Published in R.D. Ebel and J.E. Petersen “The Oxford... Government Finance “, Oxford University Press, Oxford and New York,...
Transcription
Published in R.D. Ebel and J.E. Petersen “The Oxford... Government Finance “, Oxford University Press, Oxford and New York,...
Published in R.D. Ebel and J.E. Petersen “The Oxford Handbook of State and Local Government Finance “, Oxford University Press, Oxford and New York, 2012, pages 105-136 State and Local Governments: Why they Matter and How to Finance Them* by Serdar Yilmaz** François Vaillancourt ** and Bernard Dafflon** ______________________ **Respectively senior economist, World Bank; Fellow, CIRANO, Montréal; professor, Université de Fribourg 105 Introduction Intergovernmental reforms have been at the center of public sector change in countries across the globe. Indeed, the World Bank’s World Development Report on Entering the 21st century asserts that two forces will influence development policy in the first part of this century: globalization (the continuing integration of countries) and localization (self-determination and the devolution of power). 1 This force of “localization”, which in the international literature, is also referred to as “decentralization” is that of the division of public-sector functions among multiple types of government, central and subnational. Such “decentralization” is occurring in unitary and federal states alike. Established theories in economics and political science have articulated the efficiency and accountability gains that could accrue from decentralization reforms. The efficiency and accountability gains promised by a well-designed intergovernmental system range from the internalization of spillover effects, 2 alleviation of information asymmetry and better accountability due to the proximity of principals and agents 3 and the improvements in service delivery and revenue policy that may result from the fiscal competition among local governments. 4 This chapter begins with a generalized discussion of the “assignment” or functional roles among different types of governments and then proceeds to develop the theoretical case for decentralized governance. Recognizing that a well-designed assignment of functions among different types of governments will likely result is both vertical and horizontal fiscal imbalances, the paper concludes with a conceptual examination of the critical role of intergovernmental transfers in making intergovernmental finance “work”. Distribution of Functions among Types of Governments A seminal contribution to the theory of the intergovernmental assignment of functions was provided by Richard Musgrave in his Theory of Public Finance (1959) in which he identifies three functions of the public sector: (i) macroeconomic stabilization; (ii) income redistribution; (iii) efficient allocation of resources. At a very general level, the central (in the US, federal) government should have basic responsibility for macroeconomic stabilization and income redistribution functions. There are two tools in carrying out stabilization policy: fiscal and monetary policy tools. The three usual pro-centralization arguments for fiscal policy tools are that: 5 (i) local economies are small open economies and, as such, there is little chance that stabilization efforts initiated at the local level will bear fruit in this same local arena ; (ii) given that fact, local government units could adopt a do little or nothing strategy, letting the other local governments implement stabilization policies whose effects would spill over in their favor; and, finally, (iii) limiting local deficits and debt may be appropriate since financial markets do not function properly to create budget 106 discipline at that level of government. 6 With respect to this last point, there are two perspectives amongst public finance experts, known respectively under the terms of budget “discipline” versus “accountability.” The second tool, monetary policy, is obviously one that can only be used by the central government Bank (in the US, the Federal Reserve Bank). Thus stabilization is assigned to the center. Under Musgrave’s model, policies to adjust the inter-individual/household distribution of a nation’s income and wealth—thus the distribution function— also accrue to the center. This is because the full (assumed) mobility of both the rich and the poor between subnational entities makes it impossible for one such government to have more generous policies towards the poor than others if it has to finance these policies itself, be it only at the margin. Either the poor will move in, overloading the expenditure side and thus requiring higher taxes, or the rich will move out, starving the revenue side or both will happen simultaneously. A subnational government could have pro-rich policies to encourage outmigration of the poor but this would be undesirable from an overall allocative perspective. This would lead to segregation and thus to a reduction in the externalities that emerge from mixing in a given territory various groups. 7 However, there are costs associated with mobility. In small countries (Belgium for example), differences in ethno-linguistic-religious characteristics between subnational entities could reduce mobility from one jurisdiction to another one. In large countries, distance and its associated costs may be a barrier to mobility even if there are no differences in these characteristics. Thus it is conceivable to have some income distribution at the sub-national level. This leaves the efficiency argument--i.e., allocation policy--as the traditional raison d’être for the subnational role. This is not to say that an efficient provision of public goods requires only subnational activity. Indeed, that is not the case. Some social goods are such that the incidence of their benefits is nationwide (i.e. national defense, certain aspects of public health, some court systems), while others are geographically limited (i.e. street lighting, water distribution, public safety, fire protection). The recognition that the allocation function is an intergovernmental responsibility applies not only to broad functions of services, but also, within functions. Elementary education provides a good example. A nation as a whole has a strong interest in a well educated population, as does the locality where the service is provided. This fact, when combined with the spatial limitations of benefit incidence, suggests that the responsibility for education can be broken into several sub-functions: e.g., financing (primarily a local function, especially with respect to capital expenditures); setting the curricula (a national “core” curriculum supplemented by local options); teacher certification (the center may wish to set minimum standards); staff hiring, firing, and salary determination (local); and textbook selection (local). The Allocation Function 107 In the “next seminal” contribution to the theory of intergovernmental finance, Wallace Oates (1972) presented four criteria for assigning functional responsibilities across types of governments: i. Economies of scale will vary across goods and services. Economies of scale are significant in broadcasting, for example, where the unit cost per viewer drops by half (absent the need for additional broadcasting equipment) when the number of viewers of a given program doubles, but negligible in the provision of individualized health services such as surgical treatment. The existence of significant economies of scale constitutes an argument for a regional (e.g., special purpose) government to provide a particular good or service. ii. Heterogeneity of preferences and of circumstances also matters. Groups living in different parts of a country may display strong heterogeneity of preferences or may be faced with different environments in terms of climate or topography. They may therefore prefer or may need different amounts (more or less) of services, a different quality of service (for a given amount), or a different language for delivering public services. Decentralization is appropriate to address these needs if these groups are separated by borders that match those of sub national governments. iii. The presence of externalities, negative or positive, has an impact. If some of the activities of one government have important external effects on the individuals or businesses located in other jurisdictions or on other governments type, then these activities should be well coordinated among the affected governments. iv. Emulation, also referred to as competition, which helps increase or introduce best practices in government, requires at least two, and probably more, units involved in a given activity. This is an argument for decentralizing government activities. In general, governments have three mechanisms to carry out their assigned functions: “functional” spending made through the budgetary process, regulation of private activities, and tax expenditures. The most common one is the functional budget–that is, the cost and accounting financial plan of the government’s spending. 8 Regulations are both complements and substitutes to budgetary spending. This typically involves mandating private agents to spend on tasks that the government would otherwise carry out or requiring them to modify their behavior, with or without a cost to them. For example, business firms are often required to dispose of their “own” solid waste, with such regulations accompanied by fines for noncompliance. The use of such regulation to impose costs on households or firms cannot be justified on the basis that it reduces budgetary government spending. Regulation can, however, be appropriate if they are designed to “internalize” what are appropriately private sector costs of production and/or privately-generated external costs to society as a whole. 9 They are also appropriate if private provision is less expensive than public provision due to low density for example; hence requiring septic tanks can be a substitute for a sewer system in a given jurisdiction with both high and low density parts. The distinction between national and local regulations is relevant. Noise regulations are by nature mostly local while the regulation of greenhouse gases is a national one in scope. 