Library

Document Details
DOWNLOAD
Language:
Type:
Conference Paper
Topic:
Publication Year:
2002
Centralization vs. Decentralization: The Devil is in the Details (draft)

Forum of Federations Forum des fédérations Centralization v. Decentralization : The Devil Is In The Details I – Introduction II – Intergovernmental Transfers Types of transfers What Allocation Formula for Block Grants? Table 1 – Transfers to Local Governments, Definition of Amount and Allocation Criteria, Brazil, Bolivia, Uruguay Impacts of Transfers Upon Local Taxation Figure 1 – Local Taxes as a Function of Transfers Table 2 – Tax Effort as a Function of Transfers per Capita, Parana Municipalities, 1996 Centralization v. Decentralization : The Devil Is In The Details Rémy Prud’homme & Anwar Shah (Draft, June 23, 2002) I – Introduction Those who expect to witness a fight between a pro-decentralization writer and an anti-decentralization writer will be dissapointed. Such a debate took place a decade ago, with some people more sensitive to the potential virtues of decentralization, and some other people (Prud’homme 1994, Tanzi 1995) more sensitive to its potential dangers. The “pro” emphazied the gains to be expected from a decentralized system, particularly in terms of allocative efficiency. The “anti” expressed doubts about th occurrence of some of these gains, and raised concerns about possible negative consequences, particularly in terms of macro-economic policy and of income redistribution. The debate was probably useful is throwing some light on the issue, in showing that decentralization could contribute to economic growth, in suggesting that it was not a panacea and should be pursued with a degree of prudence, in stimulating additional empirical research. Repeating this debate here and now, however, would be useless, for two main reasons. First, we have learned that decentralization is a multi-dimensional, many-faceted, concept. It cannot be reduced to a simplistic ratio, such as the ratio of sub-national government expenditures to total government expenditures (the so-called expenditures decentralization ratio), or the ratio of subnational taxes to total taxes (the so-called tax decentralization ratio). These are indeed important dimensions of decentralization, but decentralization is much more than that. It involves a large number of interrelated rules, institutions, constraints, mechanisms, practices, etc., which, in addition, are constantly changing over time. Even tax and expenditures decentralization ratios, that look like simple, quantitative, unambiguous, easy-to-measure, indicators, can be misleading. Not all local taxes are equally local. Shared taxes, so common in Latin America, or taxes raised by local governments at a rate decided by the central government, frequently found in Africa, for instance, are marquedly different from “true” local taxes, the rate of which is decided by the local government themselves. Adding the local share of shared taxes with true local taxes is adding apples and oranges1. The same is true of expenditures undertaken by local governments. In certain cases, they are genuinely local, that is decided by locally elected councils. But in other cases, (mandated education and health expenditures for instance) expenditures made by local governments have been decided and financed by the central government, with local governments being a mere passive instrument without any effective power. Adding expenditures of this type with genuine local expenditures is adding pears and mangos. Even from the simple viewpoint of tax and expenditures ratios, appraising the true degree of decentralization of a country therefore would require an in-depth understanding of the pecularities of its tax and expenditure system. On the basis of IMF statistics on tax and expenditures decentralization ratios, it is not possible to give a meaningful answer to such simple questions as: is Brazil more decentralized than Argentina, or less?; is Bolivia more decentralized than Uruguay, or less? This viewpoint, obviously, is not the only one that matters. Voting procedures and electoral rules are equally important. Are mayors elected for 3, 4, 5, or 6 years? Can they be reelected, and for how many terms? Can mayors be also members of the national Congress? How much are they paid, and who decides on their wages? Are they directly elected by popular vote, or by an elected council? Is this council elected by means of a proportional system, or of a majority system? Administrative rules are another area of importance. Is there a competent local bureaucracy? Is is stable, or does it change with the politicians? Is local participation systematically organized? Is it a widespread practice? How large are local jurisdictions? Do the borders of municipalities coincide with the borders of the socio-economic agglomerations? If not, what mechanisms, if any, exist for horizontal cooperation between municipalities? Are accounting nomenclatures and practices standardized, by law or practice? How is organized the auditing and control of subnational governments? Borrowing regulations and the nature of intergovernmental transfers (to be discussed below) are other areas that characterize a decentralization system. All this means that advocating “decentralization”, or even “more decentralization ” (or “less decentralization”), accros the board has not much meaning. The consequences of decentralization, both positive and negative, will depend much more upon the precise, specific, nature of a decentralization system than upon its “magnitude”. Decentralization is a qualitative, not a quantitative, phenomenon. It all depends upon the details. The second reason why a pro-anti decentralization debate would be sterile is that decentralization has been a political must. In most countries, it has been politically driven. The demand for decentralization came from politicians, perhaps from the grass roots, but certainly not from economists trying to improve the delivery of public services and the maximising tax-cum-expenditure packages. Decentralization redistributes power, and gives people a voice, in the form of a chance to vote. It contributes to political stability. A decentralized country is less likely to become a dictatorship than a centralized one. This is the main justification for decentralization. It is a very powerful one. And it is one that has economic implications, because some political stability is indeed a prerequisite for economic efficiency, stabilization, and redistribution. The issue at a given moment in a given country therefore is not wether decentralization is good or bad, and wether it should be increased or decreased. Rather, the issue is: since decentralization will take place anyway, what kind of decentralization is most appropriate? What matters, and what is worth discussing, are the details of decentralization schemes. The rest of this paper tries to illustrate this idea by focusssing on two important dimensions (or mechansisms or components) of a decentralization system: intergovernmental transfer systems, and subnational borrowing. The paper also briefly deals with two related issues namely the instutional arrangements for intergovernmental relations and bureaucratic culture under decentralized decision making. 1 The local share of a shared tax is considered a transfer, rather than a local tax in this paper. II – Intergovernmental Transfers The issue of intergovernmental transfers does not fit well in a pro v. anti decentralization debate. In a “pure” decentralization model —the one used by theorists— there are hardly any transfers, since taxes and expenditures are jointly decentralized. Indeed, the benefits of decentralization are contingent upon this joint decentralizaton. Each local government will choose the amount of tax-cum-expenditures, as well as the structure of taxes and of expenditures, that maximizes the welfare of its citizens-voters. Transfers do not even enter the picture, except in lateral and minor ways, such as the correction of spillover effects. In practice, however, intergovernmental transfers dominate the decentralization picture. In most cases, the central government subsidizes both intermediate level governments and local governments, and intermediate levels governments also subsidize local governments. In most countries, and not only in Latin America, transfers are by far the most important source of revenue of local governments. Decentralization is not about spending money raised locally, but in spending money received from above. How the money is spent, that is, how decentralization is effective, in any given country, depends in a large part upon the detailed characteristics of the transfer system developed. Types of transfers Transfers can take many different forms. Three basic distinctions can be introduced: the distinction between block grants and specific grants, the distinction between discretionary and automatic grants, and the distinction between open-ended and fixed total grants. These distinctions will help understand the nature of transfers, and show that not much can be said about