Financing local government State and local governments play an important role in providing public services. Decentralizing shares of total government revenuenot including intergovernmental grants-also vary widely, rang- both spending and revenue authority can improve the allocation of resources in the public sector by linking the costs and benefits of local public services more closely. To the extent possible, subnational government should charge for services, but, where such charges are not feasible or desirable, spending must be financed from local general revenue, loans, or grants from higher levels of government. This chapter examines subnational govern- ing from 2.2 percent for Tunisia to 72.1 percent for ment finance and focuses on reforming local Yugoslavia. The figures show that subnational governments tend to be more important as providers of public services than as collectors of revenues. Chinanot in the sampleis an exception and relies heavily on subnational governments to collect revenue (see Box 7.1). In every sample country the revenue raised by subnational government from its own Patterns of subnational government finance sources falls short of its spending. These shortfalls range from 0.1 percent of GDP in Costa Rica, The Gambia, and Sri Lanka to 4.2 percent of GDP in India and are more than 2 percent of GDP in six The role of subnational government varies from country to countryfor political and historical reasons as much as for economic ones. The relative sample countries. In several countries the combined state and local fiscal imbalance was more than the total government deficit, which means that central governments operated in surplus be- revenue systems to allocate resources efficiently. importance of the subnational level of government is indicated by its share of total government spending. For a sample of eighteen developing countries for which comparable data are available, this share ranges from 2.5 percent in The Gambia to 74.9 per- cent in Yugoslavia (see Figure 7.1). To capture longer run patterns, all figures reported in this section are averages of data available for 1974 through 1986. The sample is diverse. India and Yugoslavia conduct more than half of government spending at the subnational level, while for seven other countries this share is less than one-tenth. India, Yugoslavia, and, to some extent, Brazil have high subnational spending because they are large countries with strong state governments. The subnational 154 fore intergovernmental grants were disbursed. In most countries own-source revenues at the subnational level fail to cover even current spend- ing (see Figure 7.2). When grants are included, however, subnational government achieves fiscal balance or surplus in most sample countries. Accordingly, net borrowing, which is equal to the difference between total spending and total revenue, is a relatively minor source of funds for subnational governments. Fiscal imbalances do not necessarily indicate inappropriate fiscal policies at the subnational level. A grant may really be a shared tax, as in Brazil, or a compensatory grant for the central government's repeal of a local tax, as in Bangladesh, or a transfer Figure 7.1 Size of subnational and local governments, averages for 1974 to 1986 U Subnational government spending U Subnational government revenue Local government spending Local government revenue Percentage of total government spending and revenues 80 60 40 20 0 ! 5 Latin America Africa Asia Europe, Middle East, and North Africa Notes: Only state data are available for Argentina and India. The subnational government data reported here for Indonesia are not separated into state and local data. Sources: IMF, Government Finance Statistics, and World Bank data. to finance some centrally mandated expenditure such as teachers' salaries, as in Kenya. Nonetheless, the allocation of spending and revenue responsibilities, and the resultant transfers between levels of government, may induce an inefficient use of resources that contributes to total government deficits. Expenditure and financing measures can be combined to indicate the degree of fiscal decentraliza- tion. The share of state and local governments in total government spending reveals their importance as providers of public services. The extent to which they are self-financing indicates their fiscal autonomy, because outside financing may come with conditions that limit local discretion in the use of funds. Figure 7.3 looks at both aspects for the sample of eighteen countries. Yugoslavia's posi- tion in the upper right-hand corner indicates a highly decentralized government sector. Not only do the state and local governments account for a large share of total government spending, but they are largely self-financing. The Gambia is autonomous in the financing of subnational government, yet the size of its subnational government is only a small fraction of the government sector. Colombia, Indonesia, and the Republic of Korea display the opposite pattern: the subnational government has a large role in the provision of services, but its financing comes largely from central government grants and revenue sharing. Expenditure and financing measures, however, provide only a partial indication of decentralization. Central governments that want to exert control over local finance usually have many instru155 Box 7.1 Local government finance in China The financing of China's local governments presents an interesting contrast to local government finance in and "shared-up" to the higher levels. China has no regular grant program to support capital projects or most market economies. China has three distinct levels current expenditures (all grants are on an ad hoc basis), there is no mechanism or formal program for lending of government: central, provincial, and local. They share a common set of taxes, with policy decisions left to the central government and tax assessment and collection to the local government. All tax rates and bases are set centrally, and there are no truly local taxes at either the municipal or provincial level. China's revenue sharing is primarily a division of sales and profit taxes among the central, provincial, and local governments. Taxes are collected by the local governments to local governments, and there are no guidelines for local governments to develop beneficiary financing schemes. A combination of a backlog in infrastructure needs and increased urbanization has put great pressure on local budgets. The overall elasticity of taxes on profits and sales may be relatively low, that is, revenues rise less than proportionately with increases in profits and Figure 7.2 Spending and revenue of subnational governments, averages for 1974 to 1986 El Total spending El Total revenue Current spending' Own-source revenue (including grants, excluding borrowing) Percentage of subnational government spending 140 120 C C Latin America Africa a. Data for current spending are not available for Sri Lanka. Notes:/Sources: See Figure 7.1. 