哥伦比亚的分权与财政管理

__PS 212Z POLICY RESEARCH WORKING PAPER 2122 Decentralization and Fiscal Institutionalarranqements have helped Colombia Management in Colombia manaqe the fiscal aspects of decentralization, despite the ~ William Dilliniger country's political problems. Stev'en B. Webb I The World Bank Latin America and Caribbean Region Poverty Reduction and Economic Management Sector Unit U .N,Iay 1999  POIICY RESEARCH WORKING PAPER 2122 Summary findings Colombia's political geography contrasts sharply with its subnational levels) and in preventing unsuistainable econonyv. Physical characteristics and guerilla war deficits by the subnational governments. fragment the country geographically, yet it has a long The problems have arisen because central government tradition of political centrism and macroeconomic interference prevents departments from controlling their stability. costs and because of expectations of debt bailouts. Both Recently, with political and economic decentralization, are legacies of the earlier pattern of management from there has been sorme weakening of macroeconomic the center, and some recent changes - especially about pertormance. Dillinger anLd Webb explore institutional subnational debt - may improve matters. arrangements that have helped Colombia manage the Colombia's traditional political process has had fiscal aspects of decentralization, despite the country's difficulty dealing with problems of decentralization political problems. because traditional parties are weak in internal Fiscal decentralization proceeded rapidly in Colombia. organization and have lost de facto rule over substantial Education, health, and much infrastructure provision territories. have been decentralized to the departmentos and The fiscal problems of subnational government have municipios. been contained, however, because subnational Decentralization has led to substantial but not governments are relatively weak politically and the overwhelming problems, both in maintaining fiscal central government, for the time being, has been able to balance nationally (as resources are transferred to enforce restrictions on subnational borrowing. This paper - a product of the Poverty Reduction and Economic Management Sector Unit, Latin America and Caribbean Region - is part of a larger effort in the region to examine the macroeconomic consequences of decentralization. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Steven Webb, room 18-137, telephone 202-473-8680, fax 202-522-2119, or William Dillinger, room MC2-323, telephone 202- 473-3940. Policy ResearchWorkingPapers are also posted on thieWebathttp://wxvsv.worldbank.org/html/dec/Publications/ Workpapers/home.html. The authors may be contacted at wdillinger@Qworldbank.org or swebb@worldbank.org. May 1999. (45 pages) P bhe Policy Research WoDkinag Paper Seies disseminates the ,finding-, of work i progress to encourage the exchange of ideas about de7velopmen7t issutes. An objective of the series is to get the ,findings ouit quickly, even if the presentationis are less than fully polished. The papers carryT the names of the autbors and shotuld be cited accordingly,,. The findings, interpretations, and conclusions expressed in this paiper are eiitirelv those of the authtors. They do not necessanly represent the 7view of the World Bank, its Executive Directors, or the cotrztSfies thley represent. Produced by the Policy Research Dissemination Center  Decentralization and Fiscal Management in Colombia William Dillinger and Steven B. Webb World Bank DRAFT comments welcome (swebbOlworldbank.ora or wdillinger@worldbank.org) We wish to thank Katherine Baer, Ariel Fiszbein, Marcela Huertas, Connie Luff, Guillermo Perry, Fernando Rojas, and David Yuravlivker, for their useful comments on earlier drafts; they bear no responsibility for any remaining errors. Elizabeth Forsythe edited the text. The opinions expressed in this paper are those of the authors and do not necessarily represent the opinions of the World Bank. Except as noted, the paper describes developments up until mid-1998.  EXECUTIVE SUMMARY Decentralization and Fiscal Management in Colombia Colombia is a nation of sharp contrasts. The country is geographically fragmented by physical characteristics and guerrilla war, yet it has a long tradition of political centralism and macroeconomic stability. Recently it has experienced political and economic decentralization and also some weakening of macroeconomic performance. This paper explores the institutional arrangements that have helped Colombia to manage the fiscal aspects of decentralization, despite its political problems. Fiscal decentralization has proceeded rapidly in Colombia because political democratization of municipalities and then departments began at a time when substantial expenditures already had been delegated to the previously appointed mayors and governors. The decentralization has led to substantial but not overwhelming problems, both in maintaining fiscal balance at the national level, as resources are transferred to subnational levels, and in preventing unsustainable deficits by the subnational governments. The problems have arisen because national government interference prevents departments from controlling their costs and because subnational governments have come to expect debt bailouts. Both of these are legacies of the earlier pattern of management from the center, and some recent changes, particularly in the area of subnational debt, may improve matters. The regular political process has not been dealing with the problems of decentralization, because traditional parties have weak internal organizations and illegal parties exercise de facto rule over substantial territories. The fiscal problems of subnational governments have been contained, however, because these governments are relatively weak politically, and some key control functions are regulated by law or are delegated to national agencies. Fiscal problems of the national government are mostly its own doing. 