the impact of “transfers” in general, but that this impact is very much a function of the type(s) of transfer selected. A first basic distinction can be made between transfers that are given with no strings attached, often called “block grants”, and transfers that are given with constraints on the use of the money transferred, often called “specific grants”. The commune that receives a block grant can do what it pleases with the money received. The commune that receives a specific grant must use the money received for a specific purpose decided by the central government, such as build a particular road, or pay the wages of schoolteachers. The distinction is not dichotomic. It is a matter of more or less, not a matter of either/or, and the freedom or the constraints on the use of transfer money are often relative. Consider an investment block grant, for instance. It is a transfer that must be used for investments only, not operating expenditures. The freedom to use the money received is restricted. But it remains large, since the receiving commune can invest in roads or in schools or in sanitation, as it thinks appropriate. Or take an education grant, as another example. The money received must be spent on education, not on transportation or health, but in education it can be spent on investments or on wages, as the commune wishes. Most transfers come with some strings attached, but these strings are more or less tight. Nevertheless, the distinction between block grants and specific grants is useful, and defines two different types of transfers. Which are better? There is no simple answer. Each type has advantages and drawbacks. Block grants are preferred by communes, because it gives them more freedom. Block grants are more in the spirit de decentralization. If one believes that communes know better than the central government what is good for their people, then block grants are better, and will lead to a more efficient allocation of resources. They are also easier to administer for the central government: it has “only” to decide on the amount to be allocated to each commune, and to send the money. Specific grants are often preferred by central governments, for good and also not so good reasons. If one is afraid that communes might not always make good use of the money they receive, then specific grants might be better. They will to a certain extent protect the people from misallocation of resources. One can also argue that the central government has a responsibility to ensure that all citizens benefit from certain services, such as education, and that it cannot give communes the freedom not to provide these services. On the other hand, specific grants are much more difficult to administer by the central government: they imply more complex allocation procedures, as well as monitoring. Over the past two decades, particularly in developed countries (such as the USA or France), but also in developing countries, there has been a marked shift from specific grants to block grants. A second, equally basic, distinction can be made between discretionary and automatic transfers. Discretionary transfers are decided on an case by case basis by the central government, according to the “needs” of each commune, assessed with the help of criteria that has often more implicit than explicit. Automatic transfers are formula-driven transfers; once the allocation formula has been decided upon by the central government, and in most countries the Parliament, it is applied automatically, by a computer. Which are better? In this case, there is a widely accepted answer: automatic transfers are to be preferred. Discretionary transfers could be said to be more flexible and more responsive to emergencies and special cases. But they open the door to political interference, arbitrariness and discrimination. In addition, they are unpredictable. A given commune does not know until late in the year how much it will receive, which makes responsible planning difficult or impossible. They induce communes officials to spend their time and energy lobbying to obtain more transfers rather than making good and wise expenditure decisions. Automatic transfers are also much easier to administer. Over the past decades, there has been worldwide, a clear move from discretionary to automatic transfers. As a matter of fact, purely discretionary transfers are by now very rare. A third basic distinction can also be introduced between open-ended subsidies and fixed total transfers. Fixed total transfers are transfers the total amount of which is first defined at the national level, then allocated to each commune. Most (but not all) block grants are of that type. Open-ended transfers are first defined for each commune, on the basis of their needs, or their expenditures, for instance. The sum of what is given to each commune produces the total amount of the transfer; this amount is the arriving point, not the starting point, of the procedure; it is not known at the beginning, which is why this type of transfer is called “open-ended”. Matching grants provide a typical example of open-ended transfers; the central government decides to give local governments 50% (or any other number) of their investments. Because the government cannot know exactly how much local governments will invest next year, it does not know how much it will have to give them next year. Ministries of Finance do not like open-ended transfers. Such transfers introduce a degree of uncertainty that make budget preparation and implementation at the national level more difficult. Local governments, on the other hand, like such transfers, because they know ahead of time how much they will receive as a function of their behavior. For this reason, open-ended transfers are often considered a more powerful incentive. They can orientate spending decisions (and even taxing decisions) of communes along the lines desired by the central government. If, for instance, the central government wants to stimulate communal investments in education rather than in health, it can establish an open-ended matching grant at 60% for education and an open-ended matching grant for health at 30%. These three types of typologies can of course be combined to form an eight position typology, with (1) discretionary open-ended block grants, (2) discretionary fixed total block grants (3) automatic open-ended block grants, (4) automatic fixed total block grants, (5) discretionary open-ended specific grants, (6) discretionary fixed total specific grants (7) automatic open-ended specific grants, and (8) automatic fixed total specific grants. The menu of transfers is therefore quite rich. To design a transfer system is to make a certain number of decisions about: the total amount of transfers in the country, the relative importance of block grants and specific grants, the allocation of block grants. These decisions are not independent of decisions to be taken about the other components of a decentralization program, such as the allocation of responsibilities, the allocation of taxes, or the administrative mechanisms to be put in place. The options open must nevertheless be discussed. One point seems quite clear: a transfer system should include a block grant, or at least an investment grant (a grant that must be utilized for investments only). Block grants are easy to administer. They can more easily than specific grants be utilized to compensate for differences between communes —one of the main objectives of a transfer system. They contribute to teach communes to make responsible choices between various types of expenditures. Admittedly, at least in the beginning, a number of communes may not have fully the administrative and political capability to make such choices. Yet, in the long run, communes must develop such skills. Making these choices wisely is the essence of effective decentralization, which in turn is the key to increased efficiency of the public sector. Communes must be given the means to decide what their preferred expenditures are. In the first years, some communes will perhaps make mistakes, or at least what central government officials will consider as mistakes. But these few and apparent mistakes are the price to be paid for the education of local governments. Waiting for the time when all communes are perfectly equipped to make choices, and only then give them some money, would be a sure recipe for no action. The real issue is rather: should there be also specific grants? There are pros and cons. On the “yes” side, it can be argued that specific grants ensure that some money will be spent by communes on selected issues of national interest. Specific grants can be seen as a safeguard against the potential “allocation mistakes” of communes discussed above, and that they are, for that matter, justified at least in a first phase. In many countries, the history of transfers begins with specific grants, which become block grants as the country matures. On the “no” side, it must be said that specific grants are difficult to administer. The central government (perhaps in the form of provincial