108 A third, and often much less transparent, way of spending is by the use of “tax expenditures”—revenue losses resulting from tax provisions that grant special tax relief designed to encourage certain kinds of taxpayer behavior and/or to aid taxpayers in special circumstances. Tax expenditures may take one of several forms, can be hard to measure, and are typically, particularly at the subnational level, not given the periodic budget review required for good public financial management. 10 Through these three mechanisms governments produce goods and services that are not produced by the private sector in efficient amounts because of variety of reasons. The Case for the Provision of Local Public Goods Different goods and services provided by the government have beneficiary areas with different sizes. Certain publicly provided goods and services including public goods, such as national defense and environmental protection, have a beneficiary area defined by both national boundaries and treaties obligation and explicit and implicit agreements. These kinds of goods and services are national and sometimes international in terms of their beneficiary areas. Whereas other publicly provided goods and services, such as street lighting and garbage collection, have beneficiary areas limited to local jurisdictions. Some of these publicly provided goods are called local public goods, being understood that local is non national and thus can cover regional. 11 While the theoretical argument for public goods comes from Paul Samuelson, 12 he did not make a distinction between public goods with large catchment areas such as defense and other public goods that are similar in outcomes (protection from unwarranted use of force) but more local in nature such as policing. In his seminal piece Charles Tiebout (1956) makes the distinction and examines the difference between national and local public goods. He argues that citizens having differing personal valuations on local public services and varying ability to pay the attendant taxes, they will move from one jurisdiction to another until they find the one which maximizes their personal utility. Although conceptually the distinction between national and local public goods is very intuitive, practically it is not always very clear. The question is: what kinds of services are local public goods and what kind of services are national public goods? A classical example is the difference between street lighting and national defense. Street lighting is a local public good with a defined catchment area whereas national defense is a national public good covering the whole territory of a country and perhaps that of allies. However, one should note that broad headings of public activities can be misleading. For example in health, vaccination or food inspection activities have national impacts while the provision of ambulance services has local impact. The difference in size of beneficiary areas for national and local publicly provided goods and services and regulations lends itself to decentralized 109 delivery of these goods and services. The economic case for decentralized service delivery is based on the welfare gains from an improved allocation of resources. Decentralization allows local governments to tailor the levels of public services to the particular tastes and circumstances of their communities. Local governments are closer to the people and they possess knowledge of local preferences. Decentralized governments have their raison d’être in the provision of goods and services whose consumption is limited to their own jurisdictions. Since the efficient level of output (quantity and associated characteristics) of a local public good is likely to vary across jurisdictions, decentralized provision of services increases overall economic welfare. By tailoring local service delivery to the particular preferences and circumstances of their population, local government provision of local publicly provided goods and services increases economic welfare above that which results from the more uniform levels of national government service delivery. Furthermore, differential level of unit cost of service delivery across jurisdictions provides overall welfare gains to an economy. Therefore, divergence in demand across jurisdiction and interjurisdictional cost differentials are two important sources of welfare gains from decentralization: 1. Divergences in demand for local public goods and services Figure 1 depicts the demand curves for local services of jurisdictions one (D 1 ) and two (D 2 ) per household. If we assume that local publicly provided goods and services can be provided at a constant cost per unit of MC, the optimal output levels are Q 1 in jurisdiction one and Q 2 in jurisdiction two—assuming that there are no interjurisdictional spillover effects. If the central government decides on the level of output and is unwilling to or unable, due to legal/regulatory constraints, to provide anything but a uniform output, a uniform level of output of Q C could be provided in all jurisdictions. 13 If the central provision of Q C replaces the independent local provision Q 1 and Q 2 , then the loss of social welfare from centralized provision would be the sum of triangles ABC for residents of jurisdiction one and DCE for residents of jurisdiction two. 14 A uniform centrally determined level of local public goods and services will result in lower level of social welfare than decentralized service provision, where each local government provides optimal level of local service in its jurisdiction. This is the Decentralization Theorem. 15 The greater the divergence between D 1 and D 2 —the population is more heterogeneous—the larger will be the triangles ABC and DCE—the potential gains from decentralization will be larger. And if either Q 1 or Q 2 were provided, then one triangle would shrink to zero but the other would increase. 110 Figure 1 D1 $ DC D2 D E C A MC B Q1 QC Q2 Output In a Tiebout equilibrium individuals locate in jurisdictions that satisfy their demands for local public services, thus localities are perfectly homogenous in terms of demand for local public goods and services. 16 In Tiebout’s model, mobile households “vote with their feet”: they choose their jurisdiction to reside according to the fiscal package (revenue and expenditure) best suited their preferences. Decentralization is thus more important where the demand and ensuing supply for local public goods has greater variation across local governments. In places where the population is homogeneous with similar tastes and preferences for local publicly provided goods and services, the potential welfare gains from decentralization are smaller. The impact of the mobility implicit in the Tiebout model can be illustrated with the following example. There are three areas in one country that can become three local jurisdictions, I, II and III which are susceptible of producing two local public services, either S 1 or S 2 . There are 71 residents in the country who are also beneficiaries and voters. With central provision, a majority vote gives 36 voices to S 2 ; the unsatisfied minority is 35. With devolution, residents/voters in jurisdictions I and III choose to produce for local public provision S 1 ; while S 2 is chosen in jurisdiction II. Without an exit alternative, there would be altogether 25 unsatisfied residents. The “central” versus “devolved provision” solutions correspond to that of Figure 1 that is D c and D 1 /D 2 . In the Tiebout model, unsatisfied minorities could move to a preferred jurisdiction; in the example minorities in I and III choose to move to jurisdiction II. In this solution, no one is forced to accept the provision of an unwanted local service. Of course, the cost of moving remains to be taken into consideration; also, jurisdictions I and III providing the same service should consider possible economies of scale in producing S1. 17 Jurisdiction (wards with Central provision Devolved provision No mobility Tiebout model Mobility 111 central provision, S 1 independent entities when devolved or Tiebout S2 I 10 8 II 5 16 III 20 12 majority 36 unsatisfied 35 Characteristics of Central outcome command Majority choice # in minority 10-S 1 16-S 2 20-S 1 46 8 5 12 25 Specialization respect of minorities creativity, proximity Choice Vote by moving (with their feet) Mobile population Final population 8 go to II 10 + 5=15 5 go to I 16+8+12=36 12 go to II 20 36 and 35 0 Specialization, respect of minorities, proximity, fiscal competition S1 S2 S1 Source: Authors. 2. Divergences in cost of local public goods and services Cost differentials of service delivery across jurisdictions can also be a source of welfare gains from fiscal decentralization. Figure 2 depicts a case where demand curve, D C , is the same for local publicly provided services but the marginal cost of local services differs between the two jurisdictions. MC 1 is the marginal cost of local service delivery in jurisdiction one and MC 2 is the marginal cost of local service delivery in jurisdiction two. In Figure 2 the Pareto efficient outcomes are Q 1 and Q 2 . In this case, centralized provision of a uniform level of output at the average marginal cost, Q C , results in welfare loss of triangle ABC in jurisdiction one and triangle BED in jurisdiction two. The size of the welfare loss depends on the distance between MC 1 and MC 2 . The welfare gains from decentralization will be greater if demand for local services is more price responsive (less step demand curve). Figure 2 $ DC A MC 1 C MC C B E D MC 2 112 Q1 QC Q2 Output Revenues Once the assignment of service delivery responsibilities is decided, the next question is to assign financing sources to local governments to carry out those responsibilities. In an effective local government system, local governments must have control over some of their revenues. Local governments that lack independent sources of revenue can never truly enjoy fiscal autonomy, because they may be—and probably are—under the financial thumb of the central government. A rational assignment of revenue raising powers helps each level of government control its fiscal destiny. The important question is which revenue sources can and should be assigned to which type of government. This is commonly referred to as the tax assignment problem. 