2  Decentralization and Fiscal Management in Colombia William Dillinger and Steven B. Webb World Bank Since 1986, Colombia has been decentralizing its democracy and public finances, joining Argentina and Brazil in the ranks of Latin American countries with highly decentralized public sectors. In Colombia, political decentralization, in the sense of elected mayors in 1986 and then governors in 1991, came after spending had been substantially delegated to the local levels, starting in 1968. In contrast to Argentina and Brazil, Colombia implemented decentralization long after the transition from military to civilian rule and without the complication of hyperinflation. Indeed, on the one hand, Colombia has enjoyed a tradition and reputation for sound macroeconomic management, which was associated in some minds with central fiscal control or with centrally controlled fiscal deconcentration in the early 1980s. In the 1990s, on the other hand, the expansion of political and fiscal decentralization coincided with the expansion of fiscal deficits and a failure to share in the general increase in macroeconomic stability in most other parts of Latin America. This paper analyzes the institutional features of fiscal decentralization in Colombia and its effects on macroeconomic stability. The paper identifies a number of weaknesses in Colombia's decentralization but finds that decentralization has not been the main cause of the nation's macroeconomic problems. The institutions for decentralization and overall macroeconomic management have prevented Colombia's serious political problems-guerrilla war and corruption from drug money-from causing even greater problems for economic policy and have mitigated the fiscal problems that have arisen in other countries from the decentralization process. Of course, fiscal stability is not the only measure of success for decentralization, and the extent or form of decentralization to date has not been adequate to solve Colombia's political problems. 3  Colombia is a unitary state, so the departments and municipalities have less autonomous status than Brazilian states or Argentine provinces, but the constitution and the electoral process confer substantial political legitimacy on them. Another distinguishing feature are the two or three decades of guerrilla war that have conferred de facto, but not constitutional, autonomous status on large areas of the country. This partially motivated the decentralization process in the 1980s, as the national government sought to strengthen the legitimacy of formal governrnent. The fragile legitimacy of the subnational governments (SNGs) facing guerrilla threats also limits their scope for politically feasible fiscal adjustment.' Despite several differences, Colombia has faced the same basic challenges with fiscal management as Argentina and Brazil. One of the main dangers in this area is that national government will incur excess fiscal deficits as it transfers its revenue base to subnational governments but has difficulty reducing expenditure in equal measure. The other main risk is that the combination of democratic rule and the expectation of bailouts from the center could lead subnational governments to run excessive deficits, upsetting the overall balance of the public sector and eventually the fiscal balance of the national government itself. Argentina and Brazil have faced these dangers over the past 15 years and often succumbed to them. In the 1990s both countries struggled against these tendencies, Argentina with considerable success (Dillinger and Webb 1998). Because of differences in their institutions and history, Colombia's outcomes have differed thus far from those of its larger neighbors, but some of the problems and the options in Colombia today are like those in Brazil and five to fifteen years ago in Argentina, when subnational fiscal imbalances caused problems at the national level. The central and subnational govemment deficits are the key manifestations of macroeconomic problems. The outcomes depend on the combination of fiscal and political rules for This is not to say that local govemments in guerrilla-controlled areas have not been fiscally prudent, for 4  intergovermnental relations. As background to the discussion of these rules and their fiscal outcomes in the 1990s, we start with a chronological overview. I. Historical Overview Colombia embarked on political decentralization more recently than Brazil and Argentina. Although the country has a long history of civilian rule and regional diversity, the government became increasingly centralized beginning with the 1886 constitution and up until the late 1970s. By then, the national government raised more than 80 percent of the revenue, as opposed to about 60 percent in the first half of the century (Bird 1984; Sanchez and Gutierrez 1995). The first indications of national interest in decentralization appeared in the 1968 constitution, which established a regional fund for education and health to be financed from a fixed percentage of national revenues. Because the national government appointed governors, who appointed mayors, this represented less a decentralization than a deconcentration of the budget to subnational administrative units-32 departments, 4 districts (municipalities with the status of departnents), and 1,070 municipalities. In addition, municipalities were assigned 10 percent of the then-new value added tax (IVA). Further steps toward fiscal decentralization were made in 1983 with laws that strengthened the tax authority of departrnents and municipalities. The design of this decentralization drew heavily on a report by a panel of international experts (reprinted as Bird 1984). Although Colombia remained politically centralized in the sense that elections were held only for the president and Congress, the politicians for a locality-senators and deputies-had a lot of say in the selection of governors and mayors. Political decentralization in the law began in 1986, with Law 78/86 removing the power of departmental governors to appoint mayors. The first mayoral elections took place in 1988. Over the indeed they have typically shown fiscal restraint, perhaps in part because they do not have much access to credit. 