Report No. 48620-YF Serbia Doing More with Less Addressing the Fiscal Crisis by Increasing Public Sector Productivity June 16, 2009 Poverty Reduction and Economic Management Unit South East Europe Country Unit Europe and Central Asia Region Document of the World Bank FOREWORD This report i s based on the findings o f a series o f missions to Serbia between December 2008 and May 2009. The report was prepared by William Dillinger (Task Manager) with the assistance o f Lazar Sestovic, Marina Wes and Ron Hood. Sectoral chapters were contributed by Anita Schwarz, Daniel Dulitzky, Ana Djordjevic, Juan Diego Alonso, Ivana Aleksic, Boryana Gotcheva, Sanja Madzarevic-Sujster, Carolina Monsalve and MartinHumphreys. The peer reviewers were Debbie Wetzel and Ardo Hansson. The report was produced with the participation o f a wide range of Serbian -counterparts. These included officials and staff o f the Ministries o f Finance, Health, Education, Labor, Economy, and Infrastructure, and the Health Insurance Fund, Public Enterprise Roads of Serbia and the Pension Fund. Comments from the International Monetary Fund(IMF) are also gratefully acknowledged. Serbia: Doing More with Less Addressingthe FiscalCrisis By IncreasingPublic Sector Productivity Executive Summary......................................................................................................... 1 Introduction ......................................................................................................................... 4 Recent Economic Trends ................................................................................................. 4 The IMF StandbyArrangements..................................................................................... 5 The ExistingDistribution o f Public Expenditure............................................................ 7 Pensions............................................................................................................................... 9 The Serbian Pension System ........................................................................................... 9 Directions for Reform.................................................................................................... 14 Health................................................................................................................................. 19 The Serbian Health Care System................................................................................... 20 Directions for Reform.................................................................................................... 27 Education.,........................................................................................................................ -30 The Serbian Education System...................................................................................... 31 Directions for Reform.................................................................................................... 36 Social Assistance ...................... ........................................................................................ 44 Serbia's System of Social Assistance ............................................................................ 44 Directions for Reform.................................................................................................... 48 Enterprise Subsidies .......................................................................................................... 52 Agricultural Subsidies ................................................................................................... 53 Subsidies to State Enterprises........................................................................................ 54 Roads ................................................................................................................................ -56 The Serbian Roads Sector.............................................................................................. 57 Directions for Reform.................................................................................................... 62 Railways ............................................................................................................................ 71 Directions for Reform.................................................................................................... 78 Summing up....................................................................................................................... 83 Tables Table 1: Key Economic Indicator ....................................................................... 4 Table 2: Beds per 100.000Population ................................................................. 24 Table 3: Physiciansper 100.000 Population .......................................................... Table 4: Utilizationof DZ Space. in percentoftotal square meters............................... 25 28 Table 5: Serbia's Basic Education System: Main Indicators., ..................................... Table 6: Trends inEducationExpenditure. 2005-2008.............................................. 32 Table 7: PISA Resultsfor Serbiaand a Selectionof Countries. 2003 and 2006................. 34 33 Table 8: CurrentRegulationson Class Size, by Levelof Education..................... 39 Table 9: SpendingonNon-ContributorySocial Assistance andChildBenefitsand Services in Serbia. 2005-2009 (RSD mn) ................................................... 45 Table 10: Spendingon SocialProtectionin SerbiaandEUNew Member States ................ 48 Table 11: Sources of Financingof CashMaternityBenefitsinthe EUMember States.,....... Table 12: Spendingby Types of Veteran Benefitsin 2009.......................................... 50 51 Table 13: PEPS:Trends in Source and Uses of Funds, NominalRSD, mn., ...................... 60 Table 14: CumulativeMaintenance BacklogandAnnual ExpenditureNeeds..................... 62 Table 15: EstimatedCost to Budget for CorridorX in Three Scenarios. 2009-2014............. Table 16: Cross Country Comparison ofToll Rates. 2008.......................................... 64 Table 17: AnnualVehicle RegistrationFees. 2007................................................... 68 68 Table 18: InternationalIndicatorsofRailwayProductivity,2007 .................................. Table 19: Infrastructureand RollingStock Maintenancein 2006-2007........................... 73 74 Table 20: InternationalComparisonofAverage Revenueper Traffic Unit ........................ 76 Table 21: Summary of MeasuresandFiscalConsequences ......................................... 84 Figures Figure 1: Compositionof ConsolidatedCentralGovernment (Expenditure 2008).................. 7 Figure2: Changes in CentralGovernmentProgramFunding. 2007.2009 ........................ Figure3: BenefitRatesfor FullCareer OldAge Workers Retiringat Age 65................... 12 8 Figure4: Percentageof Population> 65............................................................... Figure5: Distributionof Beneficiariesby Age in EmployeePensionFund. 2007............... 13 13 Figure6: ProjectedTrends in PensionExpenditures ................................................ Figure7: Ceilingson EarlyRetirement in HighIncomeCountries................................ 14 16 Figure8: Reduction in Benefitper Year ofEarly Retirement in HighIncome Countries. ...................................................................................... 16 Figure9: Age of Retirement inOECD Countries .................................................... Figure 10:Life Expectancyat Birth inEUand Serbia................................................ 17 21 Figure 11: DeathRatesin EUand Serbia............................................................... 22 Figure 12:Trends in SpendingofMoHand HIF(constant RSDmn of2008) .................... 22 Figure 13: Trends inHIF Spendingby Function(constant RSDths of2008)..................... 23 Figure 14: HIF Spendingby Category (planned2009)................................................ 23 ................................................................ Figure 16: Avg.Lengthof Stay for SelectedTreatmentsin DifferentLocations................. 26 Figure 15: BedOccupancy Rates(2007) 25 Figure 17: CapacityUtilizationin PrimaryHealthCare Facilities................................. Figure 18: PublicExpenditureas Percentof GDP: Serbia andEuropeanCountries............ 33 27 Figure 19:Trends inMonthlyEducation Salaries (RSD)............................................ 36 Figure20: Evolution ofthe number of student andclassesinPrimaryEducation............... 37 Figure22: Evolutionof the average class size, by levelof education.............................. Figure21: Evolutionof the number of students and classes in Secondary Education............ 37 37 Figure24: Condition ofthe Main and RegionalRoadNetwork, 2008............................. Figure23: Trends in Major Subsidy Programs........................................................ 53 58 Figure25: Trends in InjuriesandFatalities ............................................................ Figure26: InternationalComparisons of Traffic Deaths (deaths per 10.000 vehicles).......... 58 58 Figure28: Trends in PEPSNet CashFlow............................................................ Figure27: RecentandProjectedSpending on Roads................................................ 65 Figure29: FuelPriceComparisons (US cents per liter. 2006)....................................... 65 66 Figure31: Trends in Passengersand FreightCarriedby ZS......................................... Figure 30: Plannedand Actual RoadCapitalExpenditures. (RSD mn. currentprices)........... 69 71 Figure 32: Trends inTraffic Units per Staffand StaffLevels ....................................... Figure 33: InternationalComparison of Budget Support to Railways.............................. 72 76 Boxes Box1: Successful Introductionof DRGs:The Case of Hungary.................................... Box 2: PreviousEffortsto EstimateSavings from Optimization.................................... 30 Box 3: Minister or Mayor: Who Should Close Schools?............................................ 41 Box 4: Amendments to the Social Assistance Legislation in Serbia in 2002 -2005............ 43 ......................................................................... 47 Box 6: Road Safety Initiatives........................................................................... Box 5: The Development Fund 55 Box 7: Will StructuralFundsHelp Finance Serbia's Transport InfrastructureInvestments?... 59 Box 8: CrackingDown on Scofflaws.................................................................. 70 Box 9: The New EBRDLoan: ReformConditions.................................................. 77 79 Box 10: InlandWaterways ............................................................................... 82 MAP: IBRD34847 EXECUTIVE SUMMARY With its economy disrupted by the global economic crisis, the Serbian government faces tight budget constraints for several years to come. The Government has already responded by freezing wages and pension benefits and making cuts in capital works and other discretionary spending. These measures, while effective in the short term, are not necessarily sustainable over time. This report looks at more fundamental reforms in key public services, in order to identify opportunities for constraining expenditures through improvements inproductivity. Pensions. Pensions are the largest single program o f government expenditure. Costs are high, due in part to generous benefits. The pension due to a new retiree in Serbia i s equal to nearly 60 percent o f the net average wage. Due to a variety o f mechanisms that encourage early retirement, average retirement ages are lower than in most Organization for Economic Cooperation and Development (OECD) countries, and years o f contribution, shorter. Pension costs are also driven up by demographics. Due to its low birth rate, Serbia has a large number of beneficiaries relative to the number o f contributing workers. As a short term response to the fiscal crisis, the Government has decided to freeze pension levels in nominal terms in 2009 and 2010. This i s a highly desirable measure from a pension policy standpoint, as it will reduce the replacement level o f pension benefits over the next two years. The Government should resist pressures to compensate for this reduction once the fiscal crisis ends. The Government should also consider other methods o f reducing benefits on a permanent basis: limitingthe number o f years a worker can retire early, reducing pensions for early retirees (regardless of their years o f contribution) and raising the retirement age for women. Savings can also be aehieved by improving the administration o f contributions and benefits. In the longer term, the existing system o f pension indexation will substantially reduce the financial burden o f the pension system. Under current legislation, which will go back into effect once the fiscal emergency passes, pensions are to be indexedsolely to inflation. Once growth resumes and wages rise in real terms, pension receipts will outstrip pension payments, resulting in a surplus-generating system, although one that pays very low benefits relative to average wages. Nevertheless, the Government would be wise to continue to index pensions solely to inflation until the replacement level falls to more affordable levels. Health. Despite an array o f management improvements to date, the Serbianhealth care system still suffers from a wide range o f inefficiencies. The hospital bed occupancy rate is below the level o f new European Union (EU) members and considerably below the level o f older EU member states. While marginal efficiency gains can be achieved by administrative fiat, the key to fundamental improvements in health care productivity i s a 1 change inthe way it is financed. The present input-based system o f financing encourages inefficiency in the use o f resources and provides no incentive for improved service volumes or quality. The Government is now inthe process of reforming the health financing system. For primary care, it will adopt a capitation based payment system, in which primary physicians are paid according to the number o f patients on their rosters. For higher level (hospital) care, the Ministry o f Health and the Health Insurance Fund (HIF) intend to move to an output-based (Diagnostic-related Group (DRG) or prospective hospital payment) system. International experience shows that such reforms can generate substantial savings but need to be carefully designed to prevent fraud. The Government should continue the pilot testing o f these reforms and move to full implementation as quickly as possible. Education. The level o f government spending on education inSerbia is comparable to other European countries (as a percent o f GDP) but its outcomes are considerably poorer. There i s evidence that significant savings could be achieved through the rationalization o f the school network, particularly at the primary level, without reducing education quality. Serbia has too many teachers, given its present student population. As a result, many classes are inefficiently small. Consolidating under-enrolled classes by shifting students to other classes in the same school (and grade) could reduce costs by ten percent. Consolidating under-enrolled classes by shifting students to other schools within the same municipality could reduce cost another 25 percent. The latter, however, would require some schools to be closed. This is a politically difficult step. Other countries in the region have found it useful to enlist local governments in this effort. The Bulgarian government, for example, allocates funding for education to each local government on the basis o f enrollment, and allows each local government to recommend which schools and classes to close. Serbia should consider doing the same. Social assistance. Serbia spends relatively little on social assistance. Of the little that i s spent, less than one quarter i s specifically targeted to poor households. The two poverty targeted programs-the MOP and the child allowance-are nevertheless well designed although under-funded. As the recession i s likely to increase the number o f needy households, the Government should consider increasing funding for these two programs. Agriculturaland enterprisesubsidies. The Government spends a significant amount on agricultural subsidies, largely in the form o f a fixed payment per hectare. While this is less distortionary than the input- and price subsidies it replaced, it i s not particularly effective either in promoting agriculture or in reducing rural poverty. The area-basedpayment should be scaled back and means tested. The system o f subsidies to manufacturing and miningenterprises is in transition. In the past, subsidies have been used to facilitate privatization, financing severance 2 payments for workers in state- and socially-owned enterprises. As privatization has proceeded, the cost of these subsidies has declined and will continue to do so. (The Government's proposed equity investment inZastavaRIAT remains an exception.) Inthe future, the Government will have to guard against a resumption o f enterprise subsidies in the form o f tax breaks and subsidized or unrecovered loans to private investors. Roads. Serbia has a history o f under-fundingmaintenance, which has led to the deterioration o f much o f the regional road network. The Government's current priority, nevertheless, i s to upgrade segments o f Corridor X (which, despite its foreign financing, still implies a considerable Government counterpart contribution). In the short term, spending on Corridor X i s likely to be somewhat lower than originally planned,due to the time required to complete land acquisition and other preconstruction activities. The prospects for financing additional maintenance expenditures by increasing fuel taxes or tolls are limited, although better toll administration could have some impact. Resources for maintenance will instead have to come from efficiency gains, including improvements inthe prioritizationofmaintenanceprojects andwider use ofperformance contracts. Railroads. While freight traffic on the state railroad Zeleznice Srbije (ZS) has been growing, passenger traffic has been declining, and the financial performance o f the company has continued to deteriorate. Current subsidies to the state railway company are insufficient to cover its operating losses. The Government nevertheless intends to invest over RSD144 billion inupgrading rail service in Corridor X. Immediate savings could be achieved by postponing the start o f works inCorridor X and by reducing the designspeed of the proposed works from 160 km/hr to 120 km/hr. The latter step would reduce the overall cost o f the project by 60 percent. The company's recurrent operating losses could be reduced by abandoning under-used track, terminating underused passenger services, reducing staff, and raising passenger tariffs. Summingup. Overall, the immediate annual savings yielded by these proposals could be equivalent to about eight percent o f consolidated central government expenditures (in2008). The proposed increase inthe MOP and child benefits, along the with start of debt service on the road company's arrears, would reduce that figure to 6.7 percent. But this package assumes a fairly radical pace o f reforms, including the immediate consolidation o f under-enrolled classrooms within schools and among schools and an immediate 50 percent cut inarea-basedpayments to farmers. Social pressures and administrative constraints may slow the pace o f reform. In 2010, the principal expenditure savings will instead continue to come from short term controls over expenditure aggregates: the freeze on pensions and wages and cuts in discretionary spending and capital works. The impact o f the efficiency measures proposed in this report will take more time to materialize. The Government should, nevertheless, make an immediate start. While the fiscal impact of these reforms will be evident over the medium term, their most important impact will be on the quality o f public services. The reforms will stand Serbia in good stead even after economic growth resumes. 3 INTRODUCTION 1. This report responds to a specific request from the Serbian Minister of Finance. In the face of slowing economic growth, the Government 'facesthe prospect of increasing deficits in the short and medium term, due largely to growing pension obligations. The problem cannot be addressed by increasing revenues: the public sector is already oversized. The Ministry of Finance has therefore asked the Bank to identify opportunities for constraining growth in expenditures, without sacrificing the quality of public services. This report is therefore focused on identifying opportunities for efficiency gains inthe major categories of Government expenditure. RECENTECONOMIC TRENDS 2. Until very recently, Serbia's economy had been grown strongly, following significant economic reforms begun in 2000. As shown in Table 1, real GDP growth in 2007 reached 7.5 percent, the second highest since the start of the transition. Non- agricultural growth averaged 7.6 percent over the 2005-2007 period, driven by the services sector which had beengrowing at double digit rates (inparticular trade, financial services and transport and telecommunication). Output rose by nearly 50 percent between 2000 and 2007, as the corporate sector started to post profits and the banking sector was restructured. The positive supply response also appears to reflect increases inproductivity and output of recently privatized and de novo firms, as evidenced by the particularly rapid rates o f output growth in precisely those sectors which have recently undergone extensive privatization (e.g., steel, cement, rubber, tobacco, dairy, sugar, and banking) or attractedforeign investors. Table 1: Key EconomicIndicators Source: IMF.Data for 2008 are preliminary. Data for 2009 and later are projections 3. Rapid growth during 2005-07 was led by domestic aggregate demand. Overall investment levels had remained roughly constant in recent years, and slightly increased to 24.4 percent of GDP in 2007, although public investment levels had been 4 rising from 2.7 percent of GDP in 2005 to 4.7 percent in 2007. Real private sector consumption had been growing rapidly and increasedby nearly 10 percent in2007 alone. Ittook place onthe back of: a credit boom: Serbia continued to experience rapid credit growth which for 2007 is estimated to be 16.1 percent (average for the year); driven by credit to households which grew by more than 51 percent in 2007 (average for the year); expansionary fiscal policies and signijicant increases in public spending: Consolidated general government spending reached 44.9 percent of GDP in 2007, up from 42 percent in 2005, driven by increases in both capital and current spending; increases in real wage levels: Total year-on-ye'ar real gross wages increased 17.9percent in2007, on top of a 12.1percent increasein2006; rapid increases in exports: Exports increased on average 29.6 percent annually (indollar terms) over the 2005-07 period, albeit from a low base. 4. The global economic crisis has hit Serbiahard. Growthin2008 as awhole is estimated to have declined to 5.4 percent, with growth in the earlier part of the year driven by continued strong goods services sector performance, a much better agriculture season and growth in manufacturing and exports offset by declines later in the year. In the near term, the Serbian economy will continue to be affected by the international economic crisis, like other countries in the region. The Serbian economy to large extent depends on the availability of credit and investment from abroad, and receives annually about 9 percent of GDP inthe form of remittances. As of May, 2009, the IMF estimated that in 2009 Serbia will see a two percent drop in GDP. Projections for the subsequent years (2010-2013) show a slow recovery, with no growth in 2010 and only three percent growth in 2011,based on ongoing developments in the global economy as well as new fiscal policies adopted inline with the Fundprogram. (See below.) THEIMFSTANDBY ARRANGEMENTS 5. Recognizing the risks inherent in the current international financial situation Serbia requesteda precautionary stand-by arrangement (SBA) with the IMF in late 2008. InJanuary, 2009, the Fund mission andthe Serbiangovernment reachedagreement on a fifteen month economic program supported by a 402.5 million precautionary stand-by arrangement (SBA). This i s now expected to be replaced by a new, 27-monthYSBA, whose terms were broadly agreedupon inMarch, 2009. 