18 It is closely related to the assignment of expenditures, both because of the importance, at least in principle, of benefit taxation and because of the need for adequacy of revenues to finance local expenditures. An adequate assignment of revenue raising powers permits choice in the level of public spending by each type of government. Each type of government should be assigned taxes that are related to the benefits derived from its service delivery. Thus the proper assignment of taxes that are related to benefits depends on the assignment of service delivery responsibilities. In other words, to the extent possible local services should be financed by user charges and fees. This is both fair and efficient in the sense of encouraging responsible use of the nation's economic resources. 19 However, user charges, fees, and taxes related to the benefits of public spending are likely to be regressive. Therefore they are unlikely to reduce income inequality. 20 For both equity and efficiency purposes, tax rates should reflect the costs and benefits of public services. For example, taxes levied on motor vehicles and motor fuels are benefit-related and receipts from these taxes can be used to construct and maintain roads. Furthermore, such taxes can also be levied for the purpose of reducing congestion and pollution—or both when the direct value of the service provided to the user as well as social costs are factored in. The notion of subsidiarity in taxation means a given tax should be assigned to the lowest “level” of government that can implement it (or for which it can be implemented) and for which it is not inappropriate. 21 Compliance with subsidiarity principle is important to minimize the tendency toward vertical imbalance. 22 The guiding principle in designing an intergovernmental fiscal system is that subnational governments must have enough own revenues to finance the services they provide. 23 Even if such subnational governments rely on grants from a central government, if the grants are determined in an objective way and are guaranteed by the constitution or by long-standing legislation they may be 113 considered to be own revenues at least in the short term, such as the agreed period of a transfer formula. But these revenues are revenues that cannot be increased or decreased by these subnational governments since they do not have access to the base or rate so it is incomplete ownership as in the private ownership of a protected historical property. On the other hand, if grants are made at the sole discretion of the higher government, perhaps on an ad hoc, arbitrary, and unpredictable basis, and even well into the fiscal year and subject to renegotiation, subnational governments may be considered to have inadequate level of revenues. In between these extremes, other arrangements include: tax surcharges levied by one type of government (e.g., a local general purpose government) but collected by another government (e.g., the state unit) as in the case of revenue base sharing and tax-piggybacking. All of these arrangements might be seen as own revenues if there is no substantial risk for revenues accruing to the subnational governments. If subnational governments are to be truly autonomous, they must be assigned marginal sources of own revenues—that is to say they have some level of control over their revenues. 24 If they can legislate 25 and implement their own taxes and/or if they are allowed to impose surcharges on the taxes levied by the central government at rates they choose, they can influence the amount of revenues they receive at the margin—thus they control their revenues. Marginal revenue raising powers allow residents of subnational jurisdictions to choose the level of public services they want. In assigning revenue raising powers to subnational governments it is important to distinguish four aspects of revenue assignment: (a) which of government determines tax policy choices, in other words, who makes policy decision on taxes from which subnational governments receive revenues, (b) who defines the tax bases, (c) who sets the tax rates, and (d) who administers the taxes. Tax policy: The primary goal of taxation is to transfer resources from one group to another one to achieve certain development objectives without jeopardizing economic goals. Equity in taxation has both horizontal and vertical components. Horizontal equity is present when there is equal tax treatment of taxpayers who have equal capability to pay taxes. Vertical equity concerns the proper relationship between the relative tax burdens paid by individuals with different capacity to pay taxes. There is no formula for setting the proper degree of progressivity of taxation. It depends on factors such as the degree of inter individual solidarity in a society or the degree of shirking by potential recipients of transfers. But it is likely that, given the issues raised with respect to income distribution above, subnational taxes will be less progressive than national ones. In this context, which level of government decides which taxes to be levied by which level of government is an important aspect of the revenue assignment issue. However, subnational governments clearly cannot be allowed total discretion in the choice of the taxes they can levy. For example, they should not be allowed to levy import duties on international trade or trade between subnational jurisdictions or to impose taxes likely to be exported in large part. 114 Tax base: For efficiency purposes, it is important for subnational governments to have some control over tax bases. But for shared taxes, it is preferable for administrative and compliance reasons to have the same base used by all tax levying jurisdictions. 26 Tax rate: For subnational governments setting the tax rates is clearly the most important aspect of their fiscal sovereignty. Ability to set tax rates allows subnational governments to choose the level of public services while minimizing the compliance costs associated with collecting the required revenues. A possible alternative would be that for equity reason, the tax rate schedule would be set at the higher government level, whereas lower level governments have the right to fix a tax coefficient (sometimes within limits) in order to comply with their own budget needs. Tax revenue administration: In theory, revenue collection authority should be devolved to the level of government that does it best. This requires a trade-off between collection efforts, collection rates and revenues and collection and enforcement costs. While central governments may have low marginal collection costs, they may also have low marginal collections since it is not their priority or that of their tax collectors. The answer will thus depend on a thorough examination of institutional arrangements. While some argue that in allocating revenue administration function overall administrative and compliance costs for the entire national and subnational system should be taken into consideration (Mikesell 2002), this is not the sole consideration; subnational autonomy may come with higher administrative/collection costs. These dimensions of revenue raising powers have important implications on how revenues are linked to services provided. Some of the services that subnational governments provide can be described as producing generalized benefits, or benefits that cannot be closely related to taxes on the beneficiaries. The general benefits of subnational government spending may be loosely related to income or to private consumption. In these cases, unless there is some reason to believe that benefits rise more or less rapidly than income or consumption, relying on flat-rate taxes on income or consumption to finance such services may be reasonable. If people do not work where they live (or if they do not invest their savings where they live), an important tax policy choice is whether production or consumption (income earned or income spent) is the better measure of generalized benefits. If the benefits of public spending are more closely related to production or the earning of income than to consumption or the spending of income, origin-based taxes on value added and payroll taxes levied where employment occurs would be superior to destination-based value added taxes (VATs), retail sales taxes, and residence-based income taxes as measures of benefits received. Empirical evidence suggests that consumption (the spending of income) is probably more closely related to the benefits of public spending than is production (the earning of income). For example, education is usually provided 115 where people live, not where they work, and the same tends to be true for health care and social aid. Therefore, for the purpose of taxing generalized benefits of public services, residence-based income taxes are probably superior to employment-based payroll taxes, and destination (consumption)-based sales taxes are better than origin (production)-based ones. This may be difficult if personal income taxation is not available to subnational governments. In that case, payroll taxation levied in the region of employment but allocated on the basis of the place of residence of workers (share of wage bill) may be a solution most likely by the central government A variety of approaches to assigning revenues to subnational governments differ in the degree of fiscal autonomy they provide to subnational governments. In certain countries, like Canada (provincial level), Switzerland (cantonal level), and the United States (state level), independent subnational legislation and administration, provides subnational governments with very high fiscal autonomy. In these countries, subnational governments choose the taxes they levy, define the tax bases, set the tax rates, and administer the taxes. 