Report No. 42154-YF Serbia Decentralization and Local Service Delivery March 8, 2008 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank FOROFFICIAL USE ONLY ACKNOWLEDGMENTS This report was prepared by a team consisting o f Lazar Sestovic (Macroeconomic Issues), William Dillinger (Sub-national Fiscal Framework and Financing Issues), Aleksandra Rabrenovic (Local Government Stucture), Pia Schneider (Health), Toby Linden(Education) and IvanaAleksic, Abdu Muwonge, Maria Laura Sanchez Puerta and Truman Packard (Welfare Assistance), and led by Mathew Verghis. Ekaterina Stefanova managed the processing of the document, and Hermina Vukovic coordinated the team missions. Ardo Hansson provided very helpful comments which greatly improved an earlier draft. Bernard Funck and Carolyn Jungr provided strategic guidance. The report was prepared under the overall supervision of Orsalia Kalantzopoulos and Cheryl Gray. The team also benefited from helpful comments on the concept paper and during preparation from the peer reviewers Deborah Wetzel and Kai Kaiser, and from Dena Ringold and Markus Goldstein. The useful comments received from participants at the decision review meetingwere incorporated into this draft. The team appreciates discussions with staff o f the Ministries o f Finance, Health, Education and Sports, and Labor, Employment and Social Policy, the Standing Conference of Towns and Municipalities o f the Republic of Serbia and from the municipalities o f Belgrade, Indjija, Novi Sad, Kraljevo, Pecinci and Backa Topola. This document has a restricted distribution and may be used by recipients only in the performance o f their official duties. Its contents may not be otherwise disclosed without World Bank authorization. CONTENTS EXECUTIVESUMMARY .......................................................................................................................... V 1. MACROECONOMIC ISSUESINDECENTRALIZATION .......................................................... 1 A. INTRODUCTION .................................................................................................................................. 1 B. c. MACROECONOMICMANAGEMENTSUBNATIONAL AND GOVERNMENTS ............................................ 1 INSTITUTIONAL FRAMEWORK FORMACROECONOMIC MANAGEMENTSUB-NATIONAL AND GOVERNMENTS ........................................................................................................................................... D. CONCLUSIONS.................................................................................................................................... 9 6 2. FISCAL FRAMEWORK AND FINANCIALISSUES .................................................................. 11 A. INTRODUCTION ................................................................................................................................ 11 B C .. EXPENDITURE RESPONSIBILITIES AND REVENUE ASSIGNMENTS ....................................................... 12 THENEW INTERGOVERNMENTAL 14 FISCAL FRAMEWORK .................................................................. D. FINANCINGISSUES HEALTH EDUCATION................................................ INDECENTRALIZING AND 22 3. LOCAL GOVERNMENT STRUCTUREAND ADMINISTRATION ........................................ 29 A . INTRODUCTION ................................................................................................................................ 29 B. POLITICAL AND ADMINISTRATIVE STRUCTURE OF LOCAL GOVERNMENTS ..................................... 29 C. LEGAL FRAMEWORK FOR EMPLOYMENT OF LOCAL GOVERNMENT OFFICIALS .................................. 31 D. OPTIONSFOR WAGESETTING IN LOCAL GOVERNMENTS ................................................................ 35 E PREPARINGMUNICIPALITIES FOR NEW RESPONSIBILITIES INDELIVERY OF SERVICES ....................... 39 F.. CONCLUSIONS .................................................................................................................................. 41 4. DECENTRALIZATION INHEALTH ............................................................................................ 43 A .. . INTRODUCTION ................................................................................................................................ 43 B LESSONS HEALTH FROM SECTOR DECENTRALIZATION FROM OTHER COUNTRIES ........................... 43 C HEALTH SECTOR DECENTRALIZATIONSERBIA IN ............................................................................ 45 D. HEALTH SECTOR REFORMS AND DECENTRALIZATIONHEALTH IN .................................................. 47 E. CONCLUSIONSAND NEXT STEPS...................................................................................................... 53 5. EDUCATIONAND DECENTRALIZATION ................................................................................ 59 A . INTRODUCTION ................................................................................................................................ 59 B C .. DISTRIBUTION EDUCATIONFUNCTIONSINSERBIA AND INTHECOMPARATIVE CONTEXT OF ............59 MAJOR 61 D. CHALLENGESFACINGTHE SERBIAN EDUCATIONSYSTEM ................................................. SECTOR ..................................................................................................................................................... THE ROLE OF DECENTRALIZATION INHELPINGTOADDRESS PROBLEMSINTHEEDUCATION THE 69 E. CONCLUSION AND NEXT STEPS INDECENTRALIZATION .................................................................. 74 6. DECENTRALIZATION AND WELFARE ASSISTANCE TO SERBIA'S POOREST HOUSEHOLDS ........................................................................................................................................... 79 A. INTRODUCTION ................................................................................................................................ 79 cB. . SOCIAL WELFAREASSISTANCE SERBIA....................................................................................... IN 80 DECENTRALIZING SOCIAL ASSISTANCE: 87 D. FROMTHEORY PRACTICE............................................ TO MITIGATING RISKSOFANDREAPING THE GAINSOFDECENTRALI~ATION THE ................................ 95 REFERENCES ............................................................................................................................................ 97 ANNEX 1:DISTRIBUTION OFRESPONSIBILITIESFORSCHOOLSINSERBIA ............................................. 101 ANNEX 2: PILOTINGTHE DECENTRALIZATIONOF FINANCINGAND ADMINISTRATION .......................... 103 TABLES TABLE1.1:TOTALEXPENDITURE AND NET TABLE1.2: CONSOLIDATEDSUB-NATIONAL GOVERNMENTFISCALOPERATIONS. AS YoGDP ......................... LENDING. LEVELOFGOVERNMENTAS YoOFGDP...................2 BY TABLE1.3: ARREARS EPS. INUSD ............................................................................................................ 3 TABLE1.4:ARREARS TONISINUSD............................................................................................................... TO 5 6 TABLE1.5: MECHANISMSENSURINGFISCALDISCIPLINE FOR ......................................................................... 7 TABLE 2.1: POPULATIONSIZE OF SERBIAN MUNICIPALITIES ......................................................................... 11 TABLE2.2: AVERAGE POPULATIONOF MUNICIPALITIESSELECTED EUROPEAN COUNTRIES IN ..................... 12 TABLE2.3:FUNCTIONALDISTRIBUTION MUNICIPAL OF EXPENDITURES(AVERAGE, 2004-2006) .................13 TABLE2.4: SOURCES OFMUNICIPAL REVENUES (BUDGET, 2003) .................................................................. 14 TABLE2.5: IMPACTOFREFORM ONPER CAPITA REVENUEDISPARITIES .ERROR! BOOKMARK DEFINED NOT . TABLE2.6: COMPONENTS OFPROPOSEDTRANSFER SYSTEM ......................................................................... 18 TABLE2.7: TRENDS INMUNICIPAL FISCALPERFORMANCE ........................................................................... 19 TABLE 2.8: SUBSIDIES TO PUBLIC UTILITY ENTERPRISES .............................................................................. 19 TABLE 2.9: SIZE OF LOCAL GOVERNMENT POST-TRANSITION COUNTRIES IN ................................................ 22 TABLE 3.1: STATUS OF LOCAL GOVERNMENT STAFF INEUROPEAN COUNTRIES ........................................... 32 TABLE 3.2: STRUCTURE OF SALARIES OF MUNICIPAL OFFICIALS ................................................................... 36 TABLE 3.3: PRIVATE SECTOR PAY COMPARISON ........................................................................................... 37 TABLE4.1: PUBLICHEALTH EXPENDITURES BY SOURCE OF FUNDING, 2004 ................................................ 46 TABLE 4.2: INFORMATION SYSTEM FOR HOSPITAL ACTIVITY MINIMUMDATA SET ...................................... TABLE5.1: CROSS-COUNTRY COMPARISON OFTHE DISTRIBUTION EDUCATIONRESPONSIBILITIES OF .........56 61 TABLE5.2: LABOR MARKET OUTCOMES INALBANIA, BOSNIA,MACEDONIA SERBIA AND ........................... 62 TABLE5.3: ENROLLMENTDATA FOR SOUTHAND CENTRAL EUROPEAN COUNTRIES ..................................... 63 TABLE6.1: SPENDINGON SOCIAL WELFARE ASSISTANCE AND CHILDPROTECTION INSERBIA, 2005, ........................................................................................................................... 81 TABLE6.2: INCIDENCE OFTHE MOPAND CHILDALLOWANCES QUMTILE .............................................. PERCENTAGEOF GDP BY 84 TABLE6.3: SOCIAL WELFARE ASSISTANCE REPUBLICANAND MUNICIPAL IN BUDGET, 86 TABLE6.4: TABLE3 SOCIAL EXPENDITURESBY MUNICIPALITY. ......................................................... . 2005 ....................... 2004 87 FIGURES FIGURE 1.1: CHANGE INTHE LEVEL EXPENDITURES OF 2003.2005. BY TYPE. AS %GDP ................................ 5 FIGURE 2.1: VARIATIONS INMUNICIPAL CAPITA REVENUES. POST-REF0RM PER .......................................... 18 FIGURE 2.2: LOCAL EXPENDITURES Yo OF GDP AS ........................................................................................ 20 FIGURE5.1:RELATIONSHIPBETWEENMEAN MATHEMATICS PERFORMANCEAND GROSS NATIONAL INCOME PER CAPITA (PPP$) .............................................................................................................................. 61 FIGURE5.2: PARTICIPATIONRATESACROSS MUNICIPALITIES ....................................................................... 66 FIGURE5.3: MATHEMATICS PERFORMANCE .................................................................................................. 67 FIGURE5.4: CORRELATIONOF MATHEMATICS AND DISADVANTAGE ................................................. FIGURE5.5: CORRELATIONBETWEENPARTICIPATIONAND MATHEMATICS ....................................... SCORE 68 SCORE 68 FIGURE6.1: SPENDINGONMEANS 2003 ..................................................................................................................................................... 82 TESTED CASH BENEFITS SELECTEDEUROPEAN COUNTRIES, CIRCA IN FIGURE6.2: SPENDINGONMEANS-TESTED RESORT" CASHBENEFITS, "LAST SELECTED EUROPEAN COUNTRIES, CIRCA 2003 ...................................................................................................................... 83 FIGURE 6.3 :TARGETING AND COVERAGE OF NON-CONTRIBUTORYASSISTANCE SOCIAL CASH BENEFITS, EUROPE AND CENTRAL ASIA, 2003 ...................................................................................................... 83 FIGURE 6.4: CONCENTRATION CURVES FOR SELECTED SOCIAL WELFARE AND CHILD PROTECTION PROGRAMS,2003 ................................................................................................................................. 84 FIGURE6.5: CURRENT AND LIKELYFUTURE DIVISIONOF RESPONSIBILITIESFORSOCIAL WELFARE ASSISTANCEINSERBIA ......................................................................................................................... 93 .. 11 Box2.1: VOJVODINA..................................................................................................................................... 21 BOX 1.1: ARREARS OF LOCAL GOVERNMENTS LOCAL AND UTILITIES ............................................................ 5 BOX 2.2: D O HEALTH CARE FINANCING REFORMS CONFLICT WITH OTHER GOVERNMENT OBJECTIVES? ....23 BOX 3.1: PROBLEMS ENCOUNTERED FISCAL DECENTRALIZATION WITH ....................................................... 40 BOX 6.1: DESIGN BENEFITSRJSERBIA ............................................................................................................................. AND IMPLEMENTATION PROBLEMS INTHE DELIVERYOF SOCIAL ASSISTANCE CASH 85 BOX 6.2: ADMINISTRATIVE CONSOLIDATIONOF SERBIA'S CASH BENEFITS FORPOVERTY REDUCTION ........94 ... 111 EXECUTIVE SUMMARY A. INTRODUCTION 1. The delivery of public services is fundamentally decentralized: citizens receive services from schools, health centers and municipality offices rather than central government ministries. The responsibility for delivering these services, however, i s often highly centralized, with most key decisions being made in capital cities. In many countries there i s a trend to transfer responsibilities for responsibilities closer to the points o f delivery: to service providing institutions such as schools and health centers and to local governments. This i s becoming the case in Serbia as well. This report reviews some aspects o f this ongoing decentralization process in Serbia with a view to providing suggestions for better management o f the process and to improve service delivery. 2. The basic unit of local government in Serbia is the municipality. Municipalities are geographical subdivisions o f the national territory. In rural areas, a municipality may include both towns and the surrounding countryside. Belgrade and three other large cities, Novi Sad, N i s and Kragujevac, have the status o f cities. There are two autonomous provinces, Vojvodina and KOSOVO~, and a region under direct control, Central Serbia. 3. Preparationsare underway in several sectors for decentralization. The main responsibilities o f municipalities are currently for infrastructure services such as heating, water and sanitation, local transport and maintenance o f physical assets. Several ministriesare discussing plans to transfer more responsibility for delivery o fpublic sector services towards municipalities and service providing institutions. Plans appear most advanced in health, but discussions are underway in education and social protection as well as other ministries. As a result, Government's ability to deliver public services will increasingly dependon the performance o f municipalities. An important message o f this report i s that plans for decentralization should be made as part o f a broader sector reform. This is indeedthe case inhealth, and to a lesser extent in social welfare. The report also highlights the importance of reforms in the infrastructure services that are already municipality responsibilities, and ineducation. 4. In principle decentralization can improve service delivery by allowing decisions to be made closer to where services are being provided. But decentralization can also provide an avenue for fiscally irresponsible behavior and poor performance. Good design, sound management and constant adaptation are required. This report focuses first on three sets of cross-cutting issues: macroeconomic management o f decentralization the fiscal framework for municipal finance and 'Vojvodina is currently under interim UN administration although Serbia continues to finance some expenditures inthe Serb-dominatedenclaves in Kosovo includingthe payment o fteacher's salaries. municipal administration; and then discusses decentralization in three sectors o f particular interest to the World Bank -health, education and social welfare. B. MACROECONOMICINDECENTRALIZATION ISSUES 5. Fiscal performance in Serbia since 2001 can be broken into three periods. The initial period 2001-2003 was one of rapidly expanding expenditures, revenues and deficits. Sub-national governments saw a significant increase in their revenues and expenditures. Serbia undertook a significant fiscal adjustment during the second phase covering 2003-05. Much o f this was at the central government level, although there was some contribution from municipal governments. The third period (2005-06) saw an unfortunate reversal o f fiscal adjustment, at both the sub-national and Republican levels. 6. The structural adjustment by the Republican Government has not always been matched at the sub-national level. Consistent with the need to achieve a smaller and more efficient state and to harden the budget constraint on the enterprise sector, the Republican Government reduced subsidies and restructured the civil service and health sector to contain wages. Between2003 and 2005, subsidies from the central government fell by more than 1 percent o f GDP. However, spending by sub-national governments moved inthe opposite direction. Total expenditure inthe central government (excluding extra budgetary funds) fell by more than 4 percentage points o f GDP. For sub-national governments, total expenditure fell by less than half of one percentage point of GDP. If municipal governments implement the kinds o f reforms that central government has to modernize the state, resources will be released from funding subsidies and salaries and can be usedto fund investment or reduce taxes. 7. For the entire period since the beginning of transition sub-national governments taken together have been roughly in fiscal balance. In many countries, sub-national governments have generated deficits and created problems for macroeconomic management. Serbia has managed to avoid this, an important indicator o f the institutional arrangements for macroeconomic management, although it may also indicate that municipalities are generously funded relative to their responsibilities. 8. Improved operation of municipality owned companies might free resources for other uses. A significant share o f municipality expenditure appears to be covering the operating losses and capital needs of municipality owned enterprises. The data are not classified in a consistent way across municipalities, but it appears likely that the entire amount listed under subsidies i s to these enterprises; as well as some portion o f the capital expenditures. Thus in 2005, depending on the portion o f capital expenditure o f municipalities allocated to their enterprises, between 22 percent and 35 percent o f municipal expenditure were on these enterprises. These services are normally covered by tariffs, and external sources o f funds should be able to cover capital needs. This is particularly true when concessional sources o f funds appear to be available. 9. Managing the macroeconomic risks arising from the operations of subnational governments (including their companies) is likely to be increasingly challenging. Following the approval o f the new Constitution, municipalities will have vi option o f generating increasing revenue streams from sale o f land and companies. These revenues will not be governed by the Law on Public Debt and might be spent wastefully, or if spent very quickly could have a significant macroeconomic impact. Countries have taken different approaches to managing the macroeconomic impact including such measures as setting up "rainy day funds" for US states and rules requiring privatization revenueat the subnational level to be spent on capital rather than recurrent expenditure. 10. The Republican Government should use consolidated general government account as the unit of analysis in preparing the macroeconomic framework. At present, the general government balance table i s prepared only after the budgets have already beenadopted. As a result, the decision on the fiscal position i s taken on the basis o f the central government budget alone. Untilnow, this has not mattered too much, since sub-national governments have not relied on borrowing to any significant extent and thus have effectively had expenditures limited by the transfer from the central government. As the resources available to sub-national governments (including below the line revenues from privatization and other asset sales mentioned above and borrowing) are likely to rise, it will become important to take their fiscal position into account in macroeconomic management. c. FISCALFRAMEWORK AND FINANCING ISSUES 11. Currently, responsibilities of municipalities are relatively limited when compared to municipal responsibilitiesin the new member states of the EU. The majority o f the recurrent costs o f public services are paid directly by the Republican Government or the Province o f Vojvodina (the salaries o f teachers, health workers, and police; social assistance, and child allowances) or financed, at least in part, by the tariffs o f public enterprises (water supply, district heating, bus service). The core role o f municipalities i s to build, acquire, or maintain physical assets: public housing, streets and bridges, water works, district heating, plants, buses, and-in an ad hoc manner-schools, clinics and other assets under the control o f the Republic Government. In many o f the new member states, some of the services paid for by the Republican Government in Serbia are paid for by local governments. 12. The new Local Government Finance Law that came into effect in January 2007 is an important improvement in the fiscal framework governing the relationship between the central government and municipalities. Two major achievements of this law are the (i)improved transparency and predictability o f the transfer o f revenues to the central government; and (ii) more equitable distribution o f a revenues betweenmunicipalities relative to the previous system. 13. Despite the improvement brought about by the new law, there are still significant differences in municipal revenue in Serbia. Municipalities in the two richest cities, Novi Sad and Belgrade, have more than 3 times higher per capita revenue than other municipalities. There i s evidence that wealthier jurisdictions use this largesse to make unjustifiably large subsidies to public enterprises. Under the provisions o f the recently enacted Constitution, central government transfers to the Province o f Vojvodina vii may also be high. Inthe past, these transfers have largely been used to finance teachers' salaries-a expense that i s paid directly by the Government in the rest o f Serbia. However, the recent Constitutionally-mandated increase in central support to Vojvodina has not been accompanied by a commensurate offloading o f central government responsibilities, however. 14. Implementing the public debt law will pose a challenge. The public debt law provides a good framework to regulate municipal borrowing. Municipalities are permitted to borrow in the domestic and in foreign capital markets subject to: (1) a five percent ceiling on ratio of short term debt (stock) to revenues; (2) a fifty percent ceiling on the ratio o f medium and long term debt to revenue, and (3) a ceiling o f 15 percent on the ratio o f mediumand long term debt service to revenues. Judging from recent data on the stock of municipal debt, these constraints do not appear to be binding. No municipality has reached the 50 percent ceiling on the long term debt to revenue ratio. (Data on the other indicators i s not yet available.) Belgrade's debt revenue ratio i s only 28 percent. In most municipalities, the ratio i s well under ten percent. Instead, the constraint appears to be the reluctance o f commercial banks to lend and the inability of many municipalities to prepare viable projects. As municipalities start to overcome these constraints, it will be very important for the Ministry of Finance to monitor and enforce the public debt law - in many countries, sub-national governments have found creative ways to get around debt ceilings. D. LOCAL GOVERNMENT STRUCTUREAND ADMINISTRATION 15. Public administration reformat the municipallevel is needed to underpinthe sectoral decentralization efforts. The focus of decentralization has been sectoral, focused in areas such as health or education. However, if these efforts are to be successful, or to improve service delivery inthe existing competencies o f municipalities, the functioning of municipalities themselveswill needto be improved. The report points to the working conditions o f municipal officials and options for wage determination, as well as the functioning o f the municipal administrative structures. 16. Employment conditions at the local level are still covered under the (largely defunct) Law on Labor Relations in State Organs Wages are also defined within an outdated legal framework. The critical question i s how to define local official status in a manner that i s consistent with the new wage system set out for civil servants at the central government level, while allowing the flexibility that i s needed to permit decentralized decision making. 17. Employment of senior local government officials (as well as the senior officialsof municipalityowned companies) should be depoliticized. One way to move inthis direction is to allow heads of departments to keep their status beyondthe period of mayors/municipal assembly term inoffice as i s the case for the Republican Civil Service. Other measures could also be instituted including restricting reasons for dismissal o f these officers and establishing merit-based promotion procedures. ... Vlll 18. Attention needs to be paid to the organization and management structure of local officials. Currently, the structures are relatively standardized, but these need to matched to new functions as they are decentralized. For example, municipalities will needto buildcapacity to oversee the management o f health care facilities, and ifa similar process i s planned for education, municipal capacity will have to be built there. 19. The structure of municipaladministrationleads to instability. In2006 alone, 10 municipalities had mayor recall procedures instituted and mayoral elections held, while new elections had to be called for 12 municipal assemblies. Insome o f these and other municipalities, the large number o f political parties that are represented in the assemblies leads to the executive and legislative branches o f municipalities from working together effectively. As a result, key appointments such as Head o f Administration are sometimes heldup. 20. Capacity building should follow a demand driven approach. In an ideal world, when tasks are decentralized, local governments are fully capable o f handling their new responsibilities. Indeed, decentralization without adequate capacity has traditionally been considered undesirable. Although there i s merit to this position, experience is showing that local capacity expands with decentralization although managing the transition creates difficulties. The nature o f capacity building i s shifting as well, from a top-down supply driven model o f training provided by the center to a model of demand- driven capacity growth with supply side assistance. In this model, the central government creates an enabling environment, using its finance and regulatory powers to help municipal governments define their own needs, and to deploy training from many sources, including the private sector and international donors, to learn by doing as decentralization proceeds, and to establish learninggroups among municipalities. E. DECENTRALIZATIONHEALTH IN 21. The health sector has made a good start in preparing for decentralization. Critically, decentralization i s being undertaken as part o f an overall sector reform that includes a comprehensive human resource strategy. The Health Care Law o f 2006 regulates the form o f decentralization. It includes the transfer o f hospitals to provincial authorities and the four cities and o f primary health care institutions to municipalities; and i s accompanied by a transfer o f decision-making authority to municipalities for the monitoring o f population health and appointment o f directors o f primary health care institutions. These changes are to be accompanied by a shift from input-based financing o f health care towards capitation system that should provide directors o f health care facilities the ability to manage resources more efficiently than under the current system. The sector has also started to implementthe humanresource. 22. Decentralizing health care over the next five years will affect 160 Primary Health Care Centers. They provide basic primary services including General Practice (GP), gynecology and pediatrics. PHC providers are organized either as a separate entity (84 PHC) or as part o f the secondary care hospitals-(76 PHC). Patients in need for specialized primary care are referred to 60 specialized centers (Zavodi). PHC with larger catchments areas usually provide a broader range o f secondary outpatient services including ophthalmology and cardiology. In addition, there are 186 hospitals offering ix acute, psychiatric and long-term care. Out o f 92 tertiary facilities, 45 are concentrated in Belgrade, especially clinical hospital centers (4) and institutes (32), which raises concern about access to tertiary care inmarginalizedregions (Sanigest, 2006). 23. Decentralization needs to be better coordinated with organizational and provider payment reforms, to prevent negative effects on sector performance. As mentioned above, health sector decentralization i s being implementedin the presence o f major sector-level reforms, including provider payment reforms and health sector restructuring. Currently these three reforms (decentralization, payment reforms, restructuring) are being orchestrated independently and influenced by different stakeholders. They need to be combined and their potential effect evaluated to ensure they support each others effectiveness. 24. The challenge going forward in health is to implement the strategy in a manner that is consistent with the capacity of municipalities and with sufficient possibility for adaptation with experience. Clear responsibilities, roles and accountabilities will need to be defined to sequence decentralization in a manner that i s consistent with capacity; and rigorous evaluation. Providers must have some degree o f autonomy to re-organize service delivery and managing their resource-mix. Continuous training should be organized for providers about payment reforms to ensure they understand which changes will be possible and necessary to do well. 25. The cost and efficiency impact of movingfrom input-basedline-item budgets to capitation payment in PHC centers and case-based payment in hospitals should be monitored and evaluated in a pilot study. Since the move to capitation payments represents a significant change, it should be carefully monitored. The Ministry o f Health and the Health Insurance Fundcould collect monthly routine data on individual patients' health status and care seeking behaviour and the resulting case-mix; expenditures and finances in all health facilities for insured and uninsured patients in at least 50 PHC centers during the year before and after the introduction o f the payment reform. Through the payment reform, providers would be able to influence their efficiency by improving the management of drug storage; the more flexible use of staff; changing their treatment style to improve drug prescriptions; and by exploiting scale economies and treating more patients. A longitudinal provider dataset would permit unique analysis to be performed, employing econometric production and cost functions to compare the effect of capitation payment plus a small co-payment on provider productivity, cost and efficiency in HCs and on referral rates to hospitals. F. EDUCATION DECENTRALIZATION AND 26. In education, plans for reform are at an early stage. Serbia faces problems providing good quality education and i s failing to equip young people with the skills and knowledge they need to be successful in the labor market. In the PISA assessment Serbia's students came in 33'd place out o f 40 in mathematics, 36'h inreading and 34'h in science. No other European country scored below Serbia, even though many transition countries participated in the assessment. Labor market outcomes for young people are also poor. There i s a presumption that Government i s committed to decentralization but no decision has beentaken to date. X 27. The education system is becomingincreasinglyinefficient. Between 1997 and 2005, the primary school population fell by 24 percent and the secondary school population by 15 percent. However, the numbers of teachers and schools have not fallen as pupil numbers have fallen. In fact, quite the opposite i s true: in secondary education, for example, in the last five years, in Central Serbia the number o f pupils in secondary educationfell by 9.3 percent but the number o f teachers increased by 6.5 percent. There was a similar story in Vojvodina, with pupil numbers lower by 8.3 percent but teacher numbers higher by 6.8 percent. As expected, this has resulted in a sharp decline this relative short period in the number of pupils per teacher - from 13.5 to 11.5 for the country as a whole. By comparison, the OECD average is 14 pupils per teacher. There has beena similar experience inprimary education. 28. This is leading to wide variations in per student resource allocation. These historical trends have left Serbia with a large number of small schools. Primary schools with less than 200 studentsaccount for seven percent of schools but less than two percent o f pupils; and more than half o f all primary schools have satellite classesS2There are more than 300 schools with five or fewer students. It also appears that the way pupils are dividedinto classesvaries across the country; for example, schools with 100 students ina grade have betweenfour and six classes for those students. 29. A major weakness in the financing of education is that management of resources is split making consideration of trade-offs difficult. Decisions about spending on wages (and in particular on the number o f staff) are made by the central government while decisions about other inputs (like professional development, materials, maintenance, etc) are made by municipalities. This means that there i s no place in the system whereby choices are made between spending on one category o f expenditure as against another. This would make it difficult inany system to shift expendituresto more effective uses. It i s doubly unfortunate in Serbia because the decline in student numbers means that there i s the potential to capture savings on salaries where pupil numbers are declining and spend these resources on non-salary inputs and in some cases in different parts o fthe country. 30. A more equitable and efficient distribution of resources is likely to result from the introductionof a formula-based allocation to municipalitiesof funding for staff wages. In this way, municipalities with similar characteristics would receive the same level o f funding, helping address a key inequity in the current system. Given the current inequalities in funding, any formula will result in some shifts in funding across municipalities-the existing allocations would not be replicated. In addition, such a move would enable municipalities to make decisions about both staffing and non-staffing inputs,with apotential improvement inefficiency of spending. 31. The benefits of the allocation formula will only accrue if there are commensurate changes in the degree of control entrusted to municipalities. Municipalities should be able to determine how to spend the resources allocated to them Satellite classes are those in small buildingsin small villages away from the main school building; usually offering only the first four grades of primaryeducation. x i and over the factors that affect the level o f spending, including the ability to switch funds betweenwage and non-wage recurrent budgetcategories, employment o f staff, especially teachers; the setting and verification of minimumservice standards for municipalities. 32. One of the most complex choices the authorities will need to make concerns the management of the school network. The most immediate concern i s in primary education, where the numbers o f students will continue to decline and enrollment rates are already close to 100 per~ent.~ Declining school-age cohorts will provide an opportunity to reduce the number o f teachers and downsize or eliminate schools (although internal migration will require additional classrooms in some locations). The government needs to decide whether the central or the municipal government will close inefficient schools when needed. A solution to consider is a partnership between the central and local governments, with the central government providing guidance on closure decisions and some transitional funding and the municipal governments responsible for implementingplans. 33. A way forward is an approach which includes the use of pilot schemes to evaluate different options. Decentralizationis one mechanism for trying to achieve the improved quality and access outcomes that are necessary for Serbia. It i s not the only one however, and pilots provide a mechanism for evaluating which approach works best in the Serbian context. The central policy option assessed here is the extent to which responsibilities are given to schools or to municipalities. The purpose o f the pilot i s to prepare the way for a broader decentralization as part o f an overall strategy for improving the performance o f the education system. Many o f the activities needed to implement a pilot will be those which will provide the foundation for this broader reform, even if the pilot i s by definition on a smaller scale. 34. The approach suggest herewould not permit individualmunicipalitiesto opt in or out of the pilot; this will represent a departure from the usual practice in Serbia. The reason for a randomized approach to a voluntary approach is that randomization i s most likely to provide the best information on how the pilots are working, and hence to come up with program designs that are best suited to Serbia. Allowing municipalities to volunteer as pilots i s likely to lead to the most capable municipalities being evaluated, and hence not provide a full picture o f the challenges o f decentralization. These pilots would be run for two years, with country-wide implementation o f the approach which has yielded the best results taking place in the third year. A clear timeline is important to ensure that the pilots do not become a mechanism for postponing reform. G. DECENTRALIZATIONSOCIAL ASSISTANCE AND 35. Two principal challenges are to increase the poverty alleviation impact of targeted cash benefits; and to improve outreach and meet the needs of vulnerable As noted above, even though secondary school cohorts are likely to decline too, the need to increase enrollment rates places might mean there is no need to reduce the number o f secondary schools or secondary school places. xii groups. Capturing the benefits o f shifting responsibilities to local actors while mitigating the risks that arise with decentralization will be critical to meetingboth challenges. The Ministry's Strategy emphasizes the increasing role that local self governments (LSGs) and non-government agencies have to play in delivering social welfare assistance. Decentralization o f social assistance has the potential to improve welfare by bringing services closer to households; allowing a wider set o f local providers to adapt services to meet the specific needs of vulnerable groups that often vary substantially across municipalities; and improving targeting, outreach and impact monitoring by placing greater reliance on municipal authorities who are closer to beneficiaries. However, decentralization also entails risks, particularly for the most vulnerable. 36. Reforms already undertaken include an improved legislative framework for foster care, steps toward "de-institutionalization", and the establishment of a Social Innovation Fund (SIF) to encourage community-driven innovations in service delivery. The SIF is unique inthe region for its focus on supporting partnerships between local self-governments and NGOs, and its objective o f scaling-up local innovations in service delivery into national policy. With financial and technical support from the SIF, local authorities in Serbia are able to pilot alternative models o f care for vulnerable (the elderly, disabled and minors without parents), and socially disconnected groups (such as Roma, internally displaced persons, or IDPs, and refugees) in partnership with non- government providers. A number o f these innovations have been found to be successful and adopted at a national level, for example Serbia's program for care o f battered women and minors. 37. An important policy question is whether the poverty impact of transfers could be improved through consolidation of the MOP and child allowances. The division o f parallel delivery responsibilities between republican and local self government agencies imposes significant costs, including duplicative administrative structures, staff and office space. Parallel institutions for the means-tested benefits also impose costs on beneficiaries who need to present separate applications and sets o f documentationfor each benefit, and on providers who end up spendinga disproportionate amount o f time on paperwork. Consolidation o f targeted cash benefits could increase administrative quality (that is, minimize nontake up o f benefits, as well as administrative leakage), and increase administrative efficiency (that is, minimize per beneficiary and per-applicant administrative costs). 