simplicity or administrative economies of scale are also used as arguments for centralized taxes with a specific proportion of tax revenues being allocated to subordinate levels of government.

In what follows, the more specific aspects of tax assignment are dealt with by addressing the basic question of which taxes can be considered good candidates for state and local tax sources and which cannot. What characterizes a good local tax?

A Good Local Tax

A good local tax adequately supports a decentralized public expenditure system. The literature on fiscal federalism and local government finance2 generally suggests that the following criteria and considerations should form the basis for decisions on which taxes can adequately be assigned to the subcentral level and which should remain at the national level.

  • To the extent that the tax in question is aimed at, and is suitable for, economic stabilization or income redistribution objectives, it should be left to the responsibility of the central government.

  • The base for taxes assigned to the local level should not be very mobile, otherwise taxpayers will relocate from high to low tax areas, and the freedom of local authorities to vary rates will be constrained. For this reason, general consumption taxes are found at subordinate levels of government only where geographical areas are very large (for example, Canada and the United States). Thus, the more mobile a tax base, the greater the presumption to keep it at the national level.

  • Tax bases that are very unevenly distributed among jurisdictions should be left to the central government.

  • Local taxes should be visible, in the sense that it should be clear to local taxpayers what the tax liability is, thereby encouraging local government accountability.

  • It should not be possible to “export” the tax to nonresidents, thereby weakening the link between payment of the tax and services received.

  • Local taxes should be able to raise sufficient revenue to avoid large vertical fiscal imbalances. The yield should ideally be buoyant over time and should not be subject to large fluctuations.

  • Taxes assigned to the local level should be fairly easy to administer or, in other words, the more important economies of scale in tax administration are for a given tax, the stronger the argument for leaving the tax base for that tax to the national level. Economies of scale may depend on data requirements, such as a national taxpayer identification number and computerization.

  • Taxes and user charges based on the benefit principle can be adequately used at all levels of governments, but are particularly suitable for assignment to the local level, inasmuch as the benefits are “internalized” to the local taxpayers.

This set of broad criteria translates into more specific recommendations regarding which taxes should be assigned to different levels of government, that is, which taxes may be considered good local taxes and which should be left in the domain of the central government. It is generally acknowledged in the literature3 that the most obvious candidates as good local taxes are land or property taxes and, to some extent, personal income taxes. With some exceptions, turnover or consumption taxes, as well as taxes on capital income, in particular corporate income taxes, are generally considered less appropriate at the local level and in some cases also at the state level4 because of the mobility of the corresponding tax bases. This broad conclusion derived from principles of local finance seems in very general terms to conform to the financing system actually found in most countries.

The following discussion addresses these questions on a tax-by-tax basis and is intended to cover all the main taxes to which tax assignment is applied in practice (disregarding whether these taxes according to the general principles are considered appropriate at subordinate levels of government or not). The treatment of the different taxes is also intended to be in descending order of importance for subordinate level of governments, although this ordering must necessarily be somewhat subjective (see Tables 2 and 3).5