27 In fact, in some cases, subnational jurisdictions enter into competition to attract businesses and wealthy individuals to their localities. Although tax competition 28 can protect citizens from the rapaciousness of politicians and bureaucrats, excessive subnational latitude in the choice of tax bases and in tax administration can create unacceptable complexity and administrative burdens, as well as inequities and distortions in the allocation of resources. In extreme cases, tax competition exposes public finance system to inconsistency, duplication of effort, and excessive complexity of compliance and administration. These problems can occur if different jurisdictions choose radically different taxes (for example, if some levy retail sales taxes, but others levy VATs as is the case in Canada), define their tax bases in different ways (as in the case of state corporate income taxes and retail sales taxes in the United States), or administer the same taxes in different ways. 29 In other countries, local government budgets are financed through surcharges or piggyback taxes. For a surcharge, the surcharge rate is set locally but both the main tax and the surcharge are collected by the government responsible for the main tax. For a piggyback tax, the rate is set locally and the tax also collected locally; it is piggybacking on an existing tax base. It is, thus, using information already available to the taxpayer to compute a different tax. 30 Ideally, in these countries, intergovernmental finance system is less prone to the problems that occur when different subnational jurisdictions define the tax base in conflicting ways, use different apportionment formulas, and administer the tax in different ways. In a surcharge regime, although subnational governments don’t have the ability to define the tax base and administer taxes, they have the power to set surcharge rates–the most important attribute of fiscal sovereignty. Subnational surcharges provide appropriate means with their own marginal revenues in cases where revenue bases are scarce. However, an incentive problem in subnational surcharge regimes is that the central or more generally the tax collecting government collects a tax that it does not keep. Surcharges 116 should, of course, be limited to that portion of the tax base reasonably deemed to arise in, or be attributed to, the taxing jurisdiction. In some cases the tax rate of the central government is zero for a particular tax, in those situations the central government simply administers the tax of subnational governments, thereby ensuring uniformity and avoiding duplication of effort. The third approach is tax sharing. Under tax sharing arrangements subnational governments receive fixed fractions of revenues from particular national taxes originating within their boundaries. In general the sharing rates are uniform across jurisdictions, but there is always variation across taxes. This approach restricts subnational governments’ fiscal autonomy. Although individual subnational governments have autonomy over how to spend a given amount of the shared revenue, they do not have the power to alter the amount of revenue they receive. Therefore, they have limited control over the level public spending. 31 In general, there is a formula to determine the deemed origin of tax revenues. Thus the real loss of autonomy will also very much depends on whether the formula is written in the constitution (as in Switzerland), in the financial law or annually decided by parliament in the budget process. Transfers Intergovernmental transfers compose of a significant portion of subnational governments' revenues in all countries and are therefore an essential component of intergovernmental fiscal arrangements. Their design has important efficiency and equity implications on local public service provision. In general, intergovernmental transfers have three principal objectives: (i) to equalize vertically (improve revenue adequacy); (ii) to equalize horizontally (interjurisdictional redistribution) (iii) to minimize inter-jurisdictional spillovers (externalities). They may also be used to allow the federal government a presence in policy areas where it has no direct role in te provision of publicly provided goods and services given the constitutional division of powers. This is the case for education in the USA (No child left behind) or health in Canada ( Canada health transfers) but with different degrees of condionalities. Vertical fiscal imbalance is the disparity between revenue sources and expenditure needs of subnational governments. A vertical imbalance occurs when the expenditure responsibilities of subnational governments do not match with their revenue raising powers. A horizontal imbalance occurs when own fiscal capacities to carry out the same functions differ across subnational governments. Horizontal imbalances across subnational governments arise from: Unequal distribution of revenue bases, natural resources, and wealth across subnational governments, Variations in the socio-economic characteristics of population, and Differences in the geography and climate across jurisdictions that lead to disparities in subnational economic opportunities or costs in the provision of goods and services. 117 A third objective of intergovernmental transfers is to correct for interjurisdictional spillovers. Some local government services’ benefits (or costs) extend beyond the borders of the locality. For example, an outbreak of a disease in one jurisdiction will have an impact on the overall health situation in neighboring communities. Local governments may be unwilling to provide an efficient level of certain services if they believe that people who reside outside the locality will enjoy many of the resulting benefits. To ensure that local governments provide a greater amount of those services, the central government may transfer resources to them. There are two general kinds of transfers; conditional and unconditional. In conditional transfers, the transferring authority (central government) specifies the purpose of use for these funds. The use of conditional transfer funds is limited to a specific sector that is highly important to the transferring government, such as education, health, housing, environmental. These types of transfers are sometimes called specific purpose grants or categorical grants. There are different applications of conditional transfers: A. Open-Ended, Matching: For a unit of money given by the donor, the recipient should spend some amount and there is no cap on the amount of funds transferred to the recipient. Open-ended means as long as the recipient provides co-financing; the central government will contribute its share as well. B. Close-Ended, Matching: Very similar arrangements but the transferring authority puts a ceiling on the total amount to be transferred. C. Non-matching: The recipient is not required to provide co-financing. The donor gives a fixed sum of money with the stipulation that it be spent on a specific publicly provided good or service. In unconditional transfers the transferring authority places no restrictions on the use of funds. In both conditional and unconditional transfer systems, one can explicitly factor equalization into the transfer formula. The reasons for introducing equalization are multiple since intra country inter region solidarity is not explained only by economic arguments. Since no federal or decentralized country is perfectly homogenous, the different levels of taxation by the subnational governments (cantons, provinces, states collectively referred to as SNGs) do not necessarily mirror differences in preferred local public services. Local financial capacities depend on both the tax bases accessible to subnational governments and the territorial distribution of those bases. Local needs vary according to the particular preferences of the local residents; but they also depend on geographic, demographic, and socio-economic factors. They are further determined by constitutional or legal requirements as to the type and nature (quantity, quality) mandatory public goods and services that SNGs must provide. Box 1 reviews the possible origins of fiscal disparities in the relevant literature. 32 The logic behind this classification in five categories is twofold: (i) "External" items that are outside the scope of local decision should be compensated, at least partly, if they result in a significant spread in the relative fiscal position of governmental units. We call them "disparities". 118 (ii) Those items that are within the scope of decision and the fiscal management of SNGs should not be taken into consideration for equalization. They belong to the sphere of local autonomy and responsibility. We call them "differences". Box 1: Five sources of fiscal differences across sub national governments and their appropriateness for equalization Appropriate for equalization - disparities A B C Differences in the territorial distribution of revenue bases (taxes, natural resource royalties…) or access to them usually measured per head. This can result from differences in endowments (land, minerals…) location (access to sea or trade routes…) or constitutional/legal constraints (some tax bases off limits). Need differences in the number of units of standardized service required per capita owing to demographic reasons (age structure, migrant / non migrant structure) and territorial (length of shoreline or border to be patrolled). Cost differences per unit of standardized public service that arise from factors such as: differences in the quantity and composition of inputs that are necessary for producing the public service, differences in factor or input prices, differences in physical characteristics (environmental factors) and presence or absence of economies of scale in the service provision). Not appropriate for equalization -differences Differences in local preferences either for optional services, for quantities or quality above the minimum standard level in the provision of services; or for autonomy that hinders the achievement of optimal size. E Differentials that are attributable to strategic behavior on the part of the SNGs with respect to federal transfer payments; local preferences between (non-benefit) taxes and user charges (benefit taxes), including the choice – if any – among different forms of taxes Source: adapted from Dafflon and Mischler, 2008: 215 D Category A concerns resource equalization: revenue sources depend in part on the geographic location of subnational governments in the national territory (proximity of urban areas or economic centers or located at the periphery), on the kind of economic activities, and on communication networks. Within an open market economy, subnational governments cannot influence these characteristics, thus they must be treated as exogenous variables. Category B and C