38. A recent feasibility study finds that important pre-requisites are needed before consolidation will yield significant benefits. The study answers two critical questions: (i)what will the benefits be o f consolidating MOP and child allowance administrative systems?; and (ii)how should administrative consolidation be implemented so as to maximize the benefits while minimizing the costs? The study concludes that administrative consolidation on its own will not result in a significant improvement in administrative efficiency (due to what the authors call "legislative incompatibility" o f the two benefits, which introduces fundamental differences in how the benefits are targeted), and only very marginal improvements inadministrative quality. ... Xlll 39. This suggests moving forward with administrative consolidation only after the legislative incompatibilities between the two benefits are removed. This would also allow time for the other prerequisites discussed above to be put into place. The study shows that the most feasible option for administrative consolidationwould be to move all aspects of MOP administration to LSGs, except for the responsibility for making home visits and the subjective income assessment (which would remain the responsibilitiesof CSW social workers) and recommendsthat the Ministry proceedwith pilotingof the proposedmodelto further determine feasibility inthe runup to legislative reformsexpectedin2009 (Hurrell, et al, 2007). xiv 1. MACROECONOMIC ISSUES IN DECENTRALIZATION A. INTRODUCTION 1.1 Serbia's ability to achieve macroeconomic stabilization and restructure its public finances has been, and will continue to be, impacted by the policy actions of its sub-national governments. For the purposes of this paper, these include the Province o f Vojvodina and municipal governments. In many countries, the fiscal performance o f sub-national governments has created problems for macroeconomic management through the running o f deficits, accumulation of arrears, or the incurring o f debts which must eventually be covered by the central government. This chapter examines the experience to date and the institutional mechanisms for macroeconomic management at the sub-national level in Serbia. The next section o f this chapter reviews the fiscal performance of Serbian sub-national governments, especially in relation to the Republican government's ongoing efforts to restructure the public sector. A key finding i s that the composition o f public expenditures evolved quite differently at the Republican and sub-national levels, especially during the successful fiscal consolidation o f 2003- 2005, when control o f spending on wages and salaries and on enterprise subsidies at the central level was not matched at the sub-national level. The increase in subsidies to municipal enterprises, combined with the growing arrears o f such firms, points to a need to focus on their restructuring and governance. The third section reviews the institutional framework for macroeconomic management in the sub-national governments. Since 2002, a number o f important laws have strengthened the framework for regulating the finances and borrowing of sub-national governments. The capacity o f the Ministry o f Finance to monitor and report on fiscal outcomes has also improved significantly. As a result, over the last several years, sub-national governments have broadly adhered to their spending and debt limits. However, further delays in the restructuring o f municipal enterprises, coupled with the likely emergence and/or growth o f new financing sources for sub-national governments (including borrowing and the proceeds from privatization and other asset sales), could create additional spendingpressure and lead to larger deficits at the sub-national level. The fourth and final section summarizes the main conclusions o f this chapter. B. MACROECONOMIC MANAGEMENT AND SUBNATIONALGOVERNMENTS 1.2 From the perspective of fiscal performance, the period since the start of transition in Serbia in 2001 can be broken into three main phases. The first period from 2001 to 2003 was one o f rapidly expanding cash expenditures, revenues and deficits. From 2003 to 2005, Serbia underwent a period of successful fiscal consolidation, with expenditures relative to GDP being contained while overall balances moved from deficit to significant surplus. Finally, from 2005 to 2006, Serbia saw an unfortunate rebound in overall spending (including current spending) and the re-emergence o f fiscal deficits. As shown in Table 1.1, during all three periods, overall spendingmoved broadly inparallel at the central and sub-national levels, rising inthe first and third phases, and declining in the second phase. However, the composition (and hence quality) o f this adjustment often differedacross the two levels o f government, most notably inthe second period o f successful fiscal consolidation, which was primarily driven by the Republican government. The rest o f this section examines fiscal performance in these three periods ingreater detail, comparing trends at the central government level with those at the sub- national level. Table 1.l:Total Expenditure and Net Lending, by level of government as YOof GDP 2001 2002 2003 2004 2005 2006 General Government 36.2 43.4 43.7 41.4 40.6 42.5 o/w CentralGovernment 16.7 22.0 27.5 25.4 22.8 25.1 o/w Sub-nationalGovernments 3.7 6.7 7.2 6.9 6.8 7.3 1.3 During the first three years of transition, 2001 to 2003, Serbia saw an increase in consolidated fiscal expenditures, revenues, and deficits. As shown in Table 2.2, consolidated general government expenditures measured on a cash basis increased from 36.2 percent o f GDP in 2001 to 43.7 percent in 2003. This increase was due to several factors, including: (a) the (desirable) more realistic alignment of cash outlays with expenditure commitments, following the 1990s, when the near-absence of normal financing sources led to unrealistic budgeting and the accumulation of arrears (meaning that part o f the observed increase in cash outlays did not reflect increased expenditure commitments); (b) the desirable conversion o f previously hiddenquasi-fiscal deficits in public utilities into explicit subsidies to such firms (again increasing cash outlays without raising aggregate demand); (c) the increase in subsidies to socially- owned companies awaiting privatization; and (d) significantly increased transfers to households, especially for pensions. In parallel, consolidated revenues also increased significantly from 35.3 percent o f GDP in 2001 to 40.7 percent in 2003. However, as cash expenditures were increasing faster than revenues, the budget deficit (excluding grants) also increased from 0.9 percent o f GDP in 2001 to a high o f 4.3 percent of GDP in2002, before declining slightlyto 3.0 percent of GDP in2003. 1.4 During this period, a sharp increase in the consolidated revenues of sub- nationalgovernments fueled a broadlysimilar increasein their expenditures. Total sub-national government revenues and grants rose by 3.4 percentage points o f GDP from 4.0 percent in GDP in 2001 to 7.4 percent of GDP in 2003. About two-thirds o f this increase was due to higher tax revenues (both shared and own taxes), and the remaining one-third to higher transfers. Part of this increase was on account of changes introduced by the new Law on local self-government which took effect in2002. While that law also transferred some additional responsibilities to local governments, it represented a net transfer of resources to the sub-national level, reflecting the political atmosphere o f the time (especially a desire to ease the very restricted financial situation of municipalities during the 1990s). This net transfer o f resources allowed local governments to finance additional expendituresbeyondthe newly devolved responsibilities. 2 Table 1.2: ConsolidatedSub-national Government Fiscal Operations, as %GDP 2001 2002 2003 2004 2005 2006 Total revenue and grants 4.0 6.9 7.4 7.3 7.1 7.2 Total revenue 4.0 6.9 7.4 7.3 7.1 7.2 1. Current revenue 4.0 6.9 7.0 6.9 6.6 6.7 Tax revenue 2.8 4.7 5.0 4.9 3.7 3.9 Shared taxes 2.2 3.5 4.2 3.3 2.5 2.7 1. Personalincometax 0.2 0.9 1.6 1.5 2.5 2.6 2. Corporate incometax 0.0 0.0 0.1 0.0 0.0 0.1 3. Turnover (retail sales) tax 1.o 1.5 1.5 1.8 0.0 0.0 4. Payroll tax 1.o 1.1 1.o 0.0 0.0 0.0 Own-taxes 0.7 1.1 0.8 1.5 1.2 1.2 1. Propertytax 0.3 0.4 0.4 0.8 0.9 0.9 2. Other taxes 0.4 0.8 0.4 0.7 0.3 0.4 Non-taxcurrent revenues 1.1 1.o 0.7 1.o 1.2 1.1 Transfers from the other levels o f government 0.1 1.3 1.3 1.o 1.7 1.7 2.Capital revenue 0.0 0.0 0.4 0.4 0.5 0.5 3. Grants 0.0 0.0 0.0 0.0 0.0 0.0 Total expenditure and net lending 3.7 6.7 7.2 6.9 6.8 7.3 Total expenditure 3.7 6.7 7.2 6.9 6.8 7.3 Current expenditure 3.6 6.3 7.1 6.7 5.9 5.9 Expenditureon goods and services 1.7 3.6 4.7 4.5 3.6 3.5 Wages and salaries(gross) 0.7 1.9 1.8 1.7 2.1 2.1 Other purchases or goods and services 0.9 1.7 2.8 2.8 1.6 1.4 Interestpayment 0.1 0.0 0.0 0.0 0.0 0.0 Subsidies and other current transfers 1.9 2.6 2.4 2.2 2.3 2.4 Subsidies 0.5 0.8 0.6 0.6 1.5 1.1 Transfers to households 1.4 1.8 1.8 1.7 0.8 1.2 Capitalexpenditure 0.1 0.5 0.1 0.1 0.9 1.4 Overall budget balance(cash) 0.3 0.2 0.1 0.5 0.3 -0.1 Overall budget balance excluding transfers from the centralgovernment 0.2 -1.1 -1.1 -0.5 -1.5 -1.8 1.5 During the second phase between 2003 and 2005, Serbia undertook a significantconsolidated fiscal adjustment, with both levels of government (especially the central level) making some contribution. Given its successful nature, we focus mucho f our discussionon this period. More specifically, while consolidatedrevenues as a share of GDP were broadly sustained in the range between 40.5 and 41.4 percent, consolidatedexpenditures were reduced by a significant 3.1 percentagepoints of GDP, from 43.7 percent of GDP in2003 to 40.6 percent of GDP in2005. The vast bulk of this decrease came in the Republican budget (excluding extra-budgetary funds, whose spending increased), where total spending fell by a massive 4.7 percentage points o f GDP. Sub-nationalgovernments contributed a further more modest 0.4 percent o f GDP. This significant reduction of spending relative to GDP over two years was primarily achieved through lower current expenditures. During the same period, consolidated 3 capital expenditures increased slightly from 2.4 to 2.6 percent of GDP. A s argued in the World Bank's Public Expenditure Review UpdateY4this increase in capital spendingwas necessary, since Serbia has one of the lowest levels of public investments among ECA countries. 1.6 However, the specific pattern of fiscal adjustment at the sub-national level during this periodwas quite different than at the central level, tendingto undermine efforts by the Republican government to move to a modern vision of the state. As shown in Figure 1.1,both levels o f government were able to reduce spending relative to GDP on two items - purchase of goods and services and transfers to households. Spending by sub-national governments on the former decreased significantly from 2.8 percent of GDP in 2003 to 1.6 percent of GDP in 2005. Meanwhile, their spending o n transfers to households declined from a peak o f 1.8 percent of GDP in2003 to 0.8 percent o f GDP in 2005, which is even less than what sub-national governments were spending before the significant increase in spending in 2002. These transfers include financial means transferred within the education sector, health care and social assistance. 1.7 Over 2003 to 2005, the other main categories of spending moved in opposite directions at the two levels of government. As shown inFigure 1.1,duringthis period, the Republican government was able to reduce all categories of primary spending relative to GDP. Consistent with the need to achieve a smaller, more efficient state and to gradually harden the budget constraint on the enterprise sector, this included containment of the wage bill and subsidies to enterprises. Incontrast, spending on these items by sub- national governments increased over the same period, somewhat undermining the Republican government's reform efforts. Sub-national government spending on wages and salaries increased from 1.8 percent in 2003 to 2.1 percent of GDP in 2005. Their spending on subsidies to firms increased particularly significantly from 0.6 percent in 2003 to 1.5 percent of GDP in 2005. The latter increase appears to have been driven by the still soft budget constraint on municipally-owned enterprises, which used such subsidies to cover operating losses and finance some n e w investments, and which enabled these firms to continue operating while delaying restructuring. There are indications that these local enterprises increased employment significantly over the last five years. For instance, waste disposal companies increased employment by about 11.5 percent between 2001 and 2005, As discussed inBox 1.1,the fact that municipalities and local utilities also continued to accumulate some arrears despite this increase in subsidies i s further evidence of limited restructuring to date in municipal enterprises. 1.8 On the other hand, the sub-national governments were successful in increasingcapitalspending,while such spending actually declinedat the Republican government level. During this period, sub-national government capital spending rose from only 0.1 percent of GDP in 2003 to 0.9 percent of GDP in 2005, while central government capital spending fell by 1.O percentage point of GDP. The overall increase in public capital spending at the sub-national level is likely to have been even higher, as some share of expenditures classified as transfers to enterprises went to financing investment projects of local utilities. World Bank (2005). The Republic of Serbia: A Policy Agenda for a Smaller and More Efficient Public Sector. 4 Figure 1.1: Change in the Level of Expenditures2003-2005, by type, as %GDP Goods and Transfers to Capital Expenditures Wages services Subsidies households expenditures 1.o 0.o -1 .o QCentral government -2.o rnLocal governments -3.O -4.o -5.O Source: Box 1.1: Arrears of Local Governmentsand Local Utilities Local governments also appear to be modestly contributing to fiscal indiscipline through the accumulation of arrears by municipalities and the utilities which they own. In most cases, these arrears are incurred by district heating and public transport companies. The accumulation o f such arrears despite the noted increase in subsidies to municipally-owned enterprises is a particular source o f concern, as it appears to indicate that these enterprises have undergone limited restructuring and continue to operate under soft budget constraints. One important element o f such arrears is delays in payments to Republican government-owned utilities - primarily to the electricity utility EPS and the oil company NISS5According to the most recent data (Tables 1.3 and 1.4), arrears o f local governments and their utilities toward EPS were USD 15 million and arrears to N I S were USD 21.5 million (combined about 0.1 percent of GDP). The growth in these arrears primarily relates to the new delays in payments toward NIS. As an illustrative example, Belgrade public transportation company increased its debt to NIS by 70 percent in 2005. Data presented here were obtained from these two large suppliers as there is not yet regular public reporting on the financial position of local utilities. This issue needsto be addressedas a matter ofurgency. Table 1.3: Arrears to EPS, in USD December 31,2005 June 30,2006 Total arrears to EPS 13,304,065.63 15,160,083.93 olw Vojvodina (Novi Sad) 1,020,245.90 345,637.84 olw Belgrade 965,462.58 976,279.85 olw Western Serbia - Kraljevo 3,287,013.43 2,221,655.44 olw South-eastern Serbia -Nis 5,074,607.69 6,313,992.52 oiw Central Serbia -Kragujevac 2,956,736.04 5,302,5 18.29 Source: EPS. 5N I S stands for Naftna Industrija Srbije (Serbian Oil Industry). 5 Table 1.4:Arrears to NIS in USD (Jugopetrol-national oil retailer and NAP oil retailer for Vojvodina primarily) - December31,2004 December31,2005 Total arrears to NIS 9,554,309 21,540,774 o/w to Jugopetrol 8,734,3 12 21,085,768 olw to NAP 819,997 455,006 Source: NIS. 1.9 The third period (2005-2006) brought an unfortunate reversal of fiscal adjustment, including at the sub-national level. Consolidated current expenditures increased by 0.5 percent o f GDP (mainly due to an increase in transfers, part o f which went for arrears clearance and thus did not contribute to an underlying increase in transfers), with all of the increase coming at the central level. In parallel, consolidated capital expenditures increased by a significant 1.2 percentage points o f GDP, with 0.5 percentage points o f this increase at the sub-national level. Overall sub-national government spending reached 7.3 percent o f GDP in 2006, a record level for the last 6 years. As a result, Serbia again movedto a fiscal deficit o f about 1.6 percent o f GDP. 1.10 Finally, lookingover the whole period 2001-2006, sub-national governments managed to generatesmall surpluses in all years except 2006, despite the sharp 3.6 percentagepoints of GDP increase in spending. As discussed above, this was driven by a generous increase in their revenue base, which relaxed their budget constraint. A crucial role was played by the significant rise in central government transfers to sub- national governments, from 0.1 percent of GDP in 2001 to 1.7 percent o f GDP in 2006. Total sub-national current revenues are expected to remain near their 2006 level with the introduction o f the Local Government Finance Law.6 While sub-national governments ran consolidated deficits relative to own revenues since 2002, once transfers from the Republican Government are included inrevenues, they were in surplus (on a cash basis) ranging from 0.1 to 0.5 percent o f GDP in every year from 2001 to 2005 (Table 2.2). However, in2006, consolidated lower sub-national governments ran a small deficit of 0.1 percent o f GDP. This deficit was driven by Vojvodina's deficit, since municipalities continued to runa small consolidated surplus. c. INSTITUTIONAL FRAMEWORKMACROECONOMIC FOR MANAGEMENT AND SUB- NATIONALGOVERNMENTS 1.11 The mechanisms for fiscal discipline in the different levels of government vary from country to country. These cross-country differences partly reflect differences in their economic and social environments and in their historical institutional heritages. Additionally, countries have different priorities when setting policy goals-for example, balanced budget, limiting overall spending, ensuring control over public debt and its servicing, etc. Among OECD countries, one can differentiate between four models for assuring coordination between fiscal policies across different levels of ~ The structure of revenues, including the changes introduced in the Local Government Finance Law that becameeffective inJanuary 2007, is discussed inthe next chapter. 6 government usinghigher or lower levels o f administrative contr01.~The first model uses direct central government administrative control over local government budgets and borrowing. The second model relies on centrally imposed rules for the size o f the budget, deficits and their financing. The third model is based on cooperation between representatives o f different levels o f government which i s institutionalized through some form o f coordination body or agreement. The fourth most liberal model primarily relies on market forces to ensure fiscal discipline and the achievement o f macroeconomic goals. Table 1.5 (Joumard and Kongsrud, 2003) shows which countries rely on which mechanism. Table 1.5: Mechanisms for EnsuringFiscal Discipline Administrative Centrally Imposed Formalized Relianceon Market Control Rules Cooperation France, Greece, Ireland, Brazil, Finland, Australia, Austria, Canada, Czech Japan, Korea, Hungary, Italy, New Belgium,Denmark, Republic,Mexico, Luxembourg,Turkey, Zealand,Norway, Germany, Iceland, Switzerland, United UnitedKingdom Poland, Portugal, Slovak Netherlands, Spain States Republic, Sweden Source: OECD. 1.12 Serbia uses centrally imposed rules on borrowing, with indirect influence on local government budgets. This indirect influence i s primarily exerted through the new Law on public debt adopted in 2005, which sets clear rules for borrowing for all government entities, including sub-national governments (discussed in the next chapter). The Debt Department within the Republican Treasury collects information on local government borrowing, although its prior approval i s not neededfor borrowing unless the creditor requests a sovereign guarantee. However, it can act afterwards if a given municipality i s found to have violated the law. While indirectly constrained by the limits imposed by the Law on public debt, and while sub-national budget execution has been much improved by the introduction o f the new Treasury system, local governments and their assemblies are otherwise fully responsible for formulating and adopting municipal budgets, without the needfor direct approval from the central government. 1.13 The process of preparing municipal budgets starts when local governments receive the Budget Memorandum prepared by the Republican government. This provides projections o f inflation, real growth, wage growth and other information needed for planning expenditures and shared revenues. Local governments then estimate non- discretionary expenditures (usually for activities assigned to them by law and for debt repayment). At the same time, they make projections o f own and shared revenues and o f the resulting financing gap, which would ideally (but not necessarily) be financed by transfers from the central government. The Republican MOF then collects the budget proposals o f local governments (including the implieddeficits, excluding transfers) while projecting available resources for transfers to local governments.8 Based on this information, and using strict criteria for the allocation o f transfers, M O F determines the 7 Joumard, Isabelle and Per Mathis Kongsrud - Fiscal Relations Across Government Levels, OECD Economic StudiesNo. 36,2003/1 8The amount for transfershas been fixed to 1.7percent of GDP beginning in2007. 7 transfer to each municipality. Once the republican parliament approves the law on annual transfer to lower levels o f governments, municipalities know the transfers they will receive during the next fiscal year, and can then approve their budgets in the local assembly. 1.14 While sub-national governmentshave made significant (albeit uneven) recent improvements in the presentation of expenditure data by both functional and economic classification, more needs to be done in the area of consolidated reporting. Progress has also been made in generating reports on various municipal extra budgetary funds (EBFs) used for on-lending activities and on the performance o f municipally- owned companies. However, as the classification o f expenditure does not yet follow standard practice across all sub-national governments, the process o f harmonizing budget classifications, accounting and reporting, and o f bringing sub-national extra-budgetary funds on budget, needs to be accelerated. Further progress is also required in achieving more systematic reporting on the functioning o f municipal enterprises. 1.15 Furthermore, the Republican government does not yet use consolidated general government accounts as the unit of analysis in the macroeconomic framework. At present, the general government balance table i s prepared only after the budgets have already been adopted. As a result, the decision on the fiscal position is taken on the basis o f the central government budget alone. Until now, this has not mattered too much, since sub-national governments have not relied on borrowing to any significant extent and thus have effectively had expenditures limitedby the transfer from the central government. As the resources available to sub-national governments (including below the line revenues from privatization, other asset sales and borrowing) are likely to rise, it will become important to take their fiscal position into account in macroeconomic management. This should begin with the use o f consolidated general government accounts in the budget memorandum, since the overall size o f expenditures and deficits inthe fiscal framework elaborated inthat memorandum should be a guiding tool for all levels o f government. 1.16 The capacity for planning, selecting and implementing public investmentsat the sub-national level also needs to be improved. The availability o f investment funds under the National InvestmentPlan, private-public partnerships and EUfunds all point to the need to improve such capacity. While higher investments in general are welcome, both increased borrowing and PPPs could cause significant risks. The Public Debt Law provides a good legal framework, however, the institutional capacity to manage public private partnerships needs to be improved even at the central level. 1.17 Taking a consolidated view of the fiscal position, and improving public investment capacity in municipalitieswill be important as they are increasingly able to access funds. Municipalities have already started to receive funds from the National Investment Plan. As municipalities start to borrow, and begin to privatize their enterprises or sell land and other assets, their ability to use these funds wisely will determine whether the funds are productively used. The macroeconomic impact will have to be managed at the Republicanlevel. 8 D. CONCLUSIONS 1.18 In conclusion, the main priorities in sub-national reform from a macroeconomic perspective are to improve the operations of municipality-owned companies, and to use General Government as the unit of account in determining the fiscal position. The large subsidies that continue to go to the municipality-owned companies are counter to the policies that Serbia i s undertaking at the central level. They are absorbing funds that can be used to fund other priorities or to reduce taxes. The persistence o f arrears in these companies creates quasi-fiscal deficits that will eventually have to be covered directly by sub-national governments as their founders or indirectly by the central government through increased transfer. At the central government level, the increasingly large share o f public expenditure that i s undertaken at the sub-national level points to the need to ensure that these expenditures are taken into account when determining the central government fiscal position and projecting the country's fiscal policy. 9 2. FISCALFRAMEWORKAND FINANCIALISSUES A. INTRODUCTION 2.1 Setting an appropriate structure for intergovernmental finance is a building block for effective delivery of public services at the local level. This chapter reviews the framework for inter-governmental finance. Following this introductory section, the next one reviews expenditure responsibilities and revenue assignments, followed by an evaluation o f the fiscal framework and a final section on financing issues in decentralizinghealth and education services. 2.2 The Republic of Serbia is a unitary state, as defined by the Constitution enacted by Parliament in September, 2006.' Nominally, it consists o f a core region under direct administration (termed central Serbia) and two autonomous regions-Kosovo and Vojvodina. Kosovo is currently under interim UN administration and is, in effect, detached from Serbia (although Serbia continues to finance some of the expenditures o f Serb-dominated areas o f Kosovo, including the payment o f teachers' salaries). The autonomy o f Vojvodina, incontrast, i s limited. 2.3 As specified in the Constitution, the basic unit o f local government is the municipality. Municipalities are geographical subdivisions o f the national territory. In rural areas, a municipality may include both towns and the surrounding countryside. Belgrade, as the capital and largest city, has a metropolitan government and 17 subordinate municipalities..lo Under the current (2002) Local Government Law, municipalities have elected assemblies (ranging from 19 to 75 members) whose members are elected for four year terms. The municipal executive (termed the president) is directly elected and serves a four year term. 2.4 Under the local government law, municipalities have the authority to create subordinate units o f Table 2.1: Population Size of Serbian Municipalities administrati Jurisdiction/Size Municipalities Yo of Population Yo of on (mesna Range Municipalities (000) Population zujednicu). Belgrade 17 10.3% 1,589.3 21.3% These are Novi Sad 1 0.6% 306.9 4.1 typically Nis 5 3.3% 252.1 3.4% Kragujevac 1 0.6% 175.2 2.3 village level others w/ 1OO,OOO+ 8 4.8% 1,275.5 17.1% organization 10,000-99,999 129 78.2 .Yo 4,122.5 55.2% s, created for < 10,000 4 2.4% 28.6 0.4% the purpose Total 165 100% 7,463.2.0 100% o f managing Source: Census o f Population, 2002. 9During the period under analysis, Serbia was one of two republics comprisingthe Federal Republic of Yugoslavia and later Serbia and Montenegro. Within the federal system, Serbia enjoyed considerable political, administrative, and fiscal autonomy. loFor administrative purposes, the Republic of Serbia (excludingKosovo and City of Belgrade) is divided into 24 districts(okrug) which function as deconcentratedextensionso fthe central government. a common service, such as water supply. While the mesna zajednica have the authority to collect tariffs for such services, they are not independent legal entities, and act in the name of the municipal government of which they are a part. 2.5 In population terms, Serbian municipalities are relatively large by European standards. There are 165 municipalities in the Republic of Serbia (excluding Kosovo). Seventeenof these comprise the city of Belgrade and operate as subunits of the Belgrade City Municipality. Five comprise the city of Nis. The municipalities of Kragujevac and Novi Sad also have special status as cities. The vast majority of Serbian municipalities have populations in the 10,000-100,000 range, as shown in Table 1. Only four municipalities have populations under 10,000, and two of these have populations in the 9,000-9,999 range. The average municipal population (including the municipalities comprising Belgrade and Nis) is roughly 45,000. As shown in Table 2.2, this is considerably larger than inmost continental Europeancountries. B. EXPENDITURE RESPONSIBILITIESAND REVENUEASSIGNMENTS 2.6 Although authorized to perform a wide Table 2.2: Average Population of array of functions, the primary responsibilities of Municipalities in Selected local governments are currently largely confined to European Countries urban infrastructure services." These include: (a) Country Average Pop. urban water supply and sewerage (b) district heating, Great Britain 126,128 66,300 (c) refuse collection and disposal, (d) street cleaning Lithuania Serbia* 46,570 and (in larger jurisdictions) (e) public transport. Sweden 30,040 Municipalities are also responsible for the construction Netherlands 27,559 and maintenance of streets and rural roads. These Denmark 18,760 services (with the exception of streets and roads) are Belgium 6,960 Poland performed by off-budget municipally owned 5 ,561 Finland 0,870 enterprises. While these enterprises have their own Norway 9,000 sources of revenue (i.e., tariffs), they rely on subsidies Italy 7,105 from the municipal budget to cover operating deficits. Estonia 5 ,713 Municipalities finance capital investments in these Germany 5,575 Latvia 4,400 sectors directly from budgetary resources. Spain 4,930 2.7 Municipalitiesalso play an active role in land Austria 3 ,421 development and housing construction. Although Hungary 3 ,242 Switzerland 2,468 urban land is legally defined as State property and Czech Rep. 1,659 cannot be sold outright, municipalities have the France 1,580 authority to designate vacant parcels o f State land for *Including municipalities comprising development. Municipal development agencies (known cities of Belgrade, Nis Sources: Honvath, T. (2000) Local as land and construction agencies) then extend water, Governments in Central and Eastern sewer, roads, and drainage to the site and auction off Europe "Decentralization: Experiments usufiwt rights (in the form of long term leases) to and Reforms". Republic of Serbia Statistical Office private developers. Municipalities are also directly Law on Local Self Government (February 2002) as amendedthrough December 2004. 12 involved in housing construction. Belgrade City, for example, i s building 2000 flats for public employees. Kragujevac is spending 1 million to construct new flats to house people displaced by an urbanredevelopment plan. 2.8 Spending on urban infrastructure over three years averaged 42% of municipal budgets over the last three years. Table 2.3: Functional Distribution of (The figures in Table 2.3 show the unweighted Municipal Expenditures (Average, averages o f actual expenditures in 2004 and 2005 2004-2006) and budget figures for 2006.) Almost 19 percent Function Percent was spent on land and housing development, and 9 o f Total percent was spent on subsidiesto municipal utilities Urban Infrastructure 42.4% and public transport companies and on capital Land development* 15.9% works undertaken on their behalf. The remainder Housing development 3.0% Water supply** 4.9% was spent on streets, roads, and public lighting. Sewage** 1.2% 2.9 Municipal responsibilities in the social Heating** 1.1% sectors are relatively limited. In education, local Sanitation** 1.8% Public lighting 1.5% governments are responsible for financing the Local roads, streets 9.6% operating costs o f primary and secondary schools Urban transport** 3.5% (Le., heating and lighting). They do not, however, pay the salaries of teachers or other staff. As a Social Services 18.0% result, municipal spending on education consumes Primary education 5.4% only about eight percent o f municipal budgets. Secondary education 3.O% Municipalities do have a management role in Health services 0.5% education: they appoint representatives to the Preschool, child welfare 7.1% Social welfare 2.0% school boards which, in conjunction with the representatives o f parents and teachers, appoint Other 39.7% school directors. Directors, in turn, have control Culture, tourism 8.2% over individual recruitment decisions. But neither sports 32% municipal governments or schools directors control Other 28.3% the number o f staffing positions authorized in Based on IMF classificationsystem. individual schools or other factors (such as salary *includes expenditures financed from sale of developmentsites scales) that affect the costs o f education in their **excludes expenditures financed from tariff jurisdictions. revenues of municipal enterprises but includes subsidies to municipal companies 2.10 Municipalities play a similarly small role and works undertakenon their behalf. in social assistance. While they assist the Government in determining eligibility for child allowances, they have no financial responsibility inthis area. Nor do they have any formal responsibility in the health care sector. Municipalities are, however, responsible for the provision o f preschools. While this service i s provided on a voluntary basis, it consumes a significant proportion-roughly seven percent on average-of municipal expenditure. 2.1 1 Taken together, spending on infrastructure services and social sectors consumes roughly sixty percent of municipal expenditures. Spending on culture, tourism, and sports consumes another eleven percent. Most o f the remainder i s classified as `general public services'-a category which, in IMF parlance, includes executive and 13 legislative overheads (including finance and personnel administration) and interest on public debt. Analysis o f individual municipality data suggests that spending on infrastructure increases with per capita revenue-richer municipalities have greater shares o f expenditure on infrastructure. 2.12 Serbia has a highly centralized revenue system, in which virtually all taxes are administered by the central government. Municipalities derive the majority of their revenues from shares o f these taxes. Prior to the introduction of the VAT (in January 2005) local governments derivedthe majority o f their revenues from fixed shares of the payroll and sales taxes. These revenues were retained in the jurisdiction in which they were collected. Local governments were also permittedto retain 100 percent of the property taxes collected intheir jurisdictions, along with income taxes on unincorporated businesses, real estate, farming, and other activities not subject to the payroll tax. 2.13 An additional, variable, share of the sales tax was allocated to each jurisdiction on an annual basis through the budget law. This variable share was originally intended as an equalization mechanism: additional shares were allocated to jurisdictions with unusually high costs or weak tax bases. This linkage subsequently deteriorated. This was partly because actual costs were no longer used to calculate expenditure needs. The dinar figures that were initially used to determine the relative costs o f particular services injurisdictions o f different types were later converted into a set o f coefficients (number o f inhabitants, number o f classrooms, etc.). The total amount to be allocated under the variable component o f the shared sales tax was pegged to a fixed share o f GDP.12Over time, the systemusedto determine the variable share became increasingly murky and subject to frequent and arbitrary changes. 2.14 As shown inTable 2.4, payroll taxes accounted for about 25 percent of municipal revenues in 2003. Property taxes (including both a Table 2.4: Sources of Municipal property transfer tax and a recurrent property tax) Revenues(budget, 2003) accounted for fifteen percent o f municipal revenues. Taxes 30% Other taxes accounted for another eight percent. fixed share 18% variable share 12% revenue (21 percent) from local fees and charges. Municipalities also derived a significant amount o f Taxes 25% localemployers)20% (on Fees for construction accounted for nearly half that central (on employees) 5yo total. (Note that these figures exclude the tariff Property Taxes 15% revenue o f municipal enterprises.) property transfer tax 8Yo C. THENEWINTERGOVERNMENTAL8Yo FISCAL Other Taxes recurrent property tax 7% FRAMEWORK Fees and Charges 21% construction fees 9% 2.15 Serbia has now adopted a stable and Total 100% transparent system of intergovernmental fiscal relations, As o f January 2005, the sales tax was abolished and replaced by a VAT. The Government also eliminated the local payroll tax. This required a fundamental l2Levitas, Anthony. 2005. `Reforming Serbia's Local Government Finance System' in Journal of Public Administration, 28: 147-178. 14 restructuring o f the system o f local government finance. The new system replaces the former one-based on varying shares o f the sales tax--with a system based on fixed shares o f the central government payroll tax, supplemented by transfers financed from general central government revenues. The basic structure o f the new system became effective in January 2005, when the Government increased the municipal share o f the central government payroll tax to 40 percent and established a new system o f transfers. At the outset, the bulk oftransfers-85 percent-were distributedon the basis o fphysical indicators-the number of primary and secondary school classes, the number o f children enrolled in childcare, etc. The remaining 15 percent was intended for poorer jurisdictions, and was distributedlargely on the basis o f municipal net material product. 2.16 The new Local Government Finance Law, which has been enacted by Parliament and took effect in January 2007, fine-tunes the 2005 reforms. Under the new law, municipalities derive their revenues from four principal sources: (1) local taxes (Le., taxes imposed at locally-determined rates); (2) shared taxes (taxes imposed at centrally determined rates and shared-in whole or in part-with the municipality in which they are collected); (3) non-categorical intergovernmental transfers; and (4) categorical intergovernmental transfers. 2.17 The principal source of revenue subject to local rate-setting will be the property tax. The property tax is now administered by the Government on the municipalities' behalf and i s imposed at centrally determined rates. Under the reform, local governments will obtain the power to administer the property taxi3 and will be permittedto set the rate of the tax, subject to a ceiling to be established in the property tax law. In addition, local governments will have the authority to set the rates on local administrative fees, a residence tax (pursuant to the law governing tourism), charges for the development of urban land, revenues from utility concessions and other miscellaneous sources. 2.18 The principal shared tax will be the forty percent share of the national payroll tax. This will be distributed among municipalities on the basis o f origin. In addition, the law continues to permit local governments to retain 100 percent of the revenues from certain forms o f income taxes, including the taxes on income from agriculture and forestry, private owner-operated businesses, and real estate. (Serbia has no unified income tax.) These taxes are administered by the central government and imposed at nationally uniform rates. 2.19 As tax reform moves forward in Serbia, the structure of revenue sharing can be revisited. Once a unified income tax is introduced, it can replace the payroll tax as the basis for revenuesharing. Over time, it might be possible to permit local governments to impose a variable surcharge on the personal income tax base-rather than receiving a fixed share of the national tax. This would provide them with more flexibility over their 13It is expectedthat transferringthis role to the municipalities will result in higher yields. As propertytax records are already organized on the basis of municipal boundaries and the national tax authorities are prepared to transfer the relevant staff from their offices in each municipality, there appear to be no administrative obstaclesto this reform. 15 revenues. However, the risks associated with permittinglocal governments such latitude suggest that this should be seen, at best, as a long-term reform. 2.20 The law establishes a complex system of intergovernmentaltransfers. These are divided into two major categories-categorical and non-categorical. Non-categorical transfers are intendedto provide general revenue support. Inother words, funds will not be earmarked for any particular function. The amount o f funding for the non-earmarked transfers i s fixed at 1.7 percent o f GDP. As specified in the law, the funds will be used, first, to reduce disparities in revenues among municipalities. Sufficient funds will be transferred to bring the per capita tax revenues o f all municipalities up to 90 percent o f the national average for all municipalities. The four officially-designated cities (Belgrade, Novi Sad, Kragujevac, and Nis) are excluded from this calculation. The revenues subject to equalization include all the taxes administered by the Republic government and imposed at nationally uniform rates-Le., the shared taxes. This will include the 40 percent share o f the payroll tax and other income taxes. It will not include shared fees and charges set by the national government. 2.21 Any remainingfunds will be used, first, to finance two other non-earmarked transfers: the compensation and transition transfers. The compensation transfer will be usedto compensate municipalities whose revenues from shared taxes under the post- 2004 system are less than the amount o f revenue generated by the former sales tax and the former local payroll tax (net of any increases in revenues from other sources). According to Article 39 o f the law, the amount o f the compensation transfer i s to remain permanently fixed at its initial level, adjusted for inflation. This element largely benefits the four cities. According to Government estimates, 93.5 percent o f the compensation grant would go to these four jurisdictions. Thirty five smaller municipalities would receive the remainder. 