Table 2.Distribution of Tax Revenue Among Different Levels of Government
Tax Revenue in Percentage of Corresponding General Government Tax
Total taxIncome taxProperty taxDomestic taxes on goods and servicesOther taxes
Country and YearTax as Percentage of GDPCentral governmentState governmentLocal governmentCentral governmentState governmentLocal governmentCentral governmentState governmentLocal governmentCentral governmentState governmentLocal governmentCentral governmentState governmentLocal government
Industrial countries
Federal
Australia (1991)30.679.716.93.5100.00.00.02.457.839.872.527.50.045.254.80.0
Canada (1989)34.950.940.28.963.536.50.00.016.283.840.159.40.460.032.47.6
Germany (1991)141.473.419.76.939.140.820.12.061.236.879.120.80.1100.00.00.0
Spain (1990)34.087.04.88.292.91.26.05.850.545.781.15.713.299.20.00.8
United States (1991)27.765.820.513.881.117.11.76.06.787.316.068.315.797.82.20.0
Unitary
Belgium (1990)145.695.6n.a.4.490.9n.a.9.1100.0n.a.0.097.0n.a.3.099.1n.a.0.9
France (1991)143.190.5n.a.9.5100.0n.a.0.0100.0n.a.0.0100.0n.a.0.081.9n.a.18.1
Netherlands (1991)148.597.3n.a.2.7100.0n.a.0.065.1n.a.34.9100.0n.a.0.096.3n.a.3.7
Norway (1990)45.278.8n.a.21.247.6n.a.52.437.9n.a.62.199.6n.a.0.497.5n.a.2.5
Sweden (1991)54.269.6n.a.30.424.7n.a.75.3100.0n.a.0.0100.0n.a.0.0100.0n.a.0.0
United Kingdom (1991)135.896.0n.a.4.0100.0n.a.0.099.2n.a.0.8100.0n.a.0.081.6n.a.18.4
Developing countries
Federal
India (1990)216.565.834.20.0100.00.00.033.766.30.049.250.80.088.911.10.0
Argentina (1989)214.660.439.60.034.265.80.049.250.80.085.814.20.061.039.00.0
Brazil (1991)24.565.030.94.1100.00.00.02.240.557.337.957.84.394.25.30.5
Mexico (1987)17.885.511.62.998.21.30.61.20.098.899.80.10.15.877.017.2
Unitary
Hungary (1990)48.692.4n.a.7.671.9n.a.28.1100.0n.a.0.0100.0n.a.0.0100.0n.a.0.0
Poland (1988)44.678.7n.a.21.375.9n.a.24.148.3n.a.51.785.8n.a.14.278.5n.a.21.5
Israel (1990)34.793.1n.a.6.9100.0n.a.0.012.3n.a.87.7100.0n.a.0.098.3n.a.1.7
Thailand (1990)19.795.6n.a.4.4100.0n.a.0.081.9n.a.18.192.1n.a.7.9100.0n.a.0.0
Chile (1988)20.596.2n.a.3.8100.0n.a.0.019.7n.a.80.396.5n.a.3.5100.0n.a.0.0
Kenya (1991)23.398.3n.a.1.7100.0n.a.0.00.0n.a.100.099.5n.a.0.5100.0n.a.0.0
South Africa (1990)27.494.51.24.3100.0n.a.0.025.5n.a.74.596.73.30.0100.0n.a.0.0
Zimbabwe (1986)31.396.4n.a.3.6100.0n.a.0.011.5n.a.88.598.2n.a.1.898.7n.a.1.3
Source: International Monetary Fund, Government Finance Statistics Yearbook.Note: n.a. = not applicable.

Includes supernational authorities’ share of general government total tax revenue for Belgium (1.5 percent), France (0.7 percent), Germany (0.9 percent), the Netherlands (1.4 percent), and the United Kingdom (1.2 percent).

Data for general government do not include local government.

Source: International Monetary Fund, Government Finance Statistics Yearbook.Note: n.a. = not applicable.

Includes supernational authorities’ share of general government total tax revenue for Belgium (1.5 percent), France (0.7 percent), Germany (0.9 percent), the Netherlands (1.4 percent), and the United Kingdom (1.2 percent).

Data for general government do not include local government.

Table 3.Distribution of Different Taxes Within Different Levels of Government(In percent)
Central GovernmentState GovernmentLocal Government
Country and YearIncome taxProperty taxDomestic taxes on goods and servicesOther taxesIncome taxProperty taxDomestic taxes on goods and servicesOther taxesGrants as percentage of taxes plus grantsIncome taxProperty taxDomestic taxes on goods and servicesOther taxesGrants as percentage of taxes plus grants
Industrial countries
Federal
Australia (1991)71.90.322.85.10.030.140.928.958.80.0100.00.00.029.9
Canada (1989)59.30.021.319.443.13.639.913.321.90.084.61.414.053.7
Germany1 (1991)16.10.131.552.462.56.630.90.019.288.211.40.40.044.7
Spain (1990)33.20.424.741.77.660.931.30.277.022.631.142.73.645.1
United States (1991)55.11.14.139.737.43.955.82.928.95.675.319.10.047.6
Unitary2
Belgium1 (1990)35.12.725.636.775.80.017.27.062.1
France1 (1991)19.12.830.647.50.00.00.0100.044.7
Netherlands1 (1991)33.32.323.740.60.043.90.056.190.2
Norway (1990)21.51.444.732.387.78.60.73.045.9
Sweden (1991)14.44.635.545.5100.00.00.00.021.5
United Kingdom1 (1991)39.18.533.918.40.01.60.098.485.8
Developing countries
Federal
India (1990)318.60.544.936.00.01.889.58.748.7
Argentina (1989)36.15.217.171.517.98.24.369.50.0
Brazil (1991)22.50.127.949.50.04.389.85.923.30.045.550.63.977.5
Mexico (1987)26.50.072.51.02.50.00.497.13.34.47.92.385.45.5
Unitary2
Hungary (1990)21.20.137.141.6100.00.00.00.059.4
Poland (1988)32.02.232.033.837.58.719.634.229.8
Israel (1990)42.41.039.117.50.096.00.04.050.0
Thailand (1990)26.23.645.025.10.017.083.00.032.9
Chile (1988)30.70.548.820.00.056.044.00.052.1
Kenya (1991)29.30.053.717.00.084.215.30.50.0
South Africa (1990)54.61.636.67.20.00.0100.00.091.50.0100.00.00.037.5
Zimbabwe (1986)47.60.432.819.20.077.415.76.980.8
Source: International Monetary Fund, Government Finance Statistics Yearbook.