6. The original 15 month SBA called for an overall fiscal deficit of 1.75 percent of GDP in 2009 and one percent of GDP in 2010. This was to be achieved largely through limitations on expenditures. (On the revenue side, the January SBA called only 5 for small increases in excise taxes to offset decreases in customs revenue resulting from the recently signed S A A pact with the EU.) In aggregate, Government revenue would decline slightly as a percent o f GDP. Apart from these measures, all adjustments would take place on the expenditure side. 7. By far the most important adjustment was to be a temporary suspension of pension indexation. Pension levels would reflect the two large increases granted in 2008, but would not be increased during 2009. In addition, increases in public sector wages would be limited to the projected rate o f inflation (7.9 percent). Non-essential hiring was to be suspended. Taken together, these measures were expected to confine growth in the wage bill to six percent. Public enterprises and local governments were expected to follow similar wage policies. The Government also committed itself to a freeze in most . spending on other goods and services (in nominal terms) and reductions in agricultural subsidies, including the abolition o f the per-hectare land payment to legal entities. Existing commitments to a joint venture project with Fiat and Zastava Car Company would be honored, however-an expenditure which the Government's accompanying Memorandum o f Economic and Financial Policies (MEFP) estimated would cost .5 percent o f GDP. 8. The new draft SBA calls for additional, more severe, measures. As specified in the Government's Memorandum of Economic and Financial Policies, these would include a freeze on the nominal wages o f government employees (including employees o f public enterprises) through 2010, an extension o f the nominal freeze on pensions through 2010, a freeze on hiring (with very limited exceptions), a reduction in transfers to local governments, and sharp reductions inthe discretionary budgetso f all budget users. 9. These measures, while severe, are short term palliatives. The Government will not be able to freeze pensions innominal terms indefinitely, as inflation will rapidly diminishtheir purchasing power. A long term freeze on salaries will make it difficult to attract and retain competent staff, Hiring freezes, by the same token, run the risk o f generating arbitrary gaps inpublic employment, as staff who retire are not replaced. Cuts in discretionary spending--particularly for road maintenance-would result in further deterioration o f the network. 10. While the eventual resumption o f economic growth will generate some increase in revenues, the fiscal envelope over the medium term is expected to be tight. According to the Fund's projections, Government revenues are projected to decline by five percent in 2009, and a further four percent in 2010. Revenues would only begin to recover, inreal RSD terms in2011. 11. In any event, increases in government revenues are not sustainable in macroeconomic terms. The Serbian public sector i s already oversized. Under these circumstances, the Government will therefore have to focus on continued restraint in expenditure. The aim o f this Public Expenditure Review (PER) is to advise the Government-and more specifically the Minister o f Finance-n opportunities for 6 constraining growth in expenditures, without sacrificing the quality o f services it provides. THEEXISTINGDISTRIBUTIONOFPUBLICEXPENDITURE 12. The Serbian public sector consists o f several distinct entities: (1) budgetary central government, which accounts for the majority o f spending, including defense and public security, social assistance, and most o f the costs o f education; (2) the pension and health insurance funds and the road company, each o f which has an independent source o f revenue but also receives varying levels of transfers from the central budget; (3) local governments, which have limited taxing powers (as well as the authority to impose tariffs to support their public utilities) but receive substantial transfers from the central budget; and (4) centrally owned public utilities, state-owned, and socially-owned enterprises. This report focuses on a subset o f these, namely entities: (1) over which the central government exercises direct fiscal control, and (2) that compete for central government tax revenues or their equivalent (e.g., payroll taxes). Thus it will focus on the budgetary central government, the pension and health insurance funds and the road company. Local governments will be considered only to the extent they represent a claim on the central government budget-i.e., as recipients o f intergovernmental transfers. State owned enterprises will be considered inthe same way. Compositionof Consolidated Central Government 13. Figure 1 Expenditure 2008 demonstrates the sectoral distribution o f pensions consolidated central health government education expenditures.2 As defense and interior shown, the largest single rn infrastructure item o f consolidated social assistance central government expenditure consists o f pensions. Spending on economy pensions consumed one agriculture third o f public other expenditures in 2008 Spending on health Figure 1 (largely financed from social contributions to the HIF) i s ' Although the pension funds and the health fund derive the majority o f their revenues from payroll contributions, these function, in effect as taxes. Any increase in contributions paid to these funds reduces the amount o f fiscal space available to the central government. Consolidated, as defined here, includes the expenditures o f the pension and health funds and the road agency that are financed from their respective own-source revenues. Data is from budget execution reports of central government, HIF,pension funds and PEPS. 7 the second largest item of consolidated central government expenditure, accounting for 15 percent of the total in 2008. Spending on education consumed another ten percent. Spending on security--defense and police--also consumed about ten percent. Spending on transport infrastructure (including foreign financed-spending) and social assistance each consumed about five percent o f the total. Taken together, these six major items o f expenditure accounted for 78 percent o f total consolidated expenditure in 2008. With the exception o f security-where the World Bank claims no expertise-these sectors are the focus o f this report. This report also examines spending on enterprise subsidies. Although these do not constitute a major proportion o f government spending, they represent particularly attractive targets for expenditure cuts. 14. As shown in Figure 2, the 2009 budget (as revised effective May 2009) implies sharp changes in the sectoral composition o f central government expenditure. (In the absence o f budgets for the pension funds, the health fund, and the road agency, this analysis is based on budgetary central government only.) Inabsolute terms, expenditures on pensions-Le., transfers to the pension fund-would soar. Expenditures on debt service, principally under the rubric 'repayment o f principal on domestic debt', would increase by a nearly equal amount. Expenditures on defense and police would increase, as would expenditure on social assistance and programs administered by the Ministry o f Economy. All other categories o f expenditures would see either little growth or sharp decline. Changesin Central Government Program Funding 2007-2009 I a pensions 0 debt service 0 defenseandinterior 0 socialassistance aeconomy Q education arailways 0agriculture Iother exec,legtslatrve,judiciary 0 health 0 grantsto LGs Iroads Figure2 8 PENSIONS 15. The Government's pension commitments are the primary threat to its fiscal equilibrium. Although pension reform laws in 2001, 2003, and 2005 have substantially restricted pension benefits (raising the normal age of retirement, imposing stronger eligibility criteria for access to disability pensions, instituting a point system linking contributions to benefits, and changing the basis of post-retirement indexation) pension benefits are still high by EU standards. In 2008, the government conceded two extraordinary pension increases, one inJanuary 2008 andthe other inOctober 2008. The fiscal consequencesof these increases will be fully felt (i.e., over the entire 12 months) in 2009. Under the terms of the January and March SBAs with the IMF, indexation is to be suspended through 2009 and 2010. Under the current legislative framework, however, indexation will recommence once the freeze expires and other costly elements of the pension systemwill remain inplace. THESERBIAN PENSIONSYSTEM The Serbian pension system consists of a single pension fund, comprised of three formerly separate funds, one for employees, one for the self-employed, and a third for farmers. These three funds were administratively unified as of January 2008, but will not be financially merged until 2011. The Employees' Fund is by far the largest, with 2.0 million contributors and 1.2 millionbeneficiaries. The Self-EmployedFundis significantly smaller with only 0.25 million contributors and 0.04 million beneficiaries. The Farmers' Fund has 0.22 million beneficiaries, but few contributors, drawing only 13 percent of expenditures from contribution revenue. A fourth fund exists for the military, but has yet to be mergedwith the other three. Contribution rates are set at 22 percent for old age, disability, and survivors' insurance, with an additional 12.3 percent for health insurance and 1.5 percent for unemploymentinsurance, making the total payroll tax 35.8 percent, well above western European and OECD levels, but average among central European countries. These rates represent a slight increase from the previous reform which had reduced contribution rates on pensions from 32 percent to 19.6 percent. In addition, many occupations permit workers to receive extra credit per year of contributions. These additional credits now require additional contributions from employers proportionalto the additional privilegestheir workers receive. Previouslythese additional credits were covered by the State.3 Contributionsare collectedby the tax authoritiesand then distributed to the pension funds and other social insurance funds as designated. Since the contributions are paid by employers, attributingthe contributionsto specific individuals is not straightforwardand requires employers to submit an additional form which allows the individualization of contributions. This process will be facilitated by the development of a CentralRegistry which is on the Government agendato create. Eligibility for pensions is based on age and years of service (contributions). An amendment to the pension law in 2001 established a retirement age of 63 for men and 58 for The extra credit that workers receive for particular occupations also allows them to retire early. While the increasedbenefit from the extra credit is paid for by employers, the additional cost from the longer retirementperiodis not chargedto employers. 9 women with 20 years of service and 65 and 60 with 15 years of service for men and women respectively, and a minimum age of 53 with 40 years of service for men and 35 years of service for women. These thresholds were in effect from January 2002, although those who retired earlier at younger ages continueto receive their pensions. The 2005 amendmentsto the pension law eliminatedthe option of retiring at age 63/58 with 20 years of service, but providedfor it to be phased out gradually over the period between 2008 and 2011. As of 2009 men may retire at age 64 with 18 years of service while women may retire at age 59 with 18 years of service. The 2005 law also reduced the minimum pensionable age to 50 for some occupations and introduced the option of retiring at any age with 45 years of service credit. Eligibility for disability arising from awork-related injury is immediateregardlessofpast contributionhistory. Ifthe disability is not work related, the individual must have completed 5 years of contributions, with lower eligibility requirements for those younger than age 30 at the time of disability. Disability pensions are only awarded to those incapable of performing any type of work, in line with internationalbest practice. Similarly, a minimum of 5 years of contribution is requiredbefore an individual who dies can leave a pensionto survivors. Benefits are based on a point system as defined in the April 2003 law. In this law, individuals accumulate personal points based on their contribution history and their contribution wage relative to average wage, with one point being awarded per year of service for those who earn average wage and multiple points being awarded to those whose earnings are a correspondingmultiple ofthe averagewage. In additionto the extra service awardedto particular occupational groups, all women are given an additional 15 percent service credit. The personal points are then monetized by multiplying by the general point whose value was initially set at around 1.8 percent of average wage in 2004. Thus, a person who retired after 30 years of contributions and who had paid contributionson the basis of average wage for each of those 30 years would retire with a pension equal to 54 percent of 2004's average wage. But while individuals earn a single point for contributions at average wage level per year for the first 40 years, between 40 and 45 years, the individuals earn only half a point and beyond 45 years of contributions,they earn no points. The value of the general point was initially indexed to 50 percent of inflation and 50 percent of nominal wage growth, allowing its value to erode, over time, relative to the average wage. However, in 2005, the law was amended, shifting the basis for indexation entirely to inflation, with the change to be phased in over a 4 year period. The first year of pure inflation indexationwas to occur in 2009, but (as notedabove) this has beenheld intemporaryabeyance. Pensions following retirement were also initially indexedby the same combinationof 50 percent inflation and 50 percent nominalwage growth, with the basis for indexationsubsequently shiftedentirelyto inflation, to be phasedin over a 4 year period. But the 2005 also mandatedthat the average pensionshould fall to no less than 60 percent of average wage in the first 3 years of the law's enactment. With normal wage growth, this provision would not be binding, but the extraordinary wage growth which occurred in 2007 and 2008 resulted in an extraordinary increase inpensions inJanuary 2008. Disability pensions are determinedexactly as old age pensions, but length of service is increasedby 213 of the differencebetweenthe person's age and 53 and 1/3 the differencebetween 53 and the relevant retirementage, implicitly makingthe assumption that had the individual been healthy, he would have worked 213 ofthe time until age 53, the earliest age of retirementand only 1/3 would have worked until the normalretirementage. 10 Survivors are given 70 percent of the pension ifthere is only one survivor, 80 percent to be dividedbetween-themifthere are two survivors, 90 percentifthere are three survivors and 100 percent for 4 or more survivors. Survivor pensionsare calculated on the basis of at least 20 years of contributions, regardless of the actual years of contribution. Widows receive the right to collect a benefit at age 48 unless caring for children. Widowers can collect the survivor pension at age 53, unless caring for children. These ages are being raised to 50 for widows and 55 for widowers under the 2005 legislation. The relevant age for widows in 2009 is 49 and for widowers 54. Orphans collect pensionsuntil the age of 15 unlessthey are studying in which case the age limit can be as high as 26. Disabled orphans receive the pension for life. Other family members who were dependent on the worker, such as parents and siblings may be eligible for benefits inthe absence of a spouseand children. Ifa spouse and childrenexist, these other family members can claim the differencebetween 100 percent and the benefitawardedto the immediate family. The maincharacteristics of the systemare summarized inthe table below. Serbian Pension SystemParametersin2009 ContributionRate I22%, 11%from employers and employees each, with higher employer contributionsfor those wjthkxtra creditedservice RetirementAge - 64 for menwith 18 years of service; 59 for women with 18 years of service; 53 for thbse with 40 years of service for menand 35 years of service for women; at any age with 45 years of service BenefitRate Benefitsdetermined by the point system with 1 point awarded for each day the contribution is made on the basis of average wage; Value of GeneralPoint 1% of average wage; women get an additional 15% Indexationof General Indexedto inflation Point IndexationofPensions By law, to be indexedto inflation, but suspended in 2009 and Post-retirement 2010 Disabilitypensions Calculatedas old age pensionswith credit given for 2/3 the differencebetweenage of disability and 53 and 1/3 the differencebetween 53 andnormalretirementage Survivor pensions Normally 70% of pension ifonly one survivor; givento women at 49 andmenat 54 unlesscaringfor a child; inorder to receive the right, the woman must have reached age 44 andthe man49 at the time ofthe spouse's death 16. The current pension system-even after the 2001 and 2005 reforms are fully ineffect-is, and will remain, unaffordable. This is, in part, because pension benefits are too high. Figure 3 compares the level o f pension benefits in Serbia with those o f high income OECD countries. As shown, the pension due to a new retiree in Serbia i s equal to nearly 60 percent o f the net average wage. This i s equivalent to the norm for high income OECD countries and i s higher than the rate paid in more than one-third o f them. This comparison, in fact, understates the generosity o f the Serbian pension system. New old age benefits inhighincome OECD countries are calculated for individuals who begin work and contribution at age 20, work and contribute continuously until reaching the 11 retirement age which i s typically age 65 for both males and females. In the case o f Serbia, the pension level applies to men retiringon average at age 60 and women retiring on average at age 57. Individuals are retiring with far fewer contribution years than the 45 years typical in OECD countries. Serbian pensioners are thus receiving similar pensions for fewer years o f work and contribution and receiving them for a longer duration because o f the earlier retirement age than their counterparts inthe OECDa4 Benefit Ratesfor Full Career Old Age Workers Retiring at Age 65 I Figure 3 17. The high cost o f the current level o f Serbian pensions is exacerbated by two characteristics that distinguish Serbia, to varying degrees, from the high income OECD countries. First, it i s an older country. As shown inFigure 4, o f the ECA region countries --which are already among the 'oldest' in the world--Serbia i s among the older 40 percent, with 14.4 percent o f the population aged 65 or above. Serbia also has a lower percentage o f population inworking age, typically defined as people between the ages o f 15 and 64: 67 percent. As a result, there are fewer workers to support larger numbers o f retirees than inmany countries. 18. Second, the process o f economic transition has left Serbia with a relatively highproportiono f the older population receiving benefits from the pension system and a relatively low proportion o f the working age population contributing to it. Typically, Furthermore, Serbia also typically calculates average pension to wage ratios using all types of pensioners, old age, disability, and survivors. The benefit levels shown above only apply to full career old age pensioners. If all type of pensions were included in the OECD numbers, together with pensions for early retirees and for those with incomplete careers, far fewer countries would be able to reach this 70 percent level. In Serbia, only about 50 percent of pensioners are old age pensioners, with 26 percent receiving disability pensionsand 24 percent receiving survivor pensions. Since survivors receive pensions which are typically 30 percent lower than the old age pension and the disability pensions are also typically 15 percent lower than the old age pensions, the large numbers o f these types of beneficiaries also drives down the averagepensionto wage ratios. 12 socialist countries had relatively high formal labor force participation rates. These high formal labor force participation rates o f the past translate into high coverage rates among the current elderly, which translate into relatively higher pension expendituresregardless o f the particular demographics o f the country. At the same time, the transition process itself dramatically changed the formal labor markets as large state-owned enterprises which offered lifetime employment were privatized, downsized, and split into smaller enterprises, which needed to respond flexibly to an evolving economic environment. This led to rising unemployment and the growth o f an informal labor market. While new privately owned enterprises have sprung up, enforcing pension contributions legislation against them has been difficult. The fall inthe number o f formal sector workers has led to low coverage among the working age population which translates into low revenue collection. Percentageof Population> 65 20 18 16 14 12 108 2 64 Figure 4 19. Pension eligibility criteria are also too generous. This i s largely due to a variety o f arrangements that allow a person to receive benefits before reaching the age o f 65. Figure 5 shows the distribution o f the beneficiaries o f the Employee Pension Fund. As shown, only 60 percent o f the beneficiaries are aged 65 or older. O f the remaining 40 percent, 45 percent are receiving old age pensions, 32 percent disability pensions, and 23 percent survivor pensions. Distributionof Beneficiaries by Age in EmployeePension Fund, 2007 20. The large number o f underage pensioners arises, 9% first, from the low retirement age that prevailed in the past. As noted above, the retirement age i s gradually being increased and will top out at 65 (for men) and 60 (for women) for persons with at least 15 years o f service. 