5  next three years, the government enacted sweeping legislation aimed at transferring many public service responsibilities to the newly elected mayors and providing them with additional fiscal resources (Laws 12/86, 24/88, and 29/89; Decree 24/88). This legislation largely bypassed departments, leaving them mainly supervisory or coordinating roles. Some municipalities developed the capacity to manage the new responsibilities, demonstrating that it could be done (Fiszbein 1997; World Bank 1995). But many did not develop the capacity, at least not quickly, which motivated government to shift to a more deliberate, step-by-step process. A restructuring of decentralization arrangements in 1991 granted more political autonomy and service responsibilities to the departments and made the fiscal authority of both departments and municipalities conditional on a certification process (Ahmad and Baer 1997; Ferreira and Valenzuela 1993; Sdnchez and Gutierrez 1995; Wiesner Duran 1992). The new 1991 constitution established the direct election of governors. Subsequent implementing legislation (Law 60) partially reversed the functional decentralization of social services to municipalities. For municipalities, the revisions during the early 1990s represented a loss of management autonomy but a gain in transfer revenues. The 1991-94 reforms also reduced the national governnent's discretion in the distribution of transfers. The increased transfers and the assignment of oil royalties gave subnational entities about half of public sector current revenues. The degree of true fiscal decentralization was less than it appears in the accounts, however, because the national government decided how the departments and to some extent the municipalities would have to use their main transfers. The national executive and congressmen each had different motives for decentralization but eventually reached a pragmatic agreement on action. Presidents from both parties, starting in the early 1980s, saw decentralization as a way to help rebuild the shattered legitimacy of the formal state. In 1991 a constitutional convention, mandated by a referendum called by the president using emergency powers, wrote a new constitution establishing decentralization. The nontraditional parties, representing interests long excluded from national power, had the majority of delegates and 6  used the opportunity to pass power and resources to the subnational levels where they held some of the executive offices. Also, the Liberal Party, which had favored decentralization since the nineteenth century, had a plurality of delegates-about one-third-many more than the traditionally centrist Conservative party (Buenahora 1997). Congress decided to live with decentralization because it gave local administrations the power to execute spending, and these local entities were usually allied with their congressmen, acting as their machines in many cases (Archer and Shugart 1997; Willis, Garman, and Haggard 1997). Table 1. Growth, Inflation, and Fiscal Balances in Colombia, 1987-97 (percent) Indicator 1987-90 1991-94 1995-97a Real growth (annual average) 4.5 4.3 3.5 Inflation (annual average) 15 9 7 Overall fiscal balance as a percentage of GDP Central -1.0 -1.1 -4.2 Department -0.2 -0.3 -0.2 Municipal -0.1 -0.3 -0 .1 Primary balance as a percentage of GDP Central 0.1 0.3 -2.2 Departrnent -0.1 -0.1 0.1 Municipal 0.0 0.0 0.2 a. 1995 only for department and municipal balances. Sources: World Bank database; Central Bank of Colombia, table 4. As the system operated in the late 1990s, strengths emerged, such as increased political participation, and so did weaknesses, such as mismatched spending authority and excessive borrowing by some subnational units. Overall, however, the fiscal deterioration occurred mostly at the national level. See table 1. II. Explanatory Variables Because the Colombian experience with political and fiscal decentralization is relatively short and an economic crisis has not yet tested the system, we evaluate the system on dimensions that have proven critical in Argentina and Brazil for determining whether subnational govermments contribute to unsustainable overall public sector deficits. The explanatory variables fall into two 7  main groups: fiscal institutions and intergovemmental political relations. In each area we identify as hypotheses the conditions we would expect to lessen the likelihood that fiscal decentralization will lead to excessive deficits. It is rare for a country to meet all of the conditions, but severe problems arise if too few of the conditions are met. The challenge of this research is to discover which combinations are adequate to prevent fiscal decentralization from leading to unsustainable fiscal deficits and which are not. Fiscal institutions For fiscal institutions, the key questions concern the division of authority in three areas: revenues, spending, and borrowing. What authority do SNGs have to raise revenue and control