2.22 The transition transfer will be used to compensate local governmentsfor the subsequent changes in the transfer formulas between 2005 and 2007. The relevant article (Article 40) specifies that any municipality expected to experience a decline in revenues of over five percent as a result o f changes inthe transfer formula i s to receive a grant sufficient to cover the gap-that is, the difference betweenthe actual loss and a five percent decline. The amount o f the grant i s to decline systematically over three years. The amount inthe second year would be halfthat ofthe first; the amount inthe thirdyear would be one-quarter o f the original amount. 2.23 The law also includes a final, `fraternal' revenue sharing provision. Any local government whose shared revenues (per capita) exceeds 150 percent o f the national average will have 40 percent o f the excess deducted from its non-categorical transfers and added to the pool o f funds to be distributed to the remainingjurisdictions. This will be distributed on the basis o f a formula: 65 percent on the basis o f population, 19.3 percent on the basis o f land area, 6 percent according to the number o f children enrolled in child care, and 1.5 percent according to the number of child care facilities. The remainder will be distributed according to the number o f elementary and secondary school classes and the number o f elementary and school buildings (as a percent o f the 16 total number o f each item). Inthis way, the four officially designated cities are required to share part o ftheir revenues with poorer jurisdictions. 2.24 The law also provides for two systems of grants tied to specific functions. The first group-termed block transfers-would provide funding in general functional categories. Judging from the explanatory material attached to the draft version o f the law, these would be defined at a level o f generality such as `childcare' or `primary education'. The amount o f any such block grant will be `based on' the total cost o f the function in the year preceding its transfer to local government. A second group, termed `categorical transfers in a narrower sense' would permit line ministries to compensate local governments for `particular original and delegated functions.' The law requires any line ministry proposing such transfers to inform the Ministry o f Finance, during the budgetpreparationcycle, o fthe criteria and procedures for allocating such transfers. 2.25 The law establishes a Commissionfor IntergovernmentalAffairs. It consists o f eleven members: a president and five members appointed by the Serbian Government and five members o f the Standing Conference o f Towns and Municipalities (of whom at least three shall be local government representatives). Inter alia, the commission would review the criteria for the allocation o f non-categorical and categorical transfers, monitor the `vertical and horizontal balance' o f the system and the level o f debt incurred by local governments and prepare an annual report on the system o f intergovernmental fiscal relations with recommendations for its improvement. 2.26 The new transfer system reduces revenue disparities among municipalities. The principal achievement of the legislation is to eliminate the arbitrary and unpredictable distribution o f revenues that characterized the former system. This i s achieved by fixing the formula for determining the level and distribution o f non- categorical grants in law. While the various `grandfathering' provisions o f the municipal finance law (compensating jurisdictions that lose funds under the reforms) and the Article 43 provision add unnecessarily to the complexity o f the new system, it nevertheless constitutes a major step inthe direction o f transparency and predictability. 2.27 Among individual jurisdictions, the reforms achieve a significant reduction of disparities in per capita revenues. As shown in Table 2.5, the coefficient o f variation in per capita revenues prior to the reform was 0.55. M O F simulations suggest that Table 2.5: Impact of Reform on Per Capita it will be 0.42 once the reforms are in place. RevenueDisparities Total Total (The MOF simulations are based on estimated Revenues-Pre- Revenues- revenues o f 2006, although the reforms will not Reform (2004) Post Reform come fully into effect until 2007.) The chart Coefficient 0.553 0.42115 below illustrates the degree o f variation in per of Variation capita revenueunder the Poorest Dinars 3,105 Dinars 5,767 Richest Dinars 28,448 Dinars 32,3 19 post-reform scenario. Each bar represents the Ratio 9.16 5.60 per capita revenues o f a single municipality. With the exception o f the tenwealthiest jurisdictions, all the municipalities fall within a range o f 5000-13,000 Dinars per capita. The four cities would nevertheless remain 17 among the wealthiest jurisdictions. Novi Sad would rank first, followed by Belgrade. N i s would rank eighth; Kragujevac, nineteenth. As whole, municipalities in Vojvodina would have roughly twice the revenues, per capita, o f those in central Serbia (excluding Belgrade and Nis). Municipalities located in the districts o f Toplica and Jablanica-on the Kosovo border-would have the lowest average per capita revenues, while those in the district surrounding Novi Sadwould have the highest. I Figure 2.1: Variations in Municipal per Capita Revenues,Post-Reform V a r i a t i o n s in M u n i c i p a l P e r C a p i t a R e v e n u e s , P o s t - R e f o r m I' 1 5 0 0 0 1 0 0 0 0 2 s o o o Source: 2.28 This outcome is largely the result of Table 2.6: Components of Proposed the compensation and generalcomponents of Transfer System the transfer system. As shown in Table 2.6, Transfer component Dinars YOof million total equalization itself comes cheap. percent o f the total transfer amount i sOnly 15 Equalization 2,590.6 15 required Compensation 6,234.1 35 to bring the per capita revenues of all Transition 720.2 4 municipalities up to 90 percent of the national 5,855.1 33 average (excluding the cities). This impact i s transfer** plusArticle 43 2,392.9 13 offset by the compensation component, which Total general 8,248.0 46 increases the revenues o f the four cities. This, Total 17,792.9 100 in turn is offset by the general transfer, which * equal to 1.7 percentof GDP less allocations favors jurisdictions with below average to the three transfers above. revenues from the payroll tax. ** additionalfunds generatedfrom deduction o f revenues from richestjurisdictions 18 2.29 The aggregate fiscal impact of the reforms to date has been relatively minor. As shown inTable 2.7, municipal revenues in2005 were only two percent higher than in 2004, adjusting for inflation. They are expected to decline slightly, inreal terms, in2006. The reforms that are to take effect in 2007 are not expected to significantly change the situation. This i s intentional: the reform is meant to be revenue neutral. This result has been achieved by calibrating the percentage o f GDP that would be allocated to the non- earmarked transfers.l4 2.30 I t is not clear that the present level of transfers to municipalities is the right one. This i s an issue that i s inherent in a structure o f intergovernmental finance that depends governmentupontransfers-central Table 2.7: Trends in Municipal Fiscal Performance rather than locally (Dinar billion, constant prices of 2001) imposed taxes-to finance 2001 2002 2003 2004 2005 2006 municipal expenditures. i w e n ~ e S 31.6 60.3 68.0 73.9 75.4 74.0 (In the latter case, local C~rentexpenditure 28.4 54.6 65.6 68.0 63.3 62.0 taxpayers' willingness to as % of GDP 4.4% 7.7% 7.8% 8.0% 7.6% 6.9* pay would be the as%Republicrevenues 11.4 17.7 18.2 17.6 17.0 16.2* determining factor .) Different parties to the discussion are likely to have different views. Local governments will claim that urgent local needs justify an increase in funding. But the Ministry of Finance will have to weigh these arguments against rival claims on the national budget from national sectoral ministries and the macroeconomic constraints under which the Serbian public sector must operate. 2.31 There is no precise way to calculate the `right' level of funding for local governments. Some countries attempt to do so by defining physical norms for each o f the services provided local governments and calculating the Table 2.8: Subsidies to Public Utility Enterprises cost o f achieving them. As the Belgrade Metropolitan Government (2005 budget) Serbian experience with Subsidies to: SD mn YOof expenditures* intergovernmental transfers districtheating 1837.4 5.1% suggests, this approach is difficult SWM 1828.2 5 .O% to implementinpractice. Costs o f water supply 1325.7 3.7% public transport 4580.0 12.6% meetinga givennorm vary widely among jurisdictions. Many of the total 9571.3 26.4% *excluding additions to reserves and expenditures financed functions performed by local from land agency receipts governments are difficult to Source: quantify. l4At the outset, this percentagewas fixed at 1.175 percent of GDP-the amount that-taken together with the 40 percent share of the payroll tax-would compensate for the abolition of the sales tax and the elimination of the municipal payroll tax. Since then, the Government has reduced the rate of the central payroll tax and increased its exemption threshold. To compensate, the amount of the non-earmarked transfer has been increased to 1.7 percent of GDP. In effect, the Republic budget is absorbing the cost of the changes inthe payrolltax regime. 19 2.32 Another approach is to look for more generic mismatches between the revenues assigned to local governments and their functional responsibilities. On this basis, anecdotal evidence suggests that local government may be over-funded. Local governments in Serbia have few o f the expenditure responsibilities that normally require financing though taxes (or tax-derived intergovernmental transfers). It is the central government, not the local governments, that pays the salaries o f teachers, police, and health workers. Even social welfare payments-while administered in part by local governments-are paid directly from the central treasury. The majority o f municipal expenditure responsibilities, in contrast, might be expected to be financed from tariffs. These include urban water supply and sewerage, public transport, solid waste management, and land development. Aside from their contribution to revenues, there are strong efficiency arguments for charging consumers for these services. (There are, of course, adverse political consequences for doing so.) Data from Belgrade suggests that the municipality is usingtransfers as a substitute for tariffs. As shown inTable 2.8, over one-quarter o f metropolitan Belgrade's expenditures are spent on subsidies for public utilities (including public transport companies). If this pattern i s true o f most Serbian municipalities, it suggests that the level o f transfers to local governments i s too large. Reducing them could improve the efficiency o f resource allocation in the Serbian public sector-albeit on a relatively small scale. 2.33 A third approach is to compare Serbia with other countries. Figure 2.2 compares the scale o f local government spending in Serbia with that of the eight Eastern European countries Figure 2.2: Local Expendituresas % of GDP which recently joined the EU. L o c a l E x p e n d i t u r e s a s % G D P 2.34 At first blush, the scale o f local H U government in Serbia c z appears to be relatively PO small. In the s v comparator countries, ti 0 enw o n ment local expenditures range t i from 7.5 percent o f GDP h e a l t h r e c r e a tion (Slovakia) to 13.3 E S l a e d u c a t i o n percent o f GDP S K (Hungary). In Serbia, S B the figure is 6.3 percent. 0 0 % 5 0 % 1 0 0 % 15 0 % But these figures are not strictly comparable. As Source. shown in the breakdown o f expenditure composition, a large part of local spending in comparator counties i s devoted to education. In all the EU8 accession countries, teachers' salaries are now paid through local budgets. If the costs of education are excluded (including the non-salary operating costs)'' the scale o f local spending in Serbia-five percent-places it in the company o f Estonia, Slovakia, and Lithuania, 15To maintain consistency with the IMF definitions used for the comparator countries, local government spendingonpre-schoolis also excludedfrom the figures on Serbianlocal governmentspending. 20 somewhat higher than Latvia and Slovakia but still below the levels of Poland, Czech Republic or Hungary (Table 2.9). It should be emphasized that this does not suggest that the level of local funding in Serbia i s ideal. It merely shows that Serbian practice is consistent with the (questionable?) practices of neighboringcountries. Box 2.1: Vojvodina Vojvodina occupies a somewhat anomalous role inthe structure of local government-an intermediate tier of government serving only part of the national territory. Organized out of the part of Serbian territory that was formerly under the Austro-Hungarian empire, it has its own provincial assembly consisting of 120 directly elected deputies. It also has a substantial budget. Expenditures in 2005 totaled Dinar 20.17 billion, or US$146 per capita. (The average expenditure of Serbian municipalities was US$225.) This was largely financed from central government transfers (63 percent) and spent on personnel (99 percent). Legislation enacted in 2002 (Law on Establishing Particular Competencies of the Autonomous Province ) designates an extensive oversight role for Vojvodina, specifying competences in primary, secondary and tertiary education, health care (including health insurance), social welfare (including childcare, pensions, and veterans affairs), and urban planningand construction. Until2007, the majority ofthe province's expenditures were spent on teachers' salaries. In effect, the province was financing a service that was paid for directly by the central government in the rest of Serbia, an arrangement that could be considered reasonably equitable. Under the new (2006) Constitution, however, the volume of central government transfers to Vojvodina is to be substantially increased. The Constitution provides that Vojvodina will receive I seven percent of central government revenues. For 2007, this represents a near-doubling of the amount transferred to the province in 2006. No additional functional responsibilities have been allocated to the Table Box 2.1 ~ ~ Variationsin~ i ~ ~ per i ~~ ~ ~ i ~~ ~l l province. Such an increase in transfers is difficult to Capita Revenues justify, at least on equity grounds, particularly since the estimated, 2006, DINARS 28,653 municipalities of Vojvodina are already relatively wealthy. Belgrade As shown in Table Box 2.1, the average per capita Other central Serbia 8,524 revenues o f municipalities in Vojvodina (excluding Novi Novi Sad 32,3 19 Sad) were expected to be about 10 percent higher than in Other Vojvodina 9,460 central Serbia (excluding Belgrade), using the revenue simulations for 2006. 21 2.35 In the absence of definitive evidence, it cannot be argued that the existing level of funding for local government-at 1.7 Table 2.9: Size of Local percent of GDP and 40 percent of payroll Government in Post-Transition taxes-is inappropriate. But it should be Countries recognized that the issue should be revisited Year Local periodically. While stability is an important Expenditures as Yo GDP" attribute of a system o f intergovernmental Latvia 2004 3% finance, circumstances change. The economy Slovakia 2003 4% will grow, increasing the absolute size of GDP- Estonia 2001 5% based transfers as well as the yield o f payroll Slovenia 2004 5% taxes. The costs of local public services will rise Lithuania 2004 5% as a result o f inflation or Government-decreed Serbia 2005 5% Poland 2003 6% increases in public sector wages. It is Czech Rep. 2004 8% commendable that the local government finance Hungary 2002 9% law legislation establishes a mechanism for * net of education spending Source: IMF reviewing this issue: the Commission for Government Finance Statistics Yearbook 2005; Serbia MOF Intergovernmental Affairs. D. FINANCING ISSUES INDECENTRALIZING HEALTH EDUCATION AND 2.36 Reforms in health financing are off to a good start but implementation challenges lie ahead. The Government is now considering proposals to fundamentally change the financing system for health and education. The local government finance law calls for a new financing formula, which i s expected to be in place by 2009. Inthe mean time, sector specific designissues needto be addressed. 2.37 Reforms in primary health care are farthest advanced. In 2005, the M O H began implementinga rationalizationo f staff under the Human Resources Strategy.16 It i s now in the process o f preparing a Health Sector Rationalization Plan. This provides a draft investment plan for infrastructure and equipment for primary health care, identifying PHC centers that need to be closed or downsized to rationalize the health sector and reach the goals identified in the M O H action plan. Changes in the management and financing o f health care are to begin shortly thereafter. According to the Health Care Law (2005), ownership and responsibility for Primary Health Care (PHC) centers and pharmacies were to begin to be shifted from the Ministry of Health (MOH) to municipalities beginning in January 1, 2007. (The process has been delayed.) These responsibilities will include the financing of capital investment, equipment, and maintenance costs in PHC facilities. After the takeover, municipalities will appoint PHC directors, who will in turn have control over levels o f staffing and other major cost drivers. Initial participation will be voluntary. Municipalities have up to five years to assume these responsibilities. The process i s expected to be completed in2010. 2.38 Financingfor the recurrent costs of PHC clinics will continue to be provided by the HealthInsuranceFund. Inthe short term, the HIF will continue to pay the costs l6Republic of Serbia: HumanResource Strategy2006-2010. 22 o f salaries and pharmaceuticals on a line-itembasis. But once the new financing system i s inplace, these costs will beginto be financed through block (lump sum) grants. These grants will be distributed not on the basis of existing staff and facilities but on a capitation basis; that is, according to the number o f individuals (clients) registered with each PHC facility. (The payment amount would vary according to risk factors, such as age and sex.) This i s intendedto improve performance, by providing PHCs with greater flexibility over the use of resources. 1 Box 2.2: Do Health Care FinancingReformsConflict With Other Government Objectives? I The Ministry o f Health has expressed concern that proposed reforms in public funds management and government salary scales conflict with the objectives o f capitation based financing. The public funds management project adopted by the Government in May, 2006, mandates the consolidation o f the sub accounts o f individual health facilities into a single HIF account. It also provides for the centralized payment o f health facility staff, The law on the public sector wages (now before Parliament) will standardize wage rates throughout the government on the basis o f twelve pay grades (reflecting the responsibilities and complexity o f different positions) and fifteen `horizontal' pay levels (permitting salary increases within a given position, based on individual performance.) The MOH is concerned that these measures will deprive HCF directors o f the budget flexibility that capitation based financing is intended to provide, by increasing central control over the allocation o f funds and limiting PHC directors' ability to cut payroll expenses or reward good performance. These concerns appear to be unfounded. Reforms inthe management o f public funds are intended to improve the government's liquidity management (by consolidating idle balances in sub accounts). The Government does not intend to use this measure to increase its control over the allocation or disbursement o f funds. By the same token, the centralization of the payroll will merely facilitate the payment o f wages; it will not affect a PHC directors' control over staffing numbers or wage levels. The proposed public sector wage law will constrain the wages a PHC director may offer to individual staff. But the wide range o f compensation allowed by the horizontal pay scales-the difference between the lowest and highest pay level is 50 percent- suggests that PHC director will have ample latitude to reward good performance. 2.39 Capitation-based financing for primary health is widely employed in the post-transition countries of central Europe. Slovakia introduced a capitation based system for primary care in 1994. The Czech Republic introduced a capitation-based system in 1997.17Under its current system, primary physicians receive a fixed amount per registeredpatient, differentiated by age group. (0-4 year olds, for example, are worth 4.2 times as much as patients inthe 20-24 year old range). Hungary, similarly, reimburses primary health care on a capitation basis, using a points system that distinguishesamong patients infive different age groups. InLithuania, the HIF reimburseshealth facilities on the basis of the number o f patients registeredwith it. Each facility, inturn, is allowed to reimburseindividual physicians on a fee-for-service or capitation basis." 2.40 Due to their scale and complexity, these reforms will require careful sequencing in Serbia. The Government has taken the right step in starting with a 17 This followed a failed attempt to allocate fundingto individual physicianson a fee for service basis-an approachwhich fosteredan explosionof claims, not all ofthem valid. Salaries,however, remainthe predominant form of compensation. 23 centrally managed rationalization o f staffing and facilities. European experience suggests that health sector restructuring i s best undertaken by a level o f government that is relatively isolated from local pressures to retain excess staff and keep redundant facilities open. The Government's decision to introduce capitation financing only in municipalities where the rationalization process has been completed seems sensible. Othenvise, there could be dramatic mismatches between the volume o f funds provided under the capitation system and the costs of operating existing facilities, requiring massive layoffs or cuts in services. It should be noted that even after the health facility rationalization i s complete, capitation payments may still not cover operating costs. Temporary funding for severance payments or assistance with outplacement o f staff may be required duringatransitionperiod. 2.41 Reforms in the regulatory framework and in the management capacity of PHCswill also be required. Capitation shifts the financial risk for the provision of care from the HIF to the PHC directors. PHC directors are expected to respond by adjusting their input-mix to minimize costs and operate within the capitation amount. However, PHC directors may also respond by attempting to exclude high-cost clients from their rosters or shifting costs onto patients through informal charges. They may lower the quality of care they provide. Or they may demand subsidies from their municipal owners. 2.42 This suggests the need for an adequate regulatory framework and data management and analysis capabilities. For municipalities, these should include procedures to ensure transparency in the hiring o f PHC directors. PHC facilities will need procedures to manage financial operations and to collect routine data on quality, utilization, revenues and expenditures. Capitation payment requires the registration o f individuals with individual PHCs and a system for managing this information. The implementation o f these reforms should be supported with regular monitoring and evaluation o f provider and sector performance in order to correct and adjust reform measures. 2.43 Over the long term, the government will also have to establish a mechanism to adjust capitation payments to reflect changes in costs. A principal determinant o f health care costs-wage levels-is determined by the central government in Serbia. Centrally mandated wage hikes could cause severe fiscal dislocations in PHCs unless they are accompanied by corresponding increases in the level o f capitation payments. Legislation in some EU8 countries requires the government to increase capitation grants inline with annual salary adjustments. Serbia should consider a similar formulation. 2.44 The Government must make a strategic decision before proceeding with reforms in education financing. Plans to reform the organization and financing o f primary and secondary education are less advanced. The Government is, nevertheless, considering reforms that would parallel those in the health care sector. Primary and secondary education would become a municipal responsibility. Funds would be allocated to municipalities on the basis o f block grants. These would be allocated not on the basis o f staffing levels but on the basis o f enrollment. 24 2.45 The aim of this proposalis three fold. The first aim is to improve equity inthe allocation o f funding. Decisions over many years regarding the funding and location o f schools have resulted in significant disparities in education funding. Available data suggest that per-pupil spending in standard primary schools is 40 percent higher in Vojvodina than in central Serbia. Disparities among municipalities within these regions are equally wide. Under capitation financing, funding would be explicitly and transparently allocated to eachjurisdiction on an equivalent basis per student. While the level o f the capitation payment might vary according to grade level or location (urbadrural), similar students would be funded similarly. The second aim i s to increase local control over the allocation o f resources. The existing system i s based on line-item budgets imposed by higher levels of government. Capitation funding would be provided on a block grant basis, allowing school directors to allocate funds among different economic categories o f expenditure on the basis o f local priorities. The third aim i s to encourage network rationalization. At present, there i s some evidence o f an over supply o f schools and teachers. Between 1997 and 2003, the number o f pupils enrolled in primary schools declined by 24 percent. The number o f teachers (full-time-equivalent- FTE) declined by only half that amount (13.7 percent). At the secondary level, the number o f pupils declined by 15 percent, while the numbers of teachers (FTE) declined by only seven percent." Shifting to capitation-based financing would cut the level o f funds available to schools with declining enrollments. Inprinciple, this would force such schools to shed staff or shut down, freeing resources to be used in more crowded facilities. 2.46 Experience with capitation based financing in neighboring transitional economies provides useful lessons. Five o f the EU8 countries -Poland, Hungary, Lithuania, the Czech Republic and Slovakia-now employ some form o f capitation- based financing for primary education. Bulgaria has done the same. Russia i s testing it on a pilot basis. A key constraint on the implementation o f these reforms has been the unwillingness or inability of the responsible authorities to close under-enrolled schools. Under-enrollment i s particularly acute in rural areas, due to the decline o f rural economies and consequent out-migration. But rural schools are difficult to close. Parents resist bussing their children to distant towns. Community leaders oppose the closure o f what may be the only community facility inthe village. 2.47 This problem is compounded by legal, regulatory, or financial obstacles to reducing staff. In most o f these countries, the onus o f reducing staff lies with local governments whose authority o f local government to dismiss staff i s limited. In Lithuania, all teachers are civil servants, with legal protection against dismissal. Individual municipal councils do have the authority to fix the number o f positions ineach career stream and to determine the grade attached to each post. But once hired, staff are difficult to dismiss. As a result, municipalities must rely on hiring freezes and a variety There are many small schools, particularly in rural areas in Central Serbia, which combine multiple grades under a single teacher. As a result, although the pupil: teacher ratios o f these schools are not excessively high on a per pupil basis, quality is poor. 25 o f subterfugesto reduce the amounts budgetedfor staff on their payroll.20InBulgaria, the constraint on dismissal i s not the civil service law but the Ministry o f Education. Teachers are employed under the private labor law and can be dismissed without cause. Butthe MinistryofEducationmustapprove all dismissals, along with school closings. It has resisted doing so. Polish teachers are also employed under the private labor law and may be dismissed when a school i s totally or partially liquidated. Municipal governments, however, are requiredto provide six month's severance pay and re-employ the teacher at the first opportunity. 2.48 Governments have responded to these concerns by adopting capitation formulas that minimize funding cuts, particularlyin rural areas. In Lithuania, the capitation amount i s determined according to a complex formula that differentiates among both school characteristics and grade levels. For students in grades 1-4, rural elementary schools receive roughly twice as much funding per student as those in municipal capitals and big cities. Schools with fewer than 150 children receive 50 percent more than those in the 150-299 school-size bracket and nearly twice as much as those with 300 or more students. InPoland, rural schools receive a 33 percent supplementover the basic per capita amount. Towns with populations under 5,000 receive an 18 percent supplement." Hungary also supplements its standard per capita amount (equal to about US $440 per student in grades 1-8 and US $570 per student in grades 9-13) with additional funding (US $58) for primary education invillages with populations o f 3,000- 3,500 and those with fewer than 3,000 inhabitants (US $116).22 Bulgaria's capitation system scarcely changes the allocation o f funding at all.23 Such differentiation may be necessary to reflect the high cost o f transporting rural students to schools located inurban areas. Or to allow for lower pupil: teacher ratios in rural schools where bussing to an urban school i s not an option. But to the extent that it reflects only a desire to avoid school closings and staff reductions, it vitiates a principle objective o f the reform: rationalizingthe school network. 2oThey can, for example, force older staffto retire. By law, teachers over 62.5 years of age can now be put on one-year renewable contracts). Work can also be spread out. Teachers with a full 36-lesson teaching load can be forced to reducetheir work time to the 18-lesson minimum, allowing the municipality to retain staff, albeit at a lower level o f total compensation.It is also reported that municipalities "reorganize" in a way that leaves some staffwithout positions, allowingthem to be dismissed. 2' To encourage gminas to hire qualified staff, the formula continues to distinguish among teachers with different levels of qualifications,based on five centrallydeterminedpay categories. 22 The Hungarian education transfers are not intended to cover the full costs of education, nor are they earmarkedfor that purpose. 23Bulgaria's capitationformula began with a set of standards for teacher: pupil ratios. Separate standards were establishedfor five categoriesof municipalities,on the basis of average classroom size. For example, municipalitieswith small classes (less than 16 pupils per class) were assigned a teacher pupil ratio of -14. Municipalitieswith larger classroom sizes (over 22.2 pupils per class) were assigned only ,0967 teachers per pupil.Basedon these standards, a normative (or `desired') number ofteachers in each municipalitywas calculated. This was then multiplied by the average teacher's salary in each municipality in 2001, adjusted for subsequent wage increases, yielding an estimate o f the wage bill for teachers in eachjurisdiction. This total is now distributed among individual municipalities on the basis of enrollment, with adjustments for classroom size category. Because the new formula explicitly reflects differences in average classroomsizes and differences in average wages, it has little impact on the level o f salary funding in individual jurisdictions. In fact, the formula was calibrated so as to only affect municipalities with average classroom sizes below sixteen or above 22.2. 26 2.49 In Serbia, teachers do not enjoy civil service protection. Although their salaries are fixed in national legislation, they are employed under the private labor law. Intheory, they can be dismissed(and their positions eliminated) under the terms of this legislation. But in practice, this may be politically difficult or expensive. Before proceeding with capitation based financing, the Government must therefore make a strategic decision: whether to introduce capitation financing first, and leave it to the municipalities to rationalize their facilities or to rationalize the school network before decentralization takes place, and introduce capitation financing subsequently. Under the latter scenario, the Government would rationalize the distribution o f teaching facilities and staff, cutting or expanding facilities and teaching loads to achieve rough parity in terms o f per pupil costs. Capitation based financing would follow. School directors (or municipalities) would have the authority to determine staffing levels (and school closings) thereafter and would be expected to respond to changing enrollments and consequent changes in capitation payments accordingly. But the initial political cost o f rationalizationwould be borne by the Government. 2.50 There are counterarguments to this proposal. It has been suggested, for example, that municipalities are better positioned to rationalize their own school networks, as they are better informed o f the consequences, in terms o f costs and individual hardships, o f different configurations. Ifthe Government chooses to introduce capitation financing first, and leave it to the municipalities to rationalize their facilities, it i s important that municipalities have the legal and regulatory authority to close schools and reduce staff. Transitional funding may also have to be provided to allow time for individual municipalities to adjust teaching loads and reorganize facilities in response to levels o f financial support. 2.51 In sum, the immediate challenges for intergovernmental finance are to continue to review the fairness of the distribution of transfers, and to design the sequencing of health and education decentralization. Although the Law on Local Government Finance reduces the disparities in spending capacity across municipalities, significant variation remains particularly after Vojvodina's expenditures are included in the calculations. A second set of priorities are to design the sequencing o f health and education decentralization. While the issues are outlined here, they will be covered in more detail inthe chapters on health and education. 27 3. LOCALGOVERNMENTSTRUCTUREAND ADMINISTRATION A. INTRODUCTION 3.1 Decentralization in Serbia has thus far beenfunctional - individual functions such as in health or education are planned to be gradually moved down to municipalities. As this process moves forward, it becomes increasingly important that attention is paid to cross-cutting issues o f public administration at the local government level. This chapter raises four cross-cutting issues: (a) the political and administrative stability o f local governments; (b) the legal framework for employment o f local officials; (c) options for wage setting at the local level balancing the need to allow for flexibility in adapting to local conditions with the need for some consistency; and (d) the need for municipalities to prepare for their new functional responsibilities in areas such as health, education and social protection. 3.2 These issues will be discussed in the context o f the new Constitution o f Serbia, which should have a profound effect on operation o f local governments. In addition to restoration o f the property to local governments, the Constitution regulates in a quite detailed manner competences o f municipalities and creation o f municipal authorities (for example, indirect election o f the mayor by municipal assemblies in contrast to current direct election) which will necessitate changes in the legislation on local government.. Issues o f status o f local officials and administrative management, on the other hand, are left for legislators to arrange in accordance with general Constitutional principles (gender equality, prohibition o f discrimination and conflict o f interest, protection o f national minorities etc.). B. POLITICAL AND ADMINISTRATIVE STRUCTURE OF LOCAL GOVERNMENTS 3.3 The legal framework that governs the structure of municipalities can be improved. The municipal organizational structure is currently regulated by the Law on Local Self-Government adopted in 2002. While this law was an improvement over the 1991 law, implementation has led to some problems described below. The key actors include municipal assemblies as the representative body of a municipality, mayors as executive body, the municipal council as both the executive and advisory body, and the municipal administration as the administrative body. 3.4 The effective operation of the municipal assembly has been affected in several cases by political instability. The municipal assembly comprises councilors, whose number i s determinedby the Statute o f a municipality and who are elected for a four-year term by local citizens in direct elections. If the municipal assembly fails to meet in the legally prescribed period o f three months, the central government has the right to appoint an interim management o f local government and call for new elections. As the political parties in municipal assemblies are often not ready to compromise, formation o f majorities in a number o f municipal assemblies has been obstructed, affecting the appointment o f other municipal bodies and causing significant political instability. In 2006, new elections for municipal assembly were called in 12 municipalities: Zrenjanin, Uzice, Gornji Milanovac, Leskovac, Despotovac, Ub, Doljevac, Zabari, Barajevo, Kovin, Bujanovac, Medvedja and Presevo. 3.5 The key executives of municipalities are mayors. However, their authority is substantively limited in practice due to provisions o f the law that require approval o f the municipal assembly for a number o f decisions that are in mayor's competence, such as the appointment of a deputy, appointment o f members o f municipal council and appointment o f a head o f municipal administration. The position o f mayor i s especially underminedby provisions on the recall of mayors. A lack of adequate criteria for recall and relatively small number o f citizens signatures required for the recall have prompted recall procedures in a number o f municipalities, which has exacerbated problems o f stability and continuity, especially in smaller municipalities. In 2006, mayor recall procedures were conducted and new elections for a mayor carried out in 10 municipalities: Leskovac, Ub, Despotovac, Krupanj, Bor, Novi Becej, Bujanovac, Medvedja, Presevo and Kraljevo. 3.6 Current structuresin municipalityare complicatedby the unclear positionof the municipal council. Unlike the previous system where the municipal council (the so called-executive committee) had the clear status o f an executive body, the role o f municipal council in current system i s both executive and advisory. The key duty o f the council i s to propose decisions on the budget o f the municipality to the municipal assembly, supervise activities o f municipal administration, and decide in second-instance administrative proceedings on the rights and responsibilities o f the citizens and institutions. This `hybrid' function o f municipal council whose authorities are limited in practice has provoked questions about it's role and operation and prompted a discussion on the need for redefinition o f its status either inthe direction o f an executive or advisory body. 3.7 The procedure for appointment of the municipal council has proven to be difficult in practice. Members of the council are proposed by a mayor and elected by the municipal assembly through a majority o fvote. Inmunicipalities inwhich the mayor and president o f the assembly come from different parties it was almost impossible to appoint the municipal council. Inthis way, this body which was supposed to coordinate the functions o fthe major and municipal assembly has become a destabilizingforce. 3.8 The key administrative body in the municipality is the municipal administration. The municipal administration is in charge of preparation of draft regulations, implementation o f decisions o f the municipal assembly and president o f the municipality, administrative supervision and deciding in the first instance the procedure on rights and responsibilities o f citizens (issuing construction licences, etc). The municipal administration i s usually headed by a Head o f Administration. However, municipalities with more than 50,000 inhabitants are entitledto have several departments, headedby heads of departments. 30 3.9 A change in the selection and appointment process for heads of administration and heads of department would be beneficial. Appointments o f heads of administration have proven to be particularly sensitive. Heads o f departments are proposed by the mayor and appointed by the municipal assembly. In a number o f municipalities, problems with appointments o f heads o f administrations occurred due to the subsequent refusal o f municipal assemblies to appoint persons proposed by a mayor. As there were no legal provisions on how to resolve this situation, the functioning o f municipalities was adversely affected. The appointment procedure for heads o f administration and heads o f departments should be changed and replaced by a professional recruitment procedure, as will be discussed inthe next section. 3.10 The new Serbian Constitution has introduced indirect elections for mayors by the municipal assembly. The fact that the Constitution is referring to executive bodies impliesthe intentionthat both the mayor and the municipal council should have an executive function. Therefore, the municipal council i s expected to change its role from mainly advisory to its previous executive role. It will be necessary to introduce these changes inthe Law on Local Self-Government in order to comply with the provisions o f the Constitution. Finally, the new Constitution does not prescribe the election o f executive bodies o f towns and the City o f Belgrade and leaves this to be regulated by the Law. c. LEGAL FRAMEWORKFOR EMPLOYMENT OF LOCAL GOVERNMENTOFFICIALS 3.11 The status of local government staff is currently governed by the largely defunct Law on Labor Relations in State Organs. That Law was used to regulate the status o f both central government employees and local officials. However, the new Civil Service Law which came into effect in July 2006 now governs the status o f central government employees but does not apply to local government employees. Therefore, there i s a need for a new regulation o f the status o f local government staff, which could be done inthree ways: by the Labor Code; by the Public Service law; or by a special Law on the status o f local officials. These three options are discussed below. 