0 Benelaabss65 Early OM Age 0 Earb Dlsabbd 0 Early SUNNMS Figure 5 13 But it also reflects the prevalence o f provisions that allow those with 45 years of contributions to the pension system to retire at any age and to credit granted to persons working inparticular occupations. 21. The large number o f under-65 disability pensioners i s largely attributable to the previously liberal disability laws which defined disability as inability to perform the work previously performed by the individual. As noted above, the law has been strengthened to apply only to those who are unable to perform any type o f work, but those previously granted disability under laws existing in previous years cannot be constitutionally reviewed. The large number o f survivor pensioners i s largely due to provisions allowing widows and widowers to receive survivor pensions as young as 54 for men and 49 for women, rather than requiring them to wait until the legal retirement age. DIRECTIONS REFORM FOR 22. Freeze pensions,then indexto inflation To address these problems, the Serbian government has a variety o f options. First, in order to lower the average pension level to a level more inkeeping with Serbia's resources, the Government i s well advised to maintain the agreed freeze on pension levels in 2009 and 2010 and thereafter revert to the indexation method specified in the 2005 pension law. The 2009 and 2010 freeze will have a dramatic impact on pension expenditures. Thereafter, imposing an inflation-only index on both the general point (which determines the level o f pension benefits at the time o f retirement) and subsequent pension payments will slow the rate of growth in the average pension level in nominal terms and reduce the aggregate level of pension spending as a percentage o f GDP. As shown in Figure 6, the 2009 freeze combined with a reversion to inflation-only indexation thereafter would reduce aggregate pension expenditures from 14 percent o f GDP in 2010 to 12.6 percent o f GDP by 2015 and 11.3 percent o f GDP in 2020.5 Extending the freeze through 2010 would further reduce pension costs to 12.7 percent o f GDP in 2010, 11.8 percent in 2015 and 10.7 1 percent in2020. Projected Trends in Pension Expenditures 23. Over the longer term, 14.5% , however, indexation based solely 14.0% on inflation will reduce pension 13.5% 6 yK 8 ` 13 0% levels to socially unacceptable 125% &Extend freeze levels. Employees and employers 8 120% through 2010 115% flus increased would be asked to contribute 22 11 0% retiremnt age for percent o f wage over a lifetime o f 10 5% w o m n 10 0% flus phasingout employment to support a pension early retiremnt equal to only 9 percent o f the average wage. The system would - Figure 6 Ina point system, there are two parameterswhich requireindexation, the value ofthe generalpoint which affects the value of the pensionat the time of retirement and the value of each subsequent pensionpayment. Current law would index bothparametersto inflation. Source: PROSTprojectionMay 2009. 14 not be providing value added for the worker and would be insufficient to prevent poverty in old age. Over the longer term, Serbia should therefore consider moving to a mixed inflation and wage based system, in which the value o f the general point would be indexed to wages (permitting new retirees to retire with a pension that reflects prevailing wages at the time they retire) while subsequent pension payouts would be indexed to inflation. 24. Reduce pensions for early retirees Second, the Government could change the parameters o f pension benefits. It could, for example, reduce the level o f pension benefits for those taking early retirement, even if they meet the years of contribution criteria. Most countries reduce the pension for early retirees, imposing an actuarial reduction on the pension related to the number o f years that a person retires early. The concept underlying the actuarial reduction relates to maintaining the total value o f the pension package regardless o f the age o f retirement. If the normal retirement age is 65 and individuals at age 65 are expected to live 15 years, an individual retiring at age 62 will be receiving the pension for at least 3 additional years and will be receiving 20 percent more inpension benefits than the person who retires at age 65 ifthe benefit levels are the same. Most countries would then reduce the pension level for the 62-year old by 20 percent to make the value o f the retirement benefits equal to what the individual would have received at age 65. 25. Limit the number of years a worker can retire early. Most countries also limit the number o f years that a worker can retire early. Figure 7 shows the number of years o f early retirement allowed for men in selected middle and high income countries. As can be seen from the figure, most countries others limit it to 5 years or less and many do not permit any early retirement at all. Hungary and the Czech Republic, for example, limit early retirement to three years. Poland does not allow it. As noted earlier, Serbia allows early retirement as young as age 53 for menwith 40 years o f contributions, and, in effect allows retirement at any age for persons o f either sex with 45 years o f contributions. Limits on the number o f years o f early retirement are a necessary complement to reductions in the pension benefits o f early retirees. Pensions are intended to prevent poverty in old age. Allowing people to retire too early (with corresponding actuarial reductions in their pensions) would result in pension levels too low to achieve that purpose. Figure 8 shows the degree o f reduction per year o f early retirement in selected middle and high income countries. The average reduction i s about 4.5 percent per year although this still provides an incentive for early retirement,since between 6 and 7 percent is the actuarially fair reduction. InHungary, the reduction is 2.5 percent; inthe Czech Republic 3.5 percent. Serbia does not reduce the pension at all. Figure 6, above, illustrates the impact o f eliminating early retirement at the rate o f six months per year between 2011 and 2040 (in addition to freezing pensions in 2009 and 1010.) As shown, the impact would be relatively small inthe short term. Pensioncosts would fall by about 0.1 percent o f GDP. Delaying retirement would result in individuals accumulating more years o f service and thus becoming eligible for higher benefits when they do retire, partially offsetting the expenditure reduction impact o f a higher retirement age. On the other hand, those postponing retirement also make contributions in the additional years 15 that they work, reducing the amount o f support the pension system requires from the budget. The difference can be quite significant over the longer term. The savings are nevertheless not insignificant. Figure 7: Ceilings on Early Retirement in High Income Countries * ! 7 ' l- Figure 8: Reduction in Benefit per Year of Early Retirement in High Income Countries 26. Raise the retirement age for women. Even when fully phased in, the 2005 law maintains a 5 year difference in retirement ages between men and women, allowing women to retire at age 60 with 15 years o f service credit. Since women live longer than men, the differential provides a large subsidy to women retirees, and given the longer life expectancy o f women, there i s no economic reason to permit the early retirement. Most highincome countries have moved to equalize retirement ages betweenmen and women. Figure 9 shows the retirement ages in high income countries. Only in Austria, Greece, Italy, and Switzerland is there a differential between retirement ages for men and women, and the differential i s only one year inthe case o f Switzerland. Figure 6 shows the impact o f raising retirement ages for women gradually from 60 to 65 over a 10 year period from 2011-20. Again, the impact i s relatively small. The reform would not discernibly affect 16 pension costs until it is fully phased even, and even then would only reduce these costs by 0.1 percent o f GDP. Accelerating the pace o f reform, would o f course, hasten its impact. Age of Retirement in OECD Countries 0 Male Female Figure 9 27. Improve complianceand administration. Pension administration in Serbia has been complicated by the merger o f three pension funds into one. While the funds have been consolidated, all o f the information i s still kept in three separate sub-accounts which requires duplicative systems and records. Furthermore, the Tax Authority collects the pension contributions together with income taxes and other social insurance contributions and remits the money to the pension fund. Employers are obliged to submit an M-4form to the pension fund each year which lists the individual workers on whose behalf the money was paid. 28. The Government has taken the decision to create a Central Registry as a separate entity which would keep individualized records on pension contributions as well as health and unemployment contributions in addition to personal income taxes. It has only recently appointed the head o f the entity. Once fully operational, the entity should lower compliance costs for employers since a new worker would be registered with the entity and not separately with the pension fund, health insurance, unemployment insurance fund, and the tax authority. Employers are also expected to report employee data on an individual basis monthly instead o f the annual data that the social insurance funds now receive, making it easier to match contributions with the persons for whom they are intended. All o f these measures will improve the efficiency and contribution enforcement capacity o f the pension fund. 29. Nevertheless, there will be data issues which remain which involve the breakup o f the former Yugoslavia where workers have contribution histories in what are now other countries. During the various conflicts, enterprises often did not remit contributions or employment was interrupted or records lost. Enterprises undergoing 17 privatization were not always obligated by law to make contributions. All o f these factors suggest that past contribution histories will remain problematic for some time. Furthermore, the business processes o f the pension fund need to be seriously reviewed andstreamlined to improve efficiency and effectiveness inadministering pensions. 30. But as noted earlier, without a change inpensionpolicy, the current system is not attractive for new workers. Forcing workers and employers to comply with policies which are disadvantageous to them will be increasingly difficult even with the best administration, but nearly impossible without major improvements. 31. LongerTerm Reforms. In the longer term, the Serbian pension system i s projected to turn around completely. From a deficit-generating moderately generous system, it turns into a surplus-generating system which pays very low benefits. As noted earlier, the long run shows a clearly unsustainable system from the social point o f view with employees and employers being asked to contribute 22 percent o f wage over a lifetime o f employment to support a pension equal to only 9 percent of wage for a limited retirement period. The system would not be providing value added for the worker andwould be insufficient to preventpoverty inold age. 32. Moving to a mixed inflatiodwage-based indexation system for both the general point and the pensions after retirement will improve the long run benefits somewhat, but not sufficiently to make them worthwhile. A more comprehensive reform will be required. A number o f approaches could be taken. The simplest would be to de- link the indexation of the general point from the pensions after retirement and allow the general point to be linked to average wage growth while the pensions after retirement are linked to inflation. The starting value o f the general point would need to be determined such that pensions would be adequate while being fiscally sustainable and affordable. 33. Alternatively, a different pension framework could be adopted. Options include a return to a definedbenefit pension system similar to the type that existed inthe former Yugoslavia or a notional account system similar to the one adopted in Sweden, Poland, and other countries. Each o f these options would need to be costed and evaluated within the Serbian context and parameters chosen which would provide value to workers and pensioners while being affordable. A downside o f adopting a new pension framework i s that Serbia adopted a new framework in 2002 and changing frameworks frequently might give workers the impression that the pension system instead o f providing long term security, is itself a source o f instability. 34. Serbia has considered adopting a second pillar, a defined-contribution fully funded pension system alongside its existing publicly provided pension system in the past. However, adopting such a system means that some o f the contributions from current workers will be diverted from the current government-run system into workers' individual savings accounts. This diversion would result in less revenue going to the public system. Given the sizable deficit already in the public pension system, the Government cannot afford to adopt policies which will increase the level o f the deficit 18 further. If some o f the medium term measures are adopted and create sufficient fiscal space, introducing a second pillar in the future might become an option for the Government to consider. But the fiscal space currently i s not there for this even to be an option. 35. Irrespective o f the nature o f the long runsystem, there will be more elderly in the future without pension rights. While the pension system currently provides benefits to 53 percent o f those aged 65 and above, currently only 34 percent o f the working age population i s contributing, suggesting that in the future only a little more than a third o f the old age population will be eligible for pension benefits. The Government may need to consider whether the current social assistance system i s adequate for providing benefits to what will now be the majority o f the elderly who are retiring without access to any other form o f public benefits. One option would be complement the contributory pension system with either a means-tested benefit exclusively for the elderly poor or by better integrating the elderly poor into the MOP. 36. The Government will also need to further develop the private pension sector. As the population ages, the public system will be unable to provide generous benefits to retirees. People who want more in old age than what the public system will be able to provide will have to undertake additional savings. The Government can assist by providing vehicles for that saving, as they have with the private pension system, and with regulating and supervising those savings to enhance their security. HEALTH 37. The Serbian health system has made significant progress over the last 15 years. Outcomes have improved and more services are delivered at a level o f spending comparable to other countries inthe region. Some measures to regulate medical practices and improve quality o f care have also been implemented. At the same time the system faces major challenges. As in many health systems inEurope, Serbia confronts pressures for increased health spending, due to the aging o f the population, the introduction o f new (and expensive) pharmaceuticals, and the development o f new technologies. These exacerbate the fiscal pressures already confronting the system as a consequence o f the global economic crisis. Lower growth will most likely be accompanied by an increase in unemployment and poverty, potentially reducing the Health Insurance Fund's (HIF) revenue base and increasing the pool o f vulnerable groups who must be subsidized from the general budget. Under this scenario the Government will need to find ways to use resources more efficiently, by improving management and furthering the reforms that will create incentives to use resources more productively. These are explored in the following section. 19 THESERBIANHEALTH SYSTEM CARE The health care system in Serbia i s organized on the basis o f social health insurance principles, with stewardship by the Ministry o f Health. The Republic Health Insurance Fund (HIF) i s in charge o f collecting contributions, pooling resources, and purchasing services from health care providers, which are in turn majority-owned by the Ministry o f Health or municipal governments. Both employers and employees are required to contribute 6.15 percent o f their payroll/wages, with a minimumandmaximum threshold. The self-employed contribute 12.3 percent o f declared income. Farmers contribute four percent o f their property tax assessment. The Pension Fund makes a contribution on behalf o f retirees equal to 12.3 percent of their respective pensions.6 The Ministry o f Health makes a contribution on behalf o f the unemployed (for a period not to exceed six months) and a wide range o f other vulnerable groups who are exempted from paying into the system.' As the HIF has runa cash surplus inrecent years, there has been no need for additional subsidies from the Government budget to the Fund. In terms of health service delivery, Serbia has not followed the path o f other countries in the region that adopted family medicine models. Instead primary care i s provided at Community Health Centers-durnuvi zdravlja (DZs). Some DZs also provide specialist outpatient services, and about twenty DZs (out o f 159) operate low level inpatient facilities. Most DZs are now owned by municipalities or are inthe process There is some evidence suggesting that evasion is pervasive among certain groups; self-employed and farmers contributed 5 percent and 0.78 percent of HIF revenues in 2004 respectively, while workers contributed68 percent of revenues from contributions and the pension fund 24 percent (Bredenkamp and Gragnolati). ' The current estimated number of individuals exempted from contributions is 710,000 and includes the following categories:(i). children aged 15 years or younger; (ii)schoolchildren and students until the end o f their education or 26 years of age; (iii)pregnant women up to 12 months after birth; (iv) Disabled and mentally challenged people; (v) People under treatment for HIV infectionor other infectious diseases (as ' describedby the law), malignant diseases, haemophyllia, diabetes, psychosis, epilepsy, sclerosismultiplex, terminal renal failure, cystic fibrosis, systematic autoimmune disease, rheumatic fever, drug abuse, people that got iinjuredor sick related to providing emergency medical care, and people relatedto the organ or tissue donationor acceptance; (vi) Monks; (vii) Peopleon social welfare accordingto regulationsregarding social protection, military veterans, military disabled and civil disabled in war; (viii) Users of permanent monetary assistance, placed in the social institutions or other families, according to social protection regulations; (ix) Unemployed, or other people whose monthly income is below minimal income (lath; (x) members of the family whose main sustainer is serving military service; (xi) Roma people who, due to traditional way of life, do not have permanentaddress; and (xii) People registered as refigeesor internally displaced, and if their monthly income is below minimal income prescribed. According to the law, the transfer per capita to the HIF from central Budget on behalf of these groups is 12.3 percent of the minimal monthly income. In the last few years the transfers have been lower than the statutory levels. The resources transferred to HIF on behalfof vulnerable groups have representedbetween 1.88 (plan for 2009) and3.14 (2004) percentofHIFrevenues. 20 o f being transferred to them from the MoH. Hospitals and clinical hospitals are owned bythe M o H andprovide services at the secondary and tertiary care levels. 38. The last 15 years have seen significant improvements inthe management of the system. The M o H has spearheaded reforms to improve quality o f care, including: (i) reconstructionof several DZsand some hospitals andclinical centers; (ii)the the upgrading o f medical equipment; (iii) the development of capacity to produce vaccines locally in a cost-effective manner; (iv) the creation o f professional chambers (doctors, nurses, dentists, pharmacists) in charge o f licensing health professionals; ahd (v) the creation o f a National Agency for Quality and Accreditation. In terms o f data management, the HIF: (i)introduced a fully functioning IT system; (ii)developed a database o f insurees; and (iii) introduced e-invoicing to health service providers resulting in more precise and timely monitoring of expenditure. The sector also resolved a large stock of debts and arrears that burdened the system and took steps to partially rationalize the system, reducing the number of beds and staffing, increasing co-payments, and reducing the generosity o f the benefit package (e.g., by eliminating most dental services for adults from the package). The HIF also embarked upon a substantial change in the mechanism used to pay health care providers, although this has yet to be fully employed. 39. Partly as a consequence, health outcomes have improved significantly over the last decade. Serbia now has an epidemiological pattern not unlike most countries in Eastern Europe. In fact many indicators are equal or better than those inthe most recent EU member states. As shown in Figure 10, average life expectancy, at 73.7 years, is almost equal to the EU8+2 average. It i s higher than in the Baltic countries and roughly equivalent to that o f Hungary (73 years) and Slovakia (74 years). Progress in improving the health status of the population accelerated particularly after the 1990s. For example neonatal deaths sharply decreased inthe period from 1999 to 2006 and its rate i s now at a level comparable to the average o f EUmember states that joined after 2004. Standardized 85 1death rates, however, remain above the averages o f both recent 1 SO -_. -- -_...--I- EU countries and older members. -Serbia EU averages. (Figure 11.) They 1 ll 75 are roughly equal to the rates in EU members aince 2004 or the Baltic countries and Hungary 70 ____I 2W7 and considerably above the rates inSlovakia. 