Includes supernational authorities’ share of general government total tax revenue for Belgium (1.5 percent), France (0.7 percent), Germany (0.9 percent), the Netherlands (1.4 percent), and the United Kingdom (1.2 percent).

There are no state governments in unitary countries.

No data on local governments are available.

Source: International Monetary Fund, Government Finance Statistics Yearbook.

Includes supernational authorities’ share of general government total tax revenue for Belgium (1.5 percent), France (0.7 percent), Germany (0.9 percent), the Netherlands (1.4 percent), and the United Kingdom (1.2 percent).

There are no state governments in unitary countries.

No data on local governments are available.

Property Taxes

Property taxes, including in particular land taxes, have historically been widely used as subcentral taxes without any special regard to their alleged incidence. This is the outcome of the perceived advantages of the property tax as a local tax. With a property tax it is always clear which authority is entitled to the revenue it yields, which is not always so for income taxes and other taxes. Administration costs are generally found to be lower for a property tax (provided that there is a registry of properties with updated values) than for an income tax, which requires complex tax returns. The yield of a property tax can be predicted more accurately than for an income tax or a profits tax. Finally, some of the tax will be levied on businesses, which seems reasonable to the extent that businesses derive benefits from subcentral services, such as roads and other infrastructure services.

An additional argument for the use of property taxes is that, while almost all residents pay directly or indirectly (through rents) the property tax, thus avoiding free-rider problems in local service provision, this is not always the case for local income taxes. It has also been argued that property taxes are guided by the benefit principle of taxation to the extent that the corresponding spending by local governments benefits local properties by increasing their value. Against this view, it could be held that, although land and existing structures and thus the tax base cannot move in a physical sense, the tax base can do so in a fiscal sense via the capitalization of property taxes to the extent that property taxes are not used for purposes viewed as beneficial to property owners.

The main disadvantage of property taxes lies in the fact that they almost universally realize lower amounts than needed. There are many reasons for this, including the fact that it is a very visible tax (and thus politically unpopular), that it is perceived to have unwanted distributional consequences to the extent that the tax is borne by renters and not by owners of property, and that there are problems associated with the measurement of the tax base, including in particular the “correct” valuation of property, and its updating.

Some countries prefer to distinguish between residential property and commercial and industrial property, with the former being assigned to local taxation, and the latter either to local taxation with a uniform rate or to national taxation only (as is the case in some Nordic countries). In this regard, a particularly contentious issue in many countries (whether industrial, developing, or in transition) has been the taxation of agricultural land. In countries with a general income tax (including income from agriculture), a tax on land could be seen as a discriminatory surcharge on a basic factor input in one sector of the economy, rather than a local benefit tax. In countries without an income tax on agriculture, it has been argued that a land tax on agriculture impedes the development of this important foreign exchange earning sector. Whatever the merits of these arguments may be, the relatively modest tax burden on agriculture found in most countries (which is generally independent of the level of development of the countries in question) seems to reflect the political influence of this sector rather than economic principles or sound fiscal policies.

Other countries apply alternative criteria for the assignment of property taxes to different levels of government. In Brazil, for example, urban property is taxed at the municipal level, while the federal government levies and administers the tax on rural property.

More specifically, at least four important issues relate to the definition and measurement of the base upon which property taxes are levied: the coverage of the base, the use of capital or rental values, the number and nature of exemptions, and the frequency and methods of updating property values. The main issue regarding coverage has been whether land, improvements to land, and buildings should all be subject to tax. The systems applied vary substantially between countries, although most of the countries for which information is available include the unimproved value of land, the value of land improvements, and usually also the value of buildings. The efficiency and equity implications of property taxes have been intensively debated in the literature and will not be pursued further here (see McLure (1977) for an overview of the issues).

In principle, the impact of using rental values or capital values should be the same, assuming well-functioning property and capital markets. It has been argued, however, that there may be major differences in the actual outcome to the extent that rental values reflect mainly the current use of the property, while capital values are said to reflect the value of the property in the best alternative use. Also in this regard, actual methods vary between countries. Capital values are generally based upon market values, although some countries apply corrections to these market values (for example, use a specific proportion of the market value).

Most countries apply a large number of different reliefs under the property tax, for example, in the form of exemption of government property, highways, railways, and other transport or communication facilities, and mining, agriculture, and forestry industries. The subsidies implicit in this kind of treatment, not least with respect to agricultural land, have been increasingly criticized in a number of countries. As indicated above, many countries apply different tax treatment to residential and business property, with residential property usually subject to a more favorable treatment.