governments) must have some difficult-to-obtain knowledge of the situation of each commune. Specific grants involve a complicated dialogue between the specialized ministries (of education, health, agriculture, etc.) and the ministries of Finance and Interior, as well as with provincial governors. There is also the following dilemma: specific grants are either open-ended, or total amount fixed, and each of these options raises problems. If a specific grant is open-ended (for instance a grant equal to 50% of education expenditures) the allocation is simple and raises only a monitoring problem; but the total cost to the central government of the transfer is unknown and hard to predict, which make budget planning difficult, a very serious issue. If the specific grant starts with a fixed total amount, there is no problem of budget planning; but the demand for subsidies coming from all communes is likely to exceed this fixed amount, making it necessary to devise an allocation formula that will be complicated, information-demanding and potentially unfair. What Allocation Formula for Block Grants? A block grant has two major components: (1) the definition of the total amount of transfer to be granted to the various communes, and (2) the allocation formula needed to allocate this amount to each of the communes. Total amount – The total amount is of course a political decision. It could be decided every year, in the framework of the national budget, according to the standard administrative and political budgetary procedures. This has, from the viewpoint of the ministry of Finance, the advantage of flexibility, because in that case the total amount can be adjusted (upwards or downwards) as a function of the macro-economic and public finance situation. But this discretionary definition of the total amount has, from the viewpoint of communes, the major disadvantage of uncertainty. They do not know from one year to the next how much they will receive in block grant money. They cannot count on it. This makes it impossible for them to plan and to conduct policies. This is particularly serious for the communes that receive a larger share of their resources from the block grant. It affects the entire credibility of the transfer system, and of the decentralization program. This is why, in many countries, particularly in Latin America, the total amount of block grant is defined as a pre-determined share of central government taxes or (more rarely) expenditures. The amount of block grant could therefore be set equal to 10% of central government taxes. Such a ratio is of course voted by Parliament for a certain number of years. This predetermined share need not be constant: it could, for instance, be set equal to 8% in the first two years, then 9% in the next three years, etc., or any other schedule. In some countries, the block grant is defined as a share of certain taxes only: for instance, 20% of the value-added tax. There is not much to gain here. A single tax is likely to be more volatile than total tax income, which would introduce some of the uncertainty that we are trying to avoid. Some compromise solutions might also be considered, that would limit both the uncertainty for communes and the lack of flexibility for the ministry of Finance. The total amount could for instance be determined each year in a discretionary fashion within predetermined upper and lower bounds: not 10% of total taxes, but more than 9% and less than 11% of total taxes. Or communes could have a guarantee that the total amount does not decrease in real terms from one year to the next (which would most probably imply a decrease in percentage of total taxes). Allocation formula – Once the total amount of block grant is determined, there remains the task of allocating it to the participating communes. Let us assume that the n communes of the country benefit. The problem is to divide the pie into n parts. This is generally done by an allocation formula that utilizes one of more allocation criteria. The general form of the formula is the following. Let: T = total amount of transfer; Ti = transfer to commune i (with ˜iTi = T); Ki = value of criteria k for commune i (with ˜Ki = K) We calculate Ti as Ti = T*Ki/K The question is: what allocation criteria should be selected. They must be relevant and justified. They must also be such that the value of the criteria for each of the communes is known with a reasonable degree of certainty. A possible list of allocation criteria would include: – Population; all other things equal, the more populated a commune, the more transfers it deserves; – Local tax base: as mentioned above, compensating for differences in tax bases is a major objective of block grants2: communes with a low tax base should receive more than communes with a high tax base, on a per capita basis; – Needs: communes with specific needs, resulting from geography or history or demography, might receive more than others, all other things equal; the geographic size of a commune, for instance, might be said to imply heavier transport or equipment investments; or communes that have suffered from war damages can claim additional transfers; – Endowment: presently, communes are not all equally well equipped in terms of roads, school buildings, dispensaries or hospitals, etc.; those that have less might be entitled to obtain more transfer money; the problem with this criteria is that one can also argue that those communes that are well endowed need more money to maintain, repair and operate their stock of investments; – Development: poorer communes (in terms of output, not of tax bases), or, since communal income is usually not known, communes located in less developed regions of the country, might need more money than richer and/or better located communes; – Fixed costs: it can be argued that, irrespective of size, operating a commune implies fixed administrative costs, that call for a constant lump sum transfer to each commune. This list could be lengthened. But it would not be very useful. An allocation formula must remain simple. It cannot include too many allocation criteria. Complex formulas are difficult to understand, they are difficult to administer, and in the end too many criteria cancel each other. It is probably reasonable to consider only a few criteria, at least at the beginning. Population is an obvious candidate. It is easy to understand, not difficult to administer (assuming population census are reliable), and basically fair. Local tax base is another likely candidate, provided that it is known, and that tax bases are assessed in a comparable fashion countrywide (which probably implies an assessment by the central government administration). 2 The allocation formula presented above is not appropriate for this criteria; a more complicated formula, indicating to what extent one wants to compensate, must be utilized. The simplest way to combine several criteria is to divide the total block grant amount into several parts, and to allocate each part with the help of one criterion. Let us assume that only two criteria are retained: population, and local tax base. It can be decided that 60% of the total amount will be allocated prorata population, and 40% prorata local tax base. Let us call T, the total amount of block grant; Pi, the population of commune i; bi, the tax base per capita in commune i; and define bm as the tax base per capita beyond which there will not be any compensation, the allocation formula will become: Ti = 0.6*T*Pi/˜iPi + 0.4*T*(bm-bi)Pi/˜i(bn-bi)Pi when bi>bm Ti = 0.6*T*Pi/˜iPi when bi>bm One can check that with such a formula, the sum of the block grant allocated to each commune is equal to the total amount of the block grant. Table 1 below provides, for three Mercosur (or associate) countries, a brief description of existing transfer systems. It shows how crude they are, in terms of types of transfer used, of definition of total amounts of transfers, and of allocation criteria. Table 1 – Transfers to Local Governments, Definition of Amount and Allocation Criteria, Brazil, Bolivia, Uruguay Relative Definition of Total Allocation Importance Amount criteria Brazil (Parana) 1996 Fed. Municipal Fund 37% 22.5% of IPI & Income tax Complex formula Fed. Invest. Grants 14% Federal Budget Negociated State transfer 6% 50% of State auto tax Origin State transfer 44% 25% of State ICMS Complex formula with 75% Origin Bolivia 1999 Coparticipacion for communes42% 20% of (most) “national taxes” Population 1992 Matching grants for communes17% Mostly foreign aid Negociated Block grant for prefectures 8% 25% of national oil & gas tax 50%: equal; 50%: Pop Block grant for prefectures 12% 100% of royalties/natural resources. Origin Compensation Fund for pref. 4% Open-ended Compensating formula Matching grants for prefectures 17% Mostly foreign aid Negociated Uruguay 1999 Shared tax 13% 5% of oil & tobacco tax Population & area Shared tax 13% 6% of diesei oil tax Origin Shared tax 4% 10% of tax on casinos Origin Shared tax 10% 18% of rural land tax Origin Reimbursement 23% Open-ended Contrib.to pension fund Road grant 9% Discretionary