their spending? What authority does the national government have to shift spending responsibilities to the subnational governments and to reduce the transfer of resources in order to balance the national government budget? What budget constraints do SNGs face, and how are they set? Under what conditions can SNGs issue debt? Revenue autonomy. With fiscal decentralization, the SNGs usually receive certain tax bases, but for various reasons-politics, equity, and efficiency-these tax bases rarely cover all of their expenses. Thus SNGs also receive some federal transfers. One view is that subnational governments have smaller deficits if they rely more on their own tax bases (and have the power to change tax rates on the margin) because they can adjust to shocks by increasing revenue. Also, relying on one's own resources may strengthen the incentives -to control spending, and of course it reduces the burden on national government. Thus unsustainable overall public sector deficits are less likely under condition 1, subnational governments raise much of their own revenue. Furthermore, given that there are transfers from the center to the subnational governments, unsustainable deficits would be less likely to be problematic under condition 2, transfers are specified by legal formula, not ad hoc. 8  In Colombia, decentralization grew out of the deconcentration of national revenues to subnational administrative units. As in Argentina and Brazil, revenue sharing set by formula accounts for most of the transfers. The critical question was flexibility at the margin. Starting in 1968 a departmental fund for education and health was financed from a fixed percentage of national revenues, and municipalities were assigned 10 percent of the then-new IVA, which was not earrnarked. This was designed to solve the problem of ad hoc transfers to supplement inadequate sources of local revenue. Even after 1968 ad hoc transfers remained a problem, as mayors continued to ask the president for help to meet the cost of their new responsibilities. A major review of the system of intergovernmental transfers hardened the SNGs' budget constraint vis-a-vis the national government and strengthened their own revenue sources (Bird 1984). A 1983 law standardized the departmental taxes on liquor and cigarettes and ceded the national tax on vehicles to territorial governments. It also authorized a revaluation of the municipal property tax and permitted some local discretion over tax rates. Although it may seem unimportant that earmarked revenues were deconcentrated to subnational authorities appointed by the national government, congressmen at the head of local political machines actually controlled these funds and appointments. The 1991 constitution (which also made the office of governor an elected post) and Law 60 of 1993 moderately expanded the amount of revenues assigned to departments by broadening the base of the existing revenue-sharing system (the situado fiscal) to include all recurrent revenues of the govermment: the value added tax, customs, income tax, and special funds.2 They mandated a steady increase in the share of these revenues to be transferred to the departments. The share of the situado increased from 22.1 percent in 1993-net of one-time adjustments-to 23 percent in 1994, 23.5 percent in 1995, and 24.5 percent in 1996. Thereafter, the constitution committed the government to increasing the share sufficiently "to permit adequate provision of the services for 2 Revenues from special funds were excluded. 9  which it is intended." The sharing formula was to be revised by Congress every five years. For municipalities, the 1991 constitution broadened the base of the existing revenue-sharing system from the IVA to all government current revenues and committed the national government to increase the municipal share from 14 percent in 1993 to a minimum of 22 percent by 2002. Thus the 1991 constitution and Law 60 committed the national government to sharing nearly half of all its current revenues with territorial govermments and entities by 2002. (The proportion in 1998 was already 42.5 percent.) The 1991-94 reforms also reduced the national government's discretion in the distribution of transfers. Prior to Law 60, the situado was paid directly to teachers and health workers under the ministries of education and health. Law 60 changed this system to one in which the situado was transferred directly to each departnental government on the basis of a formula.3 The distribution of revenue sharing among municipalities-the participaciones municipales and the share of the value added tax-was also formula driven: 60 percent was to be distributed in proportion to the number of habitants with unsatisfied basic needs and relative level of poverty (as determined by the Central Statistical Agency), with the remaining 40 percent distributed according to population, administrative efficiency, and improvements in quality of life (all quantitatively defined in legislation). As a transition measure in 1994-98, each municipality received, as a minimum, the amount of IVA transferred in 1992, in constant prices. Also Law 60 specified that part of the transfer would be distributed equally to all municipalities, which meant that small municipalities got much more per capita and that there was a lot of incentive to form new (small) municipalities. The latter happened a lot at first, as in Brazil and Mexico, but then the problem was brought under some control by the rules for certification of municipalities. Even though this sort of 3 According to Law 60, 15 percent of the situado is to be uniformly distributed to each department and district. The distribution of the remaining 85 percent is to be based on a formula taking into account the current number of students enrolled, the number of school-age children not attending school, the number of patients seen by health units, and the number of potential patients based on population. 