3.12 Regulation of the status of local officials by the labor code has not worked well elsewhere for addressing the needs of local governments. Although in some countries the labor code governs the status o f local officials (see table 1) this solution has not proved to be satisfactory. The provisions o f the labor code are rather general and do not contain specific provisions for recruitment, promotion, transfer and termination o f employment that would be suitable for local governments. 3.13 Using a Law on Public Service for local government employees may not provide sufficient flexibility. Such a law would be designedto apply to employees in public services (such as health, education etc.). The status of public employees in health and education and other public services i s currently governed by the Labor Code. Although there have been some initiatives for the adoption o f a law on public services that would regulate the status o f employees in public services in greater detail, the Serbian Government has not shown strong interest in further regulation of their status, except for their salaries, which i s the subject o f a special law on pay in public services now under preparation. Furthermore, it i s questionable if the Law on public services 31 could satisfy the needs o f both public employees and local officials, due to substantive differences inthe position o f these two categories (such as the much stronger hierarchy in local governments, different classification and promotion system, distinct remuneration system etc.). This problem has occurred, for example, in Slovakia, where the Act on Public Service, applicable to both employees delivering public services and local and regional government staff, has not guaranteed sufficient standards o f professionalism for local governments. For these reasons, most European countries have distinct legal regimes for employees in civil service/local governments and public services (France, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovenia). 3.14 A special Law on local officials to govern the status of only local government officials might be the best option. This law should provide a unified framework for capacity development o f local staff taking into account large differences regarding institutional and fiscal capacity and economic development prospects. Inthis respect, the first issue to be raised is whether the whole staff at the local government level should have the status o f local officials, or whether a distinction should be made between positions holding `local government powers' and support services. It would be advisable that local government employees are divided between those who perform core administrative functions and others who carry out support services or who are employed in local government-owned companies. This would secure the stability of staff who perform key local government tasks, on the one hand, and enable flexibility o f employment for supporting staff, on the other hand. It i s further recommended that the scope o f local government staff be restrictive, following the practices o f most EU member states which reflect attempts for downsizing and restructuring of the public sector. Narrowing the scope o f the civil service would also free up resources that can be invested in decompression and improvements in salaries o f a smaller group o f civil servants. Table 3.1: Status of Local Government Staff in European Countries Status Of Local Government Staff Country Status of local officials is given to all local France, Slovenia, Bulgaria government staff Status of local officials is given only to local Belgium, Estonia, Finland, Germany officials holding `state powers' Status of supporting technical staff is governed by Hungary, Lithuania, Romania, Slovakia the Labor Code Status of all local government staff is governed by Latvia, Russian Federation, United Kingdom the Labor Code Source: 32 3.15 The status and recruitment of senior local government officials, especially heads of departments, should be depoliticized. As pointed out in the previous section, heads o f departments are subject to great political pressures, being appointed by municipal assemblies after the proposal o f a Mayor. It i s important that these positions are protected from political influence and be recruited by procedures o f open or internal competition, in line with the Civil Service Law. However, it should not be expected that these politically sensitive positions would be fully depoliticized and some degree o f flexibility for these positions should be allowed. Similar to Civil Service Law provisions, heads o f departments could keep their status o f appointed personnel, but their period o f appointment should be extended to 5 or 7 years in order to go beyond the period o f mayors/municipal assembly term in office. It would further be important to secure their positions by determining restrictive reasons for their dismissal. These measures should enable gradual depoliticization o f higher levels o f municipal administration. 3.16 The emphasis on law degrees may screen out good candidates for head of department positions. The Law on Local Self Government mandates legal qualification for obtaining a head o f department post. Article 49 states that the heads o f departments must have at least a universitydegree, must have passedthe state exam and must have a minimum of five years o f working experience. In this way, law degree graduates are clearly favored to candidates o f other professions. Inaddition, a state exam should not be used as a condition for selected posts in the local administration. Usage o f state exam requirements is questionable even for the central civil service and is certainly not desirable at the local level. The prescribed combination o f a law degree and a state exam seems to be particularly heavy. Therefore, these provisions should be removed from the Law on Local Self-Government and more flexible provisions on educational requirements prescribed by staffing plans (systematization Acts) o f local governments instead o f laws. 3.17 Job classification system criteria of local government officials should clearly be set out in the new legislation. The question to be posed i s whether local governments should continue to use the current classification system based primarily on educational qualifications and seniority or adopt a more advanced `position' system which much better reflects differences in the levels o f responsibility, complexity and outputs o f differentjob levels (as implementedat the central government level). Although a position based system has a number o f advantages over the seniority system, it i s questionable whether local governments would have sufficient capacity to introduce it in the short or medium term in Serbia. The introduction of such a system would require extensive training o f local government staff and necessitate the existence of a central body which would supervise introduction and implementation o f new job classification system throughout diverse local communities in Serbia. Bearing in mind these limitations, it i s recommended that the existing seniority model, based on qualifications and years in service be kept at local government level and introduced in the new Law on local officials. 33 3.18 Recruitment procedures should be based on open and internal competition. Local government posts should be accessible to any citizen who meets requirements set out by the competition. When establishing position requirements, local governments should not aim to put too high standards interms o f education and work experience, as it may be difficult to recruit such personnel, but should provide opportunities for recruitment o f younger personnel who would gradually build the necessary expertise. Special requirements, such as knowledge o f other local languages, should also be taken in due consideration in the recruitment process. If all formal requirements are met, prospective candidates should undergo specific examinationshests related to job tasks. 3.19 Promotion procedures based on merit should also be entrenched in the legislation. Promotion should be based on performance appraisal, which should be mandatory for all officials and conducted in a transparent manner. Performance appraisal procedures should be based on both top-down and bottom-up approaches, which means that the appraisal i s made by both managing and subordinated staff. In addition to opportunities for career advancement, performance appraisal should also provide a basis for receiving pay bonuses paid at the end o f the calendar year. Successive negative performance appraisals should, on the other hand, lead to termination o f employment. 3.20 Provisions on dismissal and termination of employment should be clearly specified. It would be important that these provisions would be equal for both career and appointed personnel. The Law should further contain provisions on conflict o f interest, anti-discrimination and gender equality and disciplinary measures. It would be important to emphasize the European principle that civil service employment should be the only employment o f an official and therefore any kind o f entrepreneurial activities should be prohibited. Special provisions on officials' conduct would also involve: the right to refuse unlawful orders; duty to inform supervisory bodies on unlawful orders received and liability for damage cause inthe course o f the discharge o f public duties. 3.21 A need for professionaldevelopmentof local staff should be emphasized. At the moment, there is no clear link between local government reform and capacity building. There is little motivation among staff for professional development. There i s a lack o f knowledge, skills and attitudes in local authorities to implement specific requirements o f the new legislation, relating to their new responsibilities and financial arrangements. The new Law should provide a framework to address these issues and emphasize the importance o ftraining and professionaldevelopment o f staff. 3.22 It is unlikely that a single law will be able to address all the needs of local governments. The need for a unified system for local government has to be balanced withthe needto respect differences at the local level betweenlarge, central municipalities and remote and small municipalities. Therefore, the proposed Law on local officials should not be too detailed, but should have the character o f a framework, which would flexibility for local government on selected issues. One o f the most important issues in this respect is certainly the wage system. 34 D. OPTIONS FOR WAGE SETTING INLOCAL GOVERNMENTS 3.23 The wage system of local governments is currently governed by the Law on pay in state organs and public services24and supporting secondary legislation.*' This law used to provide rules for the calculation o f salaries o f civil servants, office holders (MPs, judges), local government staff and public services employees (health, education, social protection). However, after the adoption and implementation o f the new Law on salaries o f civil servants (from January 1, 2007), this law became largely partly defunct and remained applicable to local government staff, office holders and public service employees. While the central government has already started working on both the new law on pay inpublic services and the new law on office holders, the pay system of local government officials has beenneglected. 3.24 In developing options for the local government pay law, it is useful to understand the present local government pay system. Salaries o f local government officials are calculated by formula o f multiplyinga salary base with a coefficient: Salary base * Coefficient = Basic Net Salary. A salary base i s determined by a Government decision. The current base needed for calculation o f local government salaries was determinedby a Governmental Decision of 28 September 2006. The government revises base figures each time it decides on an increase in salaries, normally annually. They include an inflation factor which results in small increases at intervals usually within the 12 month period. Coefficients are determined by the Law on pay in state organs or public services and its supporting secondary legislation. Coefficients point out the differences in educational qualifications, work experience as well as complexity o f work and responsibility and are not easily changed as they require changes o f primary and secondary legislation. 3.25 Coefficients o f elected and appointed personnel inlocal authorities are determined by setting coefficient caps, linked to coefficients o f either MPs (for elected officials, e.g. mayor) or (currently defunct) coefficients o f civil service appointed personnel (for local appointed personnel). Thus, for example, the Law stipulates that coefficients used for the calculation o f salaries o f elected personnel inthe city o f Belgrade can go up to coefficient for calculation o f a salary o f vice-president o f the National Assembly. Coefficients used for the calculation o f salaries o f appointed personnel inthe city o f Belgrade, on the other hand, can go up to coefficient for calculation o f salary o f a Head o f a special organisation etc (see Table 2). It i s interesting to note that although salaries o f civil servants are currently regulated by a new Law on Salaries o f Civil Servants, the coefficients for local officials are still linked to a largely defunct legal framework no longer applicable to the central government level. Salaries o f all staff are also subject to a length o f service supplementof 0.4 percent ofnet basic pay per year of service.26 24Law on Salaries in State Organs and Public Services (34/01). 25Decree on coefficients for calculation and payment o f salaries o f appointed and employed personnel in state organs, Official Gazette o f RS, No. 61/03, 69/03. 26This in practice amounts to amaximum of 16 percent for men and 14 percent for women. 35 Table 3.2: Structure of Salaries of Municipal Officials Elected Coefficient Salary base Basic net Allowances Net salary Decompression Civil Decompression Personnel (net) salary Basic net ratio servants ratio (without salary salaries allowances) + (net) - New allowances Civil service law In organs o f Up to coefficient territorial o fVice President 7,680.67 118,425 6.3 autonomy and o fNational 118,425 city of Belgrade Assembly = 10.50 Up to coefficient Inother cities o f president of 7,680.67 106,018 106,018 5.6 permanent. working body o f Nation. Assembly =9,40 In municipalities Up to coefficient with more than o f Members o f 7,680.67 96,995 96,995 5.2 60.000 N A inhabitants = 8.60 In other u p to 90% o f municipalities coefficient o f 7,680.67 59,448 59,448 4.6 Members o f N A = 7.14 Appointed Managem. Personnel allowance in bodies o f Up to coefficient territorial of head of Special 1,652.94 39,671 51,572 4 110,352 8 autonomy and Organization up to 30% city o f Belgrade = 24,OO Inother cities Up to coefficient o f Secretaryo f 1,652.94 35,538 46,200 3.5 98,091 7.1 Ministry up to 30% =21,50 In municipalities Up to coefficient with more than ofhead ofdistrict 1,652.94 27,274 up to 30% 35,456 2.7 87,192 6.3 60.000 = 1650 inhabitants In other u p to 90% o f municipalities coefficient o f 1,652.94 24,546 31,910 2.46 77,504 5.6 head o f district Up to 30% = 14,85 Employed Performan. Salary net Personnel bonus base for up to for up to 50 50% o f % o f employees employees Independent Professional 11.30 1,802.56 20,369 2,163 24,443 1.9 43,664 4.2 Associate Senior 3.4 Professional 10.10 1,802.56 18,206 2,163 21,847 1.68 34,93 1 Associate Professional 9.80 1,802.56 17,665 2,163 21,198 1.63 27,945 2.7 Associate Senior 9.30 1,802.56 16,764 2,163 20,117 1.55 22,356 2.2 Associate Associate 8.40 1,802.56 15,141 2,163 18,170 1.4 18,630 1.8 Highly skilled 8.30 1,802.56 14,961 2,163 17,954 1.38 15,525 1.5 worker Clerk, typist 8.20 1,802.56 14,781 2,163 17,373 1.36 13,800 1.3 Skilled worker 7.50 1,802.56 13,519 2,163 16,223 1.25 12,144 1.2 Unskilled 6.0 1,802.56 10,815 2,163 12,978 1 10,350 1 worker ' All amounts are in Serbian dinars The figures in this table exclude length o f service supplement. 36 3.26 The system leads to a complex salary structure at the local level. As Table 3.2 indicates, each group has a distinct salary base calculated with different coefficients, which are linked to coefficients o f other Government positions. Furthermore, there i s a special performance bonus base which i s added to a base pay and paid only up to 50 percent o f employed personnel. However, in the absence o f a consistent basis o f performance assessment, the payment o f such a bonus can lead to high degree o f subjectivity and unfairness. The weighted average figure perhaps represents the best indication o f the average salary at each rank, though some salaries could be much higher. This could especially be the case with employees carrying out budget and inspection, whose coefficients, in accordance with the Decree on coefficients, could be increased from 10-50 percent. This all makes the current system rather complex and un- transparent. 3.27 The present pay systemalso has the following features: 0 A difference between the lowest and highest salary (compression ratio) is relatively low, especially when compared with civil service. Although the overall compression ratio for elected personnel i s 6.3, the ratio for appointed personnel i s only 4 (compared to civil service S), while the ratio up to senior adviser level i s only 1.9 (incomparisonto civil service 4.2). 0 Pay levels are not competitive with the civil service levels (as shown in table 2) and in the Serbian market (see Table 3.3), especially for higher and mid-level positions. Table 3.3: Private Sector Pay Comparison Local Private Sector, Local Gvt. as % of Governments Lower Quartile Private Sector Lower Quartile Appointed personnelin bodiesof territorial autonomy and city of Belgrade 51,572 d a Appointed personnelin other cities 46,200 81,266 57% Appointed personnelin municipalities with morethan 60.000 inhabitants 35,456 62,978 56% Appointed personnelinother municipalities 31,910 48,805 65% IndependentProfessionalAssociate 24,443 39,640 62% Senior ProfessionalAssociate 21,847 3 1,899 68% ProfessionalAssociate 21,198 29,5 10 72% Senior Associate 20,116 29,737 68% Associate 18,170 25,538 71% Senior Clerk 17,953 25,424 71% Clerk 17,737 18,501 96% Note: Source for private sector pay: EAR pay reform team, February2006. 3.28 Under current budget constraints there is limited scope to increase salaries without corresponding savings in staff costs (through staff reductions) or other efficiencies. Central government pressure for reduction o f the wage bill levels has already triggered the process o f staff cuts in municipal administrations. Following the reduction o f the number o f civil servants in July 2005, the number o f staff in municipal 37 administrations was reduced by around 10 percent inthe course o f 2006. However, there appears to be a general recognition that local administrations are overstaffed. Inaddition, the qualification and age structure o f municipal administrations does not correspond to the needs o f modern local administration. Therefore, most municipalities envisage further staff cuts o f older personnel with primary and secondary education inorder to be able to free up resources for hiringyounger personnel with universityeducation. 3.29 Once the fiscal space is created, gradual development of an effective salary system could begin by raising the compression ratio of existing salary levels. This could be done by increasing salaries o f senior and mid-level managers by a higher percentage. This should help to retain mid and senior professional and managers in the local government. Salaries at lower levels are reasonably close to market levels and are thus less likely to need significant increases. One disadvantage of the decompression model is, however, that it would give the highest increases to managers who are currently political appointees. It would therefore be desirable to link the decompression process to earlier proposals to depoliticize the appointed positions inthe local governments. 3.30 Recent developments in some Western European countries indicate a tendency towards a more unified payment policy at the local level. Large local variations inlabor market conditions and pay levels create difficulties for recruitment and retention. This further generates inequality whereby staff in some regions receive a higher real wage than those doing the same work in other regions. Regulations that pay higher wages to staff which work inlarger administrations than those working in smaller ones are deemedunacceptable. 3.31 It is, however, important to allow for certain limited differences in pay throughout the country in linewith municipal fiscal capacity. Costs o f living are not the same throughout the country and some changes in payments should be allowed, if these positions are clearly defined as beingunder different arrangements. Any individual arrangement, however, implies that there should be control over the individual salaries and the overall wage bill. 3.32 One option is to grant local governments the right to determine basic coefficients after consultation with representatives of the employees or the local trade union. The central government may only interfere if the percentage set exceeds the financial possibilities. An important indicator in this respect would be the total wage bill of the local government administration as a share o f total local government expenditurecompared to other municipalities o f a similar size and/or budget.This system would allow raising or decreasing the general level o f payment, but would not allow for increase o f the payment level o f individual posts, in line with the job classification system. 3.33 An issue to be resolved is the payment status of local officials who perform tasks delegated from the central level. It may be argued that due to the nature of their tasks, they should be paid in accordance with central Government level legislation. However, arguments that staff working on `state functions' should be under a different 38 pay scheme than other local government staff are not generally supported. Two different pay schemes could harm the development o f the pay system and create further problems for transfers and promotions and the introduction o f any form o f performance relatedpay. E. PREPARINGMUNICIPALITIESFOR NEW RESPONSIBILITIESIN DELIVERY OF SERVICES 3.34 Municipalities need to strengthen their technical capacity to carry out their new roles. With the expected decentralisation o f provision o f services to the local level, local governments would need to buildhtrengthen their human resources capacity in several areas: social assistance, health, education and tax administration. While the institutions themselves, schools and health centers, will be responsible for delivery the services, municipalities will have oversight responsibilities. Inthe field of social assistance, local Governments are currently responsible for administering the child allowances, while the centres for social work are responsible for delivering the main targeted social assistance program, MOP (materijalno obezbedjenje porodice). One o f the decentralisation policy options under consideration is transferring responsibility for MOP to local government units (or that the MOP and child allowance are integrated into a single unique transfer), inwhich case there will be a need for strengthening o f capacity o f local governments and possibly transferring some o f the staff from the centres o f social work to local Governments units. The outset of the decentralisation process inthe health sector which has formally started in January 2007 requires municipal administrations to build capacities in discharging duties related to the primary health care development. Most municipalities currently do not have any staff responsible for the health sector. The planned transfer of founding and managing rights and financial responsibilities for primary health care institutions to local level will require the training o f existing staff o f hiring o f additional municipal administration staff which would be in charge o f designing primary health care programmes at the local level, carrying out financial management o f the primary health care institutions and effective monitoring o f their work. 0 In the education sector, the Government plans to involve local governments to participate inthe designof education curricula and inthe rationalization o f school networks. As the local governments have up to now performed largely technical functions o f financing the infrastructure costs o f primary and secondary schools, they would need to build their capacities in strategic policy planning in the respective areas. 0 Government plans for reallocation o f revenue collection o f minor taxes (such as property tax) to the local level will also affect the HR needs in local government units. As of January 1, 2007, local governments are authorized to take responsibility for collection o f the property tax and other local taxes. This shift will require the transfer of some Tax Administration staff to local government units and the building o f capacity for tax collection of existing local government 39 staff in order to allow for the smooth transfer of revenue collection functions to the local level. However, the transfer o f revenue collection authorities has already faced numerous problems and delays (Box 3.1) which has proven a need for careful and timely preparation of decentralization processes. Box 3.1: Problems Encounteredwith FiscalDecentralization The Law on local government financing, adopted in July 2006, provided for a transfer of the collection function for local taxes from the central to local level. Article 60 o f the Law on Local Government financing envisages that local governments determine, collect and control local government own revenues (property tax, local administrative charges, local communal charges, residence charges, remuneration for land use and other minor revenues) starting as o f January 1,2007. The transfer of tax collection necessitatesthe creation o f local government tax administration units, which requires the transfer of some staff from the central Tax Administration to the local level and capacity building of local government staff. However, a transition period o f six months for building o f capacities of local governments is rather short to allow for an effective transfer o f the revenue collection function. In the second half of 2006, the Tax Administration took several measures to facilitate the transfer of revenue collection function to local level: training of local government officials, creation o f unique information system/installment o f software to support the collection o f local taxes and assistance with preparation o f new systematizations o f local governments. Training on legal and practical issues o f local revenue collection was conducted in autumn 2006 for 285 employees from 80 different municipalities with the financial support o f GTZ. The great majority o f local government trainees have assessed the training as very useful and successful. The creation of a unique and comprehensive information system to support the collection o f local taxes necessitates substantive financial assistance, which has been securedthrough the National Investment Fund in the amount of 8,400,000 EUR.The installation of software in local governments has, however, suffered significant delays. The NIP Council approved funds for local software development in October 2006 and the Ministry o f Science and Environmental protection approved software development plan only in December 2006. Tax administration personnel have prepared a systematization model for all local government tax units, depending on their size. The number o f staff in these Systematization Acts is significantly higher than the number already working in local tax administration offices. It is, however, questionable if local governments will be able to properly staff new local government tax units, due to salary differences and limited wage bills. The timely devolution of revenue collection functions was hampered by the incoherence o f the legal framework governing the process o f revenue collection. Namely, the Law on Tax Procedure and Administration which regulates revenue collection at the central level is not applicable to the local level. Amendments o f the Law on Tax Procedure and Administration have been prepared, but still not adopted, due to political developments in the country. Furthermore, the Government has still not adopted a Decree on the transfer o f revenue collection functions, necessary for making the actual transfer o f employees from Tax Administration to local governments. Dueto overall delays inthe devolution oftax collection functions, local government units are obviously not yet prepared to overtake this important function. Therefore, most of them have entrusted it to the central Tax Administration, until they build sufficient capacity to take it over. Thus, 130 local governments have made an agreement with the Tax Administration that revenue collection o f local taxes would continue to be carried out by the central level until January 1, 2008. Only the city o f Belgrade has started to more intensivelyprepare for the take over of revenue collection functions, as its offices are better equipped, and it may take over the staff and functions o frevenuecollection inthe course o f 2007. 40 F. CONCLUSIONS 3.35 Serbia has made limited progress on reforming its system o f local government staffing and management, which i s posing an increasing risk to the decentralization process. While significant improvements in reforming the civil service system at the central level have occurred over the last couple o f years, the local government level has been marginalized. This may have an adverse affect on transfer o f sectoral authorities from the central to the local level, especially in view o f the uneven capacity o f local government units. 3.36 The key concerns of local government staffing and management in Serbia are related to the lack o f an adequate legal framework for human resource management, development and incentive systems. The experience o f other countries in the region shows that the success o f the decentralization process ultimately depends on capacities o f local government to absorb new functions transferred from the central level. Creation o f a proper normative framework i s a precondition for strengtheningo f institutional capacity o f local governments staff and enhancing their professionalization. 3.37 In view of priorities in local government staffing and management set out throughout this chapter, the following policy measures should be considered in the short term: 0 The Law on Local Government should be amended to conform to the new Constitution and address the weaknesses o f the current system related to the position o f the municipal council and heads o f administration 0 The Law on Local Government Officials should be prepared and adopted urgently to provide a solid normative framework for the operation o f local government staff, fostering merit-based recruitment and advancement, reducing politicization and enhancing local capacity development. 0 Local governments should identify potentials for staff cuts that would enable them to save resources for salary increases and attract qualified personnel. Salary improvements should be postponed until a substantial fraction o f employment reductions has beenattained. 0 The local government pay strategy that will lead to decompression o f the salary structure should be developed especially for positions that are most difficult to recruit and retain. Several pilot projects should be conducted in selected municipalities allowing them to increase the base pay for their staff within a firm fiscally sustainable resource envelope 0 It is important to build sectoral capacity of local government staff for the ongoing transfer o f responsibilities in health and tax administration and prepare local government units for the envisaged transfer o f authorities in the education and social protectionsectors. 41 4. DECENTRALIZATIONINHEALTH A. INTRODUCTION 4.1 The health sector has started to decentralize as part of a comprehensive sector reform strategy. Starting in January 2007 over a five-year time-span, selected decision-making powers inhealth are beingtransferred from the central to provincial and municipality governments. This i s combined with a restructuring plan that rationalizes facilities and staff to reflect current population distribution and health needs; as well as provider payment reforms to improve incentives for better management o f resources in the sector. 4.2 The expectation is that decentralization in the provision of health care will improve efficiency and equity by bringing policy makers closer to the people and by enhancing participation at the grass-root level. However, very little i s known about the practical impact of decentralization on improving the delivery o f health care services and health outcomes (Saltman, et al. 2007). However, spreading responsibilities creates a risk for duplication and inefficient administration, which threatens the performance o f care delivery, quality, and financing. Given the risks, one way forward i s to move sequentially and based on empirical investigation o f initial steps and pilots. 4.3 The chapter is organized as follows. After this introduction, Section B presents experience with decentralization from elsewhere and identifies lessons for Serbia. Section C provides an overview on the decentralization process, organization o f health care in Serbia, health financing and the availability and use o f data. Section D portrays current reforms that are intertwined with decentralization, including the restructuring o f the health sector and provider payment reforms. Section 5 concludes and suggests a design for piloting per capita funding. B. LESSONS HEALTH FROM SECTOR DECENTRALIZATION FROM OTHER COUNTRIES 4.4 It is important to combine decentralization with health financing and provider payment reforms, to prevent negative effects on sector performance. Government expenditure on health in China, India, Indonesia and the Philippines declined or stagnated after decentralization began. The fall in China and India can be attributed to three factors. First, fiscal decentralization shifts the burden to local governments without properly funding their new responsibilities. A complicated transfer system to equalize revenue and expenditureacross provinces is barely functioning and is increasing the health funding gap between poorer and richer provinces. Second, local governments have no incentive to invest inhealth as they do in infrastructure and private sector development. Third, the impact o f different types o f health services varied. More autonomy for hospitals led to buying more expensive equipment and drugs to generate local revenue, leading to greatly increased medical costs and an under-supply of preventive services such as immunization. This also happened in the Philippines and Indonesia where vaccination coverage dropped significantly after decentralization. Giving local governments and PHC facilities financial incentives through the provider payment mechanism to invest in health will improve outputs and outcomes. Setting up a transfer system to redistribute funding i s important to boost poorer regions' fiscal capacity. To make this function, responsibilities at the various levels o f government and health institutions must be clearly definedand enforced (Uchimura et al. 2006). 4.5 Poor planning of decentralization alongwith economic difficulties resulted in negative outcomes and a trend towards re-centralization in some countries. Inthe Russian Federation and Ukraine, multilayered health care systems now have fragmented responsibilities. Decentralization o f financing in Ukraine led to increasing inequalities between wealthy and poor areas (Shakarishvili, 2005). Latvia also experienced significant problems when funding for health care services through local government budgetswhich resultedinwidely differing amounts of health expenditure per inhabitant betweenregions (Saltman, et al. 2007). Consequently, several European countries and Canada started recentralizing health planning and financing to a central level, drivenby a need to control growing health expendituresand production cost through central r e g ~ l a t i o n ~ ~ . 4.6 In Europe, decentralizing capital funding to local budgets has sometimes led to hospital oversupply. Countries with a dualistic payment system, among them Serbia, use government budget funds to finance capital cost and other functions that go beyond the interest o f a purchaser organization, such as teaching students and training staff in teaching hospitals (Germany, Austria, and Switzerland), while operating expenses are paid by the insurer.Fragmented responsibility for funding capital and recurrent costs led to an oversupply o f health facilities, as capital costs are essentially free to hospitals. To make hospitals to some extent responsible for capital cost, the UK introduced capital charges into contracting arrangements, by estimating the value o f capital, based on which hospitals pay a rate o freturn to owners (Docteur et al. 2003). 4.7 Decentralization may be less effective where local governments and health facilities face capacity constraints, and where responsibilities are unclear. Experience from other ECA countries shows that unclear rules and responsibilities combined with capacity constraints at local levels create space for unaccountability and lack o f planning28.Clear definitions o f roles and responsibilities are needed. Mechanisms o f accountability should define at one level o f government the legal status, ownership, management responsibilities, planning, accreditation, and control functions. Some municipalities and facilities will need capacity building to implement devolution. Training can focus on management, planning, fund allocation, appointment and accountability procedures for PHC directors, and performance monitoring and evaluation. ''Including Scandinavia, Iceland, Canada, France, the UK and Latvia. See: Latvia Health Care Systems in '' Transition, 2003. Iceland, Health Care Systems in Transition, 2003. Rico and Leon (2005): Health care devolution in Europe. Trends and Prospects. University o f Oslo. Bach (2000): Decentralization and rivatization in municipal services: The case o f health services. ILO, Geneva. Decentralization in Healthcare. Analysis and Experiences in Central and Eastern Europe in the 1990s. (2005). Open Society Institute. Budapest. 44 A scorecard should identify municipalities that have met capacity and performance benchmarks, based on which they classify for decentralization. 4.8 Few studies have analyzed the impact on health outcomes of decentralization or the factors enabling or constraining its outcomes. An empirical study in India estimates the impact o f fiscal decentralization on rural infant mortality rates between 1990 and 1997. Findings from the random effect regression results show that fiscal decentralization plays a statistically significant role in reducing the rural infant mortality rate in India. Those results are robust to the way the decentralization variable i s measured and to different model specifications. The results also show that the effectiveness o f fiscal decentralization can be affected by other complementary factors such as the level o f political decentralization (Asfaw, et al. 2004). Based on time-series data sets o f 140 low- and middle-income countries from 1980 to 1997, another study finds that decentralization has different effects in low- and middle-income countries. In the low-income group, decentralized countries have higher coverage rates than centralized ones, with an average difference o f 8.5 percent for the measles and DTP3 vaccines. An important finding for Serbia i s that inthe middle-income group, the reverse effect i s observed, and decentralized countries have lower coverage rates than centralized ones, with an average difference o f 5.2 percent for the same vaccines (Khaleghian, 2004). Serbia may wish to consider adding to this body o f research and conduct household survey analysis or collect facility level data, to examine the impact o f decentralization o f PHC on infant health indicators. c. HEALTH SECTOR DECENTRALIZATION IN SERBIA 4.9 Health sector decentralization is governed by the Health Care Law and will take the form of devolution. It includes transfer in ownership o f hospitals to autonomous provinces and the four cities and o f Primary Health Care (PHC) centers and pharmacies to municipalities; and i s accompanied by a transfer o f decision-making authorities to local governments for the provision of PHC, monitoring o f population health, environmental health programs, the financing o f capital investment, equipment and maintenance costs in PHC facilities, and the appointment o f PHC directors by municipalities. As a result, municipalities may contract with the Health Insurance Fund (HIF) for PHC services; and they can enter in private-public partnership (PPP) agreements for the delivery o f care with the private sector. The Ministry o f Health (MOH) started drafting decrees regulating this transfer. The M O H remains responsible for policy and planning, research, education, training, and health information. Health financing functions including pooling o f funds and purchasing remains centralized with the HIF as the main purchaser of care. Devolution will result in different forms o f autonomy for sub-national authorities with respect to governance, delivery, and financial management. 4.10 Decentralizing health care over the next five years will affect 160 Primary Health Care Centers (Dom Zdravlja). They provide basic primary services including General Practice (GP), gynecology and pediatrics. PHC providers are organized either as a separate entity (84 PHC) or as part o f the secondary care hospitals-Zdravsteni centri (76 PHC). Patients in need for specialized primary care are referred to 60 specialized 45 centers (Zavodi). PHC with larger catchments areas usually provide a broader range o f secondary outpatient services including ophthalmology and cardiology. Inaddition, there are 186 hospitals offering acute, psychiatric and long-term care. Out o f 92 tertiary facilities, 45 are concentrated in Belgrade, especially clinical hospital centers (4) and institutes (32), which raises concern about access to tertiary care in marginalized regions (Sanigest, 2006). 