156 Asia Europe, Middle East, and North Africa Figure 7.3 Fiscal decentralization to the subnational level, averages for 1974 to 1986 sales. This is a structural problem, possibly exacerbated by weaknesses in the tax administration system. The high tax rates may induce tax avoidance by enterprises and perhaps even by the local governments that own them. The lack of borrowing and self-financing mechanisms creates a bias against infrastructure investments because the full cost is shifted onto the general public. Options presently under discussion include reform of the company tax, introduction of autonomous local Subnational government spending (percentage of total government spending) 80 Yugoslavia . 60 India taxes, and, in particular, the institution of a local government property tax. 40 Republic of Korea Colombia . Indonesia ments, such as approval of budgets, spending S Thailand limits on tax rates, and other fiscal constraints. Many developing countries are actively exploring ways to increase local fiscal autonomy to improve public sector efficiency and reduce total government deficits. Fiscal decentralization and the role of subnational government Romania S 20 requirements, restrictions on the use of tax bases, Argentina The Gambia Mexico. Philippines. Costa Rica Malawi Chile\ Kenya Tunisia 0 30 Brazil 50 SriLankal 70 S 90 Own-source revenue (percentage of gross revenue) Notes:/Sources: See Figure 7.1. State and local governments usually provide a range of public services that contribute substantially to raising living standards and growth. These include basic health and education; street lighting and cleaning; water, sewerage, and power; public markets and refuse collection; major transport net- tive capacity to collect revenue and prepare bud- works; and land development for business and residential purposes. Subnational government must decide how much to spend for these public services and how to finance them. Ideally each subnational government provides both the level portion of the capital resources available to them. Second, improving local government's administrative capacity can unnecessarily duplicate the number and skills of staff at the central and local levels. Decentralization should not inefficiently expand and mix of public services and the means of financing these services that most closely meet the pref- public employment. Third, the fact that public erences of individuals in its jurisdiction. In this efits or costs for other jurisdictions calls for involvement by higher levels of government. However, these problems can be addressed through appropriate grant policies and other mechanisms for strengthening local administration, as dis- way decentralization promotes efficiency by allowing a close match between public services and the multiplicity of individual preferences, and it promotes accountability and equity by clearly linking the benefits of services with their costs. This is the rationale for establishing state and local governments that are responsive to the wishes of their citizens and not simply the instruments of central government. Decentralization faces certain practical problems. First, local governments often lack the administra- gets and investment plans. In Tunisia, because municipal staff lacked expertise in project design and implementation, municipalities used only a services provided by one jurisdiction produce ben- cussed below. Assigning responsibility for services among levels of government should be as clear and simple as possible for decentralization to work. Vagueness in the division of responsibility can undermine local accountability. In Brazil, because of an unclear di- vision of functions between state and municipal 157 government, mayors can garner support for favored municipal projects from state governors instead of pushing for increased local taxes. This di- minishes the incentive for generating local revenue. A common feature of government in developing countries is joint responsibility at different levels for providing the same service in a particular jurisdiction. In Turkey, for example, something as apparently local in nature as street lighting involves all levels of government. The central government shares responsibility for funding; a national agency is mainly responsible for planning, execu- tion, and operation and maintenance; and local governments have the main responsibility for funding and secondary responsibility for plan- ning. In some caseseducation and health, for examplethe structure of service delivery is unavoidably complex. But for services that are clearly local, such as water and street lighting, it can generally be simple. Many countries would benefit from periodic reviews of the assignment of functions and finances to different levels of government. A review of in- Box 7.2 Strengthening local government finance Strengthening local government finance can improve the efficiency of the public sector and reduce the need for transfers from central to local government, particularly in urban areas. Certain revenue sources, such as user charges and property taxes, are more easily administered by local government. Strengthening these revenue sources is important for stabilization and structural adjustment in developing countries such as Indonesia and Pakistan, where inadequate reliance on subnational government finance has caused a heavy financial drain on the central government. The principles discussed in previous chapters suggest four criteria for efficiency in raising local revenue: Fiscal decentralization in Colombia Concerns about the assignment of fiscal responsibilities among central, departmental, and municipal governments in Colombia led to the appointment of the Mission of Intergovernmental Finance in 1980. The mission concluded that local resources should be relied on more for local purposes. Subsequent legislation to strengthen local government institutions and to decentralize both functions and finances marked a his- toric point in Colombia's fiscal and institutional development. The finances of subnational governments were strengthened by expanding their traditional revenue sources. The most important changes were those relating to the property tax, in particular those relating to the tax base. First, all cadastral values were updated to a 1983 base by applying an annual increase of 10 percent for each year (up to a maximum of fifteen years) since the last official valuation. Thereafter, in years when properties were not revalued, their cadastral values were to be adjusted automatically each year by a price index. Although political considerations have kept the rate of adjustment consistently below the rate of inflation, the adjustment has nonetheless slowed the erosion of this important source of local revenues. Property tax rates were also changed. The basic municipal rate can now be set between 0.4 and 1.2 per- 158 tergovernmental fiscal relations in Colombia was the basis for comprehensive local fiscal reform