cover the two determinants of expenditure equalization. Both items require more attention than differences in taxable capacity. Needs disparities when measured using simple indicators should not be influenced by subnational governments' decisions. But one must be careful as to the interrelation between specific policies and outcomes; the possible origin of needs disparities deserves careful consideration. Well-measured cost disparities in input factors most often also fall outside the SNGs' decision-making competence. But this requires using input prices from the private sector for measurement purposes. Having examined what are reasonable grounds for equalization, we turn in the next section to revenue equalization that is equalization justified under category A of Box 1. Revenue equalization In this section and in the following one, we put forward both a graphical depiction of revenue (expenditure) equalization and a discussion of issues that 119 cannot be included in such a graph. Figure 3 taken from Dafflon and Vaillancourt (2003) addresses a first set of four issues. The first issue is the level of the public revenues available to subnational governments to both satisfy the needs of their population and to be shared in a horizontal equalization scheme. The vertical axis records subnational revenue divided by population (per capita). At point A on the vertical Y-axis, the subnational government receives the average amount of public revenue per resident for that type of jurisdiction in a given country, represented by the standardized value 1.0. In this figure assume that all revenues available to satisfy the needs of the population are also available for financing an equalization scheme. But this may not be the case; for example, natural resource revenues may be treated differently for each purpose. The second issue is that equalization requires jurisdictions to be ranked according to some indicator of contribution/entitlement to equalization; this is addressed by the positioning of the various jurisdictions along the horizontal Xaxis. In Figure 3 this is revenue capacity since we are presenting revenue equalization; in figure 4 it will be cost adjusted needs. The basic revenue equalization rule is as follows: jurisdictions with higher-than-average revenue capacity should receive less (pay more); jurisdictions with lower-than-average capacity should receive more (pay less). In Figure 3, average capacity is given a value of 100. For ease of exposition, the lower value for the “poorest” jurisdiction is given a value of 30 and the highest a value of 150. In practice, ranking SNGs is not easy. There are numerous indicators that can be used for ranking purpose and good ad hoc reasons for each being deemed “best” by one actor or another such as a public finance economist, 33 a macro economist, a politician, the winning jurisdiction(s) or the losing jurisdiction(s). 120 Figure 3 A stylized representation of a revenue equalization scheme Index of per capita public revenue of subnational government before and after horizontal (H) / vertical (V) equalization G 1.15 1.10 E 1.00 A 0.85 B 0.55 C 0.40 D K 90 30 poorest Before H equalization F After H equalization J Perfect equality After H+V equalization 100 poor 125 rich Indicator of per capita financia l capacity of subnational government 150 richest Source: Dafflon and Vaillancourt, 2003: 401 Let us assume that these two issues have received an appropriate answer. The third issue is the equalization formula. To understand this, let us compare the “before” and “after” situations. With no equalization, and the possibility of identifying exactly the origin of the tax revenues, “poor” jurisdictions would certainly receive less than average per capita endowments, and “rich” ones higher than average amounts, something like the line DEG (labeled “before equalization public revenues”) in Figure 3. Any equalization formula would have to give more to “poor” jurisdictions than they would receive following the origin principle and “rich” jurisdictions would receive less, something along the CEF line. The equalizing performance is represented by the distance between lines DE and CE for beneficiary jurisdictions, and between EG and EF for the jurisdictions supporting the financial cost of equalization. Thus, for example, for the poorest jurisdiction with a revenue capacity of 30, equalization increases public revenue per capita from 0.40 (at point D) to 0.55 (at point C), but for a rich region with capacity of 125, equalization reduces public revenues available to it from 1.15 to 1.10. Of course, a balanced solution with horizontal (H) equalization requires that benefits (amounts received, represented by the surface 121 of triangle CDE) and payments (amounts contributed, represented by the surface of triangle EFG) coincide. The importance of equalization depends on the equalization formula, which gives the position of the slope CF around the central point E. The fourth issue is whether an equalization policy would introduce further limits to the redistribution formula. In Figure 3, E represents an exactly neutral position with regard to equalization: with an average financial capacity and average per capita tax revenues, a jurisdiction at this point would neither pay nor receive any equalizing amount. But the central point need not be at E. Other equalization targets are possible, and often controversial. Two specific points must be noted. First, it can be debated whether jurisdictions with just below average financial capacity should also benefit from equalization. One could argue on financial, political and equity grounds that only jurisdictions below a certain level (e.g. an indicator of capacity = 90) should qualify. Financial considerations could be one argument: at 90, the triangle equivalent to CDE would be smaller, which means smaller contributions by richer SNGs. But more crucial are political considerations; at what value does fragmentation of the nation into poor and rich jurisdictions endanger national coherence. Or, put differently how much poorer is too poor? A second related question is illustrated with the triangle BCK. The resources available after applying the horizontal equalization formula are those corresponding to line CE (above DE): the poorer a jurisdiction, the more it receives. But the horizontal equalizing payments in the example can be argued to be far from giving poor jurisdictions sufficient resources, increasing the resources for the poorest SNG from (in our example) 40% to 55% of the national average. Should they be increased? In the affirmative, what would be the appropriate limit? The example in Figure 3 ensures that poor jurisdictions receive equalizing payments so that their revenue endowment reaches at least 85% of the national average, along line BK. Since “rich” jurisdictions already pay EFG to cover CDE (equal by construction), financial resources for paying BCK come from a contribution from the center through a vertical equalization scheme. 34 But is 85% a proper level? Fragmentation, equity and incentives must be considered. But note that in Figure 3, beneficiary jurisdictions have no incentive to take initiative for their development if they are satisfied with public spending compatible with 85% of the national average per capita public revenues, and if they have no preference for autonomous revenues rather than transfers. Not all the key issues of revenue equalization are captured in Figure 3. For example, we do not discuss how equalization is financed. The questions are which revenue (tax) source is to be shared and according to which decision procedure? Several answers are possible, each with pros and cons; two are discussed below with reference to vertical equalization but the questions are also of relevance to horizontal equalization. 122 The amount is financed out of the general resources of the central government and established in its annual budget. This is a very flexible solution. But it has two main defects: (i) recipient governments are not sure that they will receive a comparable amount (in real value) from one year to another, which renders medium term planning and multi-annual policies very difficult; (ii) annual budgetary debates are subject to ad hoc political arrangements. The exact calculation of the amount is explicitly stated in the constitution or in a law, in the form of revenue sharing from at least one, but preferably several or all, specific tax sources used at the central level (the use of only one tax source for sharing purposes may result in the central government either not collecting it as vigorously as it could since its efforts in part reward sub-national governments or selecting the less buoyant tax source). In this case, the political debate on equalization takes place once when the constitution is amended or the law is passed and not on an annual basis at the time the budget is discussed. And if the tax sources are sufficiently diversified, such that they partly alleviate macroeconomic cycles, it avoids important variations in the amounts available. If they are not, it is possible to solve this potential cyclical problem with the constitution of an equalization fund that can smooth equalization payments over time. Expenditure equalization The focus of this section is on expenditure equalization, a combination of items B and C in Box 1. Figure 4 presents a stylised expenditure equalization scheme. The reader should note that contrary to figure 3, the recipients of equalization are to the right of the pivot point and not to the left. As with figure 1, one must answer four questions. First, what are the equalizable expenditures carried out by SNGs? In this figure, we assume that all expenditures are eligible for equalization. If we did not do this, then the vertical Y -axis would be drawn only for eligible expenditures which means that total SNG expenditures could be higher for some or all SNGs. As in Figure 3, we present it in per capita terms. Second how should we rank SNGs in terms for expenditure equalization? To answer this question recall that as noted by Bird and Vaillancourt (2007) average per capita expenditure differences in providing a public service reflect two factors: need differences (Box 1, B above) and cost differences (Box 1, C). Need differences—differences in the number