Pop., area, km of roads DIPRODE grant 12% Discretionary Population & area Urban dev. Grant 8% Foreign loan Population & area Sanitation grant 18% Foreign loan Ad hoc (Montevideo) Sources : Author’s field work Notes: For Brazil, only transfers to local governments in the State of Parana are considered; for Bolivia, transfers to prefectures, which are only semi-local governments, are also considered; In Uruguay, there is only one level of sub-national government The way total amounts of transfers are defined is often questionable. Most transfers are based on just one or few taxes, with the risk of erratic behaviour. Allocation criteria are even more questionable. In many cases, the amount allocated to a given local government is determined by how much of the shared tax has been raised by the central government in that particular local government. This is even the case (in Bolivia) for royalties on natural resources; this greatly favours those prefectures that are lucky enough to have such natural resources on their territory. These “origin”-based criteria tend to be unfair, and to favour the richer local governments. None of the 19 transfers considered includes the per capita tax base of local government as an allocation criteria. In theory, transfers can and should compensate for differences in tax bases. In practice, very few seem to do so. Impacts of Transfers Upon Local Taxation A different issue concerns the impact of transfers upon local taxation. There are reasons to fear that transfers undermine local taxation. The reasons are political. When the share of transfers in the resources of a local government is not large, such as 10% for instance, increasing taxes by a given percentage, let us say by 20%, will make it possible to increase local expenditures by a percentage close to 20%, 18% in our example. This may be politically attractive. Voters might be please to see public expenditures increase by 18% even though it means paying 20% more in taxes. When the share of transfers in the resources of a local government becomes too large, let us say 90%, things are completely different. To increase total expenditures by a mere 10% would require an increase of local taxation by 100%. Voters would not follow, and taking such a decision would be political suicide. This means that the motivation to increase local taxation decreases with the magnitude of transfers. At the beginning, when transfers are small, there are strong incentives to have significant local taxes. When transfers increase, such incentives tend to disappear, and the importance of local taxes decline. One can even hypothesize that there is a threshold level of transfers. Beyond this threshold, local taxation collapses. This is illustrated by figure 1, where T is this threshold level of transfers. Figure 1 – Local Taxes as a Function of Transfers Localtaxation Transfers T An econometric analysis conducted on the 367 municipalities of the State of Parana (Brazil) fully supported this view3. Taxes per capita were estimated as a function of several variables, including transfers per capita. Transfers per capita appeared significant, particularly when introduced in logistical form (the form depicted in figure 1). This made it possible to estimate the threshold value. The main findings are summarized in table 7. Municipalities that receive transfers per capita lower than 112 reais per capita raise taxes that are on average 67 reais per capita. When transfers received were higher than this threshold, taxes per capita went down to about 15 reais per capita. As table 2 indicates, when transfers increased, taxes raised decreased. As table 2 also shows, this pattern cannot be explained by the income level of municipalities. 3 To be fair, a similar study conducted on a sample of French municipalities failed to confirm this hypothesis. Table 2 – Tax Effort as a Function of Transfers per Capita, Parana Municipalities, 1996 Municipalities Number of Average Average Average Average with transfers/cap municipalities Transfers/cap GDP/cap Tax/cap Tax/GDP (s) (y) (t1) (t2) Below threshold 29 91 1,867 69.0 66.9 Above threshold 339 225 2,100 17.6 14.6 Between 112 and 159 72 138 2,172 24.0 22.4 Between 159 and 204 100 181 1,971 18,1 15.7 Above 204 166 289 2,147 14.6 10.6 Notes: The threshold is 112 Reais/cap To conclude this section on intergovernmental transfers, it appears that transfers are a basic component of decentralization systems. They are neither good not bad in themselves. But transfers can take so many different forms, with consequently so different impacts, that generalizations about transfers —and the decentralization systems they contribute to define— are bound to be of limited interest. On an example like this, one sees clearly that the policy issue is not wether decentralization should be increased or not, but rather: how can transfer systems can be improved? III. Sub-national Borrowing The capital finance needs of developing and transition countries are currently estimated to be in excess of $100 billion a year by the World Bank. Most of these investment needs are for local public infrastructure. Water and sewerage projects account for one-half of these investment needs. Local governments typically command a lion’s share in public sector investments with a low of 30% in developing countries and 70% in industrialized countries. These investments are financed by taxes, charges, reserves, capital grants, borrowing and private equity and debt in concessions/build-operate-transfer (BOT). Borrowing has traditionally served as the most important source of finance of such investments in industrial countries. This is because resort to borrowing enhances intergenerational equity as these projects are long lived and yield returns over several generations, so that the cost should equally be shared over the same generations. Such burden-sharing among generations enables local governments to undertake these large and lumpy investments. Further, this source of finance enables these governments to tailor projects so that they are consistent with local needs without being constrained by design choices made by higher level governments. In developing countries, such projects are typically financed by capital grants and on-lending from higher level governments as direct credit market access by local governments is usually not available. Since the capital finance needs of developing countries are quite large and cannot be met by traditional sources of finance, subnational credit market access represents a major challenge to finance these development needs. Credit market access at intermediate levels of government (states and provinces) in decentralized federal industrial countries usually carries few restraints. For example domestic and foreign borrowing by states/provinces in US and Canada is not subject to any federally imposed constraints. In the USA, on the contrary, income from state bonds is exempt from federal income taxes. The fiscal conservatism of these governments in financing capital needs primarily arises from limitations imposed by state constitutions and by credit market discipline. Credit market access is, however, closely controlled for both state and local governments in unitary (China, France, Indonesia, UK and Japan) and centralized federal countries (e.g., India, Pakistan and Australia until 1993) and for local governments only in decentralized federal countries (Canada, USA, Germany). In Germany borrowing by local governments is conditional on their cash flow position and subject to Laender approval. This is because an unrestrained access could potentially put the national government at risk in view of the explicit and/or implicit bailout guarantees. Such controls are also needed to limit public demands for capital investment during the boom periods and stimulate such demand during economic downturns (see Box 1 for a summary of Latin American regimes and Box 2 for a few international examples). Passive controls on subnational borrowing take many forms from broader guidelines on allowable ranges for the ratio of debt to revenues and the ratio of debt charges to own-source revenues, to more specific rules such as the “golden rule” for local debt commonly adopted in most federations. Under the golden rule, borrowing is permitted only for capital projects and local governments cannot finance current deficits from this source except to smooth over fluctuations in revenue inflows and outflows within a given fiscal year. This is the practice in Canada, USA, Germany and Switzerland. The European Union has imposed guidelines on deficit and debt limits as discussed earlier and has prohibited central banks from bailing out any governments. In Brazil, Senate Resolution 11 (1993) has restricted new state borrowing by two formal rules: (a) total debt service cannot exceed the state operating surplus during the past year or 15 percent of its revenues, whichever is less; and (b) new borrowing within any 12 month period cannot exceed the level of existing debt service or 27 percent of revenues, whichever is less. Box 1. Approaches to Subnational