and showed that such reform is feasible (see Box 7.2). The rest of this chapter focuses on reforming local revenue instruments and overcoming problems in local administration. cent, compared with the previous fixed rate of 0.4 percent. This change was important because it introduced an element of local discretion in rate-setting for the first time. The base and rates of the industry and commerce tax were also reformed. Before 1983 this taxalthough important as a revenue source, particularly in the larger municipalitieswas levied on a wide variety of bases, generally with large variations in rates depending on the type and size of the business. The reforms estab- lished a uniform tax base (based on gross sales) and reduced the variation in rates. The flow of national transfer payments to the small municipalities has substantially increased. The share of the sales tax allocated to the subnational level is to rise from 30 percent in 1986 to 50 percent by 1992. More- over, for the first time in Colombia, the distribution formula was changed to give a bigger share to small municipalities that tried to raise their own resources. A new law provided that the additional funds for municipalities above the 1986 share are to be spent solely on investment, including maintenance and debt service. Finally, the national government was given one year to transfer functions to the municipal level so as to match the transfer of additional revenues by 1992. Table 7.1 Comparison of local public spending and revenue shares by major categories in selected cities (percentage of total spending) Spending and associated revenue category Spending on general urban servicesa Local tax revenue Spending and debt service for public utiitie&' User charges and borrowing Spending on social services' Grants received Republic Colombia Kenya Ind Pakistan Brazil, of Korea, São Paulo, Bogotd, Cali, Cartagena, Nairobi, Mombasa, Seoul, Ahmedabad, Bombay, Calcutta, Gujranwala, Karachi, 1972 1974 1981 1981 1983 1984 1972 1981 1981-82 1982 1983 1982 30.2 68.7 17.3 22.6 14.0 16.9 37.2 27.1 10.3 34.1 7.9 75.6 34.0 44.2 23.2 61.4 12.4 47.0 42.2 61.3 33.1 98.4 16.8 85.9 36.3 69.1 74.5 61.2 53.0 28.0 29.1 48.9 69.8 46.4 55.4 52.3 30.8 72.0 80.4 60.1 52.1 -7.8 33.8 30.1 52.2 -16.2 -8.7 11.1 33.6 0.4 13.7 14.0 2.9 2.8 1.6 12.8 36.8 13.7 64.0 32.2 36.9 22.0 27.8 8.6 17.8 0.7 11.4 54.9 11.5 10.3 30.8 3.0 Note: For each city, spending shares (including debt service) add to 100 percent, except for rounding errors; likewise all revenue shares (including borrowing) add to 100 percent. Includes refuse collection, parks and recreation, industries, fire protection, law enforcement, general administration, employee pensions and health care, grants and transfers, and other miscellaneous services. Local taxes include other miscellaneous revenues. Includes water supply, sewerage and drainage, electricity, telephones, housing, markets and abattoirs, highways and roads, and public transport. User charges include revenues from development charges, housing schemes, and so forth. Includes education, health, and social welfare. Source: Bahl and Linn, forthcoming. The cost of providing local services should be recovered, to the extent possible, from charges on the beneficiaries. Such charges should be related to individual consumption or, where this is not possible, to a measure of individual benefit received. Services whose costs cannot be recovered from charges can be financed from general taxes- property taxes, business taxes, and sales taxeslevied within the relevant jurisdictions. If the benefits of local services spill over into other jurisdictions or produce nationwide benefits, then grants from higher level governments should finance such services in proportion to their outside benefits. Borrowing is an appropriate way to finance at least some local capital investment, provided macroeconomic fiscal balance is maintained. As these criteria indicate, the appropriate use of local revenue instruments depends on the spending responsibilities assigned to local government. Equity can dictate modifications of these criteria, such as avoiding the imposition of user charges on low-income households. In the absence of detailed nationwide information, public finance data for a sample of cities provide some insight into the composition of spend- ing by sector, and of revenue by source, for should be financed by local taxes, public utilities should be self-financing, and social services should be supported by grants. In Bogota and Cali spending and revenue have closely matched the efficiency criteria. In Cartagena grants from other levels of government have tended to support the nonsocial services. All the other sample cities collected a surplus of local taxes to help finance public utilities and social services. Calcutta has received a large share of grants, apparently mostly to support public utilities. Increasing local revenues-especially through user charges-is often desirable, but it is difficult when local governments do not have revenue authority commensurate with their spending responsibilities. Central government usually lets local government use only a few revenue sources, and even these are subject to limits. In Thailand, for example, the proportion of local expenditure financed from local sources declined between 1977 and 1982 because of central restrictions on the rate at which local taxes and charges could be levied. Local authorities could, and should, be encouraged to raise more revenue locally. This might be done through user charges, local taxes, borrowing, and grants. User charges particular local jurisdictions (see Table 7.1). These data make it possible to relate broad categories of User charges can be of two types: consumption spending with revenue sources. The four effi- related and benefit related. They account for about one-third of all locally raised revenue in a sample ciency criteria suggest that general urban services 159 / Box 7.3 Benefit financing: land readjustment in the Republic of Korea and the valorization system in Colombia In Korea urban land readjustment schemes involve the consolidation of numerous small parcels of raw land at duce municipal tax burdens. The valorization system has been used most exten- the urban periphery. Nothing is paid to the owners. sively in Colombia, and its application in the capital city, Bogota, has been carefully studied. At the height This land is serviced and subdivided for urban use and then returned to the original owners in proportion to the value of their land contribution. Some of the land is retained by the public authority, in part to meet the needs of urban infrastructure (especially roads and green spaces) and in part to provide a source of finance to defray the cost of development. The land retained for this latter purpose is sold at market prices in commercial transactions or auctions. The scale of Korea's land readjustment program has been impressive. By 1985 some 43 percent of Seoul's total built-up area was