A particularly contentious problem in a number of countries has been the frequency and method used to update property values. Thus, in most developing countries, assessment of property values and updating seem to be the major issues. The unpopularity of this type of tax may in some countries be associated with infrequent updates of values, leading to large and abrupt increases in tax liabilities when updating actually takes place. Although property valuations are generally based on market prices, problems are also encountered during certain periods and in areas with modest turnover of property. State and central governments usually perform the valuation of property in order to achieve the necessary coordination between different areas, but the way in which and frequency with which it is done vary substantially across countries.

Although most of the revenue from property taxes generally accrues to subcentral levels of government, state or local governments do not always have complete discretion over the base or the rate. Central governments typically set the rules governing valuations and their frequency and determine exemptions and other reliefs. Also, the central government may impose restrictions on the variations in property tax rates. In practice, local government discretion may be limited in other ways, for example, in the form of earmarking of property revenues, or if higher rates adversely affect grants (as in the United Kingdom before 1989). In Italy, the central government sets a minimum rate for the property tax. If a municipality does not apply the floor rate, transfers to it from the central government are supposed to be reduced correspondingly. Thus, although most of the revenue from property taxes primarily accrues to subcentral authorities in most countries, the respective central governments are generally heavily involved in formulating and administering the provisions of the taxes.

Personal Income Taxes

Most countries assign all or a large proportion of personal income taxes to the central government. Exceptions include the Scandinavian countries, Switzerland, the Baltic countries, Russia, and the other countries of the former Soviet Union. Generally, there are advantages as well as disadvantages of using personal income taxes at the subcentral level. Among the advantages is the fact that personal income taxes generally are buoyant and thus capable of raising the necessary revenue, and in addition they are believed not to fall on businesses, thereby avoiding the risk of subcentral authorities, anxious to attract new industry, indulging in tax-cutting competitions with adverse effects on services provided.

One of the main disadvantages of a local income tax as the main revenue raiser is the fact that, depending on the level of the tax threshold, many people may not pay the tax, although they receive local services.6 This could have an adverse impact on the way a decentralized system works and has been used as an argument for supplementing an income tax with other tax sources, thereby including the majority of the local constituency in the local tax net. In this regard, two schools of thought may be distinguished. First, many countries (including, for example, most Mediterranean countries and Austria) seem to place considerable weight on income redistribution and on making income tax systems easy to administer by setting a high tax threshold, thereby excluding a large proportion of the population from the tax net. In contrast, other countries (such as New Zealand, Switzerland, and the Scandinavian countries) generally put more emphasis on the inclusion of most of the population in the tax net by setting relatively low tax thresholds, so that more people share the cost of public services.

In the context of financing local governments, there seems to be a case for making a distinction between schedular and global income taxes, since schedular income taxes can in some cases be used by local jurisdictions without great difficulties, in particular if the taxes on, for example, interest income, dividend income, and wages and salaries are withheld at source and constitute the final tax paid. However, the more developed is a country, the higher is the likelihood that individuals receive income from different sources, and furthermore that these incomes are derived from different jurisdictions. This may move countries to prefer a global income tax system in which the different income sources are added together for each individual and the tax liability is adjusted according to individual circumstances.7 For such a system to work well at the local level, it requires flows of information on personal income received from other jurisdictions and thus poses the risk of tax evasion. Against this background, it may be better to leave a global income tax base with the national government, which is in a better position to acquire the necessary information.

However, such a system can be combined with revenue sharing, such as is the case in India, where the states receive about 85 percent of total income tax revenue, allocated on the basis of population, tax effort, and a measure of backwardness. In Brazil, 44 percent of the income tax revenue is transferred to lower level governments under a tax-sharing arrangement, and in Poland, in 1992, 15 percent of the personal income tax revenue was shared with local governments. As part of a recent reform of intergovernmental fiscal relations, shared personal income taxes have also been introduced in Hungary (in 1991, 50 percent of the revenue accrued to local governments). Similar tax-sharing arrangements were also important elements in the financing reform in China in 1980 (under the present financing arrangements, local governments receive all of the yields from personal income taxes).