4.1 1 Decentralizationwill not affect the flow of funds from public sources to PHC centers. The health insurance fund (HIF) will remain the main payer for health care. National Health Accounts show that the HIF contribute 92 percent o f public funds that pay for healthcare cost while the MOF provides only 6 percent and municipalities contribute a tiny proportion (see Table 4.1)29. Funds from the HIF and patients finance recurrent costs including salaries and drugs; whereas central and municipality level funds pay for capital and maintenance costs. To strengthen financial monitoring and evaluation, cost control and management accountability, the sector should at all levels invest in standardized accounting tools to monitor and evaluate the flow o f funds, financial reporting, and auditing based on which managers can be made accountable. Financial responsibilities at different levels needto be clarified. Table 4.1: Public Health Expenditures by Source of Funding, 2004 I n YOof Total Public Source of Funding Dinars Health Health insurancefund 79,5 18,386,000 92.3% Total central government 5,548,878,711 6.4% Total localgovernment 1,058,800,000 1.2% Autonomous regions 489,430,000 Municipalities 569,370,000 Total health expenditures from public sources 86,126,064,711 100.0% Source: National Health Accounts, 2004. Central government includes MOH, MOD, MOF, MOE, MOJ. 4.12 Health financing functions of the HIF including collection, pooling, purchasingand benefit package will remain centralized. Purchasers are insurers, and agencies acting on behalf of the government. Purchasing i s the transfer o f pooled funds from the HIF to providers. If health financing were to become decentralized to regional HIFs, the purchasing function would become fragmented which would weaken the insurer's purchasing power for high-quality and low cost care. Thus, following the lessons from elsewhere, health financing through the HIF should remain centralized. The effect o f decentralizing government financing on capital and maintenance needs to be monitored and evaluated as part o fthe investmentplan. 4.13 Transfers from the central budgetto local governments for PHC are largely based on historical spending. To set incentives to municipalities to develop local tax bases and be efficient in spending, financing reform measures should include tying regional equalization transfers to objective structural parameters such as "standard costs" o f providing essential guaranteed PHC services and own tax capacity. This would help 29NHA Serbia (2006): In 2004, health facilities received about 0.9 percent o f their total revenues from municipalities. 46 eliminate soft budget constraints in health, and improve incentives to promote local development. Health deficits should not be covered ex post, nor debts be assumed without involving strong accountability o f local officials. InHungary, for instance, when a hospital reaches a 10 percent deficit relative to the budgetedexpenditure, a public agent i s sent to the hospital to take over hospital management and bring back the budget into balance (Evetovits, 2007). To prevent cost shifting to the HIF or an increase in local tax rates as a consequence o f missing the health care expenditure containment by municipalities, there must be an emphasis on strong accountability of responsible local officials; the ultimate sanction being a temporary management takeover by central government. 4.14 Missing data on health indicators limit the extent to which the effect of decentralization on outcome can be examined. Most data available are aggregated at the national level. Therefore, conclusions cannot be made about differences in health status according to gender, between urban versus rural populations, or municipalities. Also, the data on childhood health indicators are not collected separately for different populations despite the recognitionthat some groups are at increased risk for malnutrition and stunted growth. Standard death rates by specific cause o f death are missing because o f the change in reporting since the separation o f Serbia and Montenegro last year. However, overall health outcomes have been improving. The life expectancy i s comparable to the life expectancy in the EU new member states (74 years). The infant mortality rate and the childhood mortality rate are steadily decreasing. Immunization for measles among children continues to be high at 95 percent. Maternal mortality rate has dropped significantly, when using the five-year averages as the indicator, from 13.88 per 100,000 live births for 1990-1994 to 8.24 per 100,000 for 1998-2002. The trend is expected to continue if there i s an increase in the number o f skilled birthing attendants and more attention is paid to vulnerable populations and levelling out inequalities (UNDP, Serbia, 2004). 4.15 Decentralization needs to be better coordinated with organizational and provider payment reforms, to prevent negative effects on sector performance. Health sector decentralization i s being implemented inthe presence o f major sector-level reforms, including provider payment reforms and health sector restructuring. Currently these three reforms (decentralization, payment reforms, restructuring) are being orchestrated independently and influenced by different stakeholders. They need to be combined and their potential effect evaluated to ensure they support each others effectiveness. D. HEALTH SECTOR REFORMS DECENTRALIZATIONHEALTH AND IN 4.16 The health sector is characterized by a misallocation of staff and resources. Some municipalities and facilities struggle with a shortage o f educated and experienced manpower. Both politicians and civil servants are often inadequately trained for management o f health functions and funds that now have to be handled at the local level. There is a general lack of accountants, and a misallocationo f health workers, with remote areas facing particular difficulties to recruit and maintain educated personnel. A total of 3,476 GPs i s covering on average 1,676 people per physician. While this number i s in line with international standards (1,500 per GP), there are regional disparities which may 47 affect access to primary care. The difference between the district with the highest number of patients per GP and the lowest is 2:l. These differences should be closer analyzed in combination with geographical access and utilization figures. A further constraint i s that some local governments may appoint staff on the basis o f residence rather than merit. Thus, inadequate staffing structures and levels and need to be co- aligned with the devolution o f functions and responsibilities, to prevent unrealistic planning and budgeting for care, failures in accounting, delays in submitting reports and poor service delivery inPHC facilities. 4.17 Local governments will find it difficult to implement rationalization measures such as restructuring and performance evaluation. Experience from other countries shows that health sector restructuring encounters resistance from the public, and i s particularly difficult for local governments to implement3'. Central financial management may help implementing rationalization plans, for example by setting financial benchmarks to municipalities. Ifthey are unable to respect health care financial balance, they must undergo restructuring plans under the control o f central government. The implementationof the restructuringplan can be supported by giving priority to enter decentralization to those municipalities that have met M O H restructuring and accreditation goals. This process can be further supported by the HIF through preferred contracting with better-performing municipalities and facilities. 4.18 The Health Sector Restructuring Strategy describes the way forward towards a restructure health sector. The M O H Action Plan31and the Health Laws pave the way for moves to rationalize the provider network and enhance productivity. The M O H Action Plan identified an excess number o f health facilities, beds and health workers. The M O H restructuring plan32includes a plan for restructuring PHC facilities in the context of the National Investment Plan (NIP), and identifies hospitals and departments that need to be merged, downsized or closed to reach the M O H objectives. It describes the number and types of institutions inthe country, the relationship between them, and interactions betweenthe public and private sectors. The Sector Restructuring Strategy fulfills a central role in this regard and will set forth the main elements and timing o f the development of the public provider network, also called Optimum Provider Network (OPN). The OPN i s defined as the minimum o f providers which can assure effective, accessible, continuous, permanent and professional health care for each geographic region. Plans foresee containment programs o f health expenditures temporarily accompanied by additional public resources, and are aimed at reducing oversupplies inhealth sector (Sanigest, 2006). 4.19 Managing the workforce will be key to implementing the strategy. The health workforce i s the most costly input factor inthe sector. Currently, staffing i s planned on a central level based on M O H staffing norms, contracted by the facility directors in accordance with the labor law and union agreements, and paid by the HIF line-item 30 Poland: Governance and accountability in a decentralized setting. December 2002. The World Bank. 3' Ministry ofHealth: Action Plan, September2005. 32 Republic o f Serbia: Health Sector Reconfiguration Strategy. Inception Report. Sanigest. August 29, 2006. 48 budget, through the M O F disbursement channel. The HIF pays all salaries but cannot influence the number of staff. The M O H started implementingits HumanResource (HR) strategy33,by reducing the health workforce following the M O H action plan34. A further spread o f responsibility for staff creates space for unaccountability and hampers the implementation of the HR strategy. Thus, HR management responsibilities should be streamlined across different levels, and be part o f facility autonomy. To ensure capacities are inplace, priority inthe devolution process should be given to municipalities that have metHRtargets andhave efficient staffmanagement structures. 4.20 Measures are being developed to move to prospective payment in PHC centers and hospitals from input-based line-item budgeting. Input-based line-item budgets set incentives to providers to employ more input factors based on which the budget is defined and operate withinthe allocated budget. Rigidbudget formulation and inflexibility provide limited decision power to health facility directors about the reallocation of funds across line-items to better respondto changes in utilization levels. As a result, directors can hire and dismiss staff within the MOH-set norms which often requires ministerial approval, and can select vendors for supplies other than pharmaceuticals but they cannot reallocate funds across budget categories - and adjust the overall staffing levels or reduce spending on salaries to increase funds for drugs - to better respondto their needs. Unsurprisingly,non-medical staff amounts to 28 percent o f all health employees in Serbia, which i s considerably higher than in the UK (16 percent) (Cochrane, 2006). Line-itemhospital budgetbased on the number of beds led to 5.8 beds per 1000 population, which is high compared to more efficient European health systems.35 High bed numbers combined with considerably longer ALOS than in European countries36,point to low productivity interms of patient caseloads per medical staff (Serbia NHA, 2006). To improve activity management and productivity, the HIF i s now preparing the shift from input-based payments to capitation in PHC centers, and to bundled case-based payment systems such as diagnosis-related groups (DRGs) in hospitals. Diagnosis-related groups classify each hospitalized case according to the diagnosis and other characteristics o f the case, and reimburse hospitals according to the resource intensity o f treatment per DRG(Cashin, et al, 2005). 4.21 Provider payment reforms will lead to capitation payment for PHC. Capitation payment provides PHC centers with a fixed payment for each individual registered, usually adjusted for factors such as age and gender. Serbia has pilot-tested simple capitation in some PHC centers and is now in the process of developing and implementing a capitation formula that includes demographic (age, gender) and geographic criteria, to be scaled up nationwide. Changes in PHC payments appear to be motivated by efforts to improve access to services, especially in rural areas. Capitation sets a financial incentive to improve productivity and operate within the capitation 33Cochrane,D.andMinistry of Health(2006): Republic of Serbia: HumanResourceStrategy 2006-2010: 34The PHC workforce must be reduced by another 1,3 18 staff to reach the 2007 staffing targets. This reductionis expectedto decreasethe annual healthworkforce wage bill by Dinars230 million. 35In 2003, the number of beds per 1,000 populationwas: 4.4 in Estonia, 2.3 in Finland, 3.9 in France, and 3.6 in Italy.Source:www.data.euro.who.int. 36In 2003, ALOS in acute hospitalswas 9 days in Serbia, 6.4 days in Austria, 3.6 days in Denmark, 4.3 days in Finland, 6.8 days in Italy, and 7.9 days in Lithuania. Source:www.data.euro.who.int. 49 amount; and it imposes the full insurance risk on providers and so discourages them from oversupplying care (Ellis and McGuire 1993). To prevent under-provision o f care, capitation payment often includes some sort o f case-mix adjustment and output-based incentives such as FFS, or performance payments for preventive care services, and strict monitoring and evaluation o f provider performance to identify adverse effects. 4.22 PHC centers and hospitals receive revenues from different sources setting an incentive for cost shifting. Health facilities receive payments from the HIF, municipalities and patients who pay out-of-pocket. Where providers have several payers, a budgetary cap set by a capitation budget or by an expenditure cap on the hospital volume is less effective, as providers may try to shift costs across different financers (Newhouse, 1996). Many hospitals use "vendor financing" when they reach their line- item ceiling to overcome shortages. The inability of hospitals to work within a budget causes the HIF and the government to finance the cost over-runs o f hospitals, and patients to eventually pay higher fees for service use. Instead o f increasing efficiency, PHC centres may also try to shift costs to municipalities or to patients by increasing co- payment levels. Capitation may also set the incentive for providing sub-quality care to decrease costs. To prevent PHC centers and hospitals from cost-shifting and from providing sub-standard quality care, they may need support in improving productivity and quality o f care through standard treatment protocols. Regular analysis needs to be conducted on provider performance, cost and the flow o f funds. 4.23 A system that pays capitation to primary health care providers and diagnosis related group based payments to hospitalssets the financial incentive to increasethe number of referrals from PHC to hospitals. The incentive under DRG payment is to increase the number o f cases treated inhospitals and shorten their average lengtho f stay. This highlights the need for monitoring and evaluation o f provider performance in all PHC facilities and hospitals, and for public reporting o f results, to ensure incentives set by the payment lead to expected results, The implementation o f a DRG system in hospitals offers a potential avenue for deriving aggregate output measures adjusted for case severity for the hospital sector. The information on individual hospital outputs could be usedto measure the impact o f decentralization andpayment reforms. 4.24 Implementing payment reforms should be paced in accordance with institutional reforms and available data, to ensure that the health care system remains stable during a period of change. Phases o f implementation can be staggered inaccordance with the adequacy of information systems and available technical support. Health centers and hospitals will have to invest in administration systems (information and billing system) to report their cases and be reimbursed by the HIF. The HIF will need an information system that computerizes the recording o f registered members and their use o fcare inoutpatient settings, and of cases byhospitals and the grouping ofcases into payment categories. The standardization o f data systems and accuracy, comprehensiveness and timeliness o f reported data needs to be monitored. Additional staff will be needed and trained inhospitals and PHC centers inthe collection o f activity and cost data and clinical coding. An agency needs to be made responsible for receiving the relevant data, applying the case mix grouper, monitoring quality standards, and benchmarking (Wiley, 2007). 50 4.25 Provider payment reforms need to be embedded in overall health strategies and undertaken in conjunction with other reforms. These include institutional and legal changes, demand side measures to steer consumers (e.g. charge lower co-payments for generics); public-private-partnerships (PPP) in health; provider autonomy allowing facility managers to adjust their resource- and output-mix to respond to the financial incentives embedded in the payment system; investment in information technology to ensure monitoring and evaluation; training o f staff; public reporting to create awareness about provider performance; and standard treatment protocols to ensure adherence to quality standards. Implicationsfor Management in Health Facilities and Municipalities 4.26 Health facility managers will have increased decision-making autonomy and could react adversely to the provider payment reform resulting in inefficiencies and unsatisfactory quality of care. It is o f importance to clearly outline autonomy, strong management skills, and monitoring ( M I S ) and evaluation systems to asses the financial and quality performance in hospitals and PHC centers. Providers' referral practice must be monitored and evaluated and managers held accountable by local stakeholders and standards governing the central level. Financial allocations from the central government to municipalities should be made based on a formula that takes into account variations in input prices, differences inhealth care priorities, needs and equity goals and supports the incentives set by the provider payment reform. 4.27 Managing PHC centers under capitation and hospitals under DRG payment requires substantial institutionalchanges. To respond to financial incentives, hospitals will have to improve productivity, which requires flexibility in the mix o f input factors including staff, drugs, equipment etc to select the mix that allows producing a case at the lowest costs while maintaining quality levels. Hospital managers will need to be held accountable for results (Widmer, 2007). 4.28 T o increase provider responsibility, particularly under capitation payment, some degree of purchasing could be decentralized to PHC providers. The United Kingdom experimented with using primary care doctors as purchasers (General Practitioner (GP) Fundholders). In 1992, volunteering GPs (GP Fundholders) received a budget to purchase pharmaceutical drugs and elective care. After some experimentation, this was extendedinthe form ofPrimary Care Trustswhich regroup all GPs and form the main purchasing agencies including for hospital care. In Central Europe, primary care fundholding arrangements are emerging. Since 2002, Estonian family practitioners receive a virtual budget representingless than 20 percent o f the total capitation fee with which they can provide care or purchase selected services. Fundholding strengthens the role o f family practitioners in deciding how resources are allocated to hospitals and specialists (Langenbrunner et al., 2005). Once the provider payment reform in Serbia i s well advanced and well functioning, the Government may consider implementingsimilar reforms. 51 Monitoring and Evaluation, and Public Reporting of Resultsto CoordinateReforms 4.29 Municipalities, health care providers and consumers need to be informed about the consequences of decentralization and effects of related reforms. Some municipalities and providers may have limited information about the additional workload decentralization will create, whether new positions need to be formed; changes in roles and responsibilities, organizational, procedural and financial changes; as well as the management and financial implications created by the payment reforms and the health sector restructuring plan. Municipalities and facility managers needto be informed about the organizational and administrative consequences of decentralization, as well as the regular performance results. Public reporting o f performance results creates transparency and peer pressures, and stimulates the interest in improving quality on the part o f providers and owners. 4.30 The lack of analytic criteria and valid data is a key reason for difficulty in determining the outcome of decentralization. Several challenges exist in terms o f assessing decentralization: (1) measuring decentralization by identifying dependent and independentvariables and then demonstrating the appropriate associations betweenthem; (2) measuring the outcomes o f decentralization in health care; and (3) comparing decentralization between countries (Saltman et al. 2007). Developing an outcome indicator which would describe the impact o f decentralization o f the PHC system on QALYs, or a cruder measure such as avoidable deaths (e.g. infant mortality rate), would require agreement on a consistent definition and methodology to measure both the avoidable deaths -- and the contribution o f PHC decentralization to this outcome aggregate health status by expected number of QALYs at a given age -- or alternatively (Hakkinen, 2007). In Serbia these challenges are further compound by a lack o f valid data. The IPH reports extremely poor record keepinginPHC with less than 30 percent o f simple examinations like Blood Pressure being recorded. In addition, despite very high levels o f requests, 30 percent of pathology and X-ray requests went unreported and more than 60 percent o f ultrasound examinations were not delivered. First visit information indicates that 75 percent of attendances at primary care level are for diseases o f the respiratory system (Sanigest, 2006), pointing to unreliable data collections by providers. Thus, an evaluation of the impact of decentralization and provider payment reforms would require additional data collection and analysis inhealth facilities. 4.31 Facility level analysis will help evaluating the impact of provider payment reformson providers' treatment behavior and prevent negativeeffects. There is little empirical evidence from transition economies on the implications of payment mechanisms for provider behavior and cost. Generally, it i s hypothesized that providers select a resource mix for treatment within the technically feasible set and subject to a budget constraint, which depends on health financing and provider payment. Findings from the US suggest that providers respondto the financial incentives embedded in the payment method and adjust their treatment intensity within a clinically acceptable range to keep cost low (Dor and Farley 1996). Inthe worst case, this treatment adjustment may lead to dumping and skimping o f patients, resulting in lower cost caused by medically inadequate quality levels; or providers might reduce those dimensions o f quality that are difficult for insurers and patients to observe (Ellis and McGuire 1996). Results from a 52 study on provider payment reform in China indicate that the change fiom fee-for-service (FFS) to a prospective global budget system inhospitals ledto a slower rate o f growth o f overall expenditures (Yip and Eggleston 2001). Another Chinese study found that provider payment reform needs to be accompanied by measures such as an essential drug list and regulation o f average length o f stay, readmissionrates and quality o f care, to limit provider response to undesirable incentives (Meng et al. 2004). Findings from a facility survey conducted in Rwanda suggest that providers adjust costs and the resource- intensity o f treatment to the expected payment source, which has a cost-containment effect (Schneider and Hanson, 2007). Considering these results, Serbia i s well advised to closely monitor and evaluate the effect o f provider payment reforms to prevent adverse effects, and to adjust the payment mechanism. 4.32 Coordinating decentralization with current provider payment and organizational reforms is of paramount importance for the overall effectiveness of and weak linkages and information flows betweenthe different actors -- PHC, hospital health sector reforms. The fragmentation o f the health care system across sub-sectors care and long-term care -- often causes concerns about the cost and quality o f care (Hiikkinenet al. 2007). Results from a study conducted by the IPH in2001 suggest that about one third o f patients were referred on to secondary care. This i s a very highreferral rate by international standards even from healthcare systems where the primary care level i s largely staffed by generalists (Sanigest, 2006). High referral rates are symptomatic o f poor organization and a lack o f well-defined referral protocols, and raise concerns about potential adverse provider responses to changes infinancial incentive that will be created by the move to capitation and DRGpayment. Improving co-ordination within health has been identified as a key element for enhancing the effectiveness o f the health care system, through higher quality o f care and cost containment. Thus, Serbia should tightly monitor and evaluate the impact on provider behavior created by the financial incentives set trough the capitation payment in PHC centers and case-based payment in hospitals. Investmentindata collection and monitoring as well as data entry staff may be neededto prevent that performance reporting becomes a strain on health staff who lack necessary IT and related data management capacity. E. CONCLUSIONSAND NEXT STEPS 4.33 Major reform strategies, combined with increased utilization for more sophisticated care and growing cost pressures have put considerable strain on both healthcare delivery and finance. The current reforms hold promise for improving system performance by changing the financial incentives to providers for improving access to better quality health care, while at the same time promoting effectiveness and the efficient use o f resources and cost containment through increased management autonomy and responsibility on a local level. However, decentralization and provider payment reform should not be seen as a remedy for reducing oversupply and rationalizing the health sector. Rather central leadership is needed for health sector rationalization including closing and merging o f facilities or departments, to ensure system level efficiency. 53 The following steps may support the effectiveness of these reform measures in Serbia: 4.34 Define clear responsibilities, roles and accountability at all levels of government and provide the necessary training and management tools. The M O H will remain responsible for accreditation and implementing the M O H action plan and related strategies. Rules must be binding and formally codified in regulations. Funding should remain centralized to assure efficient pooling. The HIF responsibilities should include funding, purchasing, monitoring and evaluation and contracting with accredited providers. There i s a need to institutionalize accounting and financial management tools at all levels to ensure transparency in the cost o f service provision. Hospitals and PHC centers need community boards, public reporting, clearly defined management autonomy, independent bank accounts, regular audits, and managers must be heldaccountable. 4.35 Providers must have some degree of autonomy to re-organize service delivery and managing their resource-mix. Continuous training should be organized for providers about payment reforms to ensure they understand which changes will be possible and necessary to do well. Provider autonomy and management should be implementedinthe following key-area~:~~ 0 Facility manager: Takes the role o f a CEO, who i s responsible and will be made accountable for cost and outcome; and who will be fired if performance targets are not met. 0 Staff management: Devolve decision about hiring and firing, remuneration and fringe benefits to facility director who selects staffing-mix that increases revenues, productivity, quality o f care, and patient satisfaction. Facility manager releases redundant staff; introduces part-time employment, and pays performance- based salaries to all staff based on patient satisfaction. 0 Financial management: Train providers on how to work under the new payment systems, how to use and invest profits and prevent financial loss; and to allocate funds efficiently to pay for recurrent costs. 0 Other input management: Identify ways in provision o f care to improve efficiency by changing the quantity and type o f drugs, supplies, and other input factors, and by working innetworks with other providers. 0 Physical assets: Disposing o f existing capital stock, including buildings and equipment, or acquiring new capital, merging of departments, and entering into private-public partnerships etc. Providers pay a capital fee based on return o f investmentto owner. 0 Organizational structure: Provider implements most efficient management structure, organization o f departments and ancillary services, and has power to 37Werner Widmer (2007): Moving from input-based to output-based payment: Requirements for hospital autonomy and impact on management. Presentation given at the World Bank workshop on Health Financingand Strategic Purchasing.Ljubljana; May 21,2007. 54 enter in PPPs and contract out services to the private sectors (e.g. laundry, cleaning, laboratory etc.). Output mix: Shift from inpatient care to outpatient and day-surgery which are less resource intense and less costly. Shorten average length o f stay and increase numberof cases treated. 0 Data and analysis: Collect and analyze cost and utilization data for unit cost analysis and process evaluation; patient satisfaction survey with each patient, use results for salary definition and publish results on webpage or local newspaper; conduct quality and performance analysis on infection rate, re-admission rate, bed turnover, productivity etc, and use results to change production o f care and input- mix (including staffing) and indecision to contract out specific services to private sector. 0 Use of surplus revenues: Use profits generated from efficiency gains for investment in quality and facility improvement, staff training, and innovation to make facility more attractive. 4.36 DecentralizePHC to municipalitiesin a stepwise approach givingpriority to municipalities that have met "decentralization readiness" benchmarks. It is recommended that scorecards be used to identify municipalities that qualify for the decentralization process. Scorecard criteria can include management capacity, staffing and facility targets as defined in the M O H action plan, accountability measures, quality targets, and accounting, monitoring and evaluation structures. The central government should only devolve powers to municipalities conditioned upon achieving scorecard results. Capacity building must support municipalities and health facilities in reaching benchmarks identified in the scorecard. A scorecard could rank districts based on the following benchmarks on the PHC provider side: 0 Human Resource (HR) strategy: Implemented all staff adjustments in PHC (including doctors, nurses, paramedical, administrative, and technical staff) and met 2010 targets identifiedinHR strategy. Management autonomy o f PHC centers established and well functioning as shown by financial and other performance results. 0 PHC director: Municipality has transparent selection procedures and contract to hire PHC director. Director is preferably not a medical doctor but trained in financial and health care management. All staff working in PHC center reports to PHC Director. 0 Staff productivity: Nurses' workload is reorganized and substantial part o f administrative work i s delegated to administrative staff, as recommended in the 2006HR strategy interimreport. 0 Capitation payment: PHC centers has structure in place to administer patients under capitation, (including population identified their preferred PHC centedgatekeeper, bank account and accountant, electronic single patient record accessible to all health care services, electronic appointment system, etc). 55 M&E: PHC center has data collection system installed to collect data on utilization, finances and cost; and send this information to HIF who conducts performance analysis. Quality: PHC center conducts quality surveys, (e.g. patient satisfaction questionnaire i s distributedto each patient visiting the PHC center). Surveys are analyzed in M O H quality supervision unit and results used to pay bonus/malus from capitation amount. Public reporting o f results to use peer pressure in quality improvement. Restructuring: PHC centers, particularly in rural areas, are refurbished with necessary equipment, pharmaceuticals and transport means, as recommended in the HR strategy interimreport. 4.37 Invest in information system technology in health facilities and the HIF and ensure the flow of data and analysis across institutions. PHC centers will have to collect data on demographic, health and the socio-economic situation o f the population registered with them. Information will be used for population profiling and case-mix adjustment. Inhospitals, DRGrequires investmentin information systems (see Table 2). Hospitals will have to invest in administration systems (information and billing system) to report their cases and be reimbursed by the purchaser. The HIF will need an information system that computerizes the recording o f cases by hospitals and the grouping o f cases into payment categories; and claims database for outpatient care, reporting utilization and quality o f care and finances. The HIF provider performance department will have to evaluate provider performance and detect under-provision or substandard quality care, or other negative effects in response to the financial incentives set by the payment and management reforms. The standardization o f data systems and accuracy, comprehensiveness and timeliness o f reported data needs to be monitored (Wiley, 2007). Table 4.2: Information System for Hospital Activity Minimum Data Set i. Hospital Number vii. Month and Year o f Admission ii. Patient Number viii. Duration o f Stay iii. Sex ix. Discharge Status iv. Age X. Main Diagnosis v. Marital Status xi. Other Diagnoses vi. Place of Residence xii. Surgical and Obstetric Procedures xiii. Other Significant Procedures Source: Miriam Wiley. The Economic and Social Research Institute. Dublin. Ireland. 4.38 Institutionalize capacity to monitor and evaluate the impact of decentralization and accompanying reforms; and use public reporting of performance results. Performance data needto be collected in all health facilities and transferred for evaluation to the central level MOH, the HIF and the IPH. Health facilities already collect and report epidemiological data to the IPH and the MOH. To 56 measure the impact o f capitation in a decentralized setting, all PHC centers will have to collect and report to the HIF claims data on a patient level that documents the use o f services and pharmaceuticals and revenues from patent co-payments. PHC centers also needto install standardized accounting to collect revenue and expenditure information on their recurrent and capital cost, which should be reported monthly to the M O H for analysis. To conduct provider performance analysis, the HIF will have to build analytical capacity and train a team who i s in charge o f analyzing provider performance based on the claims data received. Results from analysis on financial and care performance should be reported back to providers and municipalities to improve performance; and to consumers to support their choice o f preferred providers. Findings should be used in holding managers in health facilities and municipalities accountable; and by the HIF for selective contracting with health facilities. To limit adverse reactions caused by the financial incentives under payment reforms, the HIF and M O H should define a cap on the number of referrals from PHC centers to hospitals and specialists, as well as on the hospital budget and oblige hospitals to treat all patients. The effectiveness o f devolution and the implementation o f the M O H restructuring plan and payment reforms need to be evaluated regularly. Results can be used to correct and adjust reform measures, and strengthen collaboration across facilities and government levels. Publication o f the results o f comparative analyses, in particular, must not be hindered by opposition from the municipalities and powerful stakeholders. 4.39 The cost and efficiency impact of movingfrom input-basedline-itembudgets to capitation payment in PHC centers and case-based payment in hospitals should be monitored and evaluated in a pilot study. The Serbian Ministry o f Health (MOH) and the HIF could collect monthly routine data on individual patients' health status and care seekingbehaviour and the resultingcase-mix; expendituresand finances inall health facilities for insured and uninsured patients in at least 50 PHC centers during the year before and after the introduction o f the payment reform. Through the payment reform, providers would be able to influence their efficiency by improving the management o f drug storage; the more flexible use o f staff; changing their treatment style to improve drug prescriptions; and by exploiting scale economies and treating more patients. A longitudinal provider dataset would permit unique analysis to be performed, employing econometric production and cost functions to compare the effect o f capitation payment plus a small co-payment on provider productivity, cost and efficiency in HCs and on referral rates to hospitals. This would allow estimating marginal and average production and costs; as well as scale measures to identify resource capacity in health facilities. Additional analysis could be conducted on the patient experience with quality o f care. Findings should be reported back and discussed with providers to improve the effectiveness o f reforms, and used inthe adjustment process o fthe payment system. 57 5. EDUCATIONAND DECENTRALIZATION A. INTRODUCTION 5.1 The Government has yet to formalize its proposals for decentralization in education or for reforming the sector as has been done in the case of health. Several working groups have beenestablished inrecent years. Most recently, a project supported by the Council of Europe and the European Agency for Reconstruction has begun a process o f developing plans for financial decentralization in education to municipalities. This is in line with the general policy direction that municipalities should assume more responsibilities. This chapter i s intendedto provide advice to the government at an early stage o f its deliberations. 5.2 The chapter is organized as follows. The next section describes the current distribution of functions in the education sector between schools, municipalities and the Ministry of Education and Sports and provides a comparison to distribution of functions in selected comparators. The following section reviews the role of decentralization in addressing these problems and the final section suggests next steps inmoving forward. B. DISTRIBUTIONOFEDUCATIONFUNCTIONSINSERBIAAND INTHE COMPARATIVE CONTEXT 5.3 Local governments in Serbia are responsible for financing the operating costs of primary and secondary schools. This includes heating and lighting, maintenance, materials, and transportation o f pupils attending compulsory education). Salaries o f teachers and other staff are however paid by the central g o ~ e r n m e n t .With respect to ~ ~ management issues, municipalities appoint representatives to the school boards while parents and teachers elect their own representatives; each constituency group has three members. School boards are responsible for appointing school directors, with the Minister retaining the right to veto the appointment (though he cannot propose an alternative inthe event o f exercising his veto). School boards are then responsible for the recruitment of individual teaching and non-teaching staff; however, the central ministry continues to approve the number o f staffing positions inindividual schools and negotiates the salary scale with the teacher unions through a collective bargaining process. Municipalities, therefore, have very little control over the level o f salary spending in schools intheirjurisdictions which i s the major component o f the cost o f education. 5.4 Municipalities have no functions or responsibilities over the educational process in schools. The education process would cover issues such as the choice o f books used (where schools are responsible), where students attend school (where the 38 Since last year, the Ministry o f Finance (MOF) has established a database o f all education sector staff and authorizes payments to individual staff on the basis o f this information. Previously, salaries were paid by the Ministryof Education and Sports (MOES). To date, unfortunately, the MOF's salary database is not linked to the administrative databases held by the MOES. central government i s responsible) or in student assessment (where the central government i s responsible for national assessment and the schools for ongoing, classroom based assessment; evaluation o f school performance i s done through combined school level internal and national level external process). Municipalities have responsibility for spending on teachers' professional development though they are not responsible for ensuring that individual teachers develop and implement plans which would help them get promotion. Municipalities are, formally, required to spend one percent o f their education budget on professional development but there i s no regular monitoring o f this by the MOES nor do there appear to be any consequences for failing to do so. Anecdotal evidence i s that poorer municipalities do not fulfill their obligations. (Annex 1provides a more detailed accounting o f the distribution o f responsibilities.) 