65 1 60 40. Historically, the I998 1999 2000 20012W2 ZW3 2W4 2005 2006 countries o fthe former Yugoslavia Figure 10: Life Expectancy at Birth in EUand Serbia have spent a percentage of their resourceslarge on health, when compared to other countries at similar levels of development. As a percent o f GDP, Serbia's current aggregate level o f health care spending (including both private and public spending) is somewhat higher than that of the recent EU member states but 21 below that o f the older members. The aggregate level of spending has remained more or less constant, as a percent of GDP, over the last few years. 41. As in the rest of Europe, most health spending is undertaken by the public sector. Private spending on health in Serbia represents about a third of the total, and has decreased from a high of 34 percent of total spending on health in 2001 to the current level of 30 percent. These ratios are similar to those prevailing inthe EU. 400 - 42. Government spending on health in Serbia has remained roughly 1998 1999 2000 2001 2002 2003 2004 2W5 2006 constant as a percent of GDP inrecent 180000 I60000 I40000 43. Hospital care consumes 120000 a disproportionate share of total 100000 80000 health spending in Serbia. Between 60000 2005 and 2008 real spending on 40000 20000 hospitals grew about 40 percent, 0 almost twice the rate of growth of 2003 2004 2005 2006 2007 2008 overall spending, and now oMoH Budget DHIF Budget consumes half o f the HIF budget. Figuresfor 2008 are basedon preliminarybudgetfigures for the MoHandHIF. The significant fluctuations in spending by the MoH over this period reflect an increase in capital spending, financedfromthe NationalInvestmentPlan, duringthe middleofthe period. 22 - handled at lower levels o f care." It is also possible that hospitals are not usingthe most cost-effective combination o f factors inproviding care. 44. Pharmaceuticals are an 10000000000 - 9000000000 - important driver o f spending throughout 8000000000 - -.A ,--' 70000000 W - /--- - the world. A rule o f thumb dictates that 60000000 00 __ __.---,' 5000000000 -- pharmaceutical spending has been increasing at twice the rate o f growth o f 30000000 00 - 40000000 2WOOO00 00 GDP in most OECD countries. The I000000000 7 main reasons for this trend are the aging o f populations and the introduction o f Figure13: TrendsinHIFSpendingby new, expensive medicines that become Function essential for certain patients.l1 However (constant RSD ths of 2008) Serbia has managed to keep the growth o f pharmaceutical spending at bay. This has been accomplished through the use o f reference pricing, active management o f the positive drug list, and the use o f co-payments to limit demand.12 As a result, spending on drugs represents less than 15 percent o f HIF spending. 45. Spending on sick leave accounts for about four percent o f the HIF budget. (The HIF is 13% required to provide this to all its significant effort was made to reduce these costs from a high o f 7 51% spending on sick leave ~~ Figure 14: HIFSpending by Category has decreased about 30 (planned2009) percent in real terms over the last four years. However, anecdotal evidence suggests that the system is still overly generous and abused. Many countries in Europe have removed non-medical benefits from the health insurance package. lo The problem may be at the primary care level; if primary care providers do not have an incentive to provide care, they may refer patients easily to hospitals, becoming "gate-openers'' rather than "gate- keepers". Imashevaand Seiter, 2007. l2Dukic, 2007, unpublished. 23 46. As shown in Figure 14, other, smaller categories o f expenditure, including dental care and dialysis account for the remainder of HIF spending. Spending on dental care more than doubled in the same period, in spite o f a reduction in coverage, but still represents only 3 percent o f HIF spending. 47. Despite the array o f management improvements to date, the Serbian health care system still suffers from a wide range o f inefficiencies. In terms o f facilities (beds per capita) and staffing (physicians per capita) Serbia i s not out o f line. As opposed to many countries inEastern Europe or the former Soviet Union, Serbia did not builda large hospital infrastructure and concomitantly bed levels are in line with the average o f those prevailing in the EU. Between 2005 and 2008 the number o f beds was reduced by 7 percent, and the number o f beds per 100,000 i s now almost at the level o f EU15 countries. As shown in Table 2, it i s well below the level in most o f the Baltic and Visograd countries (with the exception o f Poland), The ratio o f beds to population prevailing in the EU may not be the appropriate standard, however. Countries where an explicit effort has been taken to develop day surgery systems and optimal use o f the primary care network-which i s in general more cost-effective than hospital care-have achieved significantly lower ratios-for example the UK (360), Sweden (220), Australia (260), and Ireland(290). Source: HFADB (WHO). 48. In principle, staffing levels -at least the number o f physicians- does not appear out o f line with EU countries. As shown in Table 3, the number o f physicians per 100,000 population is slightly higher in Serbia than in the newer EU member states and considerably below that o f the older ones. Again, levels in Serbia are below those o f neighboring countries, except Poland. There i s evidence, however, o f an excessive number of physicians at the primary (DZ) level, InEurope as a whole (as defined by the World Health Organization (WHO)); there i s an average o f one primary physician per 3,500 inhabitants. In Serbia, the ratio i s 1: 782. The level o f non-medical staffing-26 percent of the total in 2007--may also be too high. In the case o f DZs, on average more than 20 percent o f staff i s non-medical, with 6.4 percent administrative and 16.5 percent technical. 49. A major effort to rationalize facilities and staffing levels was begunin 2005, with the publication o fthe health sector strategy 2005-2010. The strategy provided target 24 for reductions in facilities @e., beds) and staff, It did not, however, designate specific facilities or positions to be closed. It was followed by a major voluntary dismissal program, in which staff was offered enhanced severance packages in return for their resignations. During 2005 and 2006, 14,400 staff took advantage of the program. While the program was originally intended to target non-medical staff, the HIF was ultimately compelled to offer it to all its employees. As a result, staff inkey specializations was lost, along with staff who were, infact, redundant. 50. The staffing trend has since reversed. While the health strategy envisioned further reductions of staff through 2010, total staffing levels grew from 108,975 in2005 to 111,068 in 2008. While some new hiring represents a replacement of staff lost during the voluntary dismissal program, some does not. Hospital directors are said to be under pressure to create jobs regardless of need, particularly since the start of the economic downturn. 2002 2003 2004 2005 2006 2007 Slovakia 322 319 313 Poland 230 229 224 199 203 Serbia 259 265 265 262 264 271 Hungary 319 324 334 278 304 278 ------- Latvia 299 298 311 315 315 305 Estonia 315 317 321 320 328 Lithuania 399 396 390 400 398 407 New EUmembers 260 265 267 265 253 Old EUmembers 327 329 331 336 338 51. Other productivity measures indicate a clear gap between Serbia and the EU countries. Significant progress has been made over time in improving productivity of I Countries health services,- but there is still a-gap when compared to EU countries. For example inpatient care admissions increased from 11 per 100 to almost 15 80 per 100 inthe period 1999-2006, but it 70 still fell below the 17 per 100 value in 60 EU members before 2004 and almost 50 Serbia EU members EU members CIS 21 per 100 inEUmembersjoining after before May 2004 since 2004 or 2004.13 As shown in Figure 15, the 2007 hospital bed occupancy rate (69 Figure 15: Bed Occupancy Rates(2007) percent) is below the level of the new EU members and considerably below l3 A lower admission rate may also be associated with access problems. According to the latest LSMS , six percent of the population was not insured under the compulsory health insurance, including 14 percent of those livingbelow the poverty line and 17 percent of Roma. 25 the level the older EU member states. (It is, however, roughly equal to the levels in levels in Hungary (69 percent); Estonia (71 percent) and Slovakia (68 percent). By the same token, average lengths o f stay (ALOS) are longer in Serbia that in either group o f comparator countries. If occupancy rates were to increase to levels observed in Europe, the same level o f discharges could be achieved with significantly lower bednumbers. 52. Variations among hospitals in the average length o f stay for comparable treatments suggest the extent o f inefficiency in some facilities, and by the same token, the potential for increased productivity in the system as a whole. A recent study o f six hospitals in Serbia showed that the length stay for fractures o f the femur, for example, varied from a 35 day average to a 6 day average. Somewhat smaller variations were shown for other procedures. (Figure 16.) 53. Product 4 0 0 1 KragujeVaC ILeskovac I ivity variations are Sremka Mitrovlca Krabevo also common in primary care centers. A recent study surveying 147 out o f 159 DZs concluded that the average per capita spending in the highest-spending quartile o f DZs is four times that o f Fractureof shaft of fermr Primary coxarthrosis, Coxarthrosis,unspecifled biIateraI the lowest-spending Figure 16: Avg. Length of Stay for Selected Treatments in Different quartile. Onlyvisits, nine percent of Locations moreover, are preventive. Inaddition many facilities are widely under~sed.'~ The same study estimated the production efficiency o f each DZ, defined as the ratio o f the total number o f consultations to the maximum possible number based on the inputs currently available. Figure 17 shows this ratio for each DZ, ranking them from the one with lowest to highest production efficiency ratio. Efficiency ratios for DZs vary from 0.13 to 0.86, with an average o f 0.64, More than 50 percent o f DZs produce at two thirds or less than their maximumcapacity. l4 World Bank (2008a): Serbia, Baseline Survey on Cost and Efficiency in Primary Health Care Centers before Provider Payment Reform. 26 1.oo 0 80 0 80 0 70 0.60 0.50 0 40 4 Figure 17: Capacity Utilization in Primary Health Care Facilities DIRECTIONS REFORM FOR 54. The significant progress in managing the health system and improving quality and outcomes during the last 15 years should continue, in spite of the Government's fiscal constraints. To accomplish this, the Government has to improve productivity. The analysis inhospital and primary care sectors has shown that it would be possible to maintain current levels of service provision with a smaller endowment of inputs, continuing the trend inbeds and staffreductions that was initiateda few years ago based on the recommendations of the human resources strategy. It has also been shown that, following the examples of other countries in Europe, productivity in the health sector is dependent on the incentives embedded in the payment mechanisms, and currently the system does not create the space to reward those providers with higher productivity of better quality of service. There are a number of actions that can be taken to remedy these problems. 55. In the short run, there are several immediate targets for efficiency improvements. Efforts to right-size facilities and staffing in hospitals and DZs should continue. The MoH currently plans to reduce the number of beds by 3,000. As the level of funding for health care facilities is based on the number of authorized beds, this could imply a significant reduction in costs. The HIF could also consider targeted staff reductions. This will require careful preparation, however. The HIF's previous experience with downsizing suggests that relying on voluntary buy-outs alone can result in adverse selection (only the most qualified staff leave) or inadvertent shortages in certain types of staff (anesthesiologists leave but non-medical staff stay on). Right sizing staff will therefore require a careful review of staffing needs in different types of facilities. One immediate target nevertheless presents itself. The HIF now has a 27 considerable number o f occupational therapists on its payroll. As the Fund has no obligation to provide this service beyond a statutory minimum, responsibility for this should be assumed by the private sector. There i s also evidence (cited earlier) that DZs could reduce staffing levels (as well as space) without reducing the number o f consultations they p r ~ v i d e . ' ~ 56. Several additional measures could be employed to reduced costs or generate marginal increases in revenues. Some savings could be achieved by reducing the salaries of HIF physicians who (legally) operate private practices. In December o f 2008 the M o H legally authorized the HIF to employ medical doctors who also engage in private practice. According to the new regulations doctors who want to work in private facilities have an obligation report to employers and to reduce the percentage o f time they charge accordingly. In the two months since the policy went into effect approximately 150 doctors have chosen this option and started to work part time inthe private sector. It remains to be seen what financial impact this will have and if additional incentives will be neededto motivate doctors to take advantage o f it. Small but not insignificant revenue increases could be achieved by renting out (or closing down) unused space in municipal clinics. As shown in Table 4, nearly half o f the space o f DZs i s devoted to non-clinical purposes. The HIF could also consider raising the level of copayments, although copayments were already doubled (from RSD 20 to RSD 40) effective January 2009. Source: World Bank (2008): Serbia, BaselineSurvey on Cost and Efficiencyin PrimaryHealth Care CentersbeforeProvider PaymentReform. 57. There may also be potential savings in evaluating the cost effectiveness of the benefits package. The package o f benefits offered by the HIF is not excessively generous. Nevertheless, to economize on the use o f expensive technology, Serbia might consider the example o f many EU countries and use a formal medical technology assessment and pharma-economics methods to evaluate the cost-effectiveness o f new technologies before including them in the basic benefits package. This could occur in conjunction with rules that reserve the most sophisticated medical technology for tertiary and specialized hospitals, with appropriate referral systems to ensure that patients, who need it, receive it. Also some effort (by MOF) to improve contribution administration. l5 World Bank (2008). 28 M o H argues that MoF i s less vigilant in enforcing collection o f HIF contributions than it is, for example, in enforcing VAT payments. 58. Reforming the health financing system. The key to fundamental improvements inhealth care productivity, however, i s a change in the way it i s financed. The present system of financing encourages inefficiency in the use o f resources and provides no incentive for improved service volumes or quality. At present, the budgets o f health care providers, at both the primary and secondary level, are based on the costs o f inputs. The health insurance fund pays providers on the basis of annual contracts, which specify the amounts to be spent on wages and salaries, utilities, medicines and other supplies. Allocations for staff are based on the number of authorized staff and salary coefficients. Payments for other recurrent costs are largely based on number o f beds. As a result, health care providers have a strong incentive to maximize the number o f staff and the number o f beds in their facilities. While the contracts may require reports on performance, there are typically no penalties associated with poor performance. Nor are the overall contract amounts related to the number and the severity o f the cases treated. Consequently, providers have no incentive to economize on the use o f inputs or to increase the quantity o f services they provide. 59. To create incentives for more efficient provision at both the primary and secondary level, the Government o f Serbia has initiated a reform in payment mechanisms. For primary care, the Government proposes to adopt a capitation based payment system. Under this approach, patients typically register with an individual doctor o f their choice who becomes the primary point o f contact inthe healthcare system. These doctors receive training in a broad range o f primary healthcare fields, limiting the need for referrals. To encourage physicians to register patients, the paying agent-in this case the HIF--would pay providers a standard rate for each patient on their roster. To encourage physicians to actually serve these clients, they often provide additional funding on a fee-for-service basis. 60. Design o f this reform is well underway. At present, the M o H and HIF are devising the specific formula, with assistance from the European Commission. The formula i s expected to include adjustments for age, gender, and additional incentives to provide preventive services. To prepare for the introduction o f the formula, the M o H and HIF are also providing support to DZ managers to respond to the change in incentives associated with the new payment mechanism and improve data systems and reporting in the DZs and the HIF. 61. For higher level (hospitals) care, the Ministry o f Health and the HIF intend to move to an output-based (DRG or prospective hospital payment) system care. Under this approach, hospitals are paid on a per-case basis, i.e. the average cost o f treating a patient during an entire episode. The payment can be adjusted to reflect variations across regions, hospital characteristics, and levels o f complexity. (By paying the average cost, the DRG system creates an incentive to minimize cost o f treating a certain case.) In preparation for this move, the M o H is planning to invest in hospital management 29 software in at least nine hospitals; and provide capacity building for health sector managers to adapt to the new payment mechanism. A pilot for DRGcosting has already started and full implementation of the pilot is expectedto take one or two years. 62. International experience shows that implementing such reforms can generate substantial savings. (See Box 1.) But it can also be a very difficult and lengthy process, however. Output-based systems can encourage providers to fraudulently inflate the quantity of services they provide and,may lead to declines inservice quality.I6 Box 1: SuccessfulIntroductionof DRGs: The Case of Hungary The case of Hungary is illustrative of the potential gains associated with the introductionof DRGs. Hungarybeganfull implementationof DRGs in 1993 after a five year pilot. Inits pure form, a case base payment system like DRGs would pay all hospitalsthe same amount for the same case; therefore less productivehospitalswould have to adjust their behavior in order to catch up with the more productiveones or else face losses (as their cost of producinga certain case is higherthan the amount they are paidfor it). This adjustment process takes time, however In Hungary the payment system allowed for different levels of DRGs for equivalent treatments until 1997, at which time payments for a given DRGs were equalized across all hospitals. As a result of the introduction of DRGs productivity of hospitals increased significantly. The average length of stay decreased from 12.6 to 9.5 days between 1994 and 1998, and overall spending on acute hospital care decreased 14 percent in real terms in the same period. At the same time the productivity increasedsharply, as the number of cases per 100 increasedfrom 22 to 25. This is a common pattern observed in countries following the introduction of DRGs and requires careful monitoring. In the case o f Hungary several measures had to be taken to prevent abuses by providers: (i)there is a cap on overall hospital expenditure at the national level, and as resources are exhausted the fees are recalculated (downwards); (ii)careful control on re-admissions is implementedin order to avoid charging twice for the same case; (iii)inefficiency such as provider-inducedhospitalization was reduced by charging co- payments to patients, and by monitoring and controlling provider reporting of cases; (iv) Hungary also appliedvolume control in hospitals. In general the number of discharges has to be monitoredcarefully in order to identify abuses in the system. In additionthe introduction of DRGs will require substantial changes in hospital management, as they have to be able to change the mix of inputs in order to producemore efficiently. EDUCATION 63. The level of government spending on education in Serbia is comparable to other European countries, but its outcomes are considerably poorer. There is evidence l6The World Bank financed Serbia HealthCare Projectis supportingthe development of institutional capacity inthe HIF and MOHto reviewand improvethe basic benefitpackageand the provider payment and contractingsystems. 30 that significant savings could be achieved through the rationalization of the school network, particularly at the primary level, without sacrificing the quality of and access to education. THESERBIAN EDUCATION SYSTEM The network of basic education schools in the Republic of Serbia is composed of two types of schools: regular schools and special education schools. The first type of schools includes the traditional two levels of education: primary education, within an 8-year cycle and secondary education, which usually lasts 4 years. The secondtype of schools provides education for students with special needs or special types of disabilities. Regular schools frequently take the form of a single central school and several associated satellites, which are usually smaller in size and were originally created to satisfy the needs of villages or settlements located more than 2 kilometers away from the central schools. The bulk of satellite schools (around 75 percent of them) provide only the first four grades of primary school although some provide all eight grades of primary education, and some secondary schools have satellites as well". Satellite schools comprise roughly 15 percent of total enrollment and 22 percent of total classes at the primary education level. Table 5 summarizes the main indicators for regular education schools and their distribution across the three geographic areas usually usedin education statistical reports o fthe country. The Government recently took on an additional responsibility in the education sector. Starting in school year 2006/07, it introduced a mandatory 6-month preparatory pre-school program. This is to be funded from the central government budget and is to reach all children of pre-primary-school age.18 (Prior to that time, preschool was offered on a voluntary basis by local governments and covered only 35 percent to 45 percent of children between 3 and 7 years of age.) Although mandatory, the new pre-school program is currently (2007/08) covering only 89 percent of children, with significant regional differences. I7 The creation of satellite schools is authorized inthe art. 15 (school size) of the "Manual on the Criteria for Determiningthe Price of Services inPrimary Education". The mandatory pre-school program is primarily conducted in existing kindergartens for 4 hours a day over a minimumof 6 months prior to a child's entry into primaryeducation. 