of units of standardized service required per capita—usually arise owing to demographic reasons such as the age structure of the population and different participation rates in social programs by persons of different ages. Cost differences are differences in the cost per unit of a ‘standardized’ public service. They may arise from climatic or geographic features, density or distance factors, or differences in labor cost across regions. Costs should be calculated using real (not nominal) private sector wages for equivalent inputs and not on the basis of public sector wages which may reflect such political 123 35 Plausible factors related to needs differences are the share in the total population of various age groups such as infants (post-natal care), elders (health care) and school age children, the share of population with special needs, either temporary i.e. new immigrants (language skills acquisition, integration into society) or not e.g. aboriginal population. The relevance of many of these indicators depends on the role SNG play in delivering public services. For instance, if it is the central government or the private sector that provides health care, the share of infants or elders in the population of SNGs may not be relevant in determining transfers. Various factors determining cost differences have been proposed. Some are natural ones that vary with geography such as climate (snowfall, heavy rain), frequency of natural disasters (floods, earthquakes), topography (mountainous or desert regions) and distance (remoteness from providers of inputs into public services). Others are demographic such as population density/urbanization. The difficulty is to estimate in monetary units the impact of such factors on costs. In some cases, it may be not too difficult such as using private transportation costs per km to estimate the impact of remoteness on the cost of schoolbooks being delivered to a school. But if snow removal is done only by public maintenance crews, then how does one distinguish between true differences in costs and the relative strength of public sector unions in various SNGs, assuming that each sets its own wages (not set centrally)? We thus use a cost adjusted needs index on the X-axis of Figure 4. What does this mean? Let us assume that we have two regions with identical revenue capacity (same ranking on the vertical Y -axis of Figure 3), one (A) with a proportion of 10% of older individuals in need of specific health services in its population and the other (B) with 30%. In terms of needs, B has higher needs. If the cost per % point of older population is 1 monetary unit, then B should receive 20 more units of resources than A to be able to provide the required services without having to levy more taxes than A. But if A is more mountainous than B and the cost of getting the services to the older resident is higher because of this say 1.5 per % point in A and still 1.0 in B, then the difference in cost adjusted needs is only 15; (30x1)-(101.5)=15. Hence adjusting for costs changes the relative position of these SNGs on the X-axis of Figure 2; depending on original positions and the importance of cost differentials, it could, invert the rankings. The third issue is the equalization formula. To understand how it works, let us compare the “before” and “after” situations. With no equalization “needier” jurisdictions to the right of E spend less per capita than with equalization and ‘’un-needy’’ ones to the left of E more. Note that per capita expenditures in Figure 4 are for the populations as a whole and not for the specific populations (older, immigrants, school age etc.) that may be deemed to have specific needs. Horizontal equalization in this context means than un-needy SNGs spend less overall on their residents after equalization is implemented and implicitly spend for residents of other jurisdictions. Needier jurisdictions can now spend to better 124 satisfy the needs of their residents without, given the assumption of equal revenue capacity, having to levy higher taxes while before equalization they spent less since we assumed no additional tax effort. Thus, for example, for the neediest jurisdiction with a cost adjusted needs indicator of 150, equalization increases expenditures per capita from 1.15 to 1.25, but for an un-needy region with a needs indicator of 30, equalization with its diversion of revenues reduces public expenditures it can finance from 0.7 to 0.5. As in Figure 3, a balanced solution with horizontal (H) equalization requires that benefits and costs coincide. The importance of equalization depends on the equalization formula, which gives the positions of the lines around the central point E. It is conceivable that the slopes of these two lines are not the same. Figure 4 A stylised representation of an expenditure equalization scheme SNG per capita eligible expenditures SNG expenditures after equalization 1.25 SNG expenditures before equalization 1.15 E 1.0 0.7 0.5 100 30 lowest low high 150 highest Indicator of cost adjusted needs Source: authors adapted from Dafflon and Vaillancourt (2003) The fourth issue is whether an equalization policy would introduce further limits to the redistribution formula. In Figure 4, E represents an exactly neutral position; a jurisdiction at this point would neither pay nor receive any equalizing amount. But the central point need not be at E. Other equalization targets are possible. It can be debated whether jurisdictions with just above average needs should benefit from equalization; one could argue that this would be a disincentive to become more productive or that measurement errors of needs are 125 upward biased and thus that a cushion of say 5% should be used. One could also argue on financial, political and equity grounds that only jurisdictions above a certain level (e.g. 110) should qualify. We noted earlier in Box 1 and its discussion that while disparities are eligible for equalization, differences (items D and E) should not be. We come back to this now since it is relevant to expenditure equalization. Figure 5 illustrates the difficulty of drawing the border between genuine disparities and local preferences or management abilities that result in expenditure or cost differences. The first scenario relates to the optimal size of SNGs and their capacity to realize scale economies. The second scenario illustrates the difficulty of distinguishing between higher production costs that arise from justifiable factors and those that result from X-inefficiencies. Scenario 1: Impossible economies of scale or reluctance to cooperate The jurisdictions face the usual simplified U-shaped production function for a local public good S. Start with the production function PF I for SNG1. Resident beneficiaries pay for the service on a quid pro quo basis (for simplification: one resident, one unit of local service S, one tax unit - no spillover). The efficient solution is at E for a total of N optimal residents served. The E solution shows two key results: the minimal average cost at AC 1 and the total local public expenditure at the optimal level for PF I. Consider SNG2: assume it has an identical production function PF I, but only N 2 residents. Average cost is AC 2. Why is this so? There are three plausible answers (1) The number of beneficiaries is low because of socio-demographic characteristics of the resident population in SNG2 (less school-aged children if S is a primary school) (2) SNG2 is not in a position (for topographic reasons or distance) to cooperate with neighbouring SNGs in order to increase the number of beneficiaries towards N optimal . (3) SNG2 (for reasons of differences in preferences or the desire to remain autonomous) is not willing to cooperate with neighbouring SNGs? In this last case, it should also bear the fiscal consequences of the decision and not receive equalization to make up for the difference which is not due to equalizable disparities in cost. Figure 5 expenditure Production functions for a sub-national public Average cost Production Function II = PF II (SNG3) Production Function I = PF I (SNG1 and SNG2) F B 0 E 126 Number of beneficiaries for service S D AC 2 AC 1 N2 SNG2 N optimal SNG1 Source: Dafflon and Mischler, 2008: 218 Scenario 2: Genuine cost disparities versus X-inefficiencies Now let’s look at a third local unit, SNG3 in Figure 5 with production function PF II characterized by higher costs for the whole range of production. Even with the optimal number of beneficiaries served, SNG3 cannot provide an equal level of service S at the same tax price [N optimal D > N optimal E]. If the cost difference AC 1 EDAC 2 is a genuine disparity, then the situation suggests some kind of equalization so as to restore the fiscal balance. This would not only reduce the average cost (tax price AC 2 ) of service S that residents in SNG3 face, but it should also reduce fiscally induced migration, thereby enhancing efficiency. 36 But does PF II represent the real costs or does it hide any X-inefficiencies? How can one interpret the difference ED in average costs if SNG1 and SNG3 serve the same number of beneficiaries? 37 Can SNG3 do anything about the high costs? Figure 5 thus identifies three situations that need to be examined if expenditure needs equalization is on the political agenda to be able to rank SNGs on the X axis. Cost adjusted needs can be determined in relative terms only if sources of cost differentials are clearly traced and identified. This is not simple; it requires information about the number of beneficiaries and the production function of each public service selected for equalization for several SNGs in order to set the standard cost function within a reasonably range. Such information is not always available. 38 And are costs set by beneficiary or by a production determined grouping of beneficiaries? For example, for primary school, is it average cost per pupil that matters or average cost per class with different results if regulations require a minimum and set a maximum number of pupils per class. And who determines when mergers or at least cooperation between SNGs should be required to lower average production costs; are mixed language or mixed religious classes to be mandated either centrally or by SNGs that pay horizontal equalization? From this perspective, one can see that any policy of expenditure-based equalization is a tremendous challenge. Since expenditure needs equalization is complex should one conclude that one should renounce, as the Canadian Expert Panel on Equalization recently proposed? 