covered by completed or ongo- tive importance has decreased, but it has retained a role in the financing of urban infrastructure in Bogota. One of the practical problems encountered in the ap- plication of the system has been its dependence on large financial transfers from the city's general revenue. In practice, valorization charges have not fully recovered the costs of all projects. An important reason for this is that some projects have been designed to ing schemes. The system has opened up new land for urban uses and thus has helped to increase the supply of housing and raise public funds. While it appears to improve living conditions in low-income areas, and the beneficiaries have not been expected to pay valorization charges. If a valorization program is to be main- have conveyed its direct benefits mainly to middle- and tained while pursuing subsidization of low-income neighborhoods, the resulting need for transfers from high-income landowners, some trickle-down effect may have occured. The large scale of the program probably reduced general urban land prices and rents below the levels that would have occurred in its absence. Land readjustment programs require fairly sophisti- cated methods of public land management, including effective land registration and cadastral records and land redistribution formulas. Nevertheless the admin- local general revenues should be recognized, explicitly calculated, and met. Arrears in the collection of charges have been another reason why valorization charges have failed to recover costs. For projects undertaken from 1968 to 1986, arrears amounted to 16 percent of project costs. The collection problem has arisen mainly from lack of istrative feasibility of the schemes has been amply payment by public agencies and by a few large properties. The introduction of interest charges on late pay- demonstrated in Korea. ments after 1981 reduced the recovery period for In contrast to land readjustment programs, which have been used mainly to develop new areas at the project costs. fringe of cities and towns, the valorization system often applied in Latin American countries has principally been used to finance improvements in infrastructure in built-up areas. Street improvements, water supply, and other services have been financed by this system of taxation, in which the cost of public works is allocated to affected properties in proportion to the benefits con- ferred. Valorization charges are designed to recover project costs, not to recapture all the benefits the I of its use in 1968 it contributed 16 percent to the financing of all local public expenditures, including spending by local state-owned enterprises. Subsequently its rela- project is expected to confer. The system is intended to make urban services largely self-financing and thus re- of twenty-five cities in developing countries. User charges are especially important at the local level because, being closer to beneficiaries, local public services are more amenable to such charges than services provided by higher levels of government. User charges reduce the pressure to raise revenue Finally, Bogota has demonstrated that the valorization system depends heavily on the quality of invest- ment planning and project preparation. In years in which infrastructure planning was weak, the system receded in importance even though requirements kept growing. Projects for which benefits were uncertain or not clearly concentrated among the properties bearing the valorization charge often created the greatest collection problems. The participation of beneficiaries in planning and managing the project made the resulting charges more acceptable to them. frastructure are borne by its beneficiaries. Of course, when the imposition of such charges is expressly not desired for equity reasons, as in subsidized improvement of poor neighborhoods, gen- eral taxes or grants can provide the necessary finance. from general local taxes; this can improve effi- A major benefit-related charge is the develop- ciency because the costs of public services and in- ment charge or "betterment tax": a lump sum 160 charge designed to recover the cost of infrastructure development from beneficiaries. A development charge may cover only a limited projectfor example, a neighborhood paving scheme or the construction of a sewage canalor it may cover the full development of a new area. It is usually imposed on the owners of property in the areas that are improved by public action, and it represents a tax on the increase in land values that results from new local infrastructure. It can be an equitable tax because those who benefit from development pay for its cost. Two particularly successful systems of development chargesland readjustment in East Asia and the valorization system in Latin Americaindicate their potential (see Box 7.3). The main difficulty with development charges lies in administration and collection. Jakarta's betterment tax suffered from late notification of the tax department that an area would be improved and, hence, late notification of landowners; difficulties in tracing landowners because of unregistered land transactions; and a lack of data on land values. A properly functioning property tax (discussed below) makes administering development charges far more straightforward. User charges are especially well suited to local government finance, but they have often been underutilized. Colombian cities, such as Bogota, Call, and Cartagena, were an exception because of their relatively limited dependence on local taxes and grants and much heavier reliance on user charges. Two main factors account for this. First, in the larger Colombian cities local governments provide the important public utility services (water, sewerage, electricity, and telephones), whose cost can be recovered through user charges. Second, local governments in Colombia have placed a relatively heavy emphasis on benefit charges to finance infrastructure. When not relying on user charges, local governments often require inefficient transfers from central government that strain central budgets. For example, in Pakistan increasing dependence on federal grants to finance provincial services is attributed, in part, to the provinces' low reliance on user charges. In Mexico municipalities have not had the incentive to recover the cost of investments made with grants from states or the federal government. Local taxes As the efficiency criteria indicate, another way to raise revenue locally is through local taxes. These can be either property taxes or other local taxes, such as those on industry, commerce, and the professions. PROPERTY TAX. The property tax has several advantages as a local revenue source. First, all municipalities have some taxable real estate within their boundaries. Unlike taxes on business and trade, whose bases are concentrated in major cities, the property tax can produce revenue in small outlying local governments as well as in large cities. Second, within small towns the