31 Table 5 Serbia's Basic Education System: Main indicators Type of City of Autonomous Total school Main Indicators Belgrade Province of Central Serbia Republic of Vojvodina Serbia Primary Total Number of Schools 396 1,009 3,283 4,688 N u m b e r of Satellite Schools 216 661 2,666 3,543 YoSatellites 54.5 65.5 81.2 75.6 Total Students 114,808 156,586 307,539 578,933 Total Classes 5,181 8,203 17,166 30,550 Average Class Size 22.2 19.1 17.9 19.0 Average School Size 289.9 155.2 93.7 123.5 Total Educational Staff 13,469 20,405 39,045 72,919 Total N u m b e r of Teachers 9,933 15,409 29,045 54,387 Yo Teachers 73.7 75.5 74.4 74.6 Student/Teacher Ratio 11.6 10.2 10.6 10.6 Teacher/School Ratio 25.1 15.3 8.8 11.6 Student/Educational Staff Ratio 8.5 7.7 7.9 7.9 Secondary Total Number of Schools 103 151 367 621 N u m b e r of Satellite Schools 27 40 135 202 Satellites 26.2 26.5 36.8 32.5 TotalStudents 55,137 64,330 135,562 255,029 Total Classes 2,096 2,588 5,167 9,851 Average Class Size 26.3 24.9 26.2 25.9 Average School Size 535.3 426.0 369.4 410.7 Total Educational Staff 8,238 9,293 17,981 35,512 Total N u m b e r of Teachers 6,313 7,420 14,693 28,426 Yo Teachers 76.6 79.8 81.7 80.0 Student/Teacher Ratio 8.7 8.7 9.2 9.0 Teacher/School Ratio 61.3 49.1 40.0 45.8 Student/Educational Staff Ratio 6.7 6.9 7.5 7.2 &&: Schools for studentswith specialneeds are not included Source: Own estimates on the basis of school-level database constructed for this chapter. 64. Serbia spends about five percent of GDP on public education. As shown in Figure 18 this i s similar to the average level in the Baltic countries and somewhat more than the level inBulgaria and Romania. The level of education expenditure in Serbia has varied only marginally inrecent years, both inabsolute and inrelative terms, although the trend is complicated by the history of changes inthe mandate of the ministry responsible for education, the introduction of the mandatory preparatory pre-school program, and an unusual increase in capital spending financed through the National Investment Plan (NIP).I9 As shown in Table 6, the current (2008) level of spending4.8 percent of GDP-represents a slight decline over the levels prevailingin2005 and 2006. l 9The Ministry of Education's budget is slightly overestimatedfor 2005 and 2006 when that ministry was also coveringthe area of sports. For these years, budget allocations for specific sports-relatedgovernment agencies are excluded from the tables but some shared functions (management of the system, for instance) betweeneducationand sports was not possibleto separate. 32 65. Despite this adequate level o f spending, the performance o f Serbia's education system has been disappointing. After a decade o f profound crises during the 1 9 9 0 ~Serbia started this decade with a fairly low education base. According to Census ~ data, in 2002, only 41 percent o f Public Spending on Education Serbia's population o f 15 years o f age and older had completed a secondary 7.0% 1 I school without any further education. 6.0% Also, only 4.5 percent and 6.5 percent a 5.0% o f Serbia's population had completed 4.0% post-secondary and university + s0 3.0% education, respectively". Key 2.0% indicators suggest that the education 1.O% system continues to lag OECD 0.0% comparators in several key respects. Although the net enrollment ratio" for primary education in Serbia i s close to the OECD average, it i s falling. The net enrollment ratio for secondary Figure 18: Public Expenditureas Percentof GDP: education, while rising slightly, is still Serbia and European Countries much lower than the OECD average.'2 Completion rates for primary Source: Datafor Serbia is from WB calculationbasedon education have ranged from 91 the 2005 GovernmentBudget.Data for other countriesis from EurostatEducationStatistics. (Data is for percent-93 percent since 1995, with 2005/2006.) an average o f 1-2 percentage-point improvements per year. This i s not the case with secondary education, where some significant deterioration has occurred since 1990, with fairly dramatic regional differences. In Serbia as a whole, completion rates for secondary schools have stagnated at 78 percent. By contrast, completion rates in the city ofBelgraderose from 89 percent to 98 percent between 1990 and2000. 2005 2006 2007 I 2008'3 % GDP 5.2% 5.6% I 4.8% 4.8% 66. The picture is murkier at the tertiary level due to poorer data. Net enrollment rates--the ideal indicator o f enrollment--are not available and the Republic Statistical Office does not keep track o f gross enrollment rates, either. However, the latest estimates 'O New round of census data collectionwill take place in 2011, and some preliminary testing is already beingcarried out by the Republic Statistical Office. 21 The net enrollment ratio is the percentof a given age cohort enrolledinthe appropriategrade. '2 OECDandECA data from World Bank's EdStatswebsite (http://www.worldbank.org/educatiodstats). '3 Underestimated,as dataonLSGs' contributions are not yet available. 33 o f gross enrolment rates range from 37.8 to 43 percent24 Public higher institutions constitute half o f all higher institutions in Serbia in2005, with close to 240,000 students. O f those, 93 percent study at public higher education institution^^^ and a quarter of all students are attending ISCED 5B programs, which are shorter programs with a special focus on practical, technical or occupational skills. Only a quarter of all students graduate on time, according to the criteria set by their respectiveprograms. 67. In terms of learning outcomes, Serbia's education system is performing below international averages when compared both to the OECD and neighboring countries26. Serbia took part in the Programme for International Student Assessment (PISA) in2003 and 2006. On bothoccasions, Serbia's 15-year-olds failed to demonstrate an adequate level of achievement in reading, mathematical and scientific literacy. PISA results are widely used as a proxy for students' future prospects in the labor marketq2' (Table 7.) Slovenia 504 494 519 Finland 544 548 543 547 548 563 68. Not only is Serbia faring below the OECD average of 500 points in each of the assessed areas but the results of student achievements between the two assessments did not improve. In fact, in reading, Serbia results dropped an additional 10 points. Although Bulgaria and Romania fare worse than Serbia, it is worth noting that two countries of the former Yugoslavia, Slovenia and Croatia, whose educational systems 24 UNESCO Institute of Statistics gives 43 percent of GER in higher education for Serbia in 2002, while M.Vukasovic calculated 37.8 percent. Source: Vukasovic, M. (2007), Higher Education and Social Stratfication in Serbia 1990-2005.Aveiro: Universidadede Aveiro. Master thesis. 25 Vukasovic, M. (2009), Financing Higher Education in South-Eastern Europe: Albania, Croatia, Montenegro, Serbia, Slovenia, COP, Belgrade. 26 This part of analysis relies heavily on "Quality and Equity of Education in Serbia: Educational Opportunities of the Vulnerable - PISA Assessment of 2003 and 2006 data", Aleksandar Baucal and Dragica PavloviC-BabiC, Ministry of Educationof the Republic of Serbia, 2009. 27 As OECD, the administrator of PISA, explains, PISA focuses on young people's ability to use their knowledge and skills to meet real-life challenges, rather than merely on the extent to which they have mastereda specific school curriculum. 28 In each test subject, the score for each participating country is the average of all student scores in that country. The average score among OECD countries is 500 points and the standard deviation is 100 points, which means that two-thirds of students across OECD countries score between400 and 600 points. 34 share a same origin with Serbia's, have significantly higher results than Serbia in any o f the three domains tested in PISA. According to OECD methodology, 38 PISA points are worth approximately one year o f schooling. Using this measure, Serbia i s behind the OECD average for about 60-70 points and 30-90 points behindstudent scores in Slovenia and Croatia, or at least a couple o f years o f schooling. 69. In addition to average student achievements, data from PISA 2006 allow comparisons between the countries in terms o f the level o f student achievement. Achieving level 2 (out o f 6 or 5, depending on the tested area) is deemed sufficient to classify a student as `functionally literate'. Using this measure as a standard for comparison, about 43 percent o f students in Serbia are functionally illiterate in mathematics, 38 percent are functionally illiterate in science, and half o f all 15-year olds are functionally illiterate when it comes to readingz9.This score is much worse for the poorest students in Serbia, as the poorest 20 percent have 65.2 percent in math illiteracy, 73.9 percent in reading and 59 percent in scientific literacy3'. Overall, the results per dinar investedinthe system are far lower than what the country should achieve, judging from the performance o f comparable c~untries.~' 70. As in most other countries, the lion's share o f public education expenditure inSerbia is consumed bythe salaries ofthe education staff. It ranat the level ofclose to 4 percent o f GDP from 2005 to 2008. Such high level o f education spending on compensation to education employees reflects the number o f education staff rather than the level o f their salaries in this sector. Average salaries o f primary school teachers in 2008 were about seven percent below the average for all wage earners in Serbia, and were considerably lower than the salaries o f those working in other government sectors (public administration and social insurance). Average salaries for secondary school teachers were roughly equal to the average for all wage earners and salaries for university professors (RSD56,900) considerably above that average.) While salaries in the teaching profession have increased considerably during this decade, the increase has been consistent with growth in formal sector wages throughout the economy. As shown in Figure 19, salaries in the education sector have closely paralleled the level o f average wages in Serbia as a whole since 2002. 29 Baucal and PavloviC-BabiC(2009), p.20. 30 Baucal and PavloviC-BabiC(2009), pp.22-3, 3' Baucal and PavloviC-BabiC (2009), pp.2-3 and 45-46 show that although Serbia and Croatia invest a similar proportion of the GDP in Education, Croatia's results in PISA are higher, on average, by the equivalent of 1 year of basic education. Inother words, a 15-year-old Serbianboy would needan extra year of educationto attain the level of skills that a Croatian counterparthas at his same age, despitethe fact that both countries' investmentsare identical inrelative terms. 35 Trends inAvg Wage Levels .,, 35000 &All sectors $ 0 0 30000 -+--Education -5 D 25000 Publ.adrn.& 20000 social insurance C Healthcare & 15000 social work Figure 19: Trends in Monthly Education Salaries (MD) DIRECTIONSFORREFORM 71. Rationalizing the school network. There appear to be opportunities for significant cost savings by reducing the number o f education staff, particularly at the primary level. Serbia has too many teachers, given its present student population. As a result, many classes are inefficiently small. Education o f equivalent quality could be provided with a smaller number o f teachers. The cost savings would be substantial. 72. As shown in Figure 20, primary school enrollment levels have declined dramatically over the last fifteen years, largely due to falling birth rates. The decline in the school age population i s expected to continue. The number o f primary-school-age students i s expected to decline by another 4 percent over the next 15 years A. 13 percent reduction in the number o f secondary school age children i s projected. But Serbia has been unable to reduce the number o f classes proportionately, resulting in very high teacher/pupil ratios in some areas. Similarly, while the number o f secondary school students has fallen sharply, with the number o f secondary school classes has actually increased. (Figure21.) 73. The result has been a steady fall in the size o f the average class, at both the primary and secondary level. Figures 20 and 21 provide a first glance at the evolution o f the number o f classes at the primary and secondary levels. Figure 20 shows the situation for primary education, where while the number. o f students (grades 1 through 8) has been steadily decreasing across time (23.4 percent between 1990/91 and 2006/07), the number o f classes did not fall in similar proportions (only 6.1 percent inthe same time period). In the case o f secondary education, Figure 21, the trend in enrollment has been more 36 complicated, with a sharp increase after the end o f hostilities inthe late 1990's, followed by a-st p decline. The number of classeshas, nevertheless, continued to increase. Evolution of the number of students and classes in Primary Education 34,000 1 850,000 800,000 750,000 700,0003C a4 650,000 a U 600,000 550,000 500,000 Figure 20 Evolution of the number of students , and classes in Secondary Education I1,200 330,000 11,000 -- .- 320,000 10,800 .- --- 10.600 .- 310,000 ul 30 ul U 10.400 -- A- .. 300,OW 4C 10,200 .- a c.) 10,000 -- -- 290,000 v1 9,800 280,000 9,600 9.400 OW07 270,000 1990/9 1995/96 2000101 7007/03 7003/04 --C-Classes - - 2004/05 900.5/[Xi -& Students Figure 21 I 74. Inboth cases, the average number o f students Evdutionof the awrage class size, bylevel of education per class has steadily declined. As shown in Figure 22, the reduction amounted to 18.4 percent for primary and to 10 percent for secondary education, comparing the year 1990/91 to the year 2006/07. If this decrease in the average size Figure 22 37 o f classes had led to a concomitant improvement in quality, this might have been a desirable outcome. Unfortunately, data on the trends in the performance o f Serbian students suggests otherwise. 75. The failure o f the system to adjust i s in part a result o f its own regulations: The configuration o f the system o f primary i d secondary schools in Serbia is based on an extensive system o f by-laws and ministerial decrees3',The current set o f rules fixes the key parameters that determine, inter alia: a) the type o f institutions allowed; b) the number o f positions authorized per institution; and c) the teaching norms or maximum number o f hours expected for full-time equivalent positions. With respect to the key factor in determining class size, however, it i s ambiguous. Current regulations fix a maximum class size o f 30 students for both primary and secondary There are also somewhat lower maximum classes for multi-grade classrooms in small satellite schools. In classes combining two grades, the maximum class size is 20. In classes combining three or four grades, the maximum i s 15 (see Table 8). The regulations do not, however, fix a minimum class size for classes o f any kind. Thus 30 sixth-graders could be taught in a single class o f 30 students or divided into three classes o f ten students, without violating the regulations. By the same token, small satellite schools are not required to combine grades, no matter how few pupils each grade may have. Multi-grade classrooms-a highly efficient way o forganizing education insparsely populated areas- are permittedbut not required.34 76. The absence o f minimum class size regulations is not the whole explanation for the persistence o f undersized classes, however. School administrators are free to consolidate classrooms up to the maximum size. Their failure to do so is explained by the perverse financial incentives under which they work. Serbian schools are financed on the basis o f inputs.Fundingfor salaries-the largest single item o f educational expenditure-- i s provided on the basis o f authorized teaching hours. Although the current allocation o f teaching hours may have reflected enrollment at some time in the past, school administrators have no incentive to reduce teaching hours when enrollment declines. In ~~~ 32 The Law on the Foundationsof the Education System, enacted inMay 2004, establishes the basis under which the system works and sets up the key norms that regulate the types of institution, the language of instruction, the rights of minorities, the duration of education, the curriculum implementation and the distribution of the financing of the system. The Laws on Primary Education and Secondary Education add some more detail interms of the organization aroundeach of these levels and a set of four extra Ministerial Decrees provide the specifics in terms of the determination of the price of services and the teachingnorms. 33 An exceptional class size of 34 students is allowed for schools that offer education in a certain ethnic minority's mothertongue. 34 Unfortunately, none of the school databases this study had access to presents the detail of students by grade per satellite school. However, two sources of information do present some clues as to the proportion o f multi-grade classrooms in satellite schools, unfortunately with widely different numbers. Whereas the UNICEF's database (see "Optimization of the Network of Schools in Serbia", UNICEF, 2003) shows that approximately 80 percent of the total classes in satellite schools are combined classes of 2 or more grades, the Levitas & Herczynski's database (see "The Finance & Management of Primary Education in Serbia - Findings and Recommendations", Levitas, T. and Herczynski, J., 2006) puts the number at roughly 45 percent. Inany case, the proportion of multi-grade classrooms in satellite schools seems to be fairly high. 38 fact, they have good reasonnot to do so. Fewer teaching hours mean smaller budgets and the unpleasant task of dismissing staff. Table 8: Current Regulationson Class Size, by Level of Education Maximumnumber of students allowed per class* Levelof Multi-grade teaching education Grades Standard Exceptional 3combined or 4 grades 2 grades configuration configuration- combined 1-4 15 20 30 34 Primary 5-8 Not mentioned in the L a w 30 34 Secondary 9-12 Not mentioned in the Law 30 Not mentioned in the L a w * Note that these values are maximum bounds for class sizes. Minimum values are not set in the law, because there might be schools in rural areas where this minimum may not be reached. In practice, though, these maximum values should be interpreted as minimum values, and seem to be enforced as such, especially, in the case of multi- grade teaching in small satellite schools in ruralareas. ** Refers to the case o f schools that receive minority students from other neighborhoods, towns or cities. Source: L a w on Primary School, article 27, and Law on Secondary School, article 31. 77. The savings that could result from closing under-enrolled classes can be estimated by calculating the number of classes Serbia would require if all classes were optimally sized. This can be determined by dividing the number of students in a given grade in a given municipality by a proposed minimum class size (say, 30); adding and summing the result for all municipalities in Serbia. If a minimumclass size of 30 is chosen, for example, the calculation would show that Serbiahas 11,000 more classes than it needs at the primary level; or roughly 37 percent of the total actual number.36This oversupply appears at all grade levels (1-8) and in all parts of Serbia, although it is more concentrated at the earlier grades (particularly grades 1-4) where the average proportion of excess classes hovers around 44 percent and in Central Serbia, where there are, on average, there are 50 percent more classrooms than required. A similar calculation for secondary education yields a significantly smaller oversupply, with about 1,300 excess classes, or about 13 percent of the total. 35 As an additional class would be required for those children left over after optimum-sizedclasses are filled. 36 This is the assessment that stems from the school-leveldatabase prepared for the purposes of this study. This database came as a result of an extensive compilation and consolidationof three different databases built in the last five years by different researchers: a) the database built by UNICEF for the study on the Optimization of the School Network (UNICEF, 2003); b) the database built only for primary education schools in Levitas and Herczynski (2006); and c) the data from the EducationInformation System for the school year 2006/07, which is a still incomplete database of the basic education system, but that essentially capturesroughly 90 percent ofthe current system. 39 78. An immediate, nationwide, implementation of a minimum class size rule is not feasible. The calculation above assumes that all students can be reallocated among schools anywhere within a given municipality in order to achieve optimum-sized classes. This is clearly unrealistic in the short run. It could imply that schools in remote and sparsely populated parts o f a municipality would be closed and their students transported to other, perhaps distant, parts o f the jurisdiction. It would also imply large scale reductions in staff. These are reforms that could be implemented over the course o f a few years but not over a few months. 79. More immediate efficiency gains could be achieved by rationalizing class sizes within individual schools; i.e., leaving the total number o f students in a given grade . and school unchanged, but consolidating classes within the school where the opportunity exists. It is estimated that this measure could reduce the number o f classrooms by about 2,900 or ten percent, based on a minimumclass size o f 30. 80. These estimates are based on actual central government spendingon primary education, by grade, ineach o f the 25 administrative districts into which Serbia is divided for statistical purposes. They assume that costs would decline in direct proportion to declines in the number o f classrooms but that differences inper-student spending among administrative districts (and among grades within administrative districts would remain.37.Overall, this calculation suggests that full imposition of a 30-student minimum class size within each o f the 169 municipalities o f Serbia would achieve savings of about RSD14.7 billion, or about 37 percent o f the total current budget allocated by the national government to primary education. Yet, implementing only intra-school optimization, a much more doable plan for the short term, would still achieve a significant savings o f RSD3.5 million a year, or roughly ten percent o f the Government's current spending on primary education. 37 This is why, for example, although both BraniEevo and Pcinja have an almost identical number o f total classes for grades 1-8 (988 and 983, respectively), the fact that the budget for the latter is 25 percent as big as the former makes.that every excess class in Pcinja be roughly 25 percent costlier than inBraniEevo. 