5.5 One potentialway to judge whether there is more scope for decentralization in Serbia is to compare its distribution of functions with those of other countries. The distribution o f education responsibilities and functions varies considerably across countries, with no clear patterns. Therefore, to compare Serbia with other countries, five countries have beenchosen which exemplify different appro ache^:^^ 8 England - a quasi-market, which promotes competition between schools but reduces the autonomy o f professional educators over the curriculum and assessment. 8 Scotland - a professional partnership between schools and government, with school inspectors giving more attention to school improvement than monitoring and evaluation. 8 New Zealand - education as an example new public management, with the government in the role o f funder, regulator and purchaser and managing schools through monitoring performance. 8 The Czech Republic - de-concentration and devolution to schools rather than to regions or municipalities. 8 Hungary - decentralization as the default option. 5.6 In general, the central government in Serbia is involved in more functions than in other types of countries, with the exception of the Czech Republic. An estimation o f the distribution across countries i s provided in Table 140. Serbia differs from the Czech Republic inthat there i s no regional government to which responsibilities are assigned. Serbia i s assigns relatively few roles to local government and to school professionals, compared to these other countries. 39The materialinthis section is derived from Thomas (2006). 40 The MOES in Serbia has 14 regional offices (`skola uprava') which are deconcentrated agents of the central ministry. In fact, these skola uprava seem to operatewith some small measure o f autonomy, though this discretionis not usedto the same extent in different offices. However, for the purposesof this analysis they are treated as part ofthe central government. 60 Table 5.1: Cross-CountryComparisonof the Distribution of EducationResponsibilities Central Regional Local School Parents/ government government I government professionals Community Serbia 20 (52%) 5 (11%) 23 (61%) 5 (13%) CzechRepublic 20 (53%) 6 (16%) 5 (13%) 23 (61%) 13 (11%) England 13 (35%) 8 (22%) 28 (76%) 14 (38%) Hungary 13 (34%) 9 (24%) 9 (24%) 25 (66%) 4 (11%) New Zealand 14 (38%) 0 (0%) 24 (65%) 20 (54%) Scotland ' 13 (35%) 15 (41%) 27 (73%) 4 (11%) The table records the number and percentage of functions that each level o f government performs. The percentages do not add up to 100 percent since some the responsibility for some functions is divided betweenmore than one levelof government. 2.The matrix usedto assess the distributionis the same as for Serbia inAnnex 1 ofthis chapter. Source: Thomas (2006). C. MAJOR CHALLENGES FACING THE SERBIAN EDUCATION SYSTEM 5.7 Serbia faces problems providing good quality education and is failing to equip young people with the skills and knowledge they need to be successful in the labor market. In the PISA assessment (which tests the ability o f students to perform successfully in dynamic economies, where knowledge and skills need to be applied in new settings to problems which have not been faced before) Serbia's performance was weak, coming in 33'd place out o f 40 in mathematics, 36'h inreading and 34thin science (Figure 5.1). No other European country scored below Serbia, even though many transition countries participated in the assessment; and this i s not due to economic progress since compared to other European countries. Serbia scores slightly below what one would expect givenits national income per capita. Figure 5.1: Relationship between Mean MathematicsPerformanceand Gross National Income per Capita (PPPS) Figure 1 Relationship between mean mathematics - performance and Gross National Income per capita (PPP$) 600 $ 500 0 .- 400 CI 300 5 EcmE200 100 0 0 5000 10000 15000 20000 25000 30000 35000 40000 Gross NationalIncome per capita (PPP$) 61 5.8 The labor market outcomes for young people in Serbia are poor. The overall unemployment rate for young people i s 37.4 percent and even for those with post- secondary education the unemployment rate i s above one-third (Table 5.2). In fact, the differences in employment rates barely vary for all young people with at least primary education. Amongst the whole adult population, only those with tertiary education do significantly better than the rest o f the population. However, the data for Serbia (and Macedonia) distinguishes only between those with 1-3 years o f secondary education as opposed to four years o f secondary; here the differences inlabor market outcomes are not that great. However, inAlbania and Bosnia and Herzegovina (where the data i s collected differently), it is possible to distinguishbetweenthose invocational as opposedto general education programs (Le., betweenthe type o f course rather than the length). As can be seen, those in four-year vocational programs do significantly worse than those who completed four years o f general education. Given the similarities in secondary education structure between these countries and Serbia, it i s very likely that Serbia's general schools perform better than her vocational schools. And, at present, three-quarters o f Serbia's secondary enrollment i s still invocational schools. Table 5.2: Labor Market Outcomes in Albania, Bosnia, Macedonia and Serbia Everyone Youth (25-29 years) Albania Bosnia Macedonia Serbia Albania Bosnia Macedonia Serbia Basic 55.3 41.1 21.6 37.9 61.6 48.7 21.1 37.8 1-3 Employment years/vocational 69.2 49.1 39.7 56.2 57.0 53.1 37.2 55.1 Rate 4-5 yeadgeneral 54.5 50.1 42.0 51.9 56.4 52.6 44.7 47.2 post-secondary 82.4 56.0 68.6 70.5 69.8 90.6 48.2 57.3 Total 59.1 46.1 36.3 51.0 60.9 53.1 36.1 48.8 Basic 63.5 60.5 48.8 49.5 72.2 78.4 66.2 72.9 1-3 Labor Force years/vocational 78.9 69.5 83.6 77.9 79.8 82.3 94.1 86.8 Participation Rate 4-5 yearsigeneral 65.2 69.6 73.9 70.0 76.1 74.1 83.6 72.4 post-secondary 88.6 70.9 90.1 83.5 85.5 93.4 90.3 86.9 Total 67.8 65.5 66.2 67.1 74.9 79.3 79.4 71.9 Basic 12.9 32.1 55.7 23.3 14.7 38.0 68.2 48.2 1-3 Unemployment years/vocational 12.4 29.3 52.5 27.9 28.6 35.5 60.5 36.5 Rate 4-5 yeadgeneral 16.5 28.0 43.2 25.8 25.8 29.0 46.6 34.8 post-secondary 7.0 21.1 23.8 15.6 18.4 3.1 46.6 34.1 Total 12.9 29.7 45.3 24.1 18.6 33.1 54.5 37.4 Source: 5.9 Poor education quality is likely to compound the impact on education performance of early exclusion. Pre-school education, essential for ensuring partici ation and success in education, reached only 29.7 percent o f eligible children (in 2001).$1 5.10 Serbia, along with other countries of South Eastern Europe, is behind the neighboringcountrieswhich have recentlyjoined the European Union in providing access to education. Internationally comparable data on enrollments in Serbia are only available up until 2001. Table 5.3 therefore compares Serbia's p e r f ~ r m a n c e ~ in~that year 41This is the figure for 2001, the latest for which there is data comparablewith other countries (see Annex Table A2). Morerecently,the LSMS in Serbiareporteda figure of33 percentin2005. 42When, ofcourse, Serbiawas part of Serbiaand Montenegro. 62 with other countries inthe region, and also shows the progress those countries have made in the four (or in some cases three) years since then. As can be seen, the countries o f South Eastern Europe are behind. In2001, Serbia and Montenegro's performance was in the middle of the group o f SEE countries; since then the other countries have increased their enrollment at the secondary and tertiary levels (with the exception o f Bosnia and Herzegovina in secondary education), but they still remain significantly behindthe group o f comparator countries. It seems likely, therefore, that Serbia too has some catching up to do. Table 5.3: Enrollment data for South and Central European Countries Pre-primary Basic education Total upper Higher education enrolments(net enrolments(gross secondaryeducation enrolments(gross rates, per cent of ratios, per cent o f enrolment(gross ratios, per cent of populationaged 3- relevantpopu1ation)a ratios, per cent of populationaged 19- 6) populationaged 15- 24) 1818 2001 2005 2001 2005 2001 2005 2001 2005 Czech Republic 87.2 86.4 98.6 99.2 87.9 92.4 30.9 47.3 Hungary 86.4 86.9 99.1 100.9 105.7 100.7 39.3 61.2 Poland 50.4 55.6 99.8 100.4 98.0 105.8 50.6 56.4 Slovakia 68.7 72.7 106.2 105.8 88.5 104.0 31.2 41.1 Slovenia 68.3 79.9 101.1 96.5 98.3 100.5 61.2 83.2 Estonia 80.3 85.8 103.8 106.0 90.1 93.7 62.0 65.5 Latvia 65.6 80.1 99.4 102.8 90.3 98.5 67.3 12.5 Lithuania 50.9 55.7 102.4 103.8 97.3 104.1 60.9 14.9 Bulgaria 73.6 73.7 97.1 98.6 79.0 91.3 36.9 40.1 Romania 61.7 13.8 100.0 102.3 73.2 78.9 31.6 43.1 Albania 34.3 38.4 92.5 90.6 48.4 58.7 13.9 21.7 Bosnia-Herzegovina 6.2 81.4 90.8 52.1 45.0 19.7 25.2 Croatia 36.2 49.3 95.2 94.5 81.3 86.1 29.5 36.3 FYR Macedonia 28.5 32.5 98.6 98.4 69.1 73.4 20.2 Serbiaand Montenegro 29.7 - 96.5 - 55.0 - 23.9 - 21.2 Source: 5.11 The shortcomings in the initial, formal education system are compounded by the weaknesses in the supply of quality adult education and training opportunities. As a result, those with weak or inappropriate skills and knowledge are particularly affected by these shortcomings. The legacy o f a network o f adult education institutions and organizations has not been preserved. Secondary vocational schools allow adults to take examinations incertain subjects but do not make special provision for training them. There are a few programs o f continuing education in the universities, but these are limited mainly to former students. 63 5.12 The education system is becoming increasingly inefficient. Between 1997 and 2005, the primary school population fell by 24 percent and the secondary school population by 15 percent.43 However, the numbers o f teachers and schools have not fallen as pupil numbers have fallen. In fact, quite the opposite is true: in secondary education, for example, in the last five years, in Central Serbia the number o f pupils in secondary education fell by 9.3 percent but the number of teachers increased by 6.5 percent. There was a similar story in Vojvodina, with pupil numbers lower by 8.3 percent but teacher numbers higher by 6.8 percent. As expected, this has resulted in a sharp decline this relative short period inthe number o f pupils per teacher - from 13.5 to 11.5 for the country as a whole. By comparison, the OECD average i s 14 pupils per teacher. There has beena similar experience inprimary education. 5.13 This is leading to wide variations in per student resource allocation. These historical trends have left Serbia with a large number o f small schools. Primary schools with less than 200 studentsaccount for seven percent of schools but less than two percent of pupils; and more than half o f all primary schools have satellite classes.44There are more than 300 schools with five or fewer students. It also appears that the way pupils are divided into classesvaries acrossthe country; for example, schools with 100 studentsina grade have betweenfour and six classes for those students. 5.14 Though spending per student is larger in small schools, because class sizes tend to be smaller, the results on the national assessments are not as good. Interestingly, spending on non-salary inputs is higher in small schools too and this is usually associated with better performance, but not in Serbia. Presumably this i s because this money is mainly spent on heating and utilities rather than teaching training and teaching materials which have a more direct impact on quality. This variation in spending i s compounded by the fact that the distribution o f schools o f different sizes varies across the country; in Central Serbia 65.6 percent of primary schools offer only the first four grades, compared to only 25.6 percent of schools in Vojvodina. So the access to good learning opportunities i s not evenly distributed across Serbia. This will be examined in more detail inthe next section. 5.15 Looking forward, the number of students of different age cohorts is predicted to continue to fall dramatically over the coming five years. This is especially relevant in primary education (by more than one-third). This provides an opportunity for Serbia to release resources for improving the quality and relevance o f education, as well as increase enrollment rates in secondary and tertiary education. But this will meanreducingteacher and other staffnumbers inline with the decline instudent numbers. 5.16 Educational outcomes vary widely across municipalities. Serbia has made strong strides in recent years in establishing a technically sound system o f student assessment and evaluation: this includes approved standards in mathematics and mother 43 All figures inthis paragraphare from Karavidic 2007, based on MOES data. 44 Satellite classes are those in small buildings in small villages away from the main school building; usually offering only the first four grades of primary education. 64 tongue at grades 4 and 8; national assessments o f a random sample o f students against those standards; participation in international assessments o f student performance; and school evaluation based on both internal and external assessments. However, these assessments do not yield evidence o f performance at the municipal At the end of primary education, every student who wishes to enter a four year secondary education course (at either a vocational school or gymnasium) takes the qualifying examination. Students take tests in mathematics and Serbian language.46 Since the qualifying examination i s taken by a large number of students, it i s possible to report scores at the municipal level. Unfortunately, the qualifying examination i s not yet being carried out to the same high levels o f technical excellence that characterize the national assessments described above.47Therefore, the scores received by students are not reliable measures o f comparative educational performance. As a result, it i s not possible to use the raw scores as a methodology for determiningperformance at the municipal level. Instead, two other approaches have been used in this paper. First, the participation rate i s defined as the proportion o f students in the final year o f primary education who actually take the examination. Since students who do not wish to continue in education or who wish (or feel prepared) only to go to a three-year vocational program do not take the examination, the participation rate is a measure o f how well schools and municipalities prepare students for secondary education. The results across municipalities show that, on average, 76 percent o f students who reach the end o f primary education take the examination. Municipal performance varies considerably around this mean, with 48 municipalities (30 percent) recording participation rates below 70 percent (Figure 5.2).48 45The school internal and external evaluation process will yield results at the municipal level, but the process has not been used at a sufficientnumber of schoolsto date. 46The results on this examination are combined with the marks a child gets during the last years of primary education. Students receive a score between zero and twenty for each subject. This final score is used determine to which school a child is eligible to go. Students express preferences for which schools they would like to attend. However, the Ministry o f Education and Science determines the quotas that each school and each program is able to enroll and schools with most demand for these fixed places are able to select students with higher scores on the qualificationexamination. 41Most seriously, there is no mechanism for ensuring that scores across schools are awarded consistently. There are also though a number of other weaknesses, such as the fact that students are notjudged against the same standards in a consistent way from one year to the next, the very strange distribution of marks with more students receiving high scores than low scores, and the fact that the test is essentially one of memoryrecallof facts. 48It should be noted that the participation rate records the percentage of students taking the examination amongst those who reach the final year of primary education. Approximately X percent of students drop out before they reach this stage and it seems likely that the drop out rate is higher in municipalitieswith low participationrates. 65 Figure 5.2: Participation Rates across Municipalities 1000 90 0 80 0 70 0 PartigLpd 50 0 40 0 30 0 20 0 10 0 0 0 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101 106 111 116 121 126 131 136 141 146 151 156 Municipality Ranking 5.17 The threshold score provides another measure of educational quality. The threshold score i s the number o f points above which 90 percent o f all students score. For example, inmathematics this score was 4, and inSerbianlanguage 7 (both scores out o f a maximum o f 20). Any student scoring below this level i s in the bottom 10 percent o f students inthe country. This threshold has been calculated for the purposes o f this study only - it i s not part o f the examination at all - because it i s a way to generate a reliable performance measure. The percentage of students in each municipality above the threshold score i s then calculated. The percentage o f students reaching the threshold score in each by municipality varies even more than the participation rate (Figure 5.3). Fifteen municipalities have more than half o f their students scoring below the threshold score - in other words, the majority o f students in these municipalities are getting the very worst scores in the country. And these students are, consequently, unable to get access to the secondary education program o f their choice and, in all likelihood, would not have access to good quality secondary education at all. 66 Figure 5.3: MathematicsPerformance 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101 106 111 116 121 126 131 136 141 146 151 156 Municlpalltyranking 5.18 Socio-economic factors appear to explain only part of the differences in performance across municipalities. Figure 5.4 shows that the more disadvantaged municipalities (the higher the ranking the more disadvantaged the m~nicipality),~~ in general the lower the proportion o f students in each municipality who have achievement above the threshold score. However, the relationship i s surprisingly weak compared with what one would expect from other assessments, with municipalities dispersed quite widely on either side o f the regression line and that line sloping only gently. In other words, the only a small amount o f the difference in performance on the qualifying examination can be explained by the difference in socio-economic status o f municipalities. 5.19 There is a strong correlation between the participation rate and the threshold score performance (Figure 5.5). So, students in some municipalities (those in the bottom left of Figure 4) face a double disadvantage. First, students in these municipalities do not receive a good education, so a highproportion don't even bother to sit the examination because they don't want to go to secondary education or are resigned to a 3-year rather than 4-year vocational course. But, second, even if students in these municipalities do take the examination they don't do well and so cannot go to a good secondary school. 49The measure of disadvantage here is the simple average o f municipal rankings on four measures: GDP per capita, employment rate per 1000 inhabitants, size of municipality in terms of number of students, and whether the municipality is more or less urban.All data are from the Republic Statistical Office and for the year [2005]. 67 Figure 5.4: Correlation of Mathematics Score and Disadvantage 100.0 90.0 PsrcBRtBge 70 0 60 0 50 0 40 0 30 0 20.0 10 0 0 0 0 20 40 60 BO 100 120 140 160 180 Disadvantage ranking Figure 5.5: Correlation between Participatiod and Mathematics Score 00.0 Participation rate (%) 68 D. THEROLEOFDECENTRALIZATION INHELPING ADDRESS TO THE PROBLEMSIN THE EDUCATION SECTOR 5.20 There is increasing evidence that decentralization responsibilitiesto schools can make an important contributionto improvingthe quality of education provided in schools. For example, in its analysis o f the results from the Programme for International Student Assessment (PISA) 2003, OECD has identified the following features as factors in good performance5', meaning a country's performance i s both o f a highaverage and relatively equitable?' 0 Highambitions and clear standards. o Students and schools perform better in a climate characterized by high expectations through national educational standards and the readiness by students to invest effort in learning, the enjoyment o f learning, a strong disciplinary climate, and good student-teacher relations. Access to best practice and quality professional development. o Schools and teachers need help to improve their performance. There are strategies and approaches that have proved valuable in improving performance, both from within Serbia and outside; and educational actors needto know these ideas. Diagnostic knowledge and intervention ininverse proportion to success. o The performance of the system and of schools within the system needs to take place against a background in which action will be taken in response to these findings. But not all schools need to the same level o f support; the weakest schools needmore helpto improve. 0 Devolved responsibility over teaching, the school as the centre o f action. o The study found in particular that the more autonomy schools have over deciding which courses are offered the stronger and more equitable the performance o f the system. 0 Integrated learning opportunities. o Students are not divided into different curriculum tracks, but rather all students have access to a strong core o f subjects; similarly, teaching should ensure students can apply theoretical knowledge to new and to real-life situations. 0 Individualized learning. o Individual students have different learning needs, both in terms of their starting points (which may be different indifferent areas o f the curriculum) as so See, for example, OECD, Learningfor Tomorrow s World: First Resultsfrom PISA 2003', 2004. OECD also showed that spending per student is positively correlated with performance, but, as has been found in many other studies, the correlation is not very strong; on PISA this factor explains about 30 percent of the variance across countries. S IOther researcherslooking at international assessments have come to similar conclusions. For example, Erick A. Hanushek and Ludger WoDmann, Education Quality and Economic Growth, World Bank 2007, which argues that parental choice, strong accountability and school autonomy are key reforms needed to change the incentiveson schools significantly. 69 well as the ways in which they learn most effectively. Their motivation will come from teachers responding to this diversity. This means that the professional development o fteachers 5.21 This agenda for the improvement of quality implies relatively little role for municipalities; the main locus of decision-makingon quality should be at the school level, within the context of clear and high standards set by and appropriate support from the central authorities. For Serbia, perhaps the most significant area in which schools could be given additional responsibility i s in the choice o f textbooks and other curriculum materials. This would empower schools and especially teachers to use their professionaljudgment to reach nationally-defined standards. Moreover, experience from other countries suggests that the quality o f the materials produced increases when there i s an active private publishingindustryresponding to the needs and interests o f teachers. 5.22 Introducing autonomy over the selection of materials needs to be linked to the professionaldevelopment of teachers. This will help ensure that teachers have the knowledge they need to make informed choices about materials and fitting teaching approaches to particular lessons. Funding for professional development i s already one area o f responsibility for municipalities, and it makes sense that decisions about how professional development resources are spent takes place at the municipal or school level as different teachers have different needs. Giventhe low levels of spending inthis area at the moment, it does not make much sense to develop the responsibility for spendingthis money to schools. 5.23 A network of regional training centers has begun to be established. These centers recognize the funding role of municipalities with respect to professional development by giving the municipality in which a training center i s based the responsibility for running it. The training itself i s delivered by private providers, who use the facilities ofthe center. 5.24 I t is important to find ways for these centers to offer services to all schools and teachers in a region. At the moment, these centers (which have beenestablished in NiS, CaCak, and Uiice) offer services only to schools inthe municipality inwhich they are based. And schools have identified at least some o f their training needs through the school development planning process. There i s a need for clear arrangements for other municipalities to finance activities at the regional centers. Moreover, there will be some regional or national priorities which will need to be addressed alongside local initiatives (for example, where national or international assessments reveal particular widespread deficiencies). One option for reconcilingthese approaches would be to develop a regional board with representatives from schools and teachers to advise the regional centers and to advise municipalities and schools in turn on how professional development resources mightbe spent.52 ''See Thomas 2006. 70 5.25 A major reformof secondary educationwill be needed to improvethe quality of learning. This will needto address major structural issues to do with the appropriate curriculum for secondary school students. In particular this means changes to the way vocational education i s provided, with much less emphasis on ractical work and more on core subjects like mathematics, language, and the sciences.5PIn general, more students should be in four-year courses, and these courses need a larger and stronger core o f general education subjects. These changes to curriculum will need to be accompanied by a major increase in teachers' professional development as well as new teaching equipment,materials and textbooks. 5.26 A major weakness in the financing of education is that management of resources is split making consideration of trade-offs difficult. Decisions about spending on wages (and in particular on the number o f staff) are made by the central government while decisions about other inputs (like professional development, materials, maintenance, etc) are made by municipalities. This means that there i s no place in the system whereby choices are made between spending on one category o f expenditure as against another. This would make it difficult in any system to shift expenditures to more effective uses. It is doubly unfortunate in Serbia because o f the decline in student number means that there is the potential to capture savings on salaries where pupil numbers are declining and spend these resources on non-salary inputs and in some cases in different parts of the country. However, with the separation in spending responsibilities, there are few incentives to reduce the number o f teachers; if the central ministry does this the result would mean taking the politically difficult decisions which will result only inreductions to its overall education budget. Municipalities, on the other hand, currently have an incentive to keep as many people employed as possible since they are not responsible for paying wages. 5.27 The most substantial progress toward a more equitable and efficient distributionof resources is likely to result from the introductionof a formula-based allocationto municipalitiesof funding for staffwages (school directors, teachers and pedagogicalstaff). Inthis way, municipalities with similar characteristics would receive the same level of funding, helping address a key inequity in the current system. Given the current inequalities in funding, any formula will result in some shifts in funding across municipalities-the existing allocations would not be replicated. Inaddition, such a move would enable municipalities to make decisions about both staffing and non- staffing inputs, with the potential improvement inefficiency o f spending. 5.28 However, the benefits of the allocation formula will only accrue if there are commensurate changes in the degree of control entrusted to municipalities. Municipalities should be able to determine how to spend the resources allocated to them and over the factors that affect the level o f spending: 0 The ability to switch funds betweenwage and non-wage recurrent budget categories. 0 The opportunity to switch resources between uses is a key tenet o f decentralization, as it provides local governments discretion over how to meet 53The teaching and learning approaches in gymnasia will also need to change, in particular addressingthe overloaded curriculumand the pedagogy ofteachers. 71 their responsibilities inthe education sector. Inthe short term, it should provide an incentive to close small schools and reallocate the funds saved. However, there i s risk that savings may be used to increase the number o f teachers and other staff rather than increase spending on other inputs vital for improving quality. The authorities are invited to consider setting up national standards to regulate these issues (see below). Employment o f staff, especially teachers. The decision about how many staff to hire is essential in determining the education budgets for m~nicipalities.~~ If funding for teachers' salaries is allocated to municipalities on a capitation basis, municipalities must have sufficient control over their wage bills to respond accordingly. The settingand verification o fminimumservice standards for municipalities. The central government will need to promulgate a wide range o f regulations setting the framework for municipalities in areas including health and safety, organization o f schools and finance. Inmany areas, there are existing regulations but these are out o f date. Decisions will need to be made about what new standards are appropriate, who i s responsible for upgrading facilities that currently do not meet these standards, and the sanctions for failure to observe standards. Moreover, Serbia does not currently have standards for the educational performance o f municipalities; these will need to be developed so that those municipalities who are not providing an adequate education can be supported to improve. 0 The distribution of responsibilities between municipalities and schools. The authorities need to clarify the distribution o f responsibilities, with schools, their boards and the municipalities in charge o f staffing decisions, decisions over spending, curriculum and textbook choices and setting class schedules. 5.29 One of the most complex choices the authorities will need to make concerns the management of the school network. The most immediate concern i s in primary education, where the numbers o f students will continue to decline and enrollment rates are already close to 100 per~ent.'~ Declining school-age cohorts will provide an opportunity to reduce the number o f teachers and downsize or eliminate schools (although internal migration will require additional classrooms in some locations). The government needs to decide whether the central or the municipal government will close inefficient schools when needed. A solution to consider i s a partnership between the central and local governments, with the central government providing guidance on closure decisions and some transitional funding and the municipal governments responsible for implementingplans. 54A more radicalstep would be to give municipalities(or even schools) control over the salaries and other terms and conditions of employment. Hanushek and WoRmann (2007) argue that quality of education is higher in these circumstances. However, such a step is not recommended for Serbia at the present time, given the significant disruptionthis will cause, coupled with other significant changes that would be taking place. 55As noted above, even though secondary school cohorts are likely to decline too, the need to increase enrollment rates places might mean there is no need to reduce the number of secondary schools or secondaryschool places. 72 5.30 The decline of the size of the secondary school cohorts offers an opportunity to increaseenrollments and restructure secondary educationwithout, at least in the short term, the need to close schools.56 The role o f municipalities inthis change inthe structure o f secondary education i s complex. Approximately half o f municipalities (83) have only one or two secondary schools, and a further 15 have none. As a municipality takes on more responsibilities in education, it may be tempted to increase the number o f secondary schools it runs in response to parental demand and to the national policy to increase enrollment. Municipalities currently have the power to suggest new programs, butthese have to be approved by the MOES. This process so far has not beensufficiently quick to enable schools to respond to the demand o f their communities; and as the curriculum shifts approvals will increase innumber. This threatens to clog up the system and not allow it to adjust. However, it does not make sense for all municipalities to offer all types o f secondary programs. Therefore, the process by which changes are approved needs to be re-examined so that there i s the right balance between prompt response to local needs, speed and national goals. 5.31 The curriculum changes in secondary schools will have implicationsfor the way municipalitiesspend their budgets. Since there will be a major demand for new curriculum materials and for teacher training which municipalities pay for even if the central government makes the infrastructure investments;municipalities therefore need to understand and buy into these reform ideas. 5.32 The central government will need to alter the way in which it provides funding for schools. Increases inenrollments insecondary education will not be uniform across the country, given that rural areas currently have significantly lower enrollments than urban areas. Unless there is a significant increase in resources to the education sector, this will imply that there i s some needto shift resources from areas o f the country where pupils numbers are declining or not rising so quickly, to other areas. The most transparent and equitable way o f doing this will be to shift to a funding formula driven by the numberofpupilsineachmunicipality. 5.33 Using new databases to inform analysis and disseminate results will help improve accountability. Serbia has begun to establish an effective mechanism for monitoring the quality o f education provided in the system as a whole through a system o f national and international assessments o f student coupled with assessment o f schools. The next step is for the ministry to focus efforts on how this information and analysis is usedto inform public debate about how to improve the quality o f education and, thereby, to inform policy making and the allocation o f budget resources. There i s an important responsibility also for the central institutes (`zavods') to identify and disseminate good practice. Moreover, the central government will have to identify mechanisms to help the weakest schools and municipalities through additional resources (with some guarantees that these are used in appropriate ways), targeted professional development, and strong 56There will be a major challenge however as to what to do with those teachers who currentlyteach narrow vocational courses that would be eliminated in the restructured secondary curriculum. Some will, through intensive professional development, be able to teach other subjects. Others though will need to be made redundantto allow room inthe salary budget for new teachers. 73 incentives for improved performance. Finally, the MOES i s aware o f the need to complete the establishment o f an effective education information system so that consistent and reliable administrative data is available to and used by schools, municipalities and the ministry. This system has been well-designed and the MOES i s taking the right steps to increase the incentives schools have to provide timely and reliable data. The main area o f concern i s the capacity o f the MOES to manage the database, as staffing inthe relevant unit i s not sufficient or withthe right mix o f skills and knowledge. 5.34 School report cards are now being piloted in three municipalitieswhich will provide some information for parents and the local community about what is happening in their schools. This is a very good development and, as it rolls out across the country, it will be important to learn how this information i s being understood by parents and the communities and what they are doing with this information. 5.35 There is as yet no mechanism for holding municipalities accountable for the decisions they make regarding education. As we have seen, based on international experience, they are not likely to be able to have a very effective direct role in improving quality in schools. However, given their role in allocating resources, municipalities do have an important indirect role ineducation quality. And the earlier analysis showed that average performance o f students varies considerably and unfairly across municipalities. It is therefore important to find a way to provide reliable and useful information at the municipal level, on issues where their actions can make a difference. The Ministry o f Education and Sports i s considering the development o f `municipal scorecards' (to be developed under the proposed Delivery o f Improved Local Services Project) which would be an important instrument, once good measures o f municipal performance are available to be published.57 E. CONCLUSIONAND NEXT STEPSINDECENTRALIZATION 5.36 Plans for decentralization in education, along with decisions on broader sector reform, are not as advanced as in health. The education systemhas to meet the challenge o f providing high quality education and increasing access. Significant differences inresource allocation across schools and municipalities mean that educational experience varies widely for students across Serbia. A number o f complex issues needto be considered. Each ofthese has beenresolveddifferently inother countries. Serbia will need to learn from the experience and find the set o f solutions that will work most effectively toward increasing efficiency and improving quality. This implies a key role o f monitoring and evaluation and for consultation with key stakeholders. 57 As one indicationo f the difficulty; while the number of students is available at the municipal level, but this cannot easily be converted into an enrollment rate (i.e., the proportionof children in school compared with the total populationo f the relevant age group) because some children living in a municipality may be attending school in another municipality-this is especially true at secondary education. More promising, would be to improve the quality of the qualifying examination so it does provide reliable measures of performance, to expand the sample size of the national assessments and/or to focus on spending by municipalities. 74 5.37 A way forward is an approach which includes the use of pilot schemes to evaluate different options. Decentralization is one mechanism for trying to achieve the improved quality and access outcomes that are necessary for Serbia. It i s not the only one however, and pilots provide a mechanism for evaluating which approach works best in the Serbian context. Annex 2 sets out in detail a proposal for how a pilot might be organized, but inbrief the pilot could work inthe following way: 0 A group o f LSGs would be randomly selected for each o f two options: this randomization means that policy makers can be confident that any results are likely to hold for the country as a whole when decentralization is introduced nationally. These municipalities will be matched with a group o f municipalities o fthe size number where policies will continue as before. It i s important to note that this approach would not permit individual municipalities to opt in or out o f the pilot; this will represent a departure from the usual practice in Serbia. The reason for a randomized approach to a voluntary approach i s that randomization i s most likely to provide the best information on how the pilots are working, and hence to come up with program designs that are best suited to Serbia. Allowing municipalities to volunteer as pilots is likely to lead to the most capable municipalities being evaluated, and hence not provide a full picture of the challenges of decentralization. These pilots would be run for two years, with country-wide implementation o f the approach which has yielded the best results taking place in the third year. A clear timeline is important to ensure that the pilots do not become a mechanism for postponing reform. 0 The first step i s to identify the most important effects which are to be expected from decentralization. These might be: o Indicator 1:Management practices inthe school andmunicipality o Indicator 2: Enrollment levels, both overall and including different sub- populations o f students o Indicator 3: Efficiency ofuse ofhumanand financial resources o Indicator 4: Drop out or grade repetition rates These aspects would be measured carefully before the pilot begins to establish a baseline and then again during the pilot and towards the end o f the pilot to see which policy option has the biggest impact. The central policy option assessed here is the extent to which responsibilities are given to schools or the LSGs. The proposed division o f responsibilities here is meant to be indicative rather than prescriptive; the MOES may decide to pilot a different balance o f responsibilities. But it is important that two approaches are chosen which are both realistic possibilities and which are sufficiently different that their impact can be measured. Inthe end, the goal o f piloting two different options i s to give the MOES the chance to try different policy options before scaling up to cover all LSGs. 75 0 In both options, a per-pupil allocation for determining funding would be used. The MOES (in consultation with LSGs and with the approval o f the Intergovernmental Fiscal Commission) develops a funding formula based primarily on the number o f pupils in each school. This formula could include allowance for a small number o f objective factors which increase the cost o f education such as pupils with special educational needs, sparsely populated communities, and language minorities. Ideally, the actual additional costs o f these pupils would be derived from existing spending patterns, but the evidence suggests that existing spending does not correlate accurately with real costs. Therefore, an analysis o f the EIS data and some kind o f school mapping exercise would be needed. 