Borrowing in the Latin American Region Prohibition: Chile (domestic and foreign), Mexico (foreign), Venezuela (foreign) Administrative controls on domestic borrowing: Bolivia, Brazil, Guatemala, Nicaragua, Uruguay Administrative controls on foreign borrowing: Argentina, Bolivia, Brazil, Colombia, Dominican Republic, El Salvador, Guatemala, Honduras, Nicaragua, Uruguay Debt limits: Argentina, Brazil, Colombia, Costa Rica, Ecuador, Mexico, Peru, Uruguay Golden Rule: Bolivia, Colombia, Ecuador, El Salvador (on foreign debt), Mexico, Nicaragua, Peru Market discipline only: El Salvador (domestic debt), Panama Source: Freire, Huertas and Darche (2000) More active controls on such borrowing include centrally specified limits on capital spending by each municipality as in the UK; project submission and approval as in the province of Ontario, Canada; approval for bond finance as in Japan; approval of amount of borrowing and rates as in Denmark (usually restricted to energy and urban renewal projects only) and France; and seeking community mandate on borrowing plans through popular referenda as infrequently done in United States and Canada and routinely required in Switzerland. In developing countries, central controls are even more extensive and crude and most of these countries do not allow credit market access to local governments. In India and Pakistan even borrowing at state level requires central approval as long as states and provinces owe any debt to the federal government. Net Box 2: Approaches to Sub-national Capital Finance – International Examples Pakistan: Pakistan’s Annual Development Plan (ADP) process typifies a highly centralized system. The ADP process begins with a municipality’s submission of a project proposal to the provincial government, where it is subjected to technical review; if technically approved, it is then included in a larger pool of projects eligible for financing. Financing decisions are made annually, and begin with an estimate of overall resource availability by the central government’s ministry of finance. The provincial government then makes a tentative match of resources with projects; then forwards it recommendations to the central Government’s annual plan coordination committee; which approves size and sectoral allocation of the overall package, and then submits it to the national economic council, presided over by the president. This lengthy process does succeed in eliminating technically unsound projects, and matches resources to projects, but incorporates no mechanism for weighing the degree of local commitment to investments projects. Mexico: Mexico’s National Solidarity Program (PRONASOL) is funded from an earmarked share of the national budget. Allocations are distributed among states by formula, with a fixed proportion earmarked for allocation by mayors. Allocations to municipalities are based in part upon political considerations. But within a given recipient municipio, the allocation of funds among projects draws upon a well developed system of negotiation between the mayor and community groups, in which PRONASOL funding is made conditional upon the community’s willingness to provide counterpart contributions in cash or in kind. While mayors have the latitude to vary the terms of each project agreement, the matching requirement is universal. Colombia: Colombia’s municipal credit institution, Financiera de Desarrollo Territorial (FINDETER), does not lend directly to municipal governments but operates as an autonomous discount agency to private sector and state-owned commercial banks that make the loans, appraise the projects, and monitor performance. Under the control of the finance ministry, it has been relatively insulated from political pressures. The system’s funding does not rely on government budgetary appropriation but rather on bonds, recycling of it loans, and foreign credits from bilateral and multi-lateral sources. Canada: Provincial governments have free and uncontrolled access to borrowing on domestic and international markets. Municipal borrowing is subject to provincial scrutiny and approval. Once approval is granted, municipalities are free to borrow from the private sector. Although no additional role is required, provincial authorities can provide a variety of additional assistance including loan guarantees, transfers to cover loans, assistance in marketing of municipal debt, and loans. Some provinces borrow for the purpose of relending to small local governments United States: Both state and local governments have unrestricted access to capital markets. Municipalities, or municipal agencies, issue bonds. Creditworthiness of municipal offerings is determined by private rating companies. Federal and state governments promote bond issues through income tax exemptions on interest payments. Australia: The Australia Loan Council (ALC), established in 1927 as a central credit allocation mechanism for sub-national borrowing was seen as an outmoded institution for the 21st century. The ALC now works as a coordinating agency for sharing information on federal-state-local fiscal positions and macroeconomic strategies. States are required to justify their borrowing plans for consistency with own fiscal needs and overall macro strategy for the nation as a whole. If these requirements are met, state are free to access financial markets for raising the required funds. European Union: Maastricht Treaty imposes two quantitative guidelines to ensure price stability oriented monetary policy. These are: (a) the deficit must be less than 3% of GDP and (b) the debt/GDP ratio must not exceed 0.6 (60%). Further in the event of default, there should not be any bailout by member central banks or the European Central Bank. Sources: World Bank (1994, July), and Shah (1996) federal lending to states in India and provinces in Pakistan in 2001-2002 was close to zero or negative as states/provincial debt service payments equaled or exceeded new loans. In view of the above constraints, local borrowing in most industrial countries is primarily from domestic markets and higher level governments. Only local governments in Canada, Denmark and Norway have foreign debt obligations in access of 10% of their total debt. In developing countries, state and local debt obligations are primarily owed to the central government. A significant degree of tax decentralization and secured sources of revenues through formula-based transfers is, however, opening up possibilities of global market access to sub-national governments especially in Latin America. Over the last few years, a handful of local governments ranging from the city of Rio de Janeiro in Brazil to Argentina’s provinces of Buenos Aires and Mendoza have sold hundreds of million dollars of notes to American and European investors and many other governments are eager to issue bonds on the global market. The Buenos Aires bond was oversubscribed when it was floated in April 1997 (see Freire, Huertas and Darche, 2000). In a decentralized fiscal environment, subnational government access to credit markets poses significant risks for macro stabilization policies of the national government as the possibility of imposing credit rationing and direct controls are significantly constrained by the constitutional division of powers. These risks are disproportionately higher if there is a strong dependence of subnational governments on central sources of revenues. In those circumstances, a bailout risk would be much higher but the market would fail to capitalize such risks in view of its anticipation of a central government bailout. For example, past bailout practice and pledging of central transfers in Argentina create expectations on the part of commercial investors that provinces cannot fail. Decentralized fiscal systems rely upon a combination of credit market discipline, moral suasion and agreed upon rules to impose financial discipline on subnational governments. Which system works better is an empirical question worthy of further research. The available evidence nevertheless points to a superior performance of decentralized systems in restraining subnational debt. Central controls as imposed in France, Spain, UK , India, Pakistan and Australia (till 1992 under the old Australian Loan Council) failed to keep subnational debt in check as intergovernmental gaming led to weaker discipline and the possibility of central bailouts encouraged less rigorous scrutiny by the financial sector (see Box 2). Decentralized federations, on the other hand, rely on a combination of guidelines, intergovernmental cooperation and market discipline to keep local government debt within sustainable limits. Intergovernmental cooperation or moral suasion is achieved through executive federalism as in Canada, or multilateral information exchange through the New Australian Loan Council as in Australia, or through bilateral negotiations as in Denmark. The cornerstone of financial discipline under a decentralized fiscal system is the market discipline enhanced by an enabling public policy environment that stresses tax decentralization with greater reliance on