Notwithstanding these considerations, and to the extent that the administrative capabilities are present at the national level, there is a fairly easy and cost-effective way of taxing a global income tax base in local jurisdictions, namely for the local jurisdictions to use the same statutory tax base as for the national income tax (that is, overlapping taxes or piggybacking). This solution, which reduces administrative as well as compliance cost, is actually used in a number of countries (such as, for example, the Nordic countries and Canada, where the provincial tax is levied as a percentage of the federal tax).8 However, although it introduces an additional complexity to the tax and thus offsets at least in part some of the administrative savings, some of the countries applying this system (for example, the United States) also use specific tax reliefs in their state and local tax systems. Thus, the extent to which countries using overlapping income taxes coordinate the taxes levied at different levels varies considerably. In the Scandinavian countries and Canada, for example, there is a high degree of coordination, while coordination is lacking in Switzerland and the United States.

Generally, because the system requires a fairly advanced administrative system with up-to-date recording of taxpayers’ residence, overlapping personal income taxes are generally seen only in developed countries. Combined with an efficient equalization system, such a system is seen, in the countries that apply it, to ensure that variations in tax rates across jurisdictions reflect similar differences in locally determined service levels. Even in industrial countries, however, the administrative recording requirements have been used as an argument against the workability of local income taxes (which, for example, is the case in the United Kingdom).

A special case of overlapping personal income taxes (or partly overlapping income taxes, if some differences in tax bases are allowed) arises when local income taxes, as in the United States, are deductible from federal income tax liability (deductibility is not applied in the majority of countries using overlapping personal income tax systems). The rationale of such a system is the protection it provides for the taxpayer against excessive aggregate marginal tax rates as a result of high local income taxes. An unwarranted side effect may, however, be the incentive for local governments to expand their expenditures, partly financed—at the margin—by nonresidents. It may also reduce the overall level of progressivity of the tax system.

Taxes on income deriving from the activities of small business establishments or from agriculture may often be imposed as efficiently by local governments as national governments, and in some cases local governments may even possess more information than national governments. However, since record keeping by small establishments is often modest or even absent, taxation of such business income has in many cases to rely on presumptive income, based for example on gross sales, on the floor space in which the activity takes place, or on other criteria (for example, in Hungary, the local business tax is levied on the gross turnover of businesses at a maximum rate of 0.3 percent). Taxes on income from small businesses, from self-employed, and from agriculture, with the revenues accruing mostly or solely to local governments, are well known in a number of Central and Eastern European economies in transition, including Poland and Romania, as well as in a number of developing countries, such as India.

Notwithstanding which level of government actually receives the revenue of personal income taxes, practice differs substantially across countries with regard to which level is responsible for the assessment and for the collection of the income taxes at the subordinate level. National or central government responsibility, or—at the most—state responsibility, seems, however, to be the main rule owing principally to the economies of scale involved in the administration of these taxes.

Sales Taxes

The popularity of assigning property taxes—and to some degree also income taxes—to subordinate levels of government is attributable in part to the fact that, with these taxes, differences in tax rates between areas are unlikely to cause serious problems owing to the relative immobility of the tax bases. In contrast, different sales tax rates between different jurisdictions can drive consumers (or rather their purchases) away from high tax areas, as is perhaps best reflected in the serious cross-border trade problems between countries with different tax systems and tax levels (such as between Canada and the United States, and between Ireland and the United Kingdom). A distinction must be made, however, between single-stage sales taxes, such as excises and retail taxes, and multistage sales taxes, such as turnover taxes and value-added taxes (VATs).

Retail sales taxes and excises levied on the final sale to the consumer can be given to local jurisdictions as a revenue source, provided that they do not levy these taxes with highly different tax rates. If they do, citizens will be encouraged to shop in other jurisdictions. The main factors determining the extent to which this will take place are the vicinity of other jurisdictions, the cost of travel, and the value of the goods purchased.9 Another constraining factor for the use of such taxes at the local level with anything but a modest level of tax rates is the risk of tax evasion, which may be relatively more serious for these (single-stage) sales taxes, especially under high tax rates. However, the existence of, for example, both state and municipal sales taxes in many countries must reflect the fact that these caveats are not universally perceived as serious. Thus, in India, the main revenue source of the states is the sales tax. Turnover taxes and some excises are also important provincial revenue sources in Argentina.

A case can be made for distinguishing between excises on goods, which generally should be assigned to the central level to minimize tax exporting, and excises on selected services, consumed locally, and thus much less prone to tax exporting. Some countries, such as India, assign selected excises to the central government (combined with a tax-sharing scheme), and other excises to state and local governments. A number of countries use local excises or special taxes on automobiles or on fuels, which could be regarded as benefit taxes associated with the costs to local governments of maintaining roads. Municipalities in Brazil are allowed to levy a 3 percent tax on retail sales of fuels and gas. In Poland, own sources of revenue for local governments include a tax on automobiles.

Some countries combine earmarking sales taxes with tax assignment to different levels of government. In Russia, for example, a system of regional and federal road funds is in place, financed in part by excises on fuel and on vehicles, supplemented by taxes on registration and ownership of vehicles.