property tax base is broad; thus the burden of such a tax can be distributed across a large segment of the population, and significant revenue can be raised at low tax rates. Third, because property values are enhanced by the provision of local government services, property taxation based on accurate property valuation can recover the cost of services directly from the beneficiaries. Relatively little of the tax on residential property will be shifted from property owners to others, whereas the tax on commercial property can be shifted to consumers. Finally, the burden of the property tax on the poor can be re- duced or eliminated by reducing or exempting taxes on properties below a certain value. Despite these theoretical advantages, data for ur- ban local governments in selected developing countries indicate that property taxes account for only 5 to 25 percent of recurrent receipts of local government, except in Africa, where reliance on property taxation is generally higher. Moreover, the performance of the property tax over time has been mixed. Property tax revenues have declined in real terms in more than half of the Asian and Latin American countries. The countries experiencing the greatest revenue erosion were generally those with high rates of inflation, particularly in Latin America. In most of Africa, however, property tax revenues have grown in real terms. An obvious way to increase property tax yields is to raise the tax rate. This is often a bad approach, however. It exaggerates inequities by increasing the burden on taxpayers whose properties are on the tax rolls and who pay their taxes in full. A better approach is to raise more revenue from delinquent taxpayers by collecting more efficiently. Examining payment records to identify major de- linquents, introducing a systematic method of chasing them (such as reminders and warnings), and conspicuous enforcement of penalties can together significantly increase collected arrears and produce longer term improvements in the system. In Delhi, for example, better property tax 161 collectionincluding rebates for timely payments, penalties for late payments, better facilities for making payments, and an improved management information systemincreased revenue by 16 percent in 1985-86 and 96 percent in 1986-87. Another major reason for improved collections was that tax- payers had to pay their assessment before being able to appeal it in court. Improving collections works best when the tax rolls are reasonably complete and valuations are accurateat least in relative terms. Otherwise, in- tax system needs continual updating. General revaluation of properties, which can be very expensive, can be done at five-year intervals if changes in property characteristics are recorded promptly and price increases are captured by indexing the tax base or increasing the rate. Experience with fundamental property tax reform in Brazil and the Phffippines demonstrates the need to address all aspects of the problem (see Box 7.4). Brazil, especially, shows that an efficient property tax can be achieved only if intergovern- creases in the efficiency of collection may exagger- mental fiscal relations provide the right in- ate existing inequities. More basic reforms may then be necessary. The maintenance of accurate property tax records, usually termed the fiscal cadastre, is a particularly important step in basic reform, because the tax base is constantly changing in developing countries because of rapid urban growth and high inflation. An accurate cadastre centivesan important general rule for all countries. also aids in planning and providing local services, especially infrastructure, and in administering and collecting development charges. Valuation is a critical step in basic property tax reform. One approach is to use the annual rental value, which in many developing countries is de- termined by asking renter occupants how much they pay in rent and extrapolating from that to owner-occupied property. The method is simple, but open to inaccuracy and misrepresentation, especially for owner-occupied structures. A second approach is based on the capital value, which is calculated either by estimating the value of the land and the cost of replacing the structure or by extrapolating from recent sales of comparable property. (A fundamental policy decision must be made on whether to tax only land or buildings and improvements as well.) In practice a pragmatic mix of the two approaches will be needed depending on the availability of rental and sales data. Valuations also require technical expertise and time. To economize on scarce skills, local governments can draw on a central valuation agency for information and technical assistance, as was done in Malaysia. Setting the property tax rate is the next key step in basic reform. The rate should be high enough to make the property tax worth collecting, and local governments should be allowed some flexibility in setting it, as in Colombia's fiscal reform (see Box 7.2). Flexible rates can ensure that property tax revenues are maintained between valuations: by adjusting rates, local governments can keep property taxes constant in real terms. When property is revalued, the rate can be reset to where it was at the previous valuation. Once established, a well-functioning property 162 OTHER LOCAL TAXES. Taxes on industry, com- merce, and the professions are usually another important source of local revenue. These taxes may take several different forms, including a turnover (gross sales) tax, a tax on the value of gross business assets, or specific levies tailored to the type of enterprise. Local taxes on industry, commerce, and services have accounted for more than 80 per- cent of local revenue in San Salvador and more than one-half in La Paz. In the Philippines, business license taxes are the second largest source of locally raised revenue. The revenue from business taxes can grow substantially, because with urbanization the number and size of business establishments increase and the taxable base grows. A flat charge or exemption eliminates the need to assess small firms. Few local governments in developing countries levy broadbased sales taxes partly because central governments prohibit it and partly because sales taxes are difficult to administer. To avoid adminis- trative problems, a higher tier of government might assess and collect a sales tax and then remit part or all of it to the local authority. For example, the sales tax could be a local surtax on the central government tax, with the central government acting as collector. The local government could set the rate, determine special exemptions and other details, and pay the central government a collection charge. Local autonomy would be sacrificed only in choosing the tax base and determining collection efficiency. One problem with this approach and the reason it is rarely adoptedis that central governments are generally unwilling to let the lo- cal authorities share in so lucrative a