locally raised revenues, central bank independence, disengagement of governments from ownership of commercial banks, no bailouts by the central bank or by a higher level government and requirements for public dissemination of information on public finances. Societal conservatism as in Switzerland introduces an added discipline. The 1996 State debt crisis in Brazil should not have come as a surprise to an informed observer. Brazil opted for a decentralized fiscal constitution but failed to adopt appropriate policies and develop relevant institutions to ensure market discipline in such environment. It allowed states to own commercial banks and borrow from these in a relatively unconstrained fashion while holding open the possibility of a federal government bailout in the event of default. Only recently has Brazil moved to create an enabling framework for credit market discipline for subnational borrowing (see also Ter-Minasian, 1996). Other initiatives to control state/local debt in the past included: sale or rigid controls over state owned banks; privatization of utilities; downsizing; and restructuring and harmonization of the state value added tax (ICMS) to limit its potential for state industrial policy (see Afonso and Lobo, 1996). . More recently, the Law of Fiscal Responsibility has imposed controls on both the existing stock of debt and new borrowing operations. Senate debt ceilings are now strictly enforced by the Federal Ministry of Finance on state and local governments. Any temporary departure from such ceilings must be corrected within one year and during this period, the federal discretionary transfers and guarantees are not available to the non-compliant jurisdiction and it also cannot issue any new debt. The Law prohibits intergovernmental lending, and contracting new debt during the last six months of the tenure of any administration. It further permits central government to garnish central transfers and tax shares against loan repayments (see Dillinger 2001). Facilitating Local Access to Credit Local access to credit requires well functioning financial markets and credit worthy local governments. These pre-requisites are easily met in industrial countries. In spite of this, traditions for assisting local governments by higher level governments are well established in these countries. An interest subsidy to state and local borrowing is available in the USA as the interest income of such bonds is exempt from federal taxation. Needless to say, such a subsidy has many distortionary effects: it favors richer jurisdictions and higher income individuals; it discriminates against non-debt sources of finance such as reserves and equity; it favors investments by local governments rather than autonomous bodies and it discourages private sector participation in the form of concessions and BOT alternatives. Various US states assist borrowing by small local governments through the establishment of municipal bond banks (MBBs). MBBs are established as autonomous state agencies that issue tax exempt securities to investors and apply the proceeds to purchase collective bond issue of several local governments. By pooling a number of smaller issues and by using superior credit rating of the state, MBBs reduce the cost of borrowing to smaller communities (see World Bank, 1996 and El Daher, 1996). In Canada, most provinces assist local governments with the engineering, financial and economic analysis of projects. Local governments in Alberta, British Columbia and Nova Scotia are assisted in their borrowing through provincial finance corporations which use the higher credit ratings of the province to lower costs of funds for local governments. Some provinces, notably Manitoba and Quebec, assist in the preparation and marketing of local debt. Canadian provincial governments on occasion have also provided debt relief to their local governments. Autonomous agencies run on commercial principles to assist local borrowing exist in western Europe and Japan. The Municipal Finance of Finland is owned by the association of local governments and provides debt pooling for municipal governments. Similarly Kommun Invst of Sweden is owned by the association of local governments but privately managed to provide credit to local governments. The Credit Communal de Belgique of Belgium is jointly owned by central and local governments and deposits are the main source of finance. The DEXIA in France is privately owned and raises resources entirely though bond issues. The Banco de Credito Local in Spain is also privately managed and uses bond finance. In Denmark, local governments have collectively established a cooperative municipal bank. In UK the Public Works Loan Board channels central financing to local public works. An important lesson arising from industrial countries’ experience is that municipal finance corporations operate well when they are run on commercial principles and compete for capital and borrowers. In such an environment, such agencies allow pooling of risk, better utilize economies of scale and bring to bear their knowledge of local governments and their financing potentials to provide access to commercial credit on more favorable terms (see McMillan, 2002). In developing countries, undeveloped markets for long term credit and weak municipal creditworthiness limit municipal access to credit. Nevertheless, the predominant central government policy emphasis is on central controls and consequently less attention has been paid to assistance for borrowing. In a few countries such assistance is available through specialized institutions and central guarantees to jump start municipal access to credit. Ecuador, Indonesia, Jordon, Morocco, Philippines, Pakistan, Turkey and Tunisia have established municipal development banks/funds/facilities for local borrowing. These institutions are quite fragile, not likely to be sustainable and are open to political influences. Interest rate subsidies provided through these institutions impede emerging capital market alternatives. Furthermore these institutions fail to smooth the transition to a market based capital finance system. Colombia and the Czech Republic provide a rediscount facility to facilitate local access to commercial credit (see Tables 3 and 4 for deils on functions and sources of finance of selected municipal development funds). Thailand has established a guarantee fund to assist local governments and the private sector in financing of infrastructure investments (see Gouarne, 1996, Peterson 2000). Concluding Remarks on Subnational Borrowing In conclusion, the menu of choices available to local governments for financing capital projects are quite limited and available alternatives are not conducive to developing a sustainable institutional environment for such finance. This is because macroeconomic instability and lack of fiscal discipline and appropriate regulatory regimes has impeded the development of financial and capital markets. In addition, revenue capacity at the local level is limited due to tax centralization. A first transitory step to provide limited credit market access to local governments may be to establish municipal finance corporations run on commercial principles and to encourage the development of municipal rating agencies to assist in such borrowing. Tax decentralization is also important to establish private sector confidence in lending to local governments and sharing in the risks and rewards of such lending. Central government bailouts and guarantee of subnational debt should however, be ruled out. Transparency in local budgeting and independent credit rating agencies are also essential to smooth the transition to a market based approach to sub-national lending. Table 3: Functions Performed by Municipal Development Funds Source: Peterson (2000) Table 4 : Sources of Funds by Municipal Development Funds Source: Peterson (2000) IV. Institutional arrangements for intergovernmental relations Adherence to federalism principles or “getting prices right” or even “getting the rules of the game right” as discussed earlier is a necessary but not a sufficient conditions for the success of decentralized decision making. Complementary formal and informal institutions are needed to ensure that all players in the game adhere to agreed upon set of ground rules and deviant behavior is properly dealt with. In the following, we discuss selected aspects of this consideration. Institutions and processes of intergovernmental coordination Federal countries require both formal and informal institutions of intergovernmental coordination. In some federal countries, areas of potential conflict among different levels of government is minimized through clear separation of national and sub-national responsibilities (the so-called layer-cake model of federalism as practiced in Australia, Canada, India and Pakistan) and the two levels interact through meetings of officials and ministers (executive federalism) and in Australia, India and Pakistan through federal unilateralism. Some countries place a greater premium on a common response through shared or joint tasks