Sales taxes levied at the manufacturing level should, as a general rule, be assigned to the upper tier of government and to subordinate levels of government only where geographical areas are large.

There seems to be broad albeit not universal consensus in the literature that VATs are most appropriately assigned to the central level of government. This dictum rests on the fairly extensive administrative capabilities required to operate the tax (a requirement that is generally best met by central governments) in combination with the need to make the VAT neutral with respect to the spatial allocation of production and consumption, implying that—generally—the VAT should conform to the destination principle.10 Implementation of this principle requires, however, border control between jurisdictions if the tax is to be levied by individual provinces or states. This would in most countries be neither feasible nor desirable because of the administrative costs implied and because of the impediments to the free flow of goods and services it would create. In addition, a subnational VAT system would pose problems with regard to which provinces or states should receive the revenues from VAT on imports, and which should bear the burden of VAT refunds on exports.11 Following this kind of reasoning, comprehensive VATs should be left solely with the national government, as is, in fact, the case in most countries. In some countries (for example, China, Germany, and Russia), central VAT revenues may be shared with subnational levels, although this raises the same kind of problems referred to above, if the tax sharing is based on the derivation principle.

Similar considerations on different aspects of VAT design constitute important elements in the ongoing tax reform discussion in India, which contemplates introducing a comprehensive VAT to replace existing excises and sales taxes, with the aim of sharing the revenue between the three levels of government. However, one of the main questions is whether such a system could function properly without fundamental changes in the present system of intergovernmental fiscal relations in India. According to Bird (1993), it could prove difficult to establish consensus on a formula distributing the VAT proceeds in a context of sharp regional inequalities as the one currently prevailing in that country. Bird also questions the rationale behind sharing the proceeds of any particular tax, because it would seem doubtful that the central government would go through the pain of increasing tax revenues that will accrue in large part to other governments. A more satisfactory alternative—according to Bird—would be to share with the states a fixed share of aggregate central tax revenues.

Brazil offers an example of a VAT assignment system that is generally believed to have had detrimental effects on economic performance. All three levels of government in Brazil are assigned taxing powers on consumption, but with different tax systems, and with the tax covering the widest base, the VAT-type ICMS assigned to state governments and not to the federal government. Furthermore, a large fraction of the federal government consumption tax (the IPI) is transferred to lower levels of governments under a tax-sharing arrangement. This particular design is believed to encourage tax competition between entities of government and to foster tax evasion, which is, furthermore, exacerbated by a large number of different tax rates and exemptions (see Chapter 18 for details).

Corporate Profit Taxes

There seems to be almost universal agreement that the taxation of larger businesses, and in particular corporate profit taxes, should be left to the national level and to provinces or states only where these are very large (as in Canada). This reflects the fact that the economic activities of corporations are typically much more diversified and complex, with factor inputs originating from a number of jurisdictions (and possibly also from abroad), and with sales similarly going to a multitude of jurisdictions. Depending on the nature of the specific markets in question, local taxes on corporate profits would to a large degree be exported or shifted to other jurisdictions in a nontransparent way, thus rendering the associated tax burden almost imperceptible to local citizens. In addition, a high local-tax rate may lead the business entity in question to move the tax base to other jurisdictions, either by physically moving the corporation or by adjusting the internal transfer pricing arrangements.

Leaving the taxation of corporate profits entirely in the hands of local governments would thus create serious informational problems because of the administrative issues associated with the allocation of taxable profits between different jurisdictions in cases of enterprises with economic activity spread over many localities. But also in this case these problems could, at least in part, be overcome by some form of overlapping tax bases between the national and the local level (piggybacking), although the room for tax-rate variations is much smaller for the corporate profits tax than for personal income taxes. In Canada, the base is harmonized to a considerable degree between the provinces and the federal level (although provinces do have the possibility of providing individual investment incentives), while provinces have the flexibility to vary rates. In Brazil, the states can levy a 5 percent surcharge on the corporate income tax.

Some countries have, with the above-mentioned problems in mind, chosen instead to allocate a fixed portion of the profit tax revenue originating within each jurisdiction to the local governments under a tax-sharing arrangement. Thus, Russia allocates 25 percentage points of the 38 percent tax rate on corporate profits to regional governments, while the remaining 13 percentage points remain with the federation.12 Of the corporate tax revenue in Poland in 1992, 5 percent was shared with local governments. In Nigeria, a special system is in force according to which the federal authority has the legal jurisdiction over the company tax, but the states nevertheless collect the tax and retain the proceeds.