40 Box 2: PreviousEffortsto EstimateSavingsfrom Optimization The optimization of the primary school network is not a new topic in Serbia. In fact, probably the most significant contribution along these lines comes from a publicationfrom UNICEF(2003). This reportwas basedon a comprehensive database coveringevery school in the country. It identified ten strategies for optimizing the school network, including' closingdown satellite schools (or in some cases, expanding the range of classes they offer), increasingthe transportationof students to nearby schools, and usingmobileteachers. It did not work out their cost implications,however. A subsequentstudy by the firm Development Alternatives, Inc. (Levitas & Herczynski, 2006), focused exclusively on the primary education level and did provide some costing for a subset of proposed strategies. These included: (i)applying a minimum class rule only to schools with more than 400 students or (ii)either closingsmall satelliteschools or closingclasses insatellite schoolsthat did not reach a certain number of students. The first strategy was found to have more impact, although savings were estimated to equal only 1.2 percent-2.5 percent of total spending on primary education. The present study finds considerably larger savings, largely because: (i) it applies the optimal class size rules to all 4,688 primary schools, rather than the 518 schools with more than 400 students; and (ii)it assumes that students can be shifted among schools within a municipality (The previous study focused only on class size optimization within individual schools.) The present study also uses somewhat better data.* * Serbiastill lacks an updated census of primary education. This study uses a custom data base, derivedfrom the latest but still incomplete EducationInformation System (EIS), complementedby the databases prepared for the UNICEF (2003) and Levitas& Herczynski(2006) studies. 81. Some ofthe savings from optimizing class sizes among schools would be lost to increased transportation costs. The imposition of a minimum class size rule could imply a significant increase in the number of students traveling to school by bus. Transport costs are shared by both the National Government and by the Local Self- Governments (LSGs), with the latter covering only the transportation of students living four or more kilometers from the school they attend. Due to data constraints, the precise amount of this increase in costs cannot be determined. A theoretical calculation, basedon the geographic characteristics of each administrative district and municipality, nevertheless, suggeststhat transportation costs would increase by about 50 percent. 82. Regulations governing the education of ethnic minorities will also reduce the savings from classroom rationalization. Ethnic minorities in Serbia have the right to an education intheir mother tongue, providedat least 15 studentsof a given ethnic group are enrolled for grade 1(primary education) or for grade 9 (secondary educati~n).~'Given the existence of more than 21 ethnic minorities in Serbia, jointly comprising 18 percent of the total population, this provision is likely to prevent the formulation of 30-student classes insome areas. 38 Right established in article 5 of both the Law on Primary Education and the Law on Secondary Education. 41 83. Devising a strategy for network optirnizationSchoo1 rationalization implies a considerable rearrangement o f facilities and students along with a considerable reduction in staff. Staffing reductions would fall more heavily on some schools than others, due to different rates o f demographic decline in different jurisdictions and current variations in class/pupil ratios. The Ministry o f Education has devised a proposal for rationalizing the school network, which may involve, in addition to staff reductions, the closing o f schools (particularly satellite schools), the consolidation o f classes, and reductions in non-teaching staff. In principle, the Ministry's rationalization plan i s to be implemented over three years, starting in the school year 2009/10. Implementation i s to begin in Belgrade and the other largest cities o f Serbia and gradually move towards the periphery. This focus on Belgrade and other large cities i s likely to be most cost effective, according to the analysis prepared for this report. 84. The Ministry also has to resolve a somewhat thornier implementation issue: who will do the rationalizing? One approach would be to proceed on a top-down basis, with the Ministry deciding which facilities to close and which classes to consolidate, and reassigning or dismissing teaching staff accordingly. An alternative, recently used in Lithuania and Bulgaria, i s to decentralize such decisions to local governments. Under the approaches adopted in those countries, funding for most o f the recurrent costs o f education (including salaries) is provided to each municipality as a lump sum on the basis o f enr~llment.~'As a result, municipalities with sparsely enrolled classes experience a drop in funding and are forced to undertake the difficult downsizing decisions the Ministrymightprefer to avoid. 85. Changes in financial arrangements, however, would not be sufficient. Local governments must also have the authority to respond to declining levels o f funding, by consolidating classrooms, closing schools, and dismissing teachers. In some Eastern European countries, this has been a sticking point. Local governments are not permitted to close schools or reduce teaching positions without the permission o f the Ministry o f Education. This can paralyze reform. While the Ministry o f Finance i s trying to reduce funding, the Ministry of Educationrefuses to allow cuts inexpenditures. 86. Some Eastern European countries, including Bulgaria and Lithuania, have resolved this problem by conceding considerable autonomy over the determinants o f education costs to their local governments. While salary scales are set nationally, local governments have the authority to consolidate classrooms, dismiss teachers and--subject to certain limitations--close schools. Some central governments also provide funding for severance payments and assistance in transportation. (See Box 3). Serbia would be well advised to learn from these examples. 39 Adjustments in the formula are normally made for jurisdictionsfacing inherently higher unit costs, such as those in sparselypopulated or mountainousareas. 42 Box 3: Minister or Mayor: Who Should Close Schools? Closinga school is the most difficult decision anyone involvedin education policy-makinghas to make. It strips a community of an important gathering point. It may imply the dismissal of teachers who live in the community and the bussing of students-including very young students-to places that parents regard as distant. Dropout rates also tend to rise in conjunction with school closings. As a result, school closures are hugely unpopular and highly costly in political terms. Under these circumstances, central Governments are well advised to share part of the burden of school closings with local governments. Engaging local governments has an addition benefit: it draws on their localknowledge to assist in identifyingthe right schools to be closedand workable solutions for addressingthe needs of affected students. This is Bulgaria's experience. Starting in January 1, 2007, the Bulgariangovernment shifted the system for financing primary education from one based on inputs (i.enYteachers) to one based on enrollment. The bulk of financing for education to municipalities now comes in the form of a large ear-marked grant based on the number of students in the municipality. Municipalities can decide for themselves how many schools they want to maintain. Not surprisingly, many municipalities quickly decided they could do with one, two or three fewer schools and started approaching the ministry of education with requests for school closures. The ministry of education in Bulgariastill plays an importantrole inthe school closingprocess. It has the final say in school closings and monitors compliance with the fairly elaborate procedures municipalities must undertake before doing so For instance, a school closure proposal has to include a discussion of how the transportation needs of affected students will be addressed and demonstrate that nearby schools have sufficient capacity to accommodate the additional students. The Government has also introduced a program to monitor dropout rates4' and established a number of additional national programs to support municipalities in the process. For instance, municipalitieswith school closures can apply for school buses, for additionalpayments to laid-off teachers, and for resourcesto refurbish"central" (or new "merged") schools. Finally, the ministry has put together a list of "protected schools: schools that cannot be closed down because no nearby schools exist to ensure access to education for all. Additional resources to finance such small (and usually located in mountainous areas) are provided. The results have been impressive. After many years of unsuccessful attempts to close schools (interruptedby brief spike in 1997 and 2000) In each of the 2006/07 and 2007/08 school years, roughly 100schools were closed. Inthe 2008/2009 school year, another 300 schools are expected to close. 87. To facilitate the rationalization process, the Ministry should also complete the development of the Education Information System so as to supply the Ministry o f Education and Ministry o f Finance with timely and accurate information on the number and size of classes, the number and characteristics of education staff, the number and characteristics o f students, the number and characteristics o f the school network. 40 World Bank (forthcoming): "Program Document for a Proposed Loan to Republic of Bulgaria for a Third Social Sector Institutional Reform Development Policy Loan". 43 The Government will also need to alter the regulations governing the hiringo f education staff and reform the bylaws regulating class sizes andteaching norms. 88. In principle, wage restraint could be a source of future savings in the education sector. Firm conclusions cannot be reached without better data, however. At this time, there i s no evidence that wages are any higher than necessary to attract and retain qualified staff. Nor i s there any evidence to the contrary-i.e., that wages could be lowered without generating recruitment and retention problems. 89. Inany case, significant restraints inwage levels, even ifwarranted, may be difficult to achieve. Education staff belongs to unions that have been very successful in achieving real wage increases in the past. The Government negotiates wage agreements with the three education unions on an annual basis. The December 2007 protocol, covering wage increases for 2008, called for five percent increases in March, September and October, respectively. The figure applied to all educational staff, both teaching and non-teaching. An additional increase was grantedto non-teaching staff in order to narrow the gap between teaching and non-teaching staff. Based on expected rates o f inflation for 2008, this represents a real increase o f nearly seven ~ercent.~'There i s some risk, in fact, that any savings derived from network rationalization will be exhausted on wage increases. Recent salary-setting negotiations between education trade unions and the government explicitly link the two: the unions were willing to accept downsizing only if it is tied to wage increases. While the economic crisis has temporarily changed the content for wage negotiations, the Government may find' it difficult to restrain future wage increases once the economy recovers. SOCIALASSISTANCE 90. Serbia spends relatively little on social assistance. O f the little that is spent, less than one quarter is specifically targeted to poor households. The two poverty targeted programs-the MOP and the child allowance-are nevertheless well designed although under-funded. As the recession i s likely to increase the number o f need. SERBIA'S SYSTEM OF SOCIAL ASSISTANCE Serbia operates over 20 social assistance programs, with multiple social objectives. Broadly, the objectives of the programs can be divided into three groups: (1) poverty reduction; (2) populationgrowth, and (3) assistanceto vulnerable groups such as veterans and the disabled. Differentprograms pursue these objectivesto various degrees. Serbia has only one explicit anti- poverty program-the means-tested Material Support for Low Income Households (MOP) program. But it also means-tests other programs such as the pro-natalist child allowance. Other programs-such as aid to veterans and people with disabilities--arecategorical and not explicitly 4' Inany event, the final increase, stipulated inthe December2007 protocol and set for October2008, was conditioned upon network rationalization. As this did not occur, the final increase was suspended until further negotiationsbetweenthe Ministry ofEducation andthe three teacher unions were completed. 44 means tested, but may have positive distributional impacts due to the characteristics of their recipients. Pensions, discussed earlier inthis report, have a particularlysignificant positive impact on poverty. 91* Overall, Serbia spends relatively little on social assistance. (Table 9.) Spendingas a whole averaged less thantwo percent of GDP inthe 2005-2009 period. As a share of GDP, this is lower than the average spending inthe OECD (2.5 percent, 2006) and in the EU countries (2.5 percent, 2006), and comparable with the spending of ECA countries with similar level of economic development. Compared to the countries of former Yugoslavia, where the non-contributory benefits have similar structure and objectives, Serbia spends slightly more than Macedonia and Croatia, and less than Bosnia and Herzegovina. Spending on programs targeted to the poor is particularly low. As of 2008, only 7 percent of social spending was allocated to the MOP, with another close to 16 percent allocated to the child allowance program. Together, spending on the two programs equaled only 0.44 percent of GDP, a figure far below the level of the majority o f ECA countries. Although public outlays on social assistance programs as a whole have increased as share of GDP, from 1.3 percent of GDP in 2006 to 1.8 percent in 2008, allocations for the MOP and child allowances have declined. Spending on child allowances was cut down from 0.42 percent of GDP (2005) to 0.3 percent o f GDP (2008) and the spending on MOP, from 0.16 percent of GDP to 0.14 percent of GDP, over the same period. Table 9: Spending on Non-Contributory Social Assistance and Child Benefits and Source: Ministry of Finance o f Serbia and Ministryof Labor and Social Policy of Serbia. 92. There are two reasons for this trend. First, Serbia's rapid economic growth over the last decade has reduced poverty levels significantly--from 13.4 percent in 2002 to 6.6 percent in 200742--therebyreducing the number of potential beneficiaries. Second, legislative changes restrained the scope of social assistance benefits and the number of eligible beneficiaries. This was accomplished through two major items of legislation: the I 42Living StandardsMeasurementStudy. Serbia 2002-2007. 45 Law on Financial Support for Families with Children, enacted inmid-2002 and amended in 2005, and the Law on Social Protection and Provision of Social Safety to Citizens, amended inSeptember 2004.43 93* As described in Box 4,'the laws consolidated and rationalized the existing range of social programs, abolishing regional differences in benefit levels and strengthening means testing. Several provisions, nevertheless, tended to reduce benefit levels. The threshold for MOP eligibility tended to exclude large household^.^^ The shift from wages to cost-of-living as the basis for indexing eligibility thresholds also tendedto reduce the number of beneficiaries-at least at a time when wages were rising faster than inflation. And in 2005, the access for working age able-bodied persons to the MOP was restricted to 9 months in one calendar year. The introduction of changes in the child allowance brought an immediate reduction of the number of eligible beneficiaries by 200,000 or 30 percent - from 682,000 monthly average for January-May 2002 (before the introduction of the law) to 482,000 average for July-December 2002, and to a steady though more modest reduction thereafter to 405,000 beneficiaries in2008.45 94* As a whole, Serbia's non-contributory social programs neverthelessappear to be fairly well targeted, in the sense that the majority of benefits go to the poor. Over 54 percent of all benefits go to those in the poorest quintile of Serbia's population (Serbia HBS 2006). By the same token, 46 percent of the beneficiaries of these programs belong to the poorest 20 percent of the population, and close to 70 percent of them, to the bottom 40 percent of the population. The targeting of the MOP program is particularly good, with over 60 ercent of benefits going to families in the poorest quintile and to the extreme poor4! This compares favorably with child allowances, for which the poorest quintilereceives 48 percent of total benefits and the poor receive 50 percent. Coverage is an issue, however: only 7.5 percent of the poorest quintile and 27 percent of the extreme poor receive benefits from the MOP program. Coverage of the child allowance is slightly better:22 percent ofthe poorestquintile and 36 percent ofthe extremepoor receive it. 95. The level of benefits is relatively generous for those who manage to receive them. Social assistance benefits constitute 23 percent ofthe consumption ofthe poorest 43 Accordingto an impact analysis ofthe Center for Liberal-Democratic Studies in Serbia, two thirds ofthe reduction of child allowance recipients should be attributed to the increase in family incomes and assets, and one third to the program design changes. Source: Matkovic, Gordana and Bosko Mijatovic. Impact Analysis of Government FinancialAssistance for the Poor, CLDC, 2008 (mimeo). 44 The minimum income required for eligibility declines substantially with increases in the number of family / household members. For a one-person household it is 16 percent of the average wage, for two- member household - 22 percent, for three-memberhousehold 28 percent, for four-member household- 30 - percent, and a maximumof 32 percentwhen the household consists of five or more members. 45 Matkovic, Gordanaand Bosko Mijatovic. Impact Analysis of Government FinancialAssistance for the Poor, CLDC, 2008 (mimeo). 46 The 2006 HBS does not single out householdfamily income from MOP. For that reason, a proxy (broader) social benefit income category has been used to assess the coverage and targeting of the MOP, andthe data shouldbe treatedwith some caution. 46 Box 4: Amendments to the Social Assistance Legislation in Serbia in 2002 -2005 The Law on Financial Supportfor Families with Children was enacted in mid-2002 and amended in 2005. The law: introduced differentiationbetween the social policy / poverty alleviationobjectives of certain child benefitson one handand the demographic/ populationpolicy objectives of other part of the benefitsfor children; consolidated the numerous previously existing child benefits into three main programs: (a) a means-testedmonthly child allowance with a social assistance objective; (b) a birth grant for the first four children; and (c) a wage compensation program for new mothers who have worked for at least 3 monthsbeforetaking maternity leave; abolished the regional and municipal differences in the criteria for access to child protection and the regional / municipal differences of the paid benefits and introduced `centralized' / national criteriaand nation-wide benefit levels thus improvingthe targeting to the poor inthe municipalities with relativelyhigherpoverty rates; strengthenedthe meanstest by addinga test for assets to the existing incometest; changed the base for indexing the eligibility thresholds and benefit levels from average wage growth to the more slowly increasing cost of living index, and replaced the monthly indexationswith by-yearly adjustments(as of April 1andOctober 1); abolishedthe categoricalentitlementto childallowance for the third and each subsequent child (before the meanstest was applied only for the first two children); introduceda means test for the allowance for the first four children and limited the allowance entitlementto the first four children inthe family only; and linkedthe provision ofthe child allowanceto schoolattendance for the children of school. The Law on Social Protection and Provision of Social Safe@to Citizens was amended in September 2004. The amendments resultedindesign changes to the programfor material support (MOP) with cost containmentimpact. Underthe law, the regional and municipal differences in the criteria for access to MOP were abolished and substitutedwith a uniformnationalpoverty line (access threshold); similar to the child allowances,the indexationofthe eligibility thresholds for the MOP shifted from the differentiated regional or municipal average wage growth rate to a national cost-of- living index. The MOP benefit levels, as well as the levels of all other social assistance transfers startedto be indexedwith the cost of living; the access to MOP for families where the majority of the members are able-bodied and of working-age, was reduced to 9 months in one calendar year to encourage resuming employment and prevent providing the benefit during the months of increased seasonal demand for informal employment when formal incomes become more feasible but hard to estimateandverify. 20 percent o f the population. Generosity varies significantly across programs. Chilc allowances represent close to 12 percent o f the consumption o f the poorest 20 percent o f households while MOP and caregivers allowances are much more generous, representing 47 percent and 65 percent o f the incomes o f those eligible for the benefit. Social benefits . are also generous for the extreme poor accounting for over 60 percent for their consumption. 47 DIRECTIONS FORREFORM 96. PrioritizingSpending: With the projected slowdown in Serbia's economy, levels o f poverty are likely to increase, adding to the number o f beneficiaries eligible for the MOP increasing demands for wider coverage. In responding to these demands, the Government would be well advised to focus any increases in spending on programs that are most effective in targeting the poor, and freezing or limitingspending on other programs.Two programs, inparticular, are candidates for expansion. 97. The first is the MOP. The MOP is a last resort poverty gap program activated only when all other social protection mechanisms are `exhausted' but the individual or household remains poor. Eligibility for MOP i s determined by a means test taking into consideration all earnings o f the household except those from other social benefit programs. The MOP eligibility threshold is determined as a percentage o f the average wage and adjusted for household size with a steeply declining and same for children and adults equivalence scale. 