39 Or, should there be an attempt to try to design expenditure needs equalization as best as possible with imperfect knowledge, information and data? 40 127 Combined equalization Can one combine the two types of equalization in one formula? Yes, but one must be clear on the sequence to do this. Recall that equalization is about providing similar levels of public services at similar levels of taxation. Hence to combine both types of equalization, the following steps seem appropriate: First, establish the needs index and compute the expenditure equalization entitlement, positive or negative of each SNG according to it; Second, correct the needs based index for cost differences and compute anew the expenditure equalization entitlement, positive or negative of each SNG according to it; Third, compute the revenue capacity index and compute the revenue equalization entitlement, positive or negative of each SNG according to it; Fourth, combine the expenditure and revenue equalization entitlements together to get total equalization Horizontal versus Vertical Equalization A last issue deserves attention, which is whether equalization is horizontal or vertical. In Figure 3, surface CDE = EFG implies that equalization is horizontal, between contributing and beneficiary SNG units; CBK, if it exits in total or partially, is vertical. In Figure 4, a balanced solution (around E) with horizontal equalization requires that benefits coincide: but the solution could be also to grant equalization to the jurisdictions with cost adjusted needs higher than average (100 points on the horizontal axis) without asking the jurisdictions with cost adjusted needs lower than average to contribute; needs equalization would then be vertical, centrally funded. Horizontal equalization is typically a "Robin Hood"-type of equalization: highcapacity SNGs directly transfer public revenues to a fund serving low-capacity SNGs. This is less conceivable for expenditure needs equalization. 41 This would imply that SNGs with relatively low needs and costs of service provision accept higher tax prices which allow subsidizing other SNGs with relatively high expenditure needs. This would distort the relative local tax prices of public services and result in allocative inefficiencies. Two further arguments against horizontal expenditure equalization are: (1) for those public services that are financed through user charges, if the “price" does not reflect benefit, consumers will face false price signals; (2) when the difference between SNG choices, X-inefficiencies and genuine disparities is not clear, SNGs might indulge in strategic behaviour with the aim of placing themselves in a more favourable equalizing position (in this case, higher costs and more needs). Vertical needs equalization can be set on expenditure standards that eliminate functions based on the benefit principle for their financing and that ignore SNGs potential strategies but this adds to complexity. 42 Conclusion This paper has shown that the decentralized provision of both public and private goods and services, either through budgetary expenditures, tax expenditures or regulation of economic behavior is often a solution that yields superior outcomes 128 than the centralized provision of these goods and services. What exactly should be provided in a decentralized fashion needs to be defined precisely. One cannot simply say education for example; one must distinguish between the various inputs in education, ranging from school curriculum to school cleaning. However, setting what goods and services should be provided in a decentralized fashion is only half of the equation. One must then ensure that the decentralized providers have the financial means of carrying out their mandate. This is done not only through own revenues but in most if not all countries, through intergovernmental transfers. The appropriate own revenue depends on the type of goods and services offered. The transfer scheme will reflect both vertical and horizontal imbalances as well as the degree of solidarity between regions. 129 References: Bird R. and F. Vaillancourt, 2007, "Expenditure-Based Equalization Transfers", in Martinez-Vazquez J. and B. Searle (Eds), op. cit., 259-289. Blindenbacher R. and A. Koller (Eds), 2003, “Federalism in a Changing World – Learning from Each Other”, McGill-Queen’s University Press, Montreal and Kingston. Boadway R. and F. Flatters, 1982, "Efficiency and Equalization Payments in a Federal System of Government: A Synthesis and Extension of Recent Results", Canadian Journal of Economics 15, 613-633. Boadway R. and A. Shah (eds), 2007, Intergovernmental Fiscal Transfers: Principles and Practice, Public Sector Governance and Accountability Series, The World Bank, Washington D.C. Boex J. and J. Martinez-Vazquez, 2007, "Designing Intergovernmental Equalization Transfers with Imperfect Data: Concepts, Practices and Lessons", in J. Martinez-Vazquez and B. Searle, op. cit., 291-344. Boothe P. and F. Vaillancourt, 2007, A Fine Canadian Compromise: Perspectives on the Report of the Expert Panel on Equalisation and Territorial Funding Financing Edmonton and Montréal; Institute for Public economics and CIRANO, 100 pages; http://www.cirano.qc.ca/pdf/Perequation_07.pdf Buchanan J., 1968, The Supply of Public Goods, Rand McNally & Company, Chicago. Bucovestsky, S., Marchand, M. and Pestieau, P. (1998) ‘Tax Competition and Revelation of Preferences for Public Expenditure’, Journal of Urban Economics 44: 367-90. Courchene T., 1998, "Renegotiating Equalization: National Polity, Federal State, International Economy", C.D. Howe Institute Commentary 113, C.D. Howe Institute, Toronto. Cremer, J., Estache, A. and Seabright, P. (1996) ‘Decentralizing Public Services: What Can We Learn from the Theory of the Firm?’ Revue d’ Economie Politique 106: 37-60. Dafflon B. and K. Beer-Tóth, 2009, "Managing Local Public Debt in Transition Countries: an Issue of Self-Control", in Financial Accountability and Management, 25 (3), August, pages 337-365 130 Dafflon B., 2007, "Fiscal Capacity Equalization in Horizontal Fiscal Equalization Programs, in Boadway R. and A. Shah, op. cit., 361-396. Dafflon B. and P. Mischler, 2007, Réforme de la péréquation intercommunale dans le canton de Fribourg, Centre d'études en Economie du Secteur Public, série Etudes et Rapports, Université de Fribourg, Fribourg. Dafflon B. and P. Mischler, 2008, "Expenditure Needs Equalisation at the Local Level: Methods and Practice", in Kim J. and J. Lotz (Eds), op. cit., 213-240 Dafflon B. and F. Vaillancourt, 2003, "Problems of equalization in federal countries” in Blindenbacher R. and A. Koller (Eds), op. cit., 395411. Kim J. and J. Lotz (Eds), 2007, Measuring Local Government Expenditure Needs, The Copenhagen Workshop, The Korea Institute of Public Finance and the Danish Ministry of Social Welfare, Copenhagen. Martinez-Vazquez J. and B. Searle (Eds), 2007, Fiscal Equalization: Challenges in the Design of Intergovernmental Transfers, Springer, New York. Mikesell, John. (2003). International Experiences with Administration of Local taxes: A Review of Practices and Issues. Paper prepared for the World Bank Thematic Group on Taxation and Tax Policy, march. Washington, DC: World Bank. Mueller, D. (1996) Constitutional Democracy. Oxford: Oxford University Press. Musgrave, R. A. (1959). The Theory of Public Finance (New York: McGraw-Hill). Oates, W. (1972) Fiscal Federalism. New York: Harcourt, Brace and Jovanovic. Oates, W. E. (1997). ‘On the Welfare Gains from Fiscal Decentralization’ Journal of Public Finance and Public Choice, Vol 2-3, pp. 83-92. Raff, H. and Wilson, J. D. (1997) ‘Income Redistribution with WellInformed Local Governments’, International Tax and Public Finance 4: 407-27. 131 Rechovsky A., 2007, "Compensating Local Governments for Differences in Expenditure Needs in a Horizontal Fiscal Equalization Program", in Boadway R. and A. Shah (eds), op. cit. , 397-424. Rossi S and B. Dafflon, 2002 : The theory of subnational balanced budget and debt control; in Dafflon B. (ed): Local Public Finance in Europe: Balancing the Budget and Controlling Debt, Edward Elgar, Cheltenham UK, Series: Studies in Fiscal Federalism and State-Local Finance, pages 15-44. Samuelson, Paul A. (1954). “The Pure Theory of Public Expenditure” Review of Economics and Statistics, Vol. 36 (4): 387-389. Ter-Minassian T. (ed), (1997), Fiscal Federalism in Theory and Practice, International Monetary Fund, Washington. Ter-Minassian, T. and J. Craig, (1997), "Control of Subnational Government Borrowing", in Ter-Minassian (ed.), op. cit., International Monetary Fund, Washington. Tiebout, C. (1956). “A Pure Theory of Local Expenditures” Journal of Political Economy, Vol. 64, pp. 416-424. World Bank, Entering the 21st Century: World Development Report, 19992000. Oxford: World Bank and Oxford University Press, 2000, Chapter 1. 132 Endnotes: 1 World Bank (2000) Oates (1972) and Mueller (1996) 3 Cremer et al. (1996); Raff and Wilson (1997) and Bucovetsky et al. (1998) 4 Tiebout (1956) 5 Rossi and Dafflon 2002: 20–25. 6 Ter-Minassian and Craig, 1997:162; Dafflon and Tóth, 2009: 337. 7 These can occur in the schooling environment for example. 8 The related technical issues of “off” vs. “on” budgeting and extra-budgetary accounts is not discussed here. For discussion see Wong with Gooptu and Martinez-Vazquez (2002). 9 Whenever one economic actor (a firm or individual) undertakes an action that has a (net) added value or cost to another economic actor, there is an externality. If these values (positive externalities) are not paid for by the recipient or net costs (negative externalities) are not paid for by the first actor, the result is inefficient resource allocation. 