revenue source because it may limit the national government's ability to raise revenue from the base. Another is that such a tax might be less visible and thus make local governments less accountable to their taxpayers. Some local governments use personal income Box 7.4 Fundamental property tax reform in the Philippines and Brazil The Philippine Real Property Tax Administration (RPTA) Project was designed to address the problems of a weak property tax system. The project's approach was to change the system of valuation from one based on owners' declarations to one based on a government inventory. The project aimed to compile in each jurisdiction a comprehensive inventory of all land parcels, which would incorporate missing parcels, eliminate duplicate claims, and include an accurate measurement of all land and building characteristics to be used in valuation. percent on low-value residential property), and fixes the date on which new general revaluations become effective. Interim indexation of values is not permitted. As a result, effective tax rates before and after RPTA were as low as 0.3 percent. Finally, the RPTA did not improve collection. On average revenue collections increased by only 1 percent in the year after the project began. Brazil's property tax reform (known by the acronym CIATA) was designed, like the RPTA, to address the technical defects of a system producing very little reve- The project was successful within its narrowly defined objectives. Average valuations increased by 50 percent. This was largely due not to the discovery of missing parcels, but to the revaluation of existing parcels, based on property characteristics gathered in the field. Contrary to the government's original diagnosis, property owners were in the habit of declaring all parcels under their ownership. Owners did, however, understate the dimensions and quality of each parcel and the improvements they had made to it. The RPTA left several problems, however. First, the nue. The objectives of the CIATA were to revise the fiscal cadastre, in order to incorporate missing parcels calculation of individual valuations remained highly arbitrary. Although the RPTA yielded more accurate data using more accurate data on the physical characteristics of individual properties. Intergovernmental relations in Brazil were an obstacle to property tax reform. Unconditional transfers provide 80 percent of the recurrent revenues of local gov- on the characteristics of individual properties, the method used by valuers to convert that data into estimates of value was not improved. Second, property tax liabilities remained low. The RPTA did not reform rate setting and related policies. The central government in the Philippines fixes the maximum tax rate on property, mandates fixed assessment ratios (as low as 15 and improve the accuracy of valuations, and to devise a better system for monitoring collection and financial reporting. The short-run effect of the CIATA on municipal reve- nues was dramatic. Municipalities typically reported increases of 100 to 200 percent in revenues in the year following implementation. Most of this increase was due to improvements in the fiscal cadastre: missing parcels were discovered, and values were recalculated ernment. Because local governments could draw on outside finances to deliver services, they had little incentive to raise property taxes for their constituents. taxation as an alternative to business taxes. Local income taxes have the same characteristics and rising demand for investment in, and maintenance of, street infrastructure and traffic management. face the same problems as those at the national The case for recouping these costs with a tax is strong. Unfortunately not all urban governments level, as described in Chapter 4: that is, they tend to be narrow in coverage and often quite primitive in administration. Thus most local "income taxes" are really poll (head) taxes, wage taxes, or procedures for limited income tax sharing with higher authorities. As with sales taxation, a local surtax on the national personal income tax may be the best way for local governments to raise revenue from the base of individual income. Taxing the ownership and use of urban motor in developing countries are authorized to tax automobiles. For instance, local authorities in Manila have been told not to levy taxes or fees on motor vehicle registration. Furthermore, even where local governments can impose taxes, they generally have not made a significant effort to tap this revenue source to its full potential. Jakarta and Seoul are among the few exceptions. Jakarta proves that automotive taxation, if turned over to local author- vehicles can be an important source of local ities and given sufficient attention, can make a revenueand a reasonably efficient and equitable one. Automobile use is growing rapidly in developing countries and is concentrated in urban areas and among the better off. The increase in use has been accompanied by greater traffic congestion and air and noise pollution, as well as by rapidly large contribution to local revenue; it accounted for 37 percent of local current revenue in 1983-84. In Seoul automobile tax revenue more than doubled between 1982 and 1986, while total local tax revenue increased 64 percent. One of the most striking features of local govern163 ment taxation in developing countries is the proliferation of minor revenue sources, including selective excise taxes and a variety of license taxes, fees, stamp, and poii taxes. While any one of these adds little revenue, they may jointly make a significant contribution. For example, local taxes and fees account for more than 20 percent of total local revenue in Indonesia. Often, however, their administrative and compliance costs are considerable, and they are subject to wide discretion by tax collec- tors. Fewer such taxes would be appropriate in most jurisdictions. Borrowing As mentioned earlier, borrowing is generally a minor source of local government revenue in developing countries. This reflects both the conservative financial policies of central and local governments and their limited sources of loan finance. In principle, however, there is no reason why local governments should not borrow to finance at least a portion of capital project costs. Debt service payments can be linked to user charges for local services gen- erated by the capital projects for which the debt was incurred. In this manner greater reliance on borrowing favors self-financing projects. This need not imply an increase in aggregate public sector borrowing. Instead it would usually mean replacing intergovernmental capital grants with internal loans from central to local government. Grant financing places the burden on central governments and general taxpayers, rather than on local deci- sionmakers and beneficiaries where it properly belongs. One way to provide local access to loans while ensuring some central control to preserve macroeconomic balance is to create a municipal