such as Germany, a federal country and the Republic of South Africa, a pseudo federal country. In these countries, in addition to executive federalism, the upper houses of parliament (Bundesrat and the Council of Provinces) play a key role in intergovernmental coordination. In countries with overlapping responsibilities (the so-called marble cake model of federalism), such as United States and Brazil, state lobby of Congress and interstate relations serve coordinating roles. In China, where growth concerns have imposed a federalism structure on a unitary country, regional communist party bosses/ governors exercise a moderating influence on otherwise monolithic orientation of the State Council. Constitutional provisions per se can also provide coordinating influences. For example, in some federal countries, constitutional provisions require that all legislation recognize that ultimate power rests with the people. For example, all legislation in Canada must conform to the Canadian Charter of Rights. In Switzerland, a confederation by law but a federal country in practice, major legislative changes require approval by referenda. In Switzerland, there is also a strong tradition of coordination through consensus initiatives by cantons. Institutional Arrangements for Fiscal Relations The structure of intergovernmental fiscal relations, especially the system of grants, must be determined by some body. There are five main alternatives. The first is for the federal government alone to decide on it. This alternative negates federalism and would not be acceptable in many countries. The second is to set up a quasi-independent body, such as a grants commission, whose purpose is to design and reform the system as practiced in Australia, India and the Republic of South Africa. This alternative is prone to more ideal solutions rather than pragmatic approaches and therefore runs the risk of presenting complex solutions and recommendations that may not be politically palatable. The third alternative is to use federal-state committees to negotiate the terms of the system as done in Canada. The fourth alternative is to have a joint intergovernmental cum inter-legislative commission such as the Finance Commission in Pakistan and the fifth alternative is to have an intergovernmental legislative body such as the upper house of the German Parliament (Bundesrat) as in Germany. The latter three systems allow for explicit political inputs from the jurisdictions involved, and therefore likely to opt for simple and feasible but less than ideal (compromise) solutions. Institutions of Accountability Institutions of accountability hold the key to the success of decentralized decision making. This entails institutions and mechanisms for citizens voice and exit , norms and networks of civic engagement (“social capital” according to Putnam), social consensus (Williamson, 1994 and Weingast, 1993 ), preservation instinct of a “stationary bandit” who monopolizes and rationalizes theft in the form of taxes (Olson, 1993), judicial accountability, vertical and horizontal accountability. The citizen voice and exit require institutions of democratic participation, and accountability provisions for elected officials. The origins and success of decentralization programs in Latin America is traceable to the democratic traditions that emerged in the continent in late 1980s. In Philippines, recently enacted local government legislation while empowering these governments have provided for regular elections and recall of elected officials for a breach of public trust (see the Republic of Philippines Act No. 7160, the Local Government Code, 1991). While norms and networks of civic engagement were reasonably well developed in pre-colonial traditionalist societies found in many developing countries such as the Panchayat Raj in Pre-British India, these institutions withered away either under the colonial rule or subsequently under centralized bureaucratic governance structures. The net result has been rise of opportunism and social distrust culminating in dysfunctional societies when formal institutions of governance failed. The African and the South Asian development fiascoes share this common underpinning. Societal consensus on economic and political rights is also conducive to accountability at all levels. According to Weingast (1993), this consensus need not take any formal expression but would work so long as a majority of people share a common belief as to the limits of governmental intervention and are willing to police those limits by withdrawing their support from a government that fails to abide by them (see Weingast, 1993, p.306). Preservation instincts of a stationary bandit also respect accountability (see Olson, 1993). This is because, the stationary bandit strengthens his grip on power, so long as economic performance is strong and citizens see their well being improved. This explains partly the success of the Asian Tigers and the failure of some South/Southeast Asian regimes. The latter regimes were controlled by “roving bandits” whose main aim was to make the loot to pad their overseas bank accounts and then disappear in a foreign haven. Judicial accountability strengthens the credibility of public commitments. This is particularly important for transition economies, where framework laws on property rights, corporate legal ownership and control, bankruptcy, and financial accounting and control are not fully developed. Interestingly enough, judicial accountability is much more difficult to enforce in a parliamentary democracy than in a presidential system which respects separation of legislative and executive functions. This is because, under a parliamentary democracy, the executive branch can override judicial accountability by amending the legislation – a game played ad infinitum in Pakistan to undermine a decentralized federal constitution. Judicial accountability is further compromised under a British style civil service organization as in India and Pakistan where divisional and district commissioners hold simultaneously executive, legislative and judicial powers. As noted by Montesquieu (1970), such a situation is ripe for the abuse of powers as “ … When the legislative and executive powers are united in the same body of magistrates, there can be no liberty;… Again, there is no liberty, if the judiciary power is not separated from the legislative and executive” (Montesquieu, 1970, p.397). Traditional Channels of Accountability The audit, inspection and control functions should be strengthened, since they tend to be quite weak in transition and developing economies. The auditor-general should be given greater authority and autonomy in exercising his mandate. At the same time, a case can be made for loosening the constraints of the central planning process in developing countries. Central plans lead to a centralization of authority, a reduction in flexibility, innovation and autonomy at the local level and delays in private sector activity. V. Aligning Operational Capacity with the Authorizing Environment: Reorienting bureaucratic culture to service delivery In developing and transition economies (DTEs) operational capacity for local governance is deficient. This deficiency, however, can be overcome in the short run by borrowing such capacities from the national governments, from other local governments and from the private sector and from the civil society. In the long run, training of staff and creating an enabling environment for competitive service delivery through partnership with the private sector and civil society can augment operational capacity. A matter of greater concern in DTEs is that the available capacity is not geared towards serving the citizen-voters. A similar bureaucratic culture prevailed in Western Europe not long ago. German philosopher Nietze characterized the government as “coldest of all cold monsters – whatever it says it lies – and whatever it has – it has stolen.” Over the years, industrial countries have shown a remarkable change in the performance of their public sectors. It is interesting to note that this change was brought about not through a system of hierarchical controls, as is the focus in most developing countries, but more through strengthened accountability to citizens at large. The elected representatives made a commitment along the lines of the oath required of the members of the city of Athens which stated that: “We will strive increasingly to quicken the public sense of public duty; That thus…we will transmit this city Not only not less, but greater, better and more beautiful Than it was transmitted to us.” This accountability for results was further strengthened by accountability of executive to the legislature branch. Overall the emphasis of these systems of accountability has been to bring about a change in both the bureaucratic culture and the incentives public employees face. This is done by steering attention away from internal bureaucratic processes and input controls (hard controls) to accountability for results (soft controls). While various countries have followed