10  arrangement is not a immediate danger to overall fiscal balance, since the amount of money distributed stays fixed, it contributed to inefficient expansion of the total public sector. Cofinancing funds, derived from a national government program for rural development in the 1970s, have evolved into a program of transfers to municipalities for capital investment needs- intermunicipal roads, municipal roads, social investment (mainly education), and rural infrastructure. The investment funding is important because it provides flexibility in usage, whereas most of the other transfers are earmarked for specific and relatively inflexible current expenses, mainly wages. In principle, they are distributed by a formula similar to that for other transfers-per capita plus the number of persons with unsatisfied basic needs and the local fiscal effort. In practice, at least until 1997, there was no coordination among the four main funds (one for each of the areas listed), and there was considerable political discretion in the allocation of projects, which became important channels for patronage. The total amounts were not fixed as a percentage of total tax revenues, like the other transfers, but rather were a budget item. Up to the mid-1990s they grew in importance, reaching 40 percent of total transfers to municipalities. Since then, they have declined somewhat, due to fiscal pressure on the national budget. In 1997 a reform unified the fimds, and converted most of them into soft loans managed by FINDETER, a government financial intermediary (Ahmad and Baer 1997; Rojas 1997). The reform improved the coordination and transparency of the investment funds, but along with discretionary transfers for universities, they remain important loopholes in the hard budget constraint for states. In addition, some taxes are under the control of subnational governments. A few SNGs receive royalties from mineral production, which go mostly to the producer departments (47.5 percent), producer municipalities (12.5 percent), and port municipalities (8 percent). These units receive a large amount of resources. Less than one-third-32 percent-of the royalties go into a fund that is redistributed across the country; no royalties go to the federal government (Sanchez and Gutierrez 1995). This innovation in the 1990s may have been aimed at strengthening the han-d of 11  local governments facing guerrilla threats, which often targeted oil production, but this rationale does not seem to have been explicit.4 Most departments have only the tax on alcohol, cigarettes, and lotteries. Municipal governments have a broader range of small excise taxes and property tax. The most important revenue sources for subnational govermments are the tax on alcoholic beverages by the departments and the tax on gross turnover of business and on property by the municipalities (Ahmad and Baer 1997; Bird and Fiszbein 1998). In summary then, Colombia does not meet condition 1, since subnational govermrents do not raise most of their own revenue. Most transfers are through revenue sharing by formula, in line with condition 2, but there are important exceptions. Expenditure autononty. Without SNGs having autonomy over their expenditure, there is really no fiscal decentralization and no macro-fiscal problem likely to come of it. Subnational spending autonomy is, of course, the way in which decentralization would have the expected benefits of increased efficiency in matching the needs and desires of a diverse population. In terms of the effects on macro-fiscal management, two aspects are usually important. One is whether the central government can dictate which functions the subnational govermnents must take on, at least in exchange for receiving transfers from the center. Where the central govermment can do this, it helps contain central spending and deficits. Where it cannot, as in Brazil, the central government may find itself with a constitutional obligation and political expectation that it will continue to provide some service even after it begins tuming over revenues or tax bases to subnational governments with the understanding that they will do the task. The other issue is whether SNGs have authority to cut costs, particularly to cut personnel, salaries, and pension benefits, which are 4Perry and Rodriguez (1991, p. 77) note that this allocation goes against standard practice in the literature and that the reasons were "mas bien de indole pragmatica y, si se quiere, politica." ("largely pragmatic and perhaps political"). 12  typically the largest single item of subnational expenditure. Where central rules constrain this ability, it is more difficult to reduce deficits and expectations of a central government bailout are higher. Thus unsustainable deficits should be less likely under condition 3, the central government can effectively delegate functions to subnational governments to go along with the delegation of revenue sources, and condition 4, subnational governments have authority to cut their costs. In Colombia in 1986, municipalities were assigned responsibility for constructing and maintaining schools and for administering the teaching staff assigned (and paid) by the national government. In the health sector, municipalities were made responsible for constructing and maintaining health care facilities. Municipalities were assigned responsibility for providing water and sewerage services and for constructing and maintaining local roads and urban transport facilities. To finance these expanded