98. As an anti-poverty program, the MOP is well targeted, inthe sense that most benefits go to poor households. There i s a strong case for scaling up the MOP and expanding the number o f households eligible to receive it. Inrelative terms the spending on MOP i s lower than similar spending in benchmark countries as the new EU member states. Even the `low spenders' on targeted social assistance Poland, Latvia and Estonia spend more than Serbia (Table 10). Table 10: Spending on Social Protection in Serbia and EUNew Member States *War veteran benefits and caregivers benefits. Source: MISSOC. 99. Expanding the number o f beneficiaries could be accomplished in part by raising the income threshold for eligibility and/or relaxing the asset test. It can also be accomplished by simplifying application procedures and through more aggressive outreach. There i s anecdotal evidence that the poorest o f the poor (mostly Roma and 48 IDPs) make limited use o f the MOP and instead rely on one-time grants from their respective municipalities. This is in part because eligibility documents are often issued for a fee set at the local level which i s not affordable for the poorest;47 people who migrate have no permanentaddress or cannot provide certain documents and cannot meet the eligibility criteria; or the approach to identifying beneficiaries is passive: social workers rely mostly on demand from applicants rather than undertaking active outreach with information dissemination and/or identification o f potential beneficiaries. At present, the Centers for Social Work, which implement the MOP, are making efforts to reach potential beneficiaries and make sure that they understand how to apply for and receive assistance. These efforts should be supported. Finally, the MOP equivalence scale can be aligned with the good OECD practices to eliminate the bias towards single- member and small families and household^.^^ 100. The second is the child allowanceprogram. The child allowance program is the social protection program for children from low- and low middle income families. It is means-tested and limitedto the first four children inthe family, aged 0-19. It is also conditional to school enrollment for those after age 7.49 As is the case with the MOP, the level o f benefits i s low. Eligibility i s limited to families in which the income per family member i s less than 20 percent o f the average monthly wage, i.e. only marginally higher than the access threshold for the MOP. The average amount of the child allowances is quite low: around 5 percent o f the average wage. 101. As in the case o f the MOP, there is a strong case for maintaining the child allowance at least its existing level o f funding and for addressing administrative barriers that prevent eligible households from accessing it. According to MLSP administrative data, around 20 percent o f the children o f MOP beneficiaries (Le., the poorest households) do not receive the child allowance, partly due to a lack o f evidence o f enrollment in school. Verifying regularly school attendance might help identify the barriers which poor children face and thus increase the number o f eligible beneficiaries. 102. Several other social programs are less effective in addressing poverty. While the government's pro-natalist policies and the special status of certain constituents (such as veterans) may justify the continuation o f these programs, the case for an expansion in fundingis less compelling. Insome cases, improvements intargeting may bejustified. 47 The central government itselfdoes not charge poor householdsfor documentationfees. 48 The MOP equivalence scale is implicit (the MOP due amount is determined in nominal terms per familyhouseholddependingon the number of its members) and varies across years. For example, in 2009, a two-member householdfamilyis receiving 1.37 of the MOP amount for one member family/household; three-member- 1.75; four-member- 1.87 and five-member-2 times the MOP for a single-memberfamily or household.Moreover, inthe case of families with more than five members, the MOP amount is `capped' at 2 times the MOP for a single member householdfamily. 49 Inparallel, the law providesfor enhancedaccess to child allowancesfor certaincategoriesof vulnerable children - children of single parents, children in need of special care, children in foster care and guardianship, by increasing the eligibility threshold for them by 20 percent of the standards one and by increasing of the extended benefit by 30 percent. As a result their share increased from 8 percent of all beneficiariesin2001, to 14percent in2008 (administrative data). 49 103. Wage compensation during maternity. The program for wage compensation (paid leave) during maternity accounts for a large share (26 percent) o f overall spending on social programs in 2008, although it accounts for only 0.5 percent o f GDP. The wage compensation is paid by the employer, who is compensated for this purpose (reimbursed) by the MLSP. Eligibility extends to working mothers in both the public and private sectors, including the self-employed. The benefits last for one year in the case of the first and second child, and two years for the third and fourth child. The size o f the compensation i s based on the salary received by the individual in the month prior to maternity and upon length o f service: mothers with up to 3 months o f work qualify for 30 percent o f their gross wage; those with record o f 3 to 6 months - for 60 percent o f their gross wage and those with more than 6 months o f work record - for 100 percent o f their gross wage. The maximum benefit is capped at five average monthly salaries. 104. This benefit is generous relative to similar types o f compensation in the countries o f the European Union, including the new member states from Central and Eastern Europe, where most such benefits are financed from insurance contributions, as indicated in Table 11, implying a close link between wage and benefit funding source. Maternity benefits (paid leave) inthese countries are generally shorter (with a maximum o f one year) and the rate o f compensation i s generally lower, starting at less than 100 percent o f the wage and in some countries declining over time to encourage mothers to return to work. Table 11: mrces of Financing of Cash Maternity Benefits in the EU Mc iber States From global From a separate I From earmarked I From public Other models insurance maternity and maternity fund health insurance contribution sickness fund tax Belgium Slovenia Czech Republic Denmark The Netherlands- Spain Germany Latvia Hungary unemployment Cyprus France Lithuania fimd(earning Malta Italy Poland security insurance) Portugal Austria Slovakia UnitedKingdom Romania Luxemburg Finland - Estonia Bulgaria Greece fund for sick leave Sweden- fund for sick leave ource: MISSOC, 2 105. In terms o f its distributional implications, the wage compensation during maternity i s regressive: higher share o f the transfer reaches the non-poor than the poor. As the benefit is based on the employee's exit salary, higher benefits accrue to women with higher salaries. As a result, in 2007, benefits received innon-poor households were 50 by 43 percent higher than those received inpoor household^.^^ The rules of the program also potentially allow abuse of the system because the benefit is: (i)paid by the Government rather than the employer; and (ii) on the employee's salary inher last based month of employment, which could be inflated. 106. These flaws could be addressed without radically altering the program. Its perverse distributional implications could be addressed by lowering the cap on the benefit from five average salaries to, for example, three average salaries. The incentive to fraud could be addressed by lengthening the period by which compensation is based from one monthto, for example, three months. There is also a case for reducing the overall cost of the program by approximating the duration and level of benefits to those more typical of other EU and Eastern European countries: i.e., three months to a maximum of one year, with adeclining levelof compensationover time. 107. Expenditures on military veteran benefits' also consume a relatively large share (29 percent) of social spending (2008). As shown in Table 12, veterans benefits include a wide range of individual components, and include payments to military personnel disabled during war or during peace, civilians who became disabled due to military conflicts and families of deceased soldiers. Currently, the state budget in Serbia provides such benefits for approximately 45,000 veterans, 27,000 of them military disabled during war or peace time, 3,000 civilian victims and 15,000 family members. Source: MinistryofFinance, 2009 budget. 108. MLSP is considering a reform inthe system of veterans benefits. A new law i s under preparation. It will be critical to calculate the cost of any reforms before the law i s enacted, in order to ensure that they are affordable under the current economic circumstances. Some savings could be achievedby establishing an electronic registry and data management system to verify eligibility under each program. Inprinciple, benefits paid to dependent family members could be means-tested. Any savings arising from ''Living Standards MeasurementStudy. Serbia2002-2007. 51 reducing such benefits would presumably not be worth the political expense o f doing so. No changes inveterans' benefits are therefore recommended. 109. The same is true o f the birth grant which has a strictly pro-natalist objective. Itprovides a grant for each child, with the amount increasing steeply with each additional offspring. The grant for a second child i s four times higher thanthat for the first; the grant for the third, 1.75 times that for the second; the fourth 1.35 times that for the third. Although the objective o f this grant could be questioned, it is relatively cheap. Overall, birth grants account for eight percent of social spending and only 0.15 percent of GDP (2008). It is therefore recommended that the birthgrant be maintained at its current level. 110. The caregivers allowance targets people with disabilities. It was received by 3.1 percent o f households in 2007 and by 7.2 percent o f the poor household^.^' The program could benefitfrom some redesign: consolidating some o f the small and narrowly targeted elements (e.g., allowances for specialized equipment, transportation, rehabilitation, communications, etc.) into one benefit which reflects the individual medical and social inclusion needs o f the recipient, and linking benefits with social care and employment services. This should be on the agenda for medium-term reform. But in fiscal terms, reform i s not urgent. The caregivers' allowance accounts for only about 10 percent o f spendingon social programs and 0.19 percent o f GDP. ENTERPRISESUBSIDIES 111. Subsidies to enterprises (including farms) constitute a significant proportion o f government expenditure in the current period. The five largest central government programs consumed about four percent the 2008 central government consolidated spending in 2008, based on 2008 budget execution data.52 But the subsidy regime i s in transition. A large proportion o f subsidies have been used to facilitate the process of privatization, financing severance payments for workers in state owned enterprises. With some exceptions, the privatization process has been largely completed, and the level o f these subsidies can be expected to decline. Under the revised 2009 budget, a considerable volume o f resources are committed to new equity investments. These represent part o f the Government's economic stimulus efforts, and might be expected to decline. What will remain are o f subsidies to SOEs that have been slow to privatize-particularly in the mining sector-subsidies to the railroad and to agriculture, and subsidies to certain classes o f private firms---such as SMEs. In the current economic climate, there is also pressureto make soft loans to private firms inorder to stimulate the economy. '`Livingfigures StandardsMeasurementStudy. Serbia2002-2007. These exclude a variety of smaller subsidy programs and implicit subsidies inthe form of write- offs of uncollectedtaxes and social contributionsfrom state owned enterprises. They also exclude subsidies from local governments to their respective enterprises, which are financed from localgovernment's general budgets. 52 Trends in Major Subsidy Programs AGRICULTURALSUBSIDIES 60,000 1__ _._-_ 50,000 112. In sectoral terms, the largest program o f subsidies 40,000 consists o f subsidies to E 0 farrnsubsidies agriculture. As shown in 30,000 0 severance paymnts Figure23, in 2008, these I 0: equity investments accounted for about 40 percent o f 20,000 { /El subventionsto SOEs total central government subsidies. Since 2004, Serbia has been phasing out price supports for specific crops. Sugar and I 2007 2008 2009 tobacco subsidies have been eliminated, although support to Figure 23 the dairying industry remains. But the largest form o f agricultural subsidies consists o f the so-called `de-coupled area payment'. This was introduced in 2007 to replace former input subsidies and reflects the structure o f subsidies under the EU's Common Agricultural Policy (CAP). It consists o f a standard payment (equal to 120 per hectare) to every registered farmer, up to a maximum o f 100 hectares. As of 2008, only full time farmers who contribute to the pension system are eligible to receive the subsidy. 113. As an alternative to input subsidies or price supports, the decoupled area payment i s relatively efficient. It does not distort prices. But it i s also expensive and -in the current Serbian context-difficult to justify. Ineffect, it functions as a poorly targeted form o f social assistance for the rural population. Because payments are not means tested, there is presumably a high level o f leakage to higher income groups. At the same time, the de-coupled area payment may be inhibiting agricultural productivity. The 100- hectare cap on benefits discourages land consolidation. Already, the average farm size in Serbia i s less than one-quarter that inthe EU.53Consolidation will be necessary to permit Serbian farmers to compete with those in the EU. International experience, moreover, suggests that investment in rural infrastructure (primarily irrigation and roads) and the dissemination o f new existing technologies are much more effective in promoting agricultural growth than subsidies. 114. In principle, there is therefore a strong case for eliminating the decoupled area payment--at least until Serbia joins the EU. At that time, the costs o f the program will be paid by Brussels. (Eliminating the program would not jeopardize Serbia's 53 The average farm size in the EU, including leased land, is 27 ha. According to the Serbian agricultural census, the average farm size in Serbia is 3.5 ha. This figure excludes leased land, however. Data from the farm registry suggeststhat the EU-comparable figure is closer to 6 ha. 53 accession prospects, as the EU requires only that the system for distributingthe subsidy be accredited prior to the distribution of EUagricultural subsidies.) The program's role in supporting the incomes o f poor farm families could be replaced by a scaled-up MOP, as noted earlier. Alternatively, the program could be transformed into an income- support program for small farms, with the area payment subject to means testing. The political obstacles to doing so should not be underestimated. According to the 2008 LFS, roughly one-quarter o f the Serbian labor force i s engaged in agriculture-an important constituency that would be expect to resist such reform. Nevertheless, the Government has managed to impose an across-the-board reduction in the area based payment in 2009 and may be able to further reduce it on a targeted basis infuture years. SUBSIDIES TO STATE ENTERPRISES 115. Support to state- and socially-owned enterprises (SOEs) that are due to be privatized constituted about 22 percent o f central government subsidies in 2008. These include severance payments financed from the Development Fund, the Transition Fund and the Solidarity Fund, and credit lines to socially owned enterprises for restructuring. A large proportion consists o f soft loans provided by the Development Fund and payments by the Transition and Solidarity Funds, which finance severance in connection with the pre-privatization staff restructuring. 116. Overall, the level o f enterprise subsidies (excluding subsidies to railroads) has increased in nominal terms over the last four years, due to increased funding for severance. Costs are expected to decline however, as privatization proceeds. In 2008, 18 o f the 20 largest SOEs were offered for sale. Nine were sold and one liquidated through bankruptcy. Altogether around 240 companies were sold through different privatization models. Indirect subsidies, in the form o f tax and contribution arrears, have also been a problem, although the level o f such subsidies i s expected to decline as privatization proceeds. According to data from the Tax Administration, at the end o f 2008 the overall stock o f tax and contribution arrears o f 987 public enterprises was RSD13.37 billion, o f which twenty percent was owed by inactive companies. 117. The Government nevertheless confronts the costs o f subsidizing or privatizing the more intractable state enterprises. In the mining industry, Resavica and Bor are the two largest remaining companies that are slated for privatization or closure. Resavica i s the largest single recipient o f central government budget subsidies in Serbia (other than Serbian railways). It i s a complex o f nine mines employing 4,195 people across 8 municipalities. Ofthe nine mines, a maximum o f one-third are viable. (One mine has been out o f operation since the late-80s.) Direct subsidies for Resavica in 2008 amounted to 22 million or 0.06 percent o f GDP. The level o f subsidies per employee i s equal to average wage inthe country as a whole. Inaddition to direct subsidies, Resavica receives implicit subsidies in the form uncollected taxes and social contributions. Rough estimates suggest that these indirect subsidies are almost equivalent to the direct ones. 54 118. Boor receives far less in terms of direct subsidies (a soft loan of RSD39 million) but still benefits from significant implicit subsidies. Bor's existing stock of debt to state and private institutions (the latter constitute a contingent liability for the state) amounted to a record 670 million in 2008. Out of this, arrears on taxes, social Box 5: The DevelopmentFund In principle, the Development Fund exists to provide subsidized financing for the programs related to economic and regional development, development of SMEs, increasing competitiveness, and so forth. The average subsidized interest rate at around 6-7 percent compares favorably with the marketaverage of some 15-17 percent in 2008. During 2007, the DevelopmentFundapproved 1,22 1 loans inthe total amount of approximately208 millionor 0.89 percent of GDP. In practice, a large part of Development Fund financing goes to socially-owned companies, where less than ten percent of loans are beingrepaid; the rest being written-off in the process of pre-privatization restructuring. These loans are used to cover operating costs-wages, electricity, gas, etc.-rather than a new investment-and reflect the unprofitability of the borrowers. In 2008, some 33 million was spent on 114 loss-makingcompanies distributed across transport, metal, metallurgy, textile, chemical, electrical machinery, non-metal, construction, wood, tourism and other industries. Source: Ministry of Economy, Development Fund. contributions, and loan repayments due to the Development Fund amounted to 180 million. The Government embarked upon the privatization of the mines in 2006. Privatization advisers were hired and the company was prepared for privatization. However, the process has been stalled since then, particularly at Resavica. The Governmentshould restartthe process, restructuringthe company, closingthe least viable mines,andlaunchinga staffdownsizingprogram. 119. By far the largest privatization expenditure proposed in the initial 2009 budget was the Government's proposed equity investment in Zastava, a key component in the proposed sale of a majority stake in the company to FIAT. Although this was expected to reduce subsidies to firm in the long run, in the short run, it represented an extremely heavy cost. (The Government's proposed equity investment and loans to =Zastava were budgeted at RSD14 billion in 2009. This was equivalent to about RSD230,000 per employee.) Due to changing economic conditions, the agreement with FIAT has not yet been signed, although FIAT has made some initial investments in the Zastava plant. Under the revised 2009 budget, the allocation for the Zastava project has been reduced to RSD4 billion, with the remainder of the funds allocated to several subsidized loan programs, including liquidity loans to export-oriented firms (RSD4.0 billion); start-up loans to small enterprises (RSD3.2 billion); and subsidized loans to consumers for purchases of domestically produced consumer durables (RSD2 million). 55 120. Looking forward, Serbia's enterprise subsidies have beenjustly criticized for distorting markets, undermining the country's long term competitiveness, and wasting money on non-econo`mic enterprises. Serbia's EU ambitions, if not the force o f these arguments, will eventually force it to scale back its remaining sector specific subsidies. The European Commission (EC) has adopted a "State Aid Action Plan for 2005-2009" which seeks a relative reduction o f the overall state aid inGDP and reorientation o f aid to address market failures, rather than supporting specific industries. Inthe new EUmember states, subsidy reform has been a key component o f the pre-accession and post-accession reform agenda. If Serbia i s to follow this example, enterprise subsidies would have to be limitedto development aid, R&D, closures, and staff reductions. 121. To this end, the Ministry o f Finance i s in the process o f drafting a Law on State Aid. This i s intended to be consistent with the EU state aid rules and the institutional arrangements and system o f ex-ante control o f state aid specified in Serbia's Stabilization and Association Agreement with the EU54.The draft law, however, has scope for improvement. Under the current draft, the proposed Serbian Commission for State Aid Control would lie within the Government, being presided over by the Minister o f Finance. All Commission members would be Government appointees, while administrative, technical and other support would be provided by the Ministry o f Finance. The draft law envisages annual reporting to the Government only, thus limiting access by the public or the legislature. 