diverse policies to achieve this transformation, the underlying framework driving these reform is uniform and firmly grounded in the results oriented management and evaluation framework (ROME). Under ROME, a results based chain provides a yardstick for measuring public sector performance. Results Oriented Management and Evaluation (ROME) Chain: Program/project => inputs=> activities=> outputs=>reach (stakeholders positively or adversely affected)=> outcome (purpose) => impact (goal)=> citizen feedback and evaluations=>Program design => program/project Most ROME related approaches have the following common elements: (a) Contracts/work programs based upon pre-specified output and performance targets and budgetary allocations; (b) Managerial flexibility but accountability for results; (c) Subsidiarity principle; (d) Incentives for cost efficiency and (f) Citizen charter, bottom-up accountability. Box 3. Making the Dog Wag Its Tail: ROME Culture ROME provides a coherent framework for strategic planning and management based upon learning and accountability in a decentralized environment. As the focus of the approach is on learning, failure to meet commitments in the short run may be tolerated but a failure to share values and persistent failure to meet commitments invite severe sanctions. ROME framework calls for competitive wages and task specialization and lack of formal tenures for public personnel. Public providers are given the freedom to succeed or fail. Instead public employees hold the jobs so long as they are able to fulfill the terms of their contracts. Persistent failures initiate the exit process. Responsiveness to citizenry and accountability for results are the cornerstone of this approach (see Box 4 for an example). A recent empirical study by Gurgur and Shah (1999) supports this view as it shows that political and bureaucratic culture and centralization of authority represent the most significant determinants of corruption in a sample of 30 countries. In view of this evidence, the ROME framework offers a great potential in developing countries to improve public sector governance by nurturing a responsive and accountable governance. Administrative decentralization is a pre-requisite for implementation of ROME. Administrative decentralization requires lack of any ex ante controls over the decision by local governments to hire, fire and set terms of employment of local staff. To improve tax collection or the delivery of local public services, local government should have the freedom to contract own taxing and spending responsibilities. Furthermore, local governments should have the authority to pass bye-laws in their spheres of responsibility without having to obtain prior clearance from the higher level government. Box 4 . An Example of a Performance Oriented Grant: Education grant to set minimum standards while encouraging competition and innovation Allocation basis among local governments: school age population Distribution to providers: Equal per pupil to both government and private schools Conditions: Universal access to primary and secondary education regardless of parents’ income. Improvements in achievement scores. No condition on the use of grant funds. Penalties for non-compliance with standards: Public censure, reduction of grant funds. Incentives for cost efficiency: Retention of savings Source: Shah (1999) Local governments in a large majority of countries do not have the authority to hire and fire senior local staff. Eastern European countries represent an important exception in this regard. The freedom to contract own responsibilities is typically available but this option in some cases, e.g. in Philippines, is constrained to the expenditure side only. Regulatory authority for municipal services is usually available to local governments in most countries. Overall, while there has been significant progress in decentralization in developing countries in recent years . Such progress has been much slower in the area of local government autonomy for own civil service. In Pakistan and Indonesia, lack of such autonomy is considered the Achilles’ heel of the devolution as the provincially/centrally transferred civil servants have no personal stake in the success of the devolution. VI. Conclusions (To be written) ANNEX TABLES Annex Text Table A1 Criteria For the Design of Intergovernmental Fiscal Transfers Autonomy: Subnational governments should have complete independence and flexibility in setting priorities and should not be constrained by categorical structure of programs and uncertainty associated with decision making at the center. Tax base sharing – allowing subnational governments to introduce their own tax rates on central bases, formula based revenue sharing or block grants – is consistent with this objective. Revenue adequacy: Subnational governments should have adequate revenues to discharge designated responsibilities. Equity: Allocated funds should vary directly with fiscal need factors and inversely with the taxable capacity. Predictability: The grant mechanism should ensure predictability of recipient shares by publishing five year projections of funding availability. Efficiency: The grant design should be neutral with respect to recipient governments choices of resource allocation to different sectors or different types of activity. Simplicity: The allocation should be based upon objective factors over which individual recipient units have little control. The formula should be easily comprehensible so that “grantmanship” is not rewarded. Incentive. The design should provide incentives for sound fiscal management and discourage inefficient practices. There should be no specific transfers to finance deficits or public sector wages. Safeguard of grantor’s objectives: The grant design should ensure that that certain well defined objectives of the grantor are properly adhered to by grant recipients. The various criteria specified above could be in conflict with each other and therefore a grantor may have to assign priorities to various factors in comparing policy alternatives. Source: Shah, Anwar (1994). The Reform of Intergovernmental Fiscal Relations in Developing and Emerging Market Economies, World Bank, Washington, DC Source: Shah (1994, 1998) Text Table A3. Special issues in state/province-local transfers General purpose transfers to local governments requires special considerations as local governments vary in population, size, area served and the type of services offered e.g. urban vs rural. In view of this, it would be advisable to classify local governments by population size, municipality type, and urban/rural distinction and have a separate formula for each class of municipalities. Some common useful components in these formulae are: equal per municipality component, equal per capita component, service area component and fiscal capacity component. The grant funds should vary directly with service area but inversely with fiscal capacity. Source: Shah (1994) . References Afonso, Jose Roberto Rodrigues and Thereza Lobo (1996). Fiscal Decentralization and Participation in Delayed Democratic Experiences. Processed, October, 1996. Bomfim, Antulio and Anwar Shah (1994). Macroeconomic Management and the Division of Powers in Brazil: Perspectives for the 1990s. World Development, vol.22, No.4, 535-542. Dillinger, William (2001). Regulations and Markets: Brazli’s Efforts to Control Subnational Borrowing El Daher, Samir (1996). Municipal Finance/Municipal Bond Markets. Processed. Freire, Maria, Huertas, Marcela and Darche, Benjamin (2000). Subnacional Governments and Credit Markets: Selected Cases. In Shahid Javed Burki and Guillermo Perry, editors, Decentralization and the Accountability of the Public Sector. World Bank, Washington, DC. Gouarne, Vincent (1996). Investing in Urban Infrastructure: Roles and Instruments. Processed. McMillan, Melville (2002), Designing Local Governments, In Anwar Shah, editor, Assessing Local Government Performance When There is No Bottomline. World Bank, Washington, DC. Peterson, George (2000). Building Local Credit Systems. World Bank,Washington, D.C. Shah, Anwar (1999). Governing for Results in a Globalized and Localized World. The Pakistan Development Review, 38:4 Part I (winter 1999), 385-431. Shah, Anwar (1998). The Quest for Right Balance and Responsive and Accountable Governance: Lessons of Decentralization Experience. World Bank Policy Research Working Paper Series. Shah, Anwar (1994). The Reform of Intergovernmental Fiscal Relations in Developing and Emerging Market Economies. World Bank, Washington, DC Shah, Anwar (1991). The New Fiscal Federalism in Brazil. World Bank Discussion Paper No. 124, Washington, DC: World Bank. Ter-Minassian, Teresa (1996). Borrowing by Subnational Governments: Issues and Selected International Experiences. International Monetary Fund, Washington, DC Wiengast, B. And W. Marshall (1988). The Industrial Organization of Congress: Why Legislatures , Like Firms, Are Not Organized Like markets. Journal of Political Economy 96:132-163. World Bank (1994). Better Urban Services. Washington, DC: World Bank.