develop- ment fund (MDF). Typically, MDFs have two broad objectives. The first is to raise additional resources for public investment. This usually begins with an initial injection of funds from a donor and a counterpart contribution, usually larger, from central or state government. In addition domestic financial markets may be tapped through bond and debenture issues. But ultimately most of these programs aim to extract additional resources from municipalities themselves through better revenue administration and cost recovery. The second objective is to improve the use of resources. This can be done by developing appraisal criteria and enlarging local capacity for sound financial analysis of the projects financed by MDFs (as proposed in Brazil and Mexico), by as164 sisting municipalities with the design and execution of their investment programs (as in Turkey and Venezuela), and by improving the operation and maintenance of local infrastructure (as in Jordan, which is discussed below). MDFs are not new. In the past thirty years they have spread rapidly through Africa, Asia, and Latin America. Historically, few have developed a capacity for sustained assistance and funding for municipal government on the scale needed. They have tended to play a narrow and passive financing role by providing little technical or financial appraisal to the investments they have funded and offering little assistance to municipalities other than capital finance. Furthermore repayment discipline has sometimes been weak. MDFs in Honduras, Kenya, and Morocco, for example, have experienced (and tolerated) substantial arrears, and the latter two made new advances to borrowers in default. Jordan's MDF has been notably successful. In 1979 the government established an autonomous Cities and Villages Development Bank to provide investment finance and assistance to municipal and village councils, with the aim of extending access to infrastructure and employment. The bank played a role in financing the spread of urban physical and social infrastructure (for example, paved roads, schools, clinics, piped water, and electricity) to virtually all settlements, however small and remote. It has also been instrumental in maintaining and improving the quality of infra- structure investment through its standards, appraisal, and monitoring. The bank has taken on an increasingly active role in financial and technical advice and in training for municipal authorities. It has introduced new criteria that will allow municipalities to forecast their debt service capacity more accurately. Efforts are under way in many developing coun- tries to strengthen existing MDFs or to set up new ones. To succeed the local governments concerned will need a buoyant revenue base to support debt service. If MDFs are allowed to recover debt service from central government transfers to local governments, they will be tempted to induce local authorities to borrow more than they can afford. Intergovernmental grants Central government transfers to subnational gov- ernments through shared taxes and grants are a critical link in distributing expenditure responsibil- ity and taxing authority between the two tiers. Box 7.5 Effects of intergovernmental grants: the experience of Nigeria in the late 1970s Nigeria embarked on a nationwide reform of its local government system in 1976. It was intended to reverse the deterioration in the treatment of local governments by the states after the country achieved independence. Specifically, many states had cancelled their grants. Under the reform, grants from both the federal and the state governments were to be substantially increased. In the first year local governments received in grants from the federal government roughly five-anda-half times the amount they had received in the previous year. This increase followed from sharp increases in revenue from oil exports and the reformed grant allocation rules. The budgeted revenue of some local governments increased 1,000 to 2,000 percent. As a result local governments greatly expanded their provision of social services and economic infrastructure. Some serious problems arose, however. State administrations sit between federal and local governments; all but one of them failed to pay its full allocation to the local governments. In many cases federal transfers were diverted to other uses without consultation. In addition virtually all the state governments approved large salaries for state-appointed local government officialsincreases, in some cases, of more than twothirds. Federal grants to states depended in part on the number of localities under state jurisdiction and thus led to a widespread demand for more local government units. The number of local governments increased Furthermore the criteria for distributing transfers placed too great an emphasis on the equality of revenue across jurisdictions, rather than on the generation of local revenue. The transfers served as a disincentive to local revenue effort. Local taxes, such as the property tax, remained grossly underdeveloped, and some states decided to suspend or abolish important revenue sources. Various explanations were given to justify abolition, but they all rested on the supposition that federal transfers made collection of local taxes unnecessary. So the cumulative effect of Nigeria's dramatic increase in transfers to local government was to reduce local governments' own contribution to local finance in relative and absolute terms. For example, the number of property tax payers in Ibadan City dropped from 27,000 persons in 1975 to 8,650 in 1979. Other cities experienced similar declines. The administrative burdens on local government increased as a result of the 1976 reform. The distribution of increased transfers through states brought local gov- ernments under stricter control. Local budgets had to be approved by the states, which often caused severe delays in program execution. In one case the budget process had to go through eight stages of approval with state officials before final approval by the state's chief executive. The substantial increase in transfers to local governments in Nigeria in the late 1970sin the wake of the steep rise in oil revenuedid not enhance local governments' autonomy or ability to act. from 301 to 781 between 1979 and 1983, which led to a severe shortage of skilled managerial and technical staff. Grants are needed because the benefits or costs of public services provided by local governments can spill over to other jurisdictions. For example, a highway constructed through one jurisdiction is likely to benefit residents in neighboring jurisdictions, or one jurisdiction's education services contribute to a productive labor force that can migrate to other jurisdictions. Some local public services, especially social services, may have national and regional benefits as well. Another objective of grants is to adjust for disparities in fiscal capacity among local jurisdictions caused by variations in resources, tax