responsibilities, Law 12/86 committed the government to increasing municipality's share of the IVA from 50 percent (by 1992). Corresponding national agencies were to be abolished. These included the Colombian Institute for School Construction, the National Institute for Municipal Development, INSFOPAL (which was responsible for water and sewerage), and urban development companies, which were responsible for the construction and maintenance of urban roads, management of urban transport, and provision of low-income housing. The situado fiscal always provided most of the departments' revenue, as the law gave them little scope to raise their own revenues: 60 percent of the situado was to be spent on education, 20 percent on health, and the remaining 20 percent on either sector. With the situado, departments must pay teachers and health workers, whose salaries are set in negotiations between the national government and the unions and whom the departments are effectively prohibited from dismissing. In theory since 1995 the departments have had discretion within the sector as to how to use the situado fiscal once the salaries have been paid, but such surpluses have been rare thus far. The national government's authority to negotiate teachers' salaries directly with the national teachers union has already provoked a funding crisis in the sector. In 1996 the government conceded an 8 13  percent real increase in salaries over three years. As the resulting wages exceeded the amount of the situado assigned for education in many departments, the government was forced to establish the Special Compensation Fund to finance the gap between the situado and the actual costs of teachers' salaries. For the municipalities, revenue sharing was also earmarked, but less strictly: 30 percent was to be spent on basic education (infrastructure, equipment, or personnel), 25 percent on health, 20 percent on water supply (except in municipalities that had already achieved 70 percent coverage), 5 percent on physical education, and the remaining 20 percent on housing, welfare, debt service, and other functions. Given the latitude for interpreting these terms, the earmarking of MUNICIPAL TRANSFERS has not been restrictive in practice. Some of the assignment of functions depends on the establishment of adequate management capacity. Legislation in 1993 (Law 60) reversed the earlier decentralization of social services to municipalities.5 In education, it reassigned responsibility for managing secondary and primary schools to the departments, except for (larger) municipalities that met established criteria for financial and managerial competency. In health, decentralization was included with ongoing reforms in the structure of public sector health provision. These measures aimed to shift public health spending from the so-called supply side-direct funding of public health care facilities-to the demand side, in the form of subsidies for national insurance. Some uncertainty remained in the allocation of responsibilities between the state health schemes and the national social security system (Bird and Fiszbein 1996). In implementing this program, territorial governments were to transform their existing facilities into public corporations, which would finance themselves by 5Due to its political composition, the constituent assembly that wrote the 1991 constitution could not provide for differential treatment of municipalities or departments according to their size and capacity, even though the differences are great. This had to be done later by law (Perry and Rodriguez 1991). 14  charging the insured a fee for service. During the transition, each level of govemrnent was to continue subsidizing its own facilities. To assume the administration of education and health under the terms provided by Law 60, each department and district was required to receive accreditation by the respective ministry of education or health. In principle, accreditation required (1) a functional information system, (2) the adoption of a development plan for each sector, (3) the approval by the departmental assembly of the rules and procedures for the distribution of funding, (4) the adoption of a plan for coverage, and (5) agreement with the respective ministry on institutional arrangements. These institutional arrangements delineated which dependency of the department would take responsibility for the service, how the assets would be transferred, and how personnel would be structured and administered. Once accredited, departments were to receive, within four years, the assets and personnel that would "permit them to comply with the functions and obligations they have assumed." In the absence of accreditation, these services were to continue being provided directly by the national government. (The law also established that departments could, after they were accredited and at the discretion of their legislative assemblies, firther decentralize education to the municipal level, subject to an accreditation process.) Initially only the departments of Valle del Cauca (Cali) and Antioquia (Medellin) were fully certified, and a few others were certified for health services only. In the mid-1990s, the number expanded to include most of the departments and all the large cities. Once accredited, departments still remained subject to national-government control over key management issues. The law prohibited any territorial government from hiring teachers without meeting the requirements of the estatuto docente. Departments were required to incorporate the terms of employment of all existing government or nationalized staff.6 Existing departmental or 6 Departmental or municipal teachers who were transferred to the national government prior to Law 60. 