Payroll Taxes

Like VATs and corporate profit taxes, different types of payroll taxes are also generally seen as an appropriate revenue source for the central government only, because different payroll tax rates could drive employers, and jobs, away from high tax areas. In addition, tax exporting is probably significant in the sense that, first, part of the tax may be shifted to prices and thus borne by consumers outside the jurisdiction that receives the revenue, and, second, the tax may be levied on employees with residence outside the revenue-receiving jurisdiction. Thus, the tax may not be visible to the local taxpaying constituency, and the relationship may be weak between tax payments and services provided by jurisdictions.13

Notwithstanding these general considerations, tax-sharing arrangements for payroll taxes actually exist in a few countries. A relatively small payroll tax is also levied by the states in Australia.

Natural Resource Taxes

Taxes on natural resources are generally perceived as poor candidates for local taxation, since normally the base of these taxes is very unevenly distributed across jurisdictions. In addition, extraction of economic rent from natural resources could be held to be a national prerogative, which should benefit the whole of the nation and not just selected fortunate regions. The taxes in question are also in many cases characterized by a high level of revenue volatility owing to price fluctuations. The associated uncertainty, it could be argued, should be absorbed by the central government, which generally has a number of alternative revenue sources at its disposition, and not by regional or local governments, which are meant primarily to conduct allocative functions (price fluctuations on oil, for example, have created sharp swings in the revenues of states in Nigeria). These theoretical considerations, however, do not take into account the important fact that, in practice, cultural and ethnic differences may be the reason for strong pressures toward regional independence, including regional control over natural resources, as is seen, for example, in Russia.

Alternatively, it could be held that, at least in part, these taxes should be considered as benefit taxes, that is, as payments for the benefits deriving from the provision by local or regional governments of the necessary infrastructure investment without which either exploitation of the natural resources would not be possible, or the return to the investments required could be significantly reduced. In other cases, the taxes may be considered as compensation for the environmental costs associated with the exploitation of natural resources. This might also constitute part of the reason why a number of countries actually operate tax-sharing schemes for natural resource taxation (see country chapters for details). In Russia, local governments in regions rich in natural resources benefit from the retention of a high share of these taxes. Previously, in Nigeria, all taxes accruing from oil production went to the states. In Argentina, a revenue-sharing scheme is in place for royalties on mineral extraction.

Import and Export Taxes

Import and export taxes, apart from being generally considered inferior to the taxes dealt with above, should always be imposed by the national government to reduce the possibility of introducing major distortions within the country through differential foreign trade taxes imposed by different jurisdictions. In fact, the large majority of countries assign import duties exclusively to the central government (Nigeria being one exception, with import tax revenue being shared). Nevertheless, in some countries, such as Russia, the formula for sharing important export tax receipts with regions from which the exports originate remains an important tax policy issue (because of the nature of the exports in question, these taxes may as well be considered special cases of taxes on natural resources). India operates a special tax on interstate sales with a maximum rate of 4 percent, and with a number of exemptions (see Chapter 21 for details).

Benefit Taxes and User Charges

In addition to what has been said above about specific sources of taxation, it is generally held that benefit taxes, license fees, and user charges should all be used to the maximum extent feasible at the local level because they are transparent, they minimize the risk of tax exporting, they generally do not involve problems of vertical or horizontal equity, and they increase economic efficiency. Although these charges are significant sources of revenue for the localities, they are generally modest compared with some of the taxes considered above.

Tax Assignment in Practice

A striking feature of the financing of subcentral levels of government is the significant variation in the level and composition of local government taxation across countries. This feature is illustrated in Tables 2 and 3, which for a fairly limited sample of countries show the attribution of total tax revenues to subsectors of general government as percentage of total tax revenue, in federal as well as in unitary countries, and the composition of the tax revenue for each subcentral level of government with respect to different types of taxes (including revenues from the tax-sharing arrangements). These tables by their nature do not indicate the actual degree of state, provincial, or local autonomy over the tax revenues, which, as discussed this chapter, may vary considerably across countries.

Nevertheless, as the tables show, most countries have more than one subcentral tax (although the tables do not distinguish between cases where revenues are solely assigned to the subcentral level and where they are shared under tax-sharing arrangements), and this holds for industrial as well as for developing countries, and for federal as well as for unitary countries. Generally, the personal income tax seems (as expected) to be of greater importance for the subcentral level in industrial countries than in developing countries, although for example in most Anglophone countries the property tax is the dominant tax, especially at the local level (this holds in Australia, Canada, the United States, Ireland, New Zealand, and, until 1990, the United Kingdom).