0 Local self-governments would also receive rationalization grants. The pilot LSGs would prepare a plan for which schools to close, according to the national guidelines on the minimum number o f students, class size and geographical factors. The MOES will provide the LSGs with grants to implement rationalization o f the school network, for example, to invest in newly required classrooms and school buses. 0 The MOES, Regional Education Offices and LSGs will facilitate teachers to find schools by setting up a nationwide vacancy information system (including on the internet), Relocation grants could be provided to teachers to encourage them to move to another location where there are teacher vacancies. 5.38 The purpose of the pilot is to prepare the way for a broader decentralization within the education sector as part of an overall strategy for improving the performance of the education system. Many o f the activities needed to implement a pilot will be those which will provide the foundation for this broader reform, even if the pilot is by definition on a smaller scale. Activities conducted to make the pilot successful will, therefore, also be the first steps inthe decentralization process. 5.39 Key steps which emerge from the proceeding sections are: 0 Government to develop public consensus amongst the various stakeholders (including schools, LSGs and the teacher unions) on decentralization and the need for and design o fthe pilot. 0 Government to identify policy options to be tested duringthe pilot 0 Establish baseline data. The MOES needs to ensure that EMIS i s operational in the pilot municipalities and carries out a survey o f schools. If supplemental survey data is needed, it will needto be collected. 0 MOES to develop and implement a methodology for identifying the costs o f education, especially the additional costs for special populations. This will be usedto develop the formula for distribution o f financing to schools and LSGs. 0 Calculations on the extent to which individual schools and municipalities would be winners and losers and how to ensure that losers do not see their allocations fall too quickly. 76 0 Preparation by LSGs of school network rationalization plans, with technical assistance as necessary. 77 6. DECENTRALIZATIONAND WELFAREASSISTANCE TO SERBIA'S POORESTHOUSEHOLDS58 A. INTRODUCTION 6.1 Social welfare assistance in Serbia is fast approaching a turning point.59 Reforms inthis area o f public policy since 2001 have beenmore rapid and systemic than in many other countries in Europe and Central Asia. The Poverty Reduction Strategy (PRS) and the Ministry o f Labor and Social Policy's (MoLSP) Strategy for the Development of the Social Welfare System provide an important base for greater improvement. As the process o f integration with the European Union (EU) intensifies, Serbia will have to continue to modernize and increase the efficiency and efficacy o f its social welfare safety net. Serbia's accelerating decentralization o f functions and resources from central government to local authorities presents both opportunities and risks to this process. 6.2 There are two principal challenges faced by policy makers: to increase the poverty alleviation impact of targeted cash benefits; and to improve outreach and tailor socialwelfare services to the needs of vulnerable groups. Capturing the benefits o f shifting responsibilities to local actors while mitigating the risks that arise with decentralization, will be critical to meeting both challenges. The Ministry's Strategy emphasizes the increasing role that local self governments (LSGs) and non-government agencies have to play in delivering social welfare assistance. Decentralization o f social assistance has the potential to improve welfare by bringing services closer to households; allowing a wider set o f local providers to adapt services to meet the specific needs o f vulnerable groups that often vary substantially across municipalities; and improving targeting, outreach and impact monitoring by placing greater reliance on municipal authorities who are closer to beneficiaries. However, decentralization also entails risks, particularly for the most vulnerable. Without sufficient attention to available local resources and the capacity o f local government, decentralization o f social welfare assistance can lead to an erosion o f benefits, as regional experience has shown, and can deepen geographic inequalities ifpoorer municipalities are unable to maintainservices. 58This chapter draws from arecently publishedreport on Social Assistance and Child Protection in Serbia, producedjointly by the World Bank and the Ministry o fLabor, Employment and Social Policy inJuly 2006 (cited World Bank, 2006). Further input is drawn from "Options for Delivering Social Assistance: A Feasibility Study" which was completed inApril 2007 by Oxford Policy Management(OPM), with support from the United Kingdom's Department for InternationalDevelopment (DFID) (cited Hurrell, et a1 in the references). Further input was drawn from policy discussions with MOLESP, the Standing Conference of Towns and Municipalities (SCTM) and others, from July 2006 to May 2007 for a World Bank financed projectto supportDelivery of ImprovedLocal Services in Serbia. 59 Social werfare assistance refers to the set o f cash benefits and welfare services targeted to poor and vulnerable groups. 6.3 In this chapter the merits of decentralization for improving social assistance are weighed against the risks to the welfare of the poorest that arise when responsibilities for cash benefits and welfare services are transferred to local governments and non-government service providers. The specific form which decentralization o f social welfare assistance will take in Serbia i s not yet clear, and the authorities are looking for models that work as they prepare to draft a new social protection law. Arguably, recent dramatic steps taken to decentralize fimctions- discussed in other chapters o f this report-have had the least impact on how social welfare assistance i s delivered. However, even if no explicit mention i s made o f social assistance in recent reform legislation, there are definite implications o f the fiscal decentralization already passed for what LSGs can provide and what Serbia's poorest households and vulnerable groups should expect o f their local authorities. Thus while many details will remain undefined in the short term, the objective of the balancing exercise inthis chapter i s to identify the general principals that should guide the decisions o f policy makers in the medium term as they examine and evaluate proposed changes to the current social assistance legislation in Serbia's decentralizing environment. 6.4 Following this introduction, the chapter is divided into four additional sections. Section I1 provides a brief descriptive sketch o f Serbia's social welfare assistance system, making frequent reference to other sources where interested readers will find greater detail on the strengths, weaknesses and comparative performance o f the system. Section I11presents arguments for why the merits and risks o f decentralization are very different when considering in-kind social services on one hand, and targeted cash benefits on the other, drawing on economic theory and the experience from other countries inEurope and Central Asian. The section venturesto outline the likely division o f labor between Republican and municipal governments, as well as local non- government agencies, that Serbia i s moving towards given the changes brought by the 2006 Law on Local SeZfGovernment Finance. Section IV concludes with specific issues policy makers charged with improving social assistance against the backdrop o f Serbia's decentralization should keep inmind. B. SOCIAL WELFAREASSISTANCE SERBIA IN 6.5 Serbia has a fully developed social protection system. This includes contribution-based health insurance, pensions, and unemployment benefits (not covered inthis chapter), "last resort" social assistancecashbenefits, and child allowances, as well as an increasingly diverse array o f in-kind social services, which typically take the form o f institutional placement and care for vulnerable groups, such as the elderly, disabled and children without parental care. This section provides a short description o f the system and a summary evaluation o f its performance6'. Since 2001, the Government of Serbia has enacted major reforms o f social assistance, including, inter alia, means-testing o f previously untargeted child allowances; and changes to the eligibility criteria and 6oI t draws on recent analytical work completed by the World Bank, in partnership with MoLESP, in June 2006 (World Bank, 2006), as well as other comparative studies o f social assistance conducted by World Bank staff and senior specialists outside the Bank (see Andrews and Ringold, 1999, Matkovic, 2006, Hurrell,et al, 2007). 80 benefit levels of the main "last resort" income support cash benefit, the materijalno obezbedenje porodice (MOP). 6.6 Reforms of social welfare services include an improved legislative framework for foster care, steps toward "de-institutionalization", and the establishment of a Social Innovation Fund (SIF) to encourage community-driven innovations in service delivery. The SIF i s unique inthe region for its focus on supporting partnershipsbetween local self-governments and NGOs, and its objective o f scaling-up local innovations in service delivery into national policy. With financial and technical support from the SIF, local authorities in Serbia are able to pilot alternative models of care for vulnerable (the elderly, disabled and minors without parents), and socially disconnected groups (such as Roma, internally displaced persons, or IDPs, and refugees) in partnership with non- government providers. A number o f these innovations have been found to be successful and adopted at a national level, for example Serbia's program for care of battered women andminors. 6.7 The annual allocation from the central government "Republican" budget for social welfare and child assistance has varied since 2001 between 0.99 and 1.7 percent of GDP. In 2005, the allocation for social welfare and child assistance amounted to1.4 percent of GDP and 3.8 percent o f the budget. Table 6.1 shows how funds have been allocated between cash benefits and services. O f this amount, what is strictly considered targeted social welfare assistance (in Table 6.1, "Social Assistance" and "Child allowance") i s about 0.77 percent o f GDP. Table 6.1: Spendingon Social Welfare Assistance and Child Protectionin Serbia, 2005, percentageof GDP Total 0.99 1.68 1.64 1.50 1.35 1.40 Child Protection 0.75 1.41 1.40 1.27 1.11 1.07 Wage compensationduringmaternity 0.21 0.24 0.35 0.40 0.41 0.43 Child allowance 0.03 0.91 0.74 0.55 0.50 0.44 Birth grant (parental allowance) 0.11 0.23 0.22 0.20 Educationalprogrambefore first grade 0.09 0.10 0.07 -- Others 0.05 0.10 0.06 0.07 -- Social Assistance 0.24 0.28 0.25 0.24 0.24 0.33 MOP 0.02 0.09 0.10 0.11 0.11 0.14 Caregiver's allowance 0.03 0.04 0.05 0.05 0.06 0.09 Social Institutions 0.10 0.05 Assistancefor occupational training -- __0.06 __0.08 -- _0.05 _0.06 _ _ 1.57 Others 0.09 0.10 0.01 0.02 0.02 0.03 Source: MoLESP, updatedfrom World Bank (2006). 81 6.8 Focusingon targeted cash benefitsin particular,comparedwith its neighbors in Europe and Central Asia, Serbia spends a relatively small sum. Figure 6.1 shows spendingon targeted cash benefits (both "last resort" income support benefits targeted to the poorest, and other targeted non-contributory benefits, such as child and care-givers allowances) as a percentage o f GDP in Serbia and the EU15 (using data from about 2003/2004, the last years for which a large set o f comparative indicators are available).61 Considering targeted cash benefits as a whole, Serbia spends exactly half o f the EU average. However, when only "last resort" income support benefits purely intended for poverty alleviation are considered, Serbia's spending looks particularly low. The spendingpicture does not change when a larger selection o f countries i s included in the comparison (see Figure 6.2). Figure 6.1: Spending on MeansTested Cash Benefitsin Selected EuropeanCountries, circa 2003 Other mans-tested cash benefits (% of GDP) Means-tested last resort income support (% of GDP) 6.9 The large Government contributionto the pension system has an important impact on poverty. Although not strictly among the "social welfare assistance'' programs Serbia has in place, contributory social insurance (primarily old age, disability and survivor pensions) play a similar role in parallel. Despite being designed as contributory social insurance, the Government o f Serbia spends between4 and 5 percent of GDP covering the deficits betweenbenefitpay-outs and income from the contributions o f current workers - the deficit of the pensionsystemtends to be lower inother countries. Furthermore, as World Bank (2006) shows, without pensions, the poverty rate in Serbia would be 30 percentage points higher. Therefore, attempts to measure the Government's efforts at alleviating poverty against regional and international spending benchmarks should be considered with this inmind. 6'Readers should note that because Serbia and Bosnia are the only two countries among those compared that have not adjusted their GDP for the non-observed economy, the ratios actually overstate spending in both countries. 82 Figure 6.2: Spending on Means-Tested "Last Resort" Cash Benefits, Selected EuropeanCountries, ................. . . . . . . circa 2003 ................... -....... . . . . . . . . . . . . . . . . . . . . ............................................................................................ 1I ~ ............................................................................................. i ~ ...................................................................................... ....................................................................................... .............................................. 6.10 The money Serbia spends on social assistance cash benefits is relatively well targeted to the poorest households. An analysis o f targeting and coverage o f cash benefits across countries in Europe and Central Asia (Figure 6.3) shows that benefits in Serbia are better targeted than in most ECA countries, with a larger share o f funds successfully directed to the poorest quintile than inPoland, Bulgaria, Hungary, Lithuania, Romania, and others (vertical axis). However, Serbia does not achieve the same extent o f coverage o f the poorest: Serbia's targeted cash benefits cover only half o f the poorest quintile (horizontal axis). For the sake of comparison, targeted cash benefits inRomania cover 90 percent o f households inthe poorest quintile. Figure 6.3: Targetingand Coverage of Non-contributorySocial Assistance Cash Benefits, Europe and Central Asia, 2003 Performanceof nonsontrlbutorybenefits,Includingprlvllogrs 0 20 u) Bo Bo 1w coverage of the poorest qulntllr (W) ECA Targeting PelTormance,WorldBsnk Source: World Bank staffestimates. 83 6.11 There are important differences in the targeting performance across cash benefit programsin Serbia. Of the main cashbenefit programs, MOP is by far the best targeted. About 66 percent of MOP spending went to the poor in 2003 (Table 6.2). In contrast, 72 percent of the child allowances went to the non-poor. Although it reaches fewer poor households, the MOP is better targeted and suffers less leakage to non-poor households. Only 16 percent of child allowance transfers went to the poorest quintile, while 54 percentof MOP benefits went to the poorest 20 percent ofhouseholds. Table 6.2: Incidence of the M O P and Child Allowances by Quintile ( O hof total transfers receivedby each sub-group) Non- Q1 (inYO) poor Poor (poorest) Q2 43 44 Q5 (richest) MOP 33.6 66.4 53.6 23.1 16.9 0.0 6.4 Child Allowance 72.1 27.9 16.1 32.6 23.4 15.3 12.2 Source: World Bank (2006). 6.12 Almost three quarters of resources for child allowances went to non-poor households in 200362.The concentration curves below show the targeting performance of the MOP relative to other social welfare transfers (Figure 6.4).63With the exception of child allowances, all transfers are progressive, that is, the poor get a greater share of transfers than their share in total consumption. The MOP is clearly the most progressive amongthese. Figure 6.4: Concentration Curves for Selected Social Welfare and Child Protection Programs, 2003 :: 100 { 'd 70 60 50 0 - 0 10 20 30 40 50 60 70 80 90 100 cumulative percentage of the population -+-Qual MOP t-ChildAllowance *SWCP Social Insurance -e- Social Rotection -t- Carer's Allowance Source: World Bank (2006). 62World Bank (2006). 63Concentration curves are graphic tools used to illustrate whether a transfer is progressive(that is, when the poor get a higher share of the transfer) or regressive (the opposite case). If the concentration curve of a program lies above the equality line, the program is highly progressive: the poor recipients get more benefits than their share in total consumption. 84 6.13 At present, decentralization of social assistance benefits and services is limited. The Republican budget covers about 90 percent o f all spendingon cash benefits and social welfare services. Almost all cash transfers are financed from the national budget. Further, most welfare services are also centrally financed, including residential care institutions, foster care allowances, and services for the disabled. However, local self governments assist the central government in administrating cash benefits. The MOP and child allowances are currently administered and delivered by central-government and municipal agencies in parallel. The MOP i s delivered by local agencies affiliated with MoLESP, the Centers for Social Work (CSWs) while child allowances are delivered by LSGs. Hurrell, et al, (2007) conducted extensive quantitative and qualitative review o f how both cash benefits were delivered at the local level, and found substantial under- coverage of the poor, as well as evidence o f leakage to the nonpoor (Box 6.1). Box 6.1: Design and Implementation Problemsin the Delivery of SocialAssistanceCash Benefitsin Serbia Using Household Budget Survey (HBS) data, the authors (of the OPM study) were able to determine whether households meet the eligibility criteria and therefore they are entitled to child allowances or MOP. Eligibility for benefits was checkedagainst whether or not the householdsreportedreceivingthem. The combinationof eligibility and whether a household received benefits was used to identify two sub-groups: those eligible households who do not receive the benefit (non take-up) and the non eligible householdswho do receive the benefit (administrativeleakage) (Figure 2) Householdswere also ranked for their level of relative poverty based on their consumption expenditure to assess the extentto which eligible householdsare poor or not (Figure 3). I I Figure2: ImplementationIssues:Eligibility Figure3: DesignIssues:Eligibility for andReceiptof Benefits Benefitsand Poverty I 1 Design leakage V A Recipient Undercover Amongthe findings with respectto implementationand designproblemsof the benefitprogramsinclude: (i) Sixty three percent o f eligible households do not receive child allowances and forty three percent of non-eligible householdsdo receivethem; (ii) The high non-take up of child allowances is related to high direct and indirect costs associated with applying for child allowances, such as payingfor documentation,transport costs, opportunitycosts and social costs in the form of stigmatizationsof certain populationgroups; (iii)Potentialbeneficiariesare under-informedaboutthe criteriafor eligibility to entitlements,thusdecidingto apply constitutes ahigh cost investmentwith a highly uncertainpay-off); (iv) Non take-up is not only a problem limited to certain minority groups, namely the Roma, but also due to misconceptionson eligibility criteria, and applicationcosts; (v) Administrativeleakage of benefits to non-eligiblehouseholdsis also a problem; (vi) By comparingthe characteristics of eligible households with those of recipients, it emerges that the discretion exercisedby social workers excludes many householdsin away not originally intended by the eligibility criteria; and (vii) In terms of design, the study finds that while the eligibility criteria are relatively well targeted to the poor with 45 percent of all eligible people are in the first decile, the coverage of eligibility criteria is indeed relatively limited. Seventy three percentof the peoplein the first decile are not eligible for MOP. Source:Hurrell, et al, 2007. 85 6.14 In addition to responsibility for delivering child allowances financed by the Republican budget, local governments have responsibility for meeting "extended rights". 64 This includes social services such as home care assistance, day care, temporary accommodation in shelters, foster care, and a range of community-based welfare services for other vulnerable and socially disconnected groups. While the bulk o f these services are financed and formulated at the central government level, municipalities and towns may opt to provide additional services or supplementary resources to beneficiaries. The level and extent to which LSGs offer these additional services, which complement national level programs delivered through the CSWs, will depend on their financial capacity. 6.15 However, although they can offer additional services, few do. Municipalities do not typically spend a significant amount o f resources on social welfare assistance. In 2004, the majority o f LSGs spent less than 2 percent o f their municipal budgetson social welfare, mostly on one-off emergency cash payments. In 2005 local governments planned to spend an average of only 1.8 percent o f their budget funds for social welfare. Furthermore, LSG spending on social welfare assistance varies widely from 0.1 percent to 5 percent o f municipal budgets. Not surprisingly, the richer cities and municipalities spend more on social welfare than poorer LSGs. Table 6.3: Social Welfare Assistance in Republican and Municipal Budget, 2005 (Billions of Dinars) Budget Funds for Share of PopulationBudget per Budget for Budget for 2005 social social (in capita social social (allocated welfare welfare in millions) (monthly welfare perwelfare per in bill. Din) (allocated the budget in dinars) capita capita in bill. Din) (monthly (Euros) in dinars) Republican 416.6 15.6 3.8% 7.5 4.63 174 2.07 budget allocation Total 106 1.9 1.8% 7.5 1.17 21 0.25 municipal budget allocation Total 522.6 17.5 3.3% 7.5 5.80 195 2.32 6.16 In 2004, the ratio of municipal contributions to the budgets of CSWs varied from 0.4 percent (Aleksinac) to over 50 percent (Kragujevac, Temerin) (Table3). The City of Belgrade financed additional and one-off benefits and services (63.1 percent), residential care (17.4 percent), and CSWs (19.5 percent) in2005, and continues to spend significantly more than other cities. 64The Law on Social Protection and Provision of Social Security of Citizens. 86 Table 6.4: Table3. Social Expendituresby Municipality, 2004 Republican Ratio of the Ratio of Social (MOLSP) Municipal Welfare Fundsin Contributionto the Contributionto the Local Budgets CSWs Budgets the CSWs (percent) Municipality Population (percent) Budgets (percent) Belgrade-the City6' 1,574,050 51.26 46.01 2.66 Kragujevac 175,182 38.47 55.29 0.21 Novi Sad 298,139 67.22 23.39 0.99 Nis 250,180 59.29 32.28 0.50 Bor 55,739 63.91 20.07 0.60 Aleksinac 57,044 94.58 0.44 1.75 Source: Annex 6, Strategy for Development of Social Welfare: Key Issues in Funding and Financing, DFID supported project on Social Policy Reform: Building and Strengthening State-Civil Society Partnershipsto ReducePoverty and Social Exclusionin Serbia, Belgrade, December2005. 6.17 In sum, Serbia's system of social welfare assistance provides a comprehensive array of benefits and services. However, for many households, particularly the poorest and socially disconnected, these remain largely promises. Serbia spends less on targeted cash transfers to the poorest than other countries in Europe and Central Asia, although the large volume o f spending on non-targeted benefits paid by a deficitory social insurance system have to be taken into account given their inevitable impact on household incomes, even among the poorest households. And while its poverty relief transfer, the MOP i s well targeted, it i s covering only a fraction o f the households who would qualify for the benefit. Furthermore, there i s a clear evidence of considerable leakage o f resources to non-poor households, particularly in the larger program o f child allowances. Finally, although they hold some responsibility for offering welfare services to vulnerable and disconnected groups, LSGs are spending only a meager fraction on social welfare assistance, and offering households only a minimum level o f service. A principal objective o f the MoLSP Strategy i s to use decentralization as a force to improve this performance. c. DECENTRALIZING SOCIAL ASSISTANCE: FROMTHEORY PRACTICE TO 6.18 Decentralizing responsibility for services to local governments can improve efficiency, flexibility, innovation and responsiveness (World Bank, 2004, Matkovic, 2006). Decentralizing responsibility for service delivery can improve outcomes by encouraging competition betweenlocal authorities and betweenan enlarged set o f private service providers in a given locality. Decentralizing services can also improve outcomes by opening the possibility o f local innovation and differentiation o f services to meet the particular needs o f specific groups, which often differ widely across regions and municipalities. Furthermore, local authorities can bring greater knowledge to service 65Seventeenmunicipalitiescomprise the city o fBelgrade. 87 delivery, which is particularly useful in accurately targeting services to the households that needthese most, reaching out to socially disconnected groups, and to closely monitor the impact o f services. Finally, by placing responsibility for services on the shoulders of locally elected officials, decentralization can lead to improved services by increasing accountability to households and the pressure these households can bring to bear on local government and non-government service providers when performance i s unsatisfactory (World Bank, 2004, Matkovic, 2006). 6.19 Decentralizationcan take three specific forms. The first, and most complete transfer o f responsibilities i s called devolution, by which central government transfers the authority for decision making, financing and management o f services to politically autonomous units o f local government. The second, more limited transfer i s called de- concentration by which national governments transfer activities and responsibilities of central ministries and agencies to local offices o f these agencies, but retain control over financing and formulation o f policy. The third and most limited transfer is called delegation, by which local governments are given purely administrative responsibilities for deliveringpublic services, but are subject to supervision by the central government. 6.20 There are at least four important political and economic dimensions to consider in designing decentralization. However, even where responsibility is decentralized to highly capable local governments, when considering social assistance in particular, two fundamental distinctions have to be made. The first fundamental distinction i s betweenfinancing and administration, and the second i s between targeted cash benefits and social welfare services. These distinctions can help determine what social assistance functions can be improved, and to what extent, by shifting responsibility to the local level, and should be kept in mind when considering arguments along four principal economic and political dimensions. 6.21 The first dimension is the negative externality of poverty and the corresponding public-good nature of social assistance. Many public goods (Le., government interventions to increase positive externalities and correct negative externalities) such as national defense and macroeconomic management, can only be effectively provided uniformly. The public-good nature o f social assistance-that is, its primary function to alleviate poverty and ensure minimum consumptions and social welfare-also requires a strong degree o f uniformity. When considering minimumlevels o f publicly-financed support, it i s difficult to justify disparities across localities or discrimination in the level and quality o f provision based on place o f residence (Bird, 2000). Some authors even speculate that allowing differences in social assistance provision could lead to a "race to the bottom" as local authorities seek to discourage settlement by the poor and vulnerable by lowering cash benefits and cutting services in their area (Matkovic, 2006). 6.22 The second dimension is the economics of insurance, risk management, and savings from scale. If considered "poverty insurance", social assistance is most efficiently financed from the largest possible "risk pool". Decentralization can fragment the risk pool, if responsibility for financing benefits is devolved and the tax base on which the social assistance system is built i s constrained. This would create smaller, less 88 efficient risk pools in which there are not an actuarially sufficient number o f "winners" (i.e. net contributors to the pool) supporting beneficiaries (i.e., net recipients from the pool) (Barr, 1992). In a related point, there are substantial savings from scale from centralizing provision o f social assistance, which could be lost. Smaller municipalities would bear higher marginal costs to deliver the same minimum package o f services than larger municipalities. These higher administrative costs could lead to lower level and quality o f benefits and services insmaller, poorer communities. 6.23 The third dimension is the argument for diversifying the set of service providers and capturing the benefits of market forces. As with education and health care, gains in efficiency and quality can be had from greater competition and innovation. The education and health policy literature advocates a "split" betweenthe purchase and provision o f publicly financed services (which are typically-but not necessarily- bundledtogether in, large centralized systems of public service provision), that allows a larger and more diverse set o f service providers to compete and for households to choose between them. Similar arguments can apply to social assistance, particularly to the provision o f assistance above a socially defined minimum package. These arguments, however, are more valid with respect to social welfare services, than to targeted cash benefits, as there i s greater scope for differentiation across service providers and thus a more obvious basis for competition (Barr, 2001). 6.24 Finally, there are important political considerations that have to be kept in mind. In contrast to services that benefit the majority o f households, such as education and health, social assistance i s targeted to the most vulnerable. Not only is this group often a minority inany given community interms o ftheir relative and absolute number, it also typically consists o f those households who are the least able to exert political pressure, due either to their poverty, social disconnection, or the prejudices o f the majority and local elites. Thus, without the right mechanisms to strengthen accountability and participation, local governments have fewer incentives to respond to demands for better funded and higher quality services for the poorest and most vulnerable. Indeed, local authorities may have strong incentives to divert funding away from social assistance to services enjoyed by the majority or more powerful groups in their constituencies. 6.25 In sum, economic and political theory presents several arguments that distinguish social welfare assistance from other services when balancing the merits and risks of decentralizing responsibilities to local government. While these arguments apply to social welfare assistance as a whole, they indicate that the distinctions - betweenfinancing and administration on the one hand, and between cash benefits and welfare sewices on the other, should not be ignored when taking decisions about what to decentralize and to what extent that is, devolution, de-concentration, or simple delegation. 6.26 To recap, the public good function of social assistance requires a substantial degree of uniformity to meet its objective of providing a minimum level of welfare. While uniform level and quality is important for a minimum package o f both cash benefits and services, above this minimum package the need for uniformity can be 89 relaxed, particularly in the provision o f services for which there i s greater need for innovation to meet locally-specific challenges. Actuarial principles suggest that targeted social assistance should be financed from the largest and most diverse risk pool available, national general revenues. As above, while critical for financing the minimum package o f social assistance benefits and services, this i s arguably more applicable to the provision o f minimum cash benefits. And although there are important gains to be had from competition between a wider set o f providers, these gains are most likely to be reaped from innovation and competition inthe provision o f welfare services, where there i s greater scope for differentiation. Finally, the exclusion suffered by the poor, vulnerable and socially disconnected, i s a strong challenge to the argument that shifting service delivery to lower levels o f government will increase the accountability o f service providers to households. The political economy o f providing social welfare assistance is, in this way, very different from the provision o f education and health, and requires carehl monitoring and regulation by central government to protect the rights o f disenfranchised citizens. 6.27 Taken together, these theoretical arguments drawn from the policy literature make a strong case for maintaining centralized financing of at least a uniform minimum package of cash benefits and social welfare services. The benefits o f local knowledge can be captured by giving local authorities responsibility for administration (inparticular, targeting, delivery, and monitoring); responding to local needs by ensuring scope for local innovation; reaping the gains from competition by diversifying service providers to include non-government agencies; and establishing structures to ensure accountability to the poor, vulnerable and socially disconnected groups. 6.28 The wide range of country experiences with decentralizing social assistance in Europe and Central Asia, generally reflect and are consistent with the arguments above. Topinska (1996) notes that the efforts to restructure social assistance in Europe and Central Asia inthe 1990s reliedto a great degree on decentralization, and particularly on increasing local participation inthe financing and administration o f benefits, as well as enlisting greater support for vulnerable groups from families and non-governmental organizations. The experience has beenmixed. This i s mainly because decentralization in the region in most cases was motivated less by the primary objective o f improving service delivery, and more by secondary objectives. These secondary objectives range from coping intimes o f economic and financial crises; devolution o f fiscal responsibility and contingencies; reactionary rejection o f any form o f central planning after the fall o f socialism; and even the appeasement o f ethnic and religious minorities with aspirations for political independence (Matkovic, 2006). Therefore, when decentralizing mandates and functions to local authorities, national governments in the region have often paid little attention to the arguments made in the policy literature, especially in the case o f social assistance. 6.29 There are several notable examples of responsibility for social assistance being fully devolved with little care or thought given to adequate financing arrangements. In Romania, decentralization of social assistance and child allowances lead to a deterioration o f outcomes. After passing a law on local public finance in 1998 which devolved responsibility for social assistance to local governments, total spending 90 on social assistance and child protection-including benefits and services for the disabled and elderly-fell by nearly 40 percentage points inreal terms between 1996 (when fiscal decentralization began) and 1999. This resulted in a 16 percent decline in cash benefits from their level in 1996. Because responsibility for providing social assistance was devolved without adequate concern for fiscal decentralization, a lack o f financial resources at the local level forced municipalities to scale back provision o f benefits and services, and in some cases, to stop providing services altogether. To repair the damage to poor households left in the wake o f Romania's decentralization in the late 1990s, the country recentralized social assistance financing in 2002. In Bosnia and Herzegovina, decentralization o f social assistance was plagued by similar problems, and lead to similar negative outcomes. Decentralized financing resulted in substantial inter-regional disparities in coverage, with poorer localities providing the least amount o f services. In the face o f resource constraints, eligibility criteria for most benefits inBosnia became ad hoc, as local welfare offices relied on their discretion to ration scarce municipal resources. 6.30 Regional experience suggests that the performance of social assistance in decentralized environments depends on the whether a country is largely homogenous in social and economic terms with small differences in regional incomes, or one where there are substantial social and income differences. Milanovic (2000) notes that inRussia, soon after the breakdown o f the Soviet Union, the central government's financial role was weakened contributing to its inability to allocate adequate financial resources for social assistance. Under this circumstance, the central government had little influence over the distribution o f social assistance across the autonomous republics, oblasts, or cities. As a result, social assistance became entirely a local matter: locally financed, and with benefit eligibility and benefit levels wholly under local control. This led to a wide dispersion o f the coverage and levels o f social assistance inthe country. 6.3 1 In Hungary, decentralization of service provision proved more successful than decentralization of cash benefits. There have been gains from innovation and competitions between providers, as predicted in the policy literature. The Hungarian social assistance system was decentralized following the 1990 Act on Local Self- Government. The Act gave almost full control over social assistance to newly formed local councils, and the central government remained in charge o f financing. While Romania provides an example o f devolution, Hungary's case is an example o f de- concentration. The transfer o f responsibility for welfare service provision brought greater flexibility, which spawned innovation and the introduction o f new services, as well as new service providers, who have proved more responsive to households. By the end o f the 1990s, with decentralization of social welfare services well entrenched, there were over 5000 non-government organizations providing services to poor and vulnerable groups, under the oversight and regulation o f central government authorities (Andrews and Ringold, 1999). 6.32 Regional experience also illustrates the benefits expected of decentralization, listed at the start of this section. In Albania decentralization was found to have had a positive impact on the delivery o f social assistance. Alderman (1998) finds that local 91 officials in Albania had more information on households' welfare than was available to central government authorities, and that the allocation o f social assistance among households by local authorities was much better targeted than the allocation o f social assistance funds by the central government. After decentralization, Alderman found considerable improvement intargeting with more than half o f social assistance was going to the poorest quintile. 6.33 A recent review of regional experience with the decentralization of social welfare assistance concludes that there are few theoretical arguments in favor of decentralizing financing. The review (Matkovic, 2006) suggested that there are many concrete examples to illustrate the dangers o f fragmenting the larger risk pool provided by the national public finance system, and forgoing the economies o f scale from centralized provision. Further, a rich, practical experience with assigning social welfare mandates and functions to local governments in Europe and Central Asia, indicates that the greatest advantages from decentralization are to be had by giving municipal authorities greater administrative responsibilities over cash benefits (in particular, delegating responsibility for targeting and other administrative functions), and much greater autonomy inthe provision o f welfare services (de-concentration). 6.34 Several important pre-conditions must be in place for decentralization-even along these clearly limited lines-to succeed in improving social assistance. Matkovic identifies these pre-conditions as: (i) the rights and eligibility requirements that for social assistance be clearly defined and knowledge o f these rights be improved among households; (ii) that structures to encourage innovation and differentiation o f services be in place; (iii) structural reforms, such as the shift away from institutional care, be that planned, legislated and well underway prior to the transfer o f functions; (iv) that quality standards be set, and mechanisms developed for these standards to be monitored by local and central government authorities; and (iv) that government agencies, at the local and national level, be strengthenedand ready to take up their new responsibilities. Matkovic concludes that decentralization o f mandates and responsibilities without these conditions in place can lead to the development of a myriad o f models and levels of provision, the lack o f standards, abuse and misuse o f resources, and ultimately to a deterioration o f social welfare assistance for the poor, vulnerable and socially disconnected. 