10 Gravelle (2005) identifies four variants: (i) exclusions, exemptions, and deductions, which reduce taxable income; (ii) preferential rates, which apply lower rates to part of all of a taxpayer’s income; (iii) credits subtracted from taxes and ordinarily computed, and (v) deferrals of tax, which result from delayed recognition of income for from allowing in the current year deductions what are properly attributable to a future year. 11 Governments also provide private goods. Indeed, in many countries, the provision of private goods dwarfs that of public goods in the budget. Education services, most of health services and some income insurance/support scheme are joint “private-public” goods: private because the service is delivered to individual and, as such, it consumption is rival and price exclusion is possible; public because the service provides externalities which are of collective nature (education is a key factor in social and cultural homogeneity, labor capacity and mobility, health and the exercise of democracy) The distinction between national and local applies to such private goods. The provision of education is more a local good while the provision of a pension scheme is more national. 12 Samuelson defined public goods (or “collective consumption goods” as he calls) as: “…[goods] which all enjoy in common in the sense that each individual’s consumption of such a good leads to no subtractions from any other individual’s consumption of that good...” (Samuelson 1954: 387). 13 We assume that the output will be a compromise between the two desired levels, a reasonable outcome in a democratic system . 14 For Qc, group D1 would be ready to pay B, but has to pay much more in C; and for a price equal to C, group D2 would rather obtain quantity Q2 but has to satisfy itself with Qc only. Note that the two jurisdictions could enter into a cooperative agreement, producing more in total that in the independent solution at a lower cost for each of them (Buchanan, 1968). 15 Oates (1972). 16 Tiebout 1956. 17 If they are neighbors, amalgamation could be on the agenda since the residents’ preferences are now identical at least for this public service. 18 Musgrave (1983) asks, “Who Should Tax, Where, and What?” The tax assignment problem is part of a larger set of questions that can be referred to as the revenue assignment problem. The latter includes the design of intergovernmental grants and the framework for borrowing by subnational governments. 2 133 19 Where strict compliance with benefit finance is infeasible because of the difficulty or undesirability of exclusion from the benefits of public spending, the principle is, nonetheless, instructive. 20 This statement is incomplete if the benefits of public spending, as well as tax burdens, are considered. Taxes related closely to marginal benefits may finance expenditures that involve substantial inframarginal benefits. These inframarginal benefits may be of special value to low-income families. Obvious examples include the provision of safe drinking water. Many consumers would probably consider themselves better off if they had access to safe water, even if they had to pay for it. The problem is often access, not cost. 21 Subsidiarity in taxation was introduced to the European Union with the Maastricht Treaty amendments to the Treaty of Rome (Article 3B). The Commission of the European Communities (1991, p. 7) explained that subsidiarity requires that “Member States should remain free to determine their tax arrangements, except where these would lead to major distortions.” 22 Vertical imbalance may exist because subnational governments have difficulty implementing many taxes, but higher levels of government can implement almost any tax that a lower level of government can implement. 23 If a subnational government legislates and collects its own taxes, protected by meaningful constitutional safeguards of its right to do so, it clearly has a source of own revenues. 24 Even if subnational governments have own revenues, they may be unable to influence the amount of revenue they receive. This is true, for example, if the central government shares revenues from certain taxes with subnational governments. In such a case, these are own revenues but not marginal revenues of subnational governments. 25 An important prerequisite for the exercise of subnational fiscal autonomy is the ability to choose statutory tax rates. An issue here is why statutory and not effective tax rates. Effective tax rates would vary if subnational governments could alter deductions, exemptions, and so on, but this would mean changing the base and not the rates. This would increase the complexity of the system and the compliance costs of firms operating in multiple jurisdictions. 26 In the case of Canada for example, provinces that use the harmonized sales tax (HST), a VAT collected by the federal government for both it and the province at a jointly agreed rate (5% Federal and a provincial rate) are allowed a maximum 5% variation in their base from the federal government base. 27 Subnational constitutions or laws may limit any of these, but self-imposed restrictions in the constitutions of subnational governments differ from restrictions imposed from above by law or as part of a national constitution. In the case of Switzerland, for example, the definition of the income tax base and tax deductions is federal and applies to the twenty-six cantons. The cantons can choose their own tax rate schedule and the amount of deduction (but cannot add one to the federal list). The communes (local government units) can only decide the tax coefficient for balancing their budget in a true piggyback tax system. 28 Note first that “tax competition” is distinct from the Tiebout model: in Tiebout, it is the best alternative “basket of local public services – taxes” that determines the localization of economic agents: it is “fiscal competition” through benchmarking and mobility. In tax competition, it is assumed that the basket of public local public services is almost identical so that only taxes may be considered. Brennan and Buchanan (1983) provide a strong argument for tax competition (see also McLure 1986). This is only part of the story, although an important part. Because those who enjoy public goods cannot be excluded from enjoying the benefits, they have little incentive to reveal their preferences for such goods. There is thus a tendency to underprovide public goods that 134 tax competition might aggravate (see Gordon 1983 for a theoretical analysis of the inefficiencies that can result from decentralization, including tax competition). Benefit taxation helps to combat this source of market failure (Wildasin 1986). Tax competition makes it difficult for subnational governments to tax mobile factors—capital and highly educated or skilled labor—and therefore to engage in progressive taxation. 29 Inequities and economic distortions can also occur if the tax systems of various subnational governments do not mesh, resulting in gaps or overlaps in taxation. Within limits, these problems—which differ in importance from tax to tax—can and should be tolerated in the interest of gaining the benefits of fiscal decentralization. Serious complexities, inequities, and distortions can be achieved without greatly compromising the fiscal autonomy of subnational governments through intergovernmental compacts among subnational governments or the imposition of uniform ground rules by a higher level of government, for example, rules governing the definition and division of the corporate income tax base. 30 In a sense US states using federal AGI as their PIT tax base do this. 31 Although all subnational governments, acting as a group, can attempt to influence their share of revenues from these taxes, no subnational government, acting unilaterally, can hope to do so. 32 For a commented review of the literature on this, see Dafflon, 2007, 363-366. 33 In recent years, public finance economists have proposed that SNGs' financial capacity would be better measured with RTS methods (for Representative Tax System). RTS measures the per capita tax potential of SNGs and not the actual tax yield. RTS methodological steps are: 1) determine which taxes have to be included in the calculation (structural versus irregular tax yield, buoyant or not, all taxes or a selection, and so on); 2) if the capacity is computed on the potential tax yield and not the tax base, which are the referred "standard" tax rates or tax rates schedules; 3) with several tax sources, decide which weight should be given to each of them (the standard method is that weights are proportional to the respective potential tax yield); 4) decide if the calculation is made on an annual basis or an average of several (three, more?) years. Another aspect of the debate is whether SNGs' financial capacity should be measured with macroeconomic data, such as regional GDP per capita or RTS. But the discussion is rather semantic: with small open economies, regional GDP appears to be a rather broad evaluation; at the local level, this calculation does not exist. RTS avoid this trap. 34 Which may well be paid for by residents of SNGs already contributing to horizontal equalization if central and SNG tax capacity are correlated. 35 Courchene, 1998; Rechovsky, 2007: 400-409. 36 Bird and Vaillancourt, 2007: 262. 37 Take the example of primary education. Suppose SNG1 and SNG3 buy the same number of books for the same number of pupils. Does SNG3 overspend on fancier books, try harder to keep up with new pedagogical trends, or teach a different language group and thus face higher unit costs for otherwise identical books? Yet if one refers to the logic behind Box 1, further questions arise. Is SNG3's choice to follow a new pedagogical path an item of laboratory federalism, a decision taken in coordination with other SNGs (in this case, equalization is acceptable), or is it an own decision following the specific tastes of the constituency (no equalization)? If language is different, is the higher government concerned with the protection of minorities (equalization is acceptable) or are language differences not an issue (no equalization)? Not only is it difficult to isolate variables that affect costs from variables that indicate differences in public good preferences, but the answers (to laboratory federalism and preoccupation with minorities protection in the example) – and therefore the justification of expenditure needs equalization – belong to the realm of politics. 38 Dafflon and Mischler, 2007: 183-185. 135 39 Boothe and Vaillancourt, 2007: 48. Boex and Martinez-Vazquez, 2007: 291. 41 Dafflon, 2007: 370-371. 42 Dafflon and Mischler, 2008: 235. 40 136