bases, and population. Grants can equalize local fiscal capaci- ties, although obtaining adequate data to do this by formula can be difficult. Excessive reliance on grants, or unexpected increases in them, can result in poor use of public finances. Nigeria's grant system expanded greatly in the late 1970s, and this had severely detrimental effects on local government finance and intergovernmental relations (see Box 7.5). Grants can encourage recipients to be less efficient. For example, grant systems in Indonesia and Sri Lanka have paid the salaries and allowances of virtually all lo- cal government employees, including primary school teachers. This cost reimbursement grant guarantees a uniform pay scale for local government employees without regard for the financial health of any single locality; its lack of cost sharing can encourage local governments to employ inefficiently large numbers of people. The experience of several developing countries, including Colombia, Ecuador, and Mexico, reveals that increasing reliance of local governments on grants can decrease the fiscal autonomy of local jurisdictions. Local governments come to view grants as substitutes for local taxes and user 165 charges. Many intergovernmental grants are simply central government subventions of local activities. They have many undesirable qualities: they are a drain on central revenue, they encourage inefficiency and worsen regional and urban-rural disparities, and they reduce the accountability of local government while giving it no assured access to funds. A formula-based grant system addresses these problems by simultaneously meeting central government objectives and providing local governments with some degree of autonomy. Formulabased allocations require the granting government to decide on the factors that are to be used in determining grant amounts. These may include, for example, indicators of need, such as population or miles of road, or indicators of local revenue capacity, such as local income or taxable economic activ- ity. The grants may be block grants, that is, general-purpose, lump-sum transfers that provide local governments with considerable autonomy, as in the Philippines. Alternatively they may be categorical, that is, restricted to particular uses, sometimes requiring approval by the central govern- ficult to implement effectively, and are open to abuse or ad hoc political bargaining. Local government administration Decentralizing spending responsibility and revenue authority can assist development, provided local administrative capacity is adequate to the tasks. An important reason for past shortcomings was the failure to deal with the difficulties of planning and implementation. Many of the reforms recommended for public spending in Chapter 5 apply at the local level also. In particular, local governments should prepare both a medium-term fiscal plan and a comprehensive annual budget. These force local governments to evaluate the needs of their jurisdictions and to make strong arguments for spending programs, either current or capital, which then serve as the foundation for strong fiscal efforts. They also help central governments to coordinate municipal de- velopment and provide a clearer picture of required intergovernmental transfers. Effective local administration is hampered by lack of finance and accounting skills because ment. Categorical grants give the central gov- skilled personnel frequently would rather work for ernment more control over the allocation of funds between sectors. Categorical grants can be designed to cover only part of the total spent locally in a sector, in which case grant funds are said to "match" local funds. The matching rate can be set to achieve an efficient level of local spending. If a matching grant is used to support spending that has spillover benefits, the the central government. One approach to local matching rate should ideally equal the ratio of spilover benefit to total benefit, although the information needed to compute matching rates precisely is very difficult to obtain. A single grant cannot accomplish all objectives. For example, if the principal objective is to equalize fiscal capacity across jurisdictions, the goals of stimulating local government tax effort and promoting local fiscal autonomy are not likely to be well served. This problem can be addressed by including different types of grants in the system, for example, pure shared taxes to provide adequate revenues for large urban areas, formula grants to equalize across jurisdictions, and matching grants to stimulate tax effort. Balancing the merits of shared taxes and formula grants is a central issue in designing grant systems. Grant systems should be kept as simple as possible. Complex systems lack transparency, are likely to introduce mutually offsetting incentives, are dif- 166 staffing is to assign central civil servants temporarily to local authorities. Also technical agencies at the center can provide direct services to local governments. Malaysia's. central government lends as- sistance in public health, sanitation, and food in- spection, especially in rural areas, and has assigned accountants to state governments. Technical assistance to local governments may be provided with either loan or grant finance. Municipal development funds can enlarge the local capacity for financial analysis and implementation of investment programs. Training local officials is essential for better local administration. Several Indonesian cities have pilot programs for local training to improve tax administration and financial management. In Nepal local officials have attended training workshops on techniques of financial analysis, current and capital budgeting, and tax collection procedures (including records systems). As a result, at least initially, locally raised revenues have increased. Another option for local governments with staffing shortages is to buy services from the private sector. Many services can be provided this way. Contracting reduces staff requirements and gives local governments the flexibility to provide a changing mix of services. Toward more efficient local government revenue sources that they are best equipped to use, such as property taxes and user charges. A The structure of government in many developing countries is inefficient. Often fiscal relations are opaque because of political expediency rather than properly designed system of grants can encourage efficient provision of local services. Local govern- lack of knowledge or skifi. This makes reform much more difficult. Nonetheless more open and transparent systems are urgently needed. Responsibility for many services can be devolved to local government. Local government should rely on the ments must be held responsible for their use of public resources to those who provide them: for user charges, the beneficiaries; for locally raised revenue, the general public; for grants, the central government; and for borrowed funds, the creditors. 167