6.35 Increasingly, the Ministry of Labor and Social Protection is likely to focus on regulation and quality assurance rather than administering and delivering cash benefits and services. Figure 6.5 charts the current and likely future division o f responsibilities for social welfare assistance inSerbia, drawing on the MoLSP strategy, as well as the theoretical arguments presentedin the policy literature. As shown, the most substantial likely change i s a shift in the role o f MoLSP out o f administering and delivering cash benefits and services and into regulation and quality assurance. In the future model, a minimumpackage o fbenefits and services would continue to be financed from Republicanbudget resources. However, administrative responsibility i s likely to be fully delegated to LSGs, who inturn would play a supervisory role over non-government service providers intheir areas. 92 Figure 6.5: Current and Likely Future Divisionof Responsibilities for Social Welfare Assistance in Serbia Financing Administration and Regulation and quality delivery assurance - Cash Republicanresources MoLSP CSWs MoLSP h benefits LSGs c) Minimum Republicanresources MoLSP MoLSP Z& weifare LSG resources LSGs . > 3 services U Additional LSG resources LSGs LSGs services Cash Republicanresources LSGs MoLSP 5 benefits Weifare Republicanresources LSGs MoLSP & LSG c1 services transfersto LSGs regulationofnon- S government providers CI Additional LSGresources/cost LSGs LSGregulationof non services recovery government providers Source; Based on World Bank (2006), MoLESP(2006) and Matkovic(2006). 6.36 An important policy question is whether the poverty impact of transfers could be improved through consolidation of the two principal benefit programs, MOP and child allowances. As explained inSection 11, the MOP is administered by the Centers for Social Work (CSWs), de-concentrated offices o f the Republican MoLSP, and the child allowances run by local self governments (LSGs). World Bank (2006) argues that this division o f parallel delivery responsibilities between republican and local self government agencies imposes significant costs, including duplicative administrative structures, staff and office space. Parallel institutions for the means-tested benefits also impose costs on beneficiaries who need to present separate applications and sets o f documentation for each benefit, and on providers who end up spendinga disproportionate amount o f time on paperwork. The report argues that decentralization offers an opportunity to redefine the role o f LSGs, possibly increasing their responsibility for delivery o f a consolidated targeted cash benefit. Consolidation o f targeted cash benefits could increase administrative quality (Le., minimize non take up o f benefits, as well as administrative leakage), and increase administrative efficiency &e., minimize per beneficiary andper-applicant administrative costs). 6.37 A recent feasibility study finds that importantperquisitesare needed before consolidation will yield significant benefits. With support from DFID, MoLSP has taken the critical first step o f launching a feasibility study of options for consolidation, including the possibility o f a single cash benefit for poor families.66 The study, carried 66 The PRS, launched in 2003, proposed"the integrationo f the social assistance and child allowances into a single unique transfer that could be called a family allowance." The idea o f a single consolidated welfare benefit which would take into account various household characteristics in determiningbenefit levels - - has been gaining currency in other countries. A single benefit would have one set of rules, administrative and institutionalrequirements, and source o f funding. It could be more transparent, low on administrative costs, and easy to communicate and understand.A single benefitcould further incorporatethe objective of providing increased support to families with children through a benefit formula that would provide higher benefits based on the number and age of children in the family. Such an approach is currently used in Bulgariawith its guaranteed minimumincome benefit system. 93 out by OxfordPolicyManagement(Box 6.2) answers two critical questions:(i) will what the benefits be of consolidatingMOP and child allowance administrative systems?; and (ii)how shouldadministrative consolidationbe implementedso as to maximizethe benefits while minimizing the costs? The study concludes that administrative consolidationon its own will not result in a significant improvement in administrative efficiency (due to what the authors call "legislative incompatibility" of the two benefits, which introducesfundamentaldifferencesinhowthe benefitsare targeted), andonly very marginalimprovementsinadministrativequality. 6.38 This suggests moving forward with administrative consolidation only after the legislative incompatibilities between the two benefits are removed. This would also allow time for the other prerequisites discussed above to be put into place. The study shows that the most feasible option for administrative consolidationwould be to move all aspects of MOP administration to LSGs, except for the responsibility for making home visits and the subjective income assessment (which would remain the responsibilitiesof CSW social workers) and recommendsthat the Ministry proceedwith pilotingof the proposedmodelto further determine feasibility inthe runup to legislative reformsexpectedin2009 (Hurrell, et al, 2007). Box 6.2: Administrative Consolidation of Serbia's Cash Benefits for Poverty Reduction What is the most feasible option for administrative consolidation? The OPM study considered several options and found that the most feasible was to move all aspects of the MOP administration to the LSGexcept for the responsibility for making home visits and the subjective income assessment, which would remain with the social workers at the csws. Why was this option considered more feasible than others? The study recognized that to the extent that social workers allow ineligible households to receive MOP (the incidence of which is unclear) administrative consolidation could reduce administrative leakage by removing the discretionary elements of the eligibility assessment from social workers to potentially more neutral LSG administrators. This option for administrative consolidation could facilitate the extension of home visits to child allowance applicants in order to confirm the validity of application information. The proposed model for administrative consolidation could also bring additional indirect benefits due to improved communication. Ifchild allowanceswere moved to the CSW or, ifMOP moved to the LSG, information on particularly the poor child allowance recipients may be communicatedfrom the LSGto the CSW. But even if this model is adopted other challenges remain which need to be addressed. Regardless of which institution takes on the administration of an additional benefit, there are several challenges that may impede the realization of the potential gains of the most feasible option. These include: (0 Likely variations in the financing, and thereforethe quality, of MOP administration across municipalities; (ii) LSGwill require additional staff- will non-social worker MOP staff simply be re-deployed?Ifnot, new staff will require training (who will instigate and pay for training?). In addition, there are concerns about the qualitieskapabilities of some LSG administrative staff to undertake tasks effectively; (iii) Improvedcoordination andcommunicationbetweenCSWandLSGwillberequired;and (iv) A lack of spare office space to accommodate the new staff is likely to be an issue in both the LSG and the csw. What more is required to achieve better targeting of benefits?The study points out that improving targeting of MOP and child allowances to the poorest households cannot be achieved by administrative change alone. Several reforms are proposed, including: (i)need to computerizeMOP administration procedures; (ii) step up efforts to reduce the direct and indirect costs of applying; (iii)issue and enforce official guidelines, norms and standards in a more systematic manner; (iv) increase the awareness of entitlements to MOP and child allowances with clear information provided regarding the precise details of the eligibility criteria and the required application documents; and (v) the legislations should be unified to ensure that the definitions are the same and that the poverty lines are the same. In addition, the documentation requirementsshould be reduced and simplified. Source: Hurrell, et al, 2007. 94 D. MITIGATING RISKSOFAND REAPINGTHE GAINSOFDECENTRALIZATION THE 6.39 Moves to decentralize provision of social assistance introduce considerable risk. Decentralization should be accompanied by measures to ensure sustainable financing; structures for greater coordinating betweencentral and local self governments; mechanisms to establish, update and monitor quality standards; and regulation that will allow diversity in service provision from the NGO and private sector. In Serbia, therefore, while there i s considerable momentum to increases local autonomy, it i s important that the central government continues to not only ensure a regulatory framework for social welfare assistance, but also ensure that the responsibilities for benefitprovision are matched with adequate fiscal support. 6.40 Serbia has already taken promising steps in managing these risks. The changes legislatedwith the 2006 Law on Local Self Government Finance are expected to reduce regional differences in local government revenues. If the financial resources earmarked for social assistance at the local level are also scaled up (and the local budget process i s transparent), and if better targeting o f the social transfers i s in place, then decentralized social assistance - the delegation o f administration and delivery o f cash benefits and de-concentration o f welfare services with new non-government service providers - i s likely to lead to improved outcomes. The remainder o f this concluding section outlines the main points that policy makers should keep inmind. 6.41 Ensuring sustainable financing. Lessons from other countries in the region underscore the need to maintain central budget financing for social welfare until a well- functioning intergovernmental fiscal system that ensures adequate, sustainable and equitable financing o f benefits and programs across municipalities i s in place. Without earmarked financing sources, there i s a high risk that funds designated for social assistance will be diverted elsewhere. There i s also a risk that the poorest municipalities with the greatest needwill have the most limitedresources available for social assistance, leading to a vicious circle in which funds are not available to those who need them the most. Financing from the national budget should be retained at least for a minimum package o f social welfare assistance (cash benefits and welfare services), that could then be complemented with additional services financed from municipal budgets. 6.42 Coordinating between central and local governments. Stronger partnerships and better coordination between the central government, LSGs, and improved coordination across sectors (e.g. social protection, employment, education, health, police, justice, etc.) i s important for improving the delivery o f social welfare assistance. Implementation o f the MoLSP Strategy will provide a good opportunity for inter-ministerial cooperation and facilitate more integrated social policy at the local level. 6.43 Implementing and monitoring quality standards. The relative roles o f central and local governments in implementing quality standards need to be clearly defined to ensure a framework for improvements through: strengthened incentives, shared interpretation o f the standards, training and education o f staff, periodic assessments, codes o f ethics, and quality management tools. Performance indicators should be used to assess services and the impact o f cash transfers. Monitoring and evaluation o f social 95 welfare programs has beenweak and irregular inthe past and the capacity o f central and local governments has to be strengthened through capacity building. 6.44 Exploiting local innovations in service delivery. In the area o f social services, there i s scope to build on the body o f existing experience o f local involvement in design and implementation. The MoLESP's Social Innovation Fund (SIF) has substantial experience in supporting innovative community based services to target vulnerable groups, including children-at-risk, the disabled, elderly and Roma. This includes innovations that take an integrated approach, combining interventions insocial assistance with the delivery of basic health and education services. Lessons from the SIF can inform the on-going effort to shift social care from institutions to community-based care, particularly inthe context o f growing involvement o f local governments andNGOs. 6.45 Engaging LSGs to ensure registration of Roma IDPs. Roma households, and particularly Roma IDPs often lack the most basic form o f documentation - residential and citizenship documents. Tackling the lack o f citizenship and residential registration i s challenging given complex associated legal dimensions. Authorities need to intensify efforts to ensure residential registration o f Roma, including those residing in settlements and in particular Roma IDPs, to ensure access to municipal services even without legalizing such settlements. LSGs have a critical role to play inthis outreach at the local level. 6.46 ...and other poor and vulnerable groups. A limitation o f the current system i s the absence of opportunities and incentives for CSW social workers to visit poor households to better respond to their needs. Lack o f information about the program i s a barrier to applying for some poor households. Outreach should be an integral part o f delivery, to facilitate access to both cash benefits and social services, and particularly to meet the special needs o f vulnerable groups such as Roma, IDPs and refugees. LSGs are uniquelypositionedto take the leadinensuringoutreach to all vulnerable groups. 6.47 All of the above should be accompanied by institutional reform measures to streamline and simplify procedures for delivery of benefits and services, in order to increase the efficiency and quality o f social welfare assistance. Such measures-at the Republican and the municipal level-would include upgrading information systems, improving survey data, and improving communication to the public to increase awareness o f the social welfare system and the rights and responsibilities o f Serbian citizens under this system. 96 REFERENCES Alderman, H. (1998), "Social Assistance in Albania: Decentralization and Targeted Transfers. LSMS Working Paper no.134, World Bank, Washington, DC. Andrews, Emily S. and Dena Ringold (1999) Safety Nets in Transition Economies: Toward a Reform Strategy, Social ProtectionUnit, HumanDevelopmentNetwork, The World Bank Asfaw, A., Klaus Frohberg, K.S. James, and Johannes Jutting, (2004): Modeling the Impact of Fiscal Decentralization on Health Outcomes: Empirical Evidence from India. ZEF - Discussion Papers on Development Policy No. 87. Center for Development Research,Bonn, Germany. Bagdy, G. (1996), "Policies for national consensus and reform", in Finding the Balance: Financing and Coverage of Social Protection in Europe, eds. van Ginneken, Occassional Papers on Social Security, InternationalSocial Security Association, Geneva, July 1996. Barr, N.(2001) The Welfare State as Pimv Bank, Oxford Bach (2000): Decentralizationand privatization in municipal services: The case of health services. ILO, Geneva. Belli, P (2002): Formal and Informal Household Spending On Health: A Multi-country Study in Central and Eastern Europe. Harvard School of Public Health. Central and Eastern European Health Network. Cambridge, Massachusetts. Bird, R. (2000) "Intergovernmental Fiscal Relations:Universal Principles, Local Applications", Working Paper 00-2, School of Policy Studies, Georgia State University Bird, Richard M. and Francois Vaillacourt, eds. (1998) Fiscal Decentralization in Developing Countries Cambridge University Press Branson, William; de Macedo, Jorge Braga and Jurgen von Hagen - Macroeconomic Policy and InstitutionsDuringthe Transitionto EuropeanUnion Membership, NBER Working Paper 6555, 1998 Cashin, C., Samyshkin, Y, O'Dougherty, S. et al (2005): Case-Based Hospital Payment Systems: A Step- by-Step Guide for Design and Implementation in Low- and Middle-Income Countries. USAID Abt Associates Inc. Zdrav ReformProject. Cichon, M; Scholz, W; van de Meerendonk, A; Hagemejer, K; Bertranou, F; and Plamondon, P; (2004), "Financing Social Protection", Quantitative Methods in Social Protection Series, International Labor Organization, Geneva. Cochrane, D.et al. (2006): HumanResourceStrategy 2006-2010. Final Report.Republicof Serbia. Docteur E, Oxley H. (2003): Health-Care Systems: Lessons from the Reform Experience. OECD Health Working PapersNo. 9. Dor A, Farley DE. 1996. Payment source and the cost of hospital care: Evidence from a multiproductcost functionwith multiple payers. Journal of HealthEconomics 15: 1-21. Ditch, J; and Barnes, H; (1996), "Social Assistance Coverage and Administration", Finding the Balance: Financingand Coverage of Social Protectionin Europe, eds. van Ginneken, Occassional Papers on Social Security, InternationalSocialSecurity Association, Geneva, July 1996. Ellis RP, McGuire T. (1996): Hospital response to prospective payment: moral hazard, selection, and practice-styleeffects. Journal of HealthEconomics 15:257-77. Ensor, T. (2004). Informal payments for health care in transition economies. Social Science & Medicine, 58(2), 237-246. Evetovits, Tamas: The experience with DRGs in Hungary. WHO/Euro and Semmelweis University. Presentedat WB workshop on HealthFinancingand Purchasingin Ljubljana, Slovenia. May 21,2007. Hakkinen Unto and Isabelle Joumard (2007): Cross-country Analysis o f Efficiency in OECD Health Care Sectors: Options For Research. OECD. ECO/WKP(2007)14. Health Care Systems in Transition (HIT): Latvia. (2001). European Observatory on Health Systems and Policies. Hurrell,A; Thompson, A; Carraro, L;and Makovic, M (2007), "Options for Delivering Social Assistance: A Feasibility Study", Main Report, Oxford PolicyManagement, Oxford Joumard, Isabelle and Per Mathis Kongsrud - Fiscal Relations Across Government Levels, OECD Economic Studies No. 36,2003/1, Kaiser, K. (2006), "Decentralization Reforms", In Coudouel, A. and Paternostro, S, Analyzing the Distributional Impact of Reforms, V01.2, The World Bank. Khaleghian, P (2004): Decentralization and public services: the case of immunization. Social Science & Medicine 59 (2004) 163-183. Langenbrunner, J., Orosz, E., Kutzin, J., Wiley, M. (2005): Purchasing and paying providers. In: Figueras, J. et al. eds.: Purchasing to improve health systems performance. European Observatory for Health Systems and Policies Series. Lewis, M. (2000).Who is paying for health care in Eastern Europe and Central Asia? Human Development Sector Unit, Europe and Central Asia Region. Washington, DC: World Bank. Matkovic, G. (2006), "Decentralization o f Social Welfare in Serbia", Center for Liberal-Democratic Studies, Belgrade, Republic o f Serbia. Meng Q, Rehnberg C, Zhuang N, Bian Y, Tomson G, Tang S. 2004. The impact o f urban health insurance reform on hospital charges: a case study from two cities in China. Health Policy 68: 197. MilanoviC Branko (2000) "The Role o f Social Assistance in Addressing Poverty", in Jeanine Braithwaite, Christiaan Grootaert and Branko MilanoviC, Poverty and Social Assistance in Transition countries, St. Martin's Press, New York Milanovic Branko (2000) "Ther Role o f Social Assistance in Addressing Poverty", in Braithwaite, Grootaert, and Milanovic, Poverty and Social Assistance in Transition Countries, St. Martin's Press, New York Rico and Leon (2005): Health care devolution in Europe. Trends and Prospects. University o f Oslo. Saltman, R. et al. (2007): Decentralization in Health Care: Strategies and Outcomes. European Observatory on Health Systemsand Policies. Sanigest (2006). Health Sector Restructuring Plan. Serbia. Inception Report. Shakarishvili, G (2005): Decentralization in Healthcare. Analysis and Experiences in Central and Eastern Europe in the 1990s. Open Society Institute. Budapest Schneider, P. and Kara Hanson (2007): The impact o f micro health insurance on Rwandan health center cost. Health Policy and Planning; 22:40-48. Serbia National Health Accounts (NHA).2006. The Ministry o f Health, Belgrade. Serbia. Shah, Anwar (2004) Fiscal Decentralization in Developing and Transition Economies. Ter-Minassian, Teresa - Decentralization and Macroeconomic Management, IMF Working Paper 971155, 1997. Uchimura H. and Johannes Jiitting (2006): Decentralization in Asian Health Sectors: Friend or Foes. OECD Policy Insights, No. 18. www.oecd.org. UNDP, Serbia, 2004. Widmer, W. (2007): Moving from input-based to output-based provider payment: Requirements for hospital autonomy and impact on management. Presentation given at the World Bank workshop on Health 98 Financingand Strategic Purchasing.Ljubljana; May 21,2007. Wiley, M. (2007): Factors for considerationwhen developing a case mix based hospital payment system. The Economic and Social Research Institute. Dublin, Ireland. www.esri.ie. Presentation given at World Bank GDLN session. April 11,2007. Yip W, EgglestonK. (2001). Provider payment reform in China: The case of hospital reimbursement in HainanProvince.HealthEconomics 10: 325-339. World Bank (2004) "Making Services Work for the Poor" World DevelopmentReport, Oxford University Press,WashingtonDC World Bank (2005). The Republic of Serbia: A Policy Agenda for a Smaller and More Efficient Public Sector. WorldBank (2006) Social Assistance and ChildProtectionin Serbia, PolicyNo. 35954-YU. 99 ANNEX1:DISTRIBUTIONOFRESPONSIBILITIES SCHOOLS I NSERBIA FOR Central Local School Parents1 government government professionals Community Organization School choice * * Student careers ** Instructiontime * Streamingcriteria Choice of texts * ** Student support activities Regular student assessment ** Professionalissues Hiring teachers * Hiring principal * * * Hiring non-teaching staff ** Dismissingteachers * Dismissingprincipal * * * Dismissingnon-teaching staff ** Duties/conditions of service, principal Duties/conditions of service, teachers Duties/conditions of service, non-teachingstaff *** Fixing salaries of principal Fixing salaries for teachers Fixing salaries for non-teaching staff Career of principal Career of teacher Career of non-teaching staff Deployment of teaching staff Deployment of non-teaching staff *** Planning and structures Creation and closure of schools * Creation/abolition of grade level Designing programmes for individual schools *** Deciding the subjects taught in schools * * Defining course content ** Setting graduate/ certification examinations Credentialing ** R Allocation to school for staff Allocation to school for non-teachingstaff ** Allocation to school for other expenditure * * Allocation to school for capital expenses * Within school use of resourcesfor staff * Within school use for current expenditure Within school use for capital expenditure ** Source: Thomas (2006). ANNEX2: PILOTINGTHE DECENTRALIZATIONFINANCING OF AND ADMINISTRATION Decentralization i s one approach to improving the quality o f education provided to all students. This Annex focuses on financing and management o f primary and secondary schools. However, in order for the goal o f educational improvement to be met many other things will need to be in place beyond a new financing system. In particular, systems are needed which provide support to schools and to teachers to help them improve, as are mechanisms to help vulnerable populations have access to and make use o f quality education services. These systems are needed regardless o f the policy options described below.67 A final note: there is no distinction drawn explicitly in this annex about the roles of the Ministry of Education and Sports inBelgrade and the Regional School Administrations (skola uprava); both are treated as part o f the central government. Clearly, as plans for decentralization are made the separate roles of different offices would need to be established. A: OVERALL APPROACH I t makes sense to test only two different policy options, given the number of municipalities and the importance o f being able to measure the differential impact o f different options.68 In the first year, a group o f LSGs would be randomly selected for each of the two options: this randomization means that policy makers can be confident that any results are likely to hold for the country as a whole when decentralization i s introduced n a t i ~ n a l l y . ~These municipalities will be matched with a group o f ~ municipalities o f the size numberwhere policies will continue as before. It i s important to note that this approach would not permit individual municipalities to opt in or out o f the pilot; this will represent a departure from the usual practice in Serbia and will require careful argumentation from the g~vernment.~' 61 And, of course, testing of different approaches to these issues is also possible and may be necessary where a single best approach is not identified. 68It would be possible to have pilot only one option, butthen ifthe pilot approachis deemedto be no better than the status quo, decentralization stops; at least until another approach is designed, piloted, and shownto be more effective. The other possibility is to try several different options from the beginning: however, there are not enough municipalities to test more than two options and get technically-robustevaluation results. 69 The number of municipalities would need to be determined. There are several considerations: from the statistical perspective it is importantto have a large enough group to draw meaningfulconclusions; but the larger the group the harder it is to providesupport and to organizethe pilot effectively. 'O Ifmunicipalitiesopt in then those participatinginthe pilot will notbe representativeofmunicipalitiesas a whole; they will likely be the most effective ones. Therefore, the Government will not be able to conclude from the impact of the pilot that decentralization to all municipalities will have the same effect. This undermines the central purpose of the pilot. Theoretically, it would be possible to allow municipalitiesto opt in and to construct a comparison group of municipalitieswith the same characteristics and, statistically, measure the difference in performance between these sets of municipalities. However, given the number of municipalities in Serbia it is very unlikely that such a comparison group can be constructed. These pilots would be run for two years, with country-wide implementation o f the approach which has yielded the best results taking place in the third year.71 Every ina selected municipality would berequiredto participateinthe pilot.73 The first step is to identify the most important effects which are to be expected from decentralization. Inthis Note, these are takento be: 0 Indicator 1: Management practices inthe school and municipality; 0 Indicator 2: Enrollment levels, both overall and including different sub- populations o f students; 0 Indicator 3: Efficiency o f use o fhuman and financial resources; 0 Indicator 4: Drop out or grade repetition rates. These aspects would be measured carefully before the pilot begins to establish a baseline and then again during the pilot and towards the end o fthe pilot to see which policy option has the biggest impact. Management practices (indicator 1) would be measured by re-running the survey of schools first conducted in 2006. Enrolment, efficiency and drop outhepetition rates (indicators 2 to 4) would be measured through the Education Information System (EIS): it is therefore essential that the EIS is operating fully in the LSGs to be included in the pilot as well as the comparisongroup before the pilot begins. 7' Another possibility is to try one policy option in 25 schools in the first year, but wait until the second year before starting the second policy option. This makes the implementationeasier in that only 25 each year need the intensive support associated with start up activities. Given that the second policy option could be run for two years, this would mean that the national roll out could not be started until the fourth year. 72 It is possible to run a pilot based on primary schools only. This is not recommended because it would complicate decision-makingconsiderably at both the national and local level. A decision about whether and if so how to include special schools in the pilot would be needed, but these are small numbers of schools and involverelatively little money. 73 It is possible to select LSGs to participate in the pilot in alternative ways; for example, by inviting them to volunteer to participate (this has been done in the health decentralization plan in Serbia). The advantage o f this approach is that ensures that LSGs in the pilot are active and willing participants. But this approach has the considerable disadvantage that the participating municipalities will not be representative of all municipalities and so the impact of the different policy options cannot be measured and policy makers cannot make an informedjudgment about which option to roll out to all municipalities. For example, one could imaginethat only more motivated or better governed municipalitieswould electto participate. Inthis case, at the end o f the pilot, it would be hard to tell if any improvements in education were due to the fact that these municipalities are better governed or because of the decentralization. This is why the randomized selection process is recommended. Obviously, this approach requires careful public discussion, and one of the issues to address will be to motivateLSGs which have been selected but might be unwilling to participatefully. In fact, this will provide usefulexperience for extending decentralization to all LSGs as many not selectedfor the pilot may also resist the full decentralization. In the case where LSGs volunteer to participate, it is possible, in theory, to match through statistical methods the characteristics of participating LSGs with a similar set of LSGs who are not participating. However, giventhe number of LSGs, the chances of finding a suitable matchedset of LSGs are low. 104 B. POLICYOPTIONS The central policy option assessed here is the extent to which responsibilities are given to schools or the LSGs. The proposed division o f responsibilities here is meant to be indicative rather than prescriptive; the MOES may decide to pilot a different balance o f responsibilities. But it i s important that two approaches are chosen which are both realistic possibilities and which are sufficiently different that their impact can be measured. Inthe end, the goal o f piloting two different options i s to give the MOES the chance to try differentpolicy options before scaling up to cover all LSGs. In both options, the same approach to determining funding allocations would be used: The MOES (in consultation with LSGs and with the approval o f the Intergovernmental Fiscal Commission) develops a funding formula which would be usedto allocate the budget for teaching and non-teaching staff.74 This formula would be based primarily on the number o f pupils in each but could include allowance for a small number o f objective factors which increase the cost o f education such as pupils with special educational needs, sparsely populated communities, and language minorities. Ideally, the actual additional costs o f these pupils would be derived from existing spending patterns, but the evidence suggests that existing spending does not correlate accurately with real costs. Therefore, an analysis o f the EIS data and some kind o f school mapping exercise would be needed. Based on this formula, the allocations for each school inthe pilot are determined. For the remaining schools, Le., those not participating inthe pilot, their allocations would be based on existing norms and standards. Local self-governments would also receive rationalization grants: The pilot LSGs would prepare a plan for which schools to close, according to the national guidelines on the minimum number o f students, class size and geographical factors. The MOES will provide the LSGs with grants to implement rationalization o f the school network, for example, to invest in newly required classrooms and school buses.76 The amount o f block grants will be determinedby the MOES based on each LSG's rationalizationplan. The grants will be provided for two years, during which LSGs are expected to complete necessary adjustments. Teachers made redundant would receive a payment according to 74 An alternative approach would be to allocate staff rather than the salaries of staff and this would be a less radical reform. This is not the preferred option because making decisions directly about money and its use is one o f the key elements to be tested in the pilot. 75 Inthe first policy option discussed below, where the municipalities make the major decisions about how to allocate money to schools, it would be possible to base the formula on the characteristics o f municipalities rather than schools. Under the second policy option, allocations per schools are needed by the MOES, so it makes sense to develop only one formula for the pilot which is based on the characteristics o f schools. 76 It may be that some resources are also neededto help LSGs prepare these rationalization plans. 105 national regulations. It makes most sense for these payments to be made by the Ministry ofLabour, Employment and Social Policy through existingmechanisms. 0 The Government would need to develop with LSGs involved in the pilot a mechanism for agreeing on the secondary school network because it i s not desirable for all LSGs to have secondary schools and especially inefficient for themto have all types of secondary schools.77 The MOES, Regional Education Offices and LSGswill facilitate teachers to find schools by setting up a nationwide vacancy information system (including on the internet). Relocation grants could be provided to teachers to encourage them to move to another locationwhere there are teacher vacancies. A final issue to resolve for the pilot is who should have ownership o f school facilities land and buildings). Option 1:Decentralization mainly to LSGs Under Option 1,the decision-making authority on education financing and administration would be decentralized from the MOES to LSGs to a great extent and to schools to a lesser extent. Major areas o f decentralization include the following: 1) Fundingfrom the MOES to LSGs: 0 Once the funding formula described above has determined the allocations per school, these allocations are combined at the LSG level to determine the allocation per municipality. 0 The Government will need to determine the effect on each municipality o f this allocation: some will be winners and some losers relative to historical budgets. With respect to the pilot LSGs, it is important to ensure that the level o f funding does not fall too much because it i s hard to adjust infrastructure and staff quickly. This will require some additional funding to cushion the impact o f the lower allocation, with a clear timetable for the phase out o f this transition phase.78This additional funding could come, at least in part, from keeping increases in allocations to other LSGs within certain bounds too.79 0 Once the funding allocations for LSGs are determined, this money i s transferred to LSGs inthe form o f an earmarked block grant, that is, money that can only be 77It is necessary to avoidthe situation in which an LSG in the pilot is involved in discussionson this issue with an LSG in the comparisongroup. '*Atypical amount would be that any given LSG receives at least 85% o f its allocationfrom the previous year. 79With respect to LSGs not in the pilot, the Government will need to determine whether the money remaining(once the allocationto LSGs inthe pilot has been done) is sufficientto cover all these allocations based on existingnorms. Ifnot, additionalresources may be needed for these LSGstoo. 106 used for educational purposes but can be used for both salary and non-salary expenditures." EachLSG then decides what additional resources to allocate from its own revenues to education, to cover non-salary costs for example. Each LSGthen decides how to spend the total amount o f money for education (i.e., from the block grant and from local revenues). It would decide how many teachers (and other staff) to employ at each school. The MOES would needto establish guidelines for staffing, butthese guidelines should provide flexibility to LSGs and not simply re-introduce the current centralized norms. The Government would needto decide whether to require or provide any guidelines to LSGswith respect to how to allocate non-salary costs (for example, also on a per capita basis). 0 School directors make a decision on which teachers and non-teaching staff to be retained/hired/fired within the approved number o f positions. 2) Teacher salaries: LSGswould pay the requirednumberofteachers andnon-teaching staff according to the national salary scale. 0 Teachers would be paid through the treasury system at the LSG level rather than, as now, from the national Treasury.'' 3) Education Innovation Fund:'2 Applications to the Fundwould come from schools, with the approval ofthe LSG. The existing (national) process for evaluatingand approving proposals will be used. Grants would be provided to schools, through the LSGTreasury system. Option 2: Decentralization Mainly to Schools Under Option 2, the decision-makingauthority on education financing and administration will be decentralized from the MOES to schools to a large extent, but LSGswill also play a role inrationalization o f school network. ~ It is important, as will be discussed below, that LSGs have the flexibility to spend this block grant on non-salary costs, as a way o f providing an incentive to make more efficient use o f staff. It may be that the central government places a cap on the amount o f salary spending, to ensure that local politicians can resist local pressures to increase salaries or the number o f employees at the expense o f other needed spending, which would have the effect o f reducing educational quality. "It would, theoretically, be possible for teachers to remain on the national treasury system but decisions about how many teachers to employ and in which schools be made by LSGs (or schools as in the second policy option). However, this is not desirable. When LSGs are responsible for paying salaries they have a strong incentive to keep careful track o f the number o f positions and employees; an incentive that does not exist at the moment as the central government pays salaries. 82The Education Innovation Fund is a national program which is open to all schools. The discussion inthis section is intended to address how grants under the Fund would work inthe context o f the pilot. 107 1) Fundingfrom the MOES to schools: Once the funding formula described in Section A above has determined the allocations per school, these allocations are combined and sent as a block grant to LSGs who pass this money directly on to schools in the amounts dictated by the formula. The Government will need to determine the effect on each school o f these allocation mechanisms: some will be winner and some losers. With respect to the pilot schools, it i s important to ensure that the level o f funding does not fall too much because it i s hard to adjust infrastructure and staff quickly. This will require some additional funding to cushion the impact o f the lower allocation, with a clear timetable for the phase out o f this transition phase. This additional funding could come, at least inpart, from keepingincreases inallocations to other schools within certainbounds too. Each LSG decides what additional resources to allocation from its own revenues to education, to cover non-salary costs for example, and adds this to the allocation from the central government and passes this to the school. The Government would need to decide whether to require or provide any guidelines to LSGs with respect to how to allocate revenues raised locally (for example, also on a per capita basis). The Government would needto establish any regulations covering the proportion o f the education budget which could be retained centrally by the LSG for its administrative costs and for centrally provided services to schools (such as financial management, counseling, or professional advice). Once the LSG has determinedhow much each school should be allocated from its own revenues. School directors and school boards make a decision on which teachers and non- teaching staff to be retained/hired/fired within the approved positions and within the overall budgetfor the school. Teachers and non-teaching staff may be directly paid by the LSGs or through schools. The MOES would need to establish guidelines for staffing, but these guidelines should provide flexibility to schools and not simply re-introduce the centralized norms that current exist. 2) Teacher salaries: i) Schools would pay the requirednumber o f teachers and non-teaching staff according to the national salary scale. ii) Teachers would be paid through the treasury system at the LSG level rather than, as now, from the national Trea~ury.'~ 83One issue to resolve would be whether schools should be given sub-accounts on the LSG treasury systems. 108 c. MAKINGITHAPPEN Since the purpose o f the pilot i s to prepare the way for a broader decentralization within the education sector, many of the activities needed to implement a pilot will be those which will provide the foundation for this broader reform, even if the pilot i s by definition on a smaller scale. Activities conducted to make the pilot successful will, therefore, also be the first steps inthe decentralization process. Key steps which emerge from the proceeding sections are: Government to develop public consensus amongst the various stakeholders (including schools, LSGs and the teacher unions) on decentralizationand the need for and design o f the pilot. Government to identify policy options to be tested duringthe pilot Establish baseline data. The MOES needs to ensure that EMIS i s operational in the pilot municipalities and carries out a survey of schools. If supplemental survey data is needed, it will needto be collected. MOES to develop and implement a methodology for identifying the costs of education, especially the additional costs for special populations. This will be usedto develop the formula for distribution o f financing to schools and LSGs. Calculations on the extent to which individual schools and municipalities would be winners and losers and how to ensure that losers do not see their allocations fall too quickly. Preparation by LSGs o f school network rationalization plans, with technical assistance as necessary. 109