15  municipal staff had to be brought into the national payment structure for teachers. Most important, all future remuneration and salary scales had to conform to national legislation, set in negotiations between the national government and the national union. After several adjustments to the rules, which themselves create some uncertainty, the spending assignment in Colombia largely meets condition 3, in that revenue decentralization is accompanied by clear assignment of responsibilities, especially to departments, and the national government is not left with any major unfinded responsibilities for service delivery. For municipalities the assignment of functions is less clear, but the requirement of demonstrating administrative capacity also serves as a check. From a deeper perspective, the prevalence of earmarking and sometimes mandating expenditure even without a corresponding transfer reduces the political value of good fiscal management by subnational government (because the unfunded mandates already claim their marginal resources some times.) Condition 4, freedom to cut costs, has not been met, at least for departments. Furthermore, the Special Compensation Fund to offset this problem overrides the automatic fonnula for transfers and thus partially violates condition 2. Borrowing constraints. Although tax and spending policies create fiscal pressures, whether they cause problems for macro-fiscal management depends on whether the SNGs face hard budget constraints, limiting their borrowing. A hard budget constraint is universally recommended as essential for getting proper fiscal behavior (Tanzi 1995; Ter-Minassian 1997; IDB 1997; Weingast 1995; Wildasin 1997). Some borrowing may be sustainable and good for development, and many national governments, such as that of Brazil, try to control SNG borrowing (Ter-Minassian and Craig 1997). Unsustainable deficits would be less likely under condition 5, central government strictly controls subnational borrowing ex ante. But how to achieve this in practice is not always clear when the subnational governments have considerable political autonomy. Pseudo-strict controls could make matters worse, if central govermnent approval creates the impression and 16  perhaps a self-fulfilling expectation that the central government has also made a guarantee (Bird and Fiszbein 1996). To run deficits, a subnational government must find a source of financing, which potentially includes contractual borrowing from private domestic or foreign banks, issuance of domestic or foreign bonds, and running up of arrears to suppliers and personnel.7 A creditor and the subnational government would only agree to finance unsustainable deficits if both sides expected to gain, most likely though some sort of federal bailout. The bailout could take many forms, including allowing the financial system (implicitly insured by the government) to count as an asset debt that is not being serviced. So unsustainable deficits would also be less likely under condition 6, central government credibly commits not to have bailouts, prohibiting explicit bailouts and forcing subnational governments to service their debts, and under condition 7, regulators force creditors to accept the losses implied by any failure to service debt. There is understandable ambivalence about the incentive effects of deductions from transfers to force subnational governments to pay debt service. In Argentina they have worked to enforce a hard budget constraint because the national government has not filled the gap with extra transfers when provinces found themselves unable to meet expenses after the deduction of debt service (Dillinger and Webb 1998). In Mexico, on the other hand, the federal govermment has repeatedly given extraordinary transfers to over-indebted states when debt service became burdensome and the automatic transfers has been a route through which the responsibility has been traced back to the federal level. In one case they reduced the moral hazard, in the other they increased it. The policy on transfers made the difference. It is also still an open question whether ex ante regulation or ex post enforcement of debt service is more effective in preventing excessive SNG borrowing. The comparison of Argentina's experience with ex post enforcement and Brazil's experience with ex ante regulation suggests that 17  the former approach works better, but of course the two are not mutually exclusive and can reinforce each other (Dillinger and Webb 1998). The administrative details may matter as much as the legal principles, because the former determine who starts bearing the cost immediately when there is some problem with payments. The effectiveness of the ultimate legal remedies depends in part on whether they are anticipated by, or are contradictory to, the initial allocation of burdens. In Colombia the controls on subnational borrowing have varied over time. In the 1980s and before, all subnational borrowing had to be approved by the Ministry of Finance, and such approval was the exception. This was natural, since the subnational entities were appointed representatives by the national government and had no political or fiscal autonomy. The ad hoc approval process gradually allowed more freedom for domestic borrowing in the late 1980s and 1990s, as the political and fiscal autonomy of subnational governments increased. (For external borrowing the control was strict.) There was no effective ex ante control of cash advances from banks. In 1997 the so-called Ley de Semaforos (the traffic-light law) brought into effect a rating system for territorial governmnents, based on the ratio of interest to operational savings and debt to current revenues (Perry and Huertas. 1997).