In some, especially federal, countries, general consumption taxes and in some cases also excises play a considerable role, particularly at the state level (for example, in Austria, Brazil Canada, Germany, India, South Africa, Spain, and the United States). A predominant feature seems to be that these taxes are used by large countries with correspondingly large subcentral areas. Also, in some of these countries, consumption tax systems take the form of tax-sharing arrangements with little or no state or local discretion, as in the case of the Austrian and German VAT.

A common feature not shown in these tables is the dominant use of personal income taxes at the subcentral level as opposed to corporate income taxes, reflecting the fact that corporate income taxes are generally considered unsuitable at the subcentral level owing to the mobility of the tax base.

Although the property tax is among the most popular subcentral taxes, not least in federal industrial countries and unitary developing countries, its revenue measured as a percentage of GDP is generally modest and seldom exceeds about 3 percent. This is probably because it is a highly visible tax, it is hard to evade, there are problems associated with the valuation of property, and it is generally perceived as a regressive tax. For these reasons, the property tax has become increasingly unpopular politically, which may also help to explain why its importance as a revenue source has declined in many countries during the last decade or so.

Concluding Remarks

The theory on fiscal federalism provides some fairly broad guidelines with regard to which taxes can appropriately be assigned to subnational levels of government and which should be kept at the central level. However, although some general patterns in accordance with these guidelines can be identified in country practices, even fairly homogeneous countries at the same level of development have in many cases chosen different solutions to these problems. One of the main reasons for this is that the historical, geographical, ethnic, and constitutional character of each country has profound implications for the range of feasible and efficient tax assignment policies.

Some lessons may nevertheless be drawn from actual country experiences. First is the importance of tax administration: a decentralized fiscal system cannot function satisfactorily without the necessary administrative capabilities at the subcentral level. In other words, the design of tax systems should clearly be adapted to the level and quality of administrative resources that have been found politically appropriate to devote to the subcentral levels of government. Generally, the more complicated the tax in question is made for other reasons (for example, for reasons of revenue or equity), the stronger the argument for placing the tax with a higher or the highest tier of government. As a reflection of this “rule,” more complex systems of taxation are generally assigned to subcentral levels of governments only in more developed countries.

Second, in addition to the crucial question of the choice of tax sources at subordinate levels of government, actual experience indicates that a decentralized system will work satisfactorily only if state, provincial, and local governments are given at least one major own source of revenue, that is, a source of revenue over which they have autonomy to determine the revenue (assuming that this system is supported by adequate equalization of tax capacities and expenditure needs). Only then can a multilevel system of government promote accountability and ultimately economic efficiency.

Finally, there are obvious potential gains as well as risks associated with decentralizing taxing powers. The gains include improved mobilization of revenue sources and the potential efficiency gains alluded to above. The risks take the form of leaving the central government in a more vulnerable position with respect to its ability to conduct effective fiscal policies, especially for stabilization purposes.

This assumes chat there are no substantial externalities associated with the provision of local services, that the tax cannot be shifted to other jurisdictions, and that an efficient equalization scheme is in place.

For general expositions of the principles of fiscal federalism, see Oates (1972) and King (1984).

Unless the areas in question are large as is the case in, for example, Canada and the United States.

As illustrated in Tables 2 and 3, the importance of different tax sources varies considerably across countries. Based on more comprehensive information than that presented here, there seems to be a broad tendency for income taxes at subordinate levels of government to increase in importance with increasing level of development, although there are some exceptions to this rule.

In some countries, such as Finland and Norway, the income tax threshold in the local tax is much lower than in the central government income tax.

However, schedular mechanisms such as withholding or minimum contributions may be widely used under a global system for ease of administration.

This particular feature may increase the revenue elasticity of the subcentral tax compared with a normal flat rate system to the extent that subcentral governments will share the gains of any bracket creep effects in the federal tax.

This disregards the problems posed by mail order systems, particularly with regard to the control and setting of tax rates (in the United States, some of these problems have been addressed by applying the rates of the destination states to mail order sales). Although based on a fairly limited sample of countries, Table 3 seems to indicate that the degree of development is also important in this regard, in that there is a tendency for sales taxes to be of larger revenue importance for local governments in developing than in developed countries.

Which means that the tax is levied by the jurisdiction in which consumption takes place, independent of the origin of the goods (that is, expotts are exempt and imports are liable to tax), as opposed to the origin principle, according to which the VAT is levied by the jurisdiction in which production takes place, that is, interstate exports are taxed and imports are not.

In China and Russia, ail import VAT accrues to the federal government, and only domestic VAT revenues are shared with the regions.

Formally, the 25 percent local rate is a maximum, but the large majority of regions are believed to apply the maximum rate.

In the majority of countries, provision of regional or local government services is related to the residency of individuals.

References

    Other Resources Citing This Publication