122. As a whole, this approach fails to provide the commission with sufficient autonomy and protection from political influence. The Government should consider establishing the Commission for State Aid Control as an independent body reporting the Parliament, with professionals recruited through open, public competition. The role o f the Commission could be strengthened by expanding the present definition o f its responsibilitiess5 and authorizing it to provide preliminary opinions on any legislative proposal or strategy concerning state aid, which would be bindingprior to their adoption by the government. ROADS 123. Spending on roads-including spending financed from tolls and earmarked tax revenues-accounted for about five percent o f consolidated central government expenditure in 2008. In the immediate future, the Government confronts the need to 54 EC ProgressReport, 2008. ss These could include responsibility for: (i)the assessment of state aid proposals and aid schemes within annual and multi-annual state aid approval plans; (ii)monitoringthe implementation and effects of state aid granted and order the recovery of unlawfully granted state aid; (iii)collection the data on the use and effects of state aid granted; (iv) cooperation with the authority responsible for state aid to agriculture and fisheries in the preparationof annual reports on state aid; (v) cooperation in the budget preparation process with the authorities responsible for the preparation of the state budget and the budgets of regional and local self-governmentunits, in compliance with the separate law; and (vi) participation inthe preparation of draft proposals for laws and other regulations concerning state aid, as well as promotion and encouragement of improvementsinthe state aid system. 56 increase maintenance spending (particularly on regional roads), restructuring the debt of the state road agency, and providing counterpart funding to a major program of highway investments inCorridor X. The prospects for financing these expenditures from increased tolls and fuel taxes are limited. Instead, the Government's most promising option would be to extend the construction schedule for the Corridor X works and seek immediate efficiency improvements in the planning and execution of works by the state roads authority. THESERBIAN ROADS SECTOR The density of the roadnetwork in Serbia is similar to that of neighboringcountries.Serbia has a road network totaling approximately 40,845 kilometers-with 5,525 kilometers of main and primary roads, 11,540 kilometersof regional and secondary roads, and around 23,780 kilometers of local roads.56Road density in Serbia is 462 km per thousand square km, in line with most regional neighbors, but lagging behind Albania (657), Croatia(506), and Montenegro(500) and substantially behind the levels found in new EU member states and in OECD countries. On a second measure, road density per one thousand inhabitants, Serbia, with 5.4 km per 1,000 inhabitants, has a higher road density than Albania, but below that all other countries in the Western Balkans. Although there has been an improvementin the condition of the main and regionalroadnetwork in the last three years, about 40 percent of the network remains in poor or very poor condition Figure 24 illustratesthe results of a survey of the main and regional road network conducted in 2008.57This shows that while the majority of the motorways and main roads are now in good condition, slightly over halfthe regional road network ranks as poor or very poor. In addition, the design characteristics of much of the existing network in terms of speed and axle loads is below European standards, increasingvehicle operating costs, and reducingsafety and competitiveness. ''This 56 correspondsto what the nationalroadagency calls roadsof Class I,I1and I11respectively. These resultsare from a survey conductedin 2008 as part ofthe roadand inventorydatabase study under the World Bank hnded Transport Rehabilitation Project and became available in March2009. The survey defined road condition using the InternationalRoughness Index (IRI), but with different boundary values for different roadclasses. 57 90% 80% 70% 60% 50% 40% 30% 20% 10% I 0% Nbtorways Main Regional ' Total Figure 24: Condition of the Main and Regional Road Network, 2008 Source: PEPS. Road safety is also a major concern in Serbia. As shown in Figure 25, the number o f fatalities and injuries from traffic accidents has been increasing fairly steadily over this decade (although peaking in 2007). The fatality rate, at 4.8 casualties per 10,000 vehicles in 2008 remains about five times higher than that o f the best performing European Union countries, and compares unfavorably to several countries inthe region, as shown in Figure 26. 1.000 24,000 950 22,000 20,000 4-"850 3 900 18,000 ,$ - 16,000 iz 800 14,000 750 22,000 700 10,000 2003 2004 2005 2006 2007 2008 Figure 25: Trends in Injuries and Fatalities Sources: PEPS, Bank estimates P O x ) O 8 0 6 0 4 0 2 0 0 0 Albania BH Montenegro Serbia FYR Poland Czech Slovenia United Sweden (2008) Macedonia Republic Kingdom (2004) Figure 26: International Comparisonsof Traffic Deaths (deaths per 10,000 vehicles) Sources:PEPS, Bank estimates. 58 Box 6: Road Safety Initiatives The road safety situation in Serbia is serious and, against the background of rapidly increasing motorization, will deteriorate further unless urgent, well-orchestrated and appropriately funded evidence-based action is taken. Currently, the main institutions with governmental responsibilities for road safety are the ministriesof infrastructure, interior, justice, education, and health, as well as PEPS and the Public Health Institute. Currently, there are a number of gaps, including: (i) limited analysis of risk factors in transport, police and health sectors; (ii)little multi-sectoral understanding as to priority areas for action; (iii) poor collection andhandling of road safety data; (iv) lack of road safety management capacity; and (v) limited coordination between the various agencies. The seriousness of the problem is fully acknowledged by all key stakeholders, as is the need for further reforms in public policy, road safety strategy, legal and compliance frameworks, and road safety management capacity. The Global Road Safety Facility financed the "Review of Road Safety Management Capacity and Proposals for Investment Strategies" (RSMCR) inAugust 2007. The draft Road Safety Law (RSL), expected to be adopted in 2009, is the first major update in legislation since the 1980s and addresses many aspects of the EU Transport Acquis and recommendations given in the RSMCR. However, road safety funding arrangements are inadequate and lack transparency. The public sector as a whole needs to devote appropriate levels of resources if Serbia i s to improve its safety performance to match current Europeanpractice. Inthis regard, at present it is difficult to estimate the amount of public funds allocated to road safety and there is a need to change this, especially if there i s to be a rise in the allocation of funds allocated on improvingroad safety. The Serbian roads agency Public EnterprisePutevi Srbije (PEPS) is responsiblefor managingthe primary and secondary (regional) roadwork (Its mandate does not extend to tertiary roads or urban streets.) PEPS was established by the passage of the Road Law in November 2005, with effect from January 1, 2006. As shown in Table 13, it is largely financed from transport-related taxes and fees. Proceeds from the national investment plan (Le., Government privatization receipts) and foreign borrowing provide much of the remainder. Transfers from the central government's ordinary budget constitutedonly five percent of PEPS' receipts. 59 124. The overall level of PEPS' reportedexpenditure has increasedonly modestly inreal terms inrecentyears. Total expenditures increasedonly fourteen percent between 2005 and 2008. Table 13: PEPS: Trends in Sourceand Uses of Funds Expenditures on new construction have 2005 2006 2007 2008 remained constant, despite Actual Actual Actual Actual additional financing from Use of Funds the National Investment Maintenanceand Plan (NIP). Spending on periodicrehabilitation 27468 28,658 26,692 24,101 maintenance appears to % of GDP II1.62% II 1.44% II 1.15% II 0.88% have fallen, although the figures are distorted by payments on arrears to contractors for work carried out in 2006 and 2007. And by the conflation of maintenance, rehabilitation and upgrading expenditures. (Total spending on maintenance, National Investment rehabilitation and Plan(NIP) 0 2,261 5,799 6,046 upgrading, in fact Foreignborrowing 5,811 5,081 4,056 3,353 Donationsand increased sharply in 2006 privatizationproceeds 7,63 1 0 113 343 and 2007, but was not Fundscarriedforward 1,234 paid for until 2007 and TOTAL 28,077 27,970 31,75 1 40,579 2008.) For 2009, PEPS' proposed business plan ~~Source: PEPS. ~ assumes a substantial increase in maintenance, rehabilitation, upgrading, and construction expenditures. This would be accompanied by increased spending on Corridor X, which will be carried out by the road agency's daughter company, Koridor 10 D.O.O. 125. Inadequate maintenancehastens the deterioration of any road at an increasing rate until reconstruction is necessary, at considerably greater expense than any short term saving inmaintenanceexpenditure. Adequate maintenancecan sustain the pavement of a road for a period far beyond the original design life, depending on traffic, ameliorating the needfor any reconstruction. By contrast, inadequatemaintenance leads to an increase inthe rateof deterioration, hasteningthe failure ofthe pavement, andengenderinganeed to reconstruct the road, a need that could have been avoided. Heggie and Vickers (1998) report that reconstructing a paved road is three to five times more expensive than maintaining it, incurrent terms, and around 35 percent more innet present value " Heggie,Ian,&PiersVickers(1988) "CommercialmanagementandJnancing of roads" WorldBank Technical paper No. 409. Washington DC. 60 126. InSerbia, routine maintenance is underfunded. From 2005 to 2008, reported spending on maintenance (including related arrears) averaged RSD44 billion, after adjusting for inflation. While, this i s close to the level estimated as necessary for the maintenance o f main and regional roads, including routine winter snow clearing5' it i s inflated by the inclusion o f upgrading expenditures. As a result, it remains unclear how much o f the reported amount has actually been spent to maintain the road network, and how much has been spent on upgrading activities, which should be more appropriately categorized as capital expenditures. (In this regard, Article 59 o f the Law on Public Roads defines the attenuation o f gradients, curve straightening, the widening o f pavement or shoulders, and the enlargement o f at grade intersections as enhanced maintenance, whereas one would more logically regard those activities as capital expenditures.) Expenditures on maintenance per se are therefore lower than the reported figure. In addition, there are significant questions about the efficiency and efficacy with which PEPS uses the funds. 127. Inaddition to ongoing maintenance, the future financing requirements of the road sector include two forms o f capital expenditure: (i) that are necessary to clear those the current maintenance backlog and return the road network to good condition; and (2) additional capital expenditure necessary to keep pace with growing demands and ensure that the country develops a road network that contributes to economic development. 128. The total costs o f capital expenditures necessary to address the backlog o f maintenance expenditure have been estimated at 1.45 billion (RSD113.1 billion). This estimate is based on the unit cost o f specific activities using PEPS' 2008 price list, together with information on the current length o f the network under PEPS' mandate, new road condition data, and a policy objective o f achieving `good' conditions for all reconstructed roads.60 Willingness to accept a lower quality standard on lower categoryholume roads, would result in a correspondingly decreased costs. The immediate budget implications o f this figure depend on the time frame over which it would be implemented. As shown in Table 14, clearing the maintenance backlog would cost 290 million (RSD22.6 bn) per year if the work were completed in five years but 145 million (RSD11 bn) if it were completed inten years. Given the other claims on the budget, the ten year time horizon would appear to be preferable. ~~ " The latter calculation was made on the basis of PEPS' 2008 price list and assumes the following maintenanceregime standard: 40 mm asphalt overlay every 5 years on motorways and main roads and the same overlay every 7 years on regional roads. 6o A lower policy objective of only aiming for `fair' condition for secondary roads would obviously reduce this estimate. 61 Table 14: Cumulative Maintenance Backlog and Annual Expenditure Needs Cumulative Source: Bank estimates. 129. With respect to expansions of the network, the Government intends to focus on Corridor X. A series of improvements inthis corridor will enable Serbia to capitalize on its geographical position as a key transit country in the Pan-European Network. The total investment required for the modernization of infrastructure on Corridor X is significant, amounting to an estimated 2.079 billion.61 Out of the total, 1,303 million would be financed by international financial institutions, including 1,088 million from the World Bank, EBRD, and EIB credit currently under preparation. The Government's contribution will nevertheless be considerable, totaling 774 million over the construction period, although over half of this sum represents transfer payments in the form of VAT, taxes and duties.62 DIRECTIONSFORREFORM 130. Reschedule works on Corridor X. The fiscal impact of the Government's contributions to Corridor X will depend to a large extent on the phasing of construction works. The Government has statedthat the original intentionwas to complete the work in Works on Corridor X would include: (i)the constructionof a second 2-lane carriageway on 118 km of motonvay between HorgoS-Novi Sad (Corridor Xb) at an estimated cost of 223 million; (ii)the constructionof a motonvay on 98 kmof the sectionof corridor betweenNiSandthe border with Bulgariaat Dimitrovgrad (Corridor Xc) at a provisional cost estimate of 749 million; (iii)the construction of a motonvay on the corridor between Leskovac and the Macedonian border (Corridor Xd) at a provisional cost estimate of 658 million; and (iv) the completionof Belgrade Bypass, which would form an important section of Corridor X estimated cost 166 million. These costs include design, expropriation and construction costs and include VAT and supervision costs. They are likely to be revised after tender documentation has been produced, and as such remain provisional.On the other hand, expropriationcosts may be overestimated, as current revisions to existing legislation mean that public land can be offered instead of financial compensation. This could lead to significant downward revisions to estimated expropriationcosts. 62 The Government has recentlyagreedwith the World Bank that no VAT will be paid for the World Bank financed sections of CorridorX. As this merelyreduces VAT revenuesthat would be paidto the treasury, it does not constitute asavings to the Government. It remains to be seen whether the same will be agreedwith the EBRD and EIB. 62 four years. This is now acknowledged to be overly ambitious, ifnot unrealistic. Inlight of the time neededfor landacquisition, the lengthofthe tendering process, overall readiness for implementation, and the scale of the construction works, the objective of implementing and completing a 1.79 billion investment program over 4 years looks ambitious, if not ~nattainable.~~Expenditures in 2009 are instead likely to be a small fraction of what is included in the Action Plan. Land acquisition takes a minimum of 6 months, and the length of the procurement process for the IF1 financed sections of Corridor X is estimated to be a minimum of 6 months. This suggests that there will be little or no construction on these sections before 2010. In addition, PEPS funds earmarked for Corridor X land acquisition have yet to be released (as of mid-March 2009) as PEPS' 2009 BusinessPlan has yet to be approved by the PEPS board. 131. Given the likely time frame for implementation, three alternative scenarios for the phasing of Corridor X expenditures are proposed. Using provisional cost estimates, these phasing scenarios presume the works are completed in a 5 to 6 year period. All three scenariosassume that: (i) in2009 full expropriation costs and 10percent of construction costs for the Dimitrovgrad Bypass included in the Action Plan 2009; (ii) WorldBank-EBRD-EIBfundedsectionswilltakeoneyear longertobuildthan all the other sections; (iii)full expropriation and construction of the 20 km section of HorgoS-Novi Sad in the Action Plan take place in 2009; (iv) construction of Levosoje- Presevo in 2009 (see Table 15). Further assumptions are made to differentiate the scenarios: Scenario 1. In this scenario it is assumed that Corridor X is built over a 6 year period, with a limited amount o f expropriation and construction taking place in 2009 due to limited funds and implementation constraints. Additional assumptions are that: (i)23 million worth ofconstruction on Belgrade Bypass in2009, with the remaining construction evenly split in 2010, 2011, and 2012. (ii)full expropriation and design costs are made in 2010 for remaining World Bank-EBRD-EIB sections; (iii)the construction profile for the World Bank-EBRD-EIB sections are staggered such that the remaining 15 percent of construction costs are paid in 2010, 30 percent in 2011, 25 percent in 2012, 18 percent in 2013, and 13 percent in 2014; and (iv) the expropriation and construction profile for the remainder of the Horgoi-Novi Sad is 25 percent a year over 2010-2013. e Scenario 2. Inthis scenario it is assumedthat Corridor X is built over a 6 year period and uses the same set o f assumptions as in Scenario 1, with the exception that the construction profile for the World Bank-EBRD-EIB sections are staggered such that 25 percent of construction costs are paid in 2010, 35 percent in 2011, 18 percent in 2012, 15 percent in 2013, and 8 percent in 2014. In other words, in Scenario 2 there is significantly more constructionworks over 2010-2011for the World Bank-EBRD-EIB sections. e Scenario 3. In this scenario it is assumed that Corridor X i s built over a 5 year period. Additional assumptions are the following: (i)two year construction period for Belgrade 63 The same conclusionwas made in the World Bank Policy Note, "Options for the Development of the Road and Rail Infrastructureon Corridor X", August 2008. The costs for the works and phasing presented inthis chapter differ from that ofthe PolicyNote, given new information availablesince August 2008. 63 Bypass, with 50 percent expropriation and construction a year in 2009 and 2010; (ii)full expropriation and design costs are made in 2010 for remaining World Bank-EBRD-EIB sections; (iii)the construction profile for the World Bank-EBRD-EIB sections are staggered such that the remaining 40 percent of construction costs are paid in 2010, 40 percent in 2011,and 10 percent in both 2012 and 2013; and (iv) the expropriation and construction profile for the remaining sections of Horgol-Novi Sad is 50 percent a year over 2010-2011. These assumptions lead to significantly higher expenditures over 2010- 2011anda large drop in2012 and 2013, 132. These alternative scenarios lead to significant differences inthe expenditures over the next two years. For 2009, they range from 125 million (RSD 9.75 bn to 201 million (RSD 15.7 bn).For 2010 they range from 248 million to 343 mi1li0n.d)~ In Scenario 3 works are completed in 2013, but with high levels o f expenditure that appear incompatible with fiscal constraints, the time frame for land acquisition and IF1 procurement process, as well as overall implementation capacities o f Koridor 10 D.O.O.. The table below shows the cost to the budget o f the Corridor X works under the three scenarios. It assumes 1.3 billion o f foreign borrowing, with loan disbursements paralleling construction on foreign-financed sections. (Debt service on foreign loans is not included.) The projected financing gap i s largest in 2010 inall three scenarios, at over 343 million (including VAT), as this is when the expropriation costs will fall disproportionately inall scenarios and these costs are not funded by foreign borrowing. 2009 2010 2011 2012 2013 2014 TOTAL Scenario I 125 248 143 130 94 33 703 Scenario 2 135 271 153 106 88 20 703 Scenario 3 201 343 178 26 26 0 703 133. RefinanceArrears. In addition to the costs of maintenance and rehabilitation works and Corridor X construction, the Government also confronts the costs o f paying down PEPS' arrears. PEPS has accumulated significant arrears to contractors. Outstanding arrears to suppliers for investment and maintenance services totaled RSD31.8 billion as o f end 2008, slightly up from RSD31.3 billion in 2007.@The bulk of outstanding arrears, RSD29.5 billion, are owed to suppliers for maintenance works. According to PEPS, these arrears reflect decisions to go ahead with planned works despite a lack o f funds. Arrears in construction works also reflect lower-than- expected funds from the National Investment Plan, but also the decision to finance projects ahead o f elections and cost overruns. PEPS i s awaiting a decision from the Ministry o f Finance regarding a sovereign guarantee that would allow it to take out a commercial loan to repay RSD20 billion, the total level o f guarantees as per the 2009 64 Figuresinclude VAT. 65 These arrears are classified as operatingliabilitiesto localsuppliers inthe IndependentAuditor's Report for 2007 FinancialAudit o fPEPS. 64 budget. An additional RSDl.6 billion Recent and Projected Spendingon Roads (excluding foreign financed spending) of arrears to Jugopetrol i s to be written off, in exchange for Government forgiveness of an equivalent amount 70,000 owed to it by Jugopetrol. The remainder of arrears is programmed to 60,000 8a- be repaidina 5 year period.66 50000 uother c w debt service 0 netconstruction 134. Taken together, the 0 paymnt of arrears immediate budgetary implications of 8 rraintenance the proposed maintenance and rehabilitation program, the construction of Corridor X, and the resolution of PEPS' existing arrears are fairly significant. As shown, in Figure 27, Figure 27 total Government spending in 2009 on roads, including PEPS expenditures financed from own source revenues andGovernment counterpart contributions to Corridor X, would be about ten percent higher, inreal terms, than in 2008.67Spending would spike in 2010 with the increase in spending on Corridor X, and would the revert to its previous trajectory, gradually increasing as debt service on- Corridor X works comes due.68 135. PEPS' financial situation would remain precarious. As shown in Figure 28, the agency's cash flow deficit, while declining, would remain substantial even under the somewhat optimistic assumption used in the Figure 28: Trends in PEPS Net Cash Flow pr~jection.~'One way o f covering the immediate gap would be to implicitly 66 Out of the total outstanding stock of arrears, RSD31.8 billion, RSD20 billion is expected to be repaid through commercial loan, RSD1.6 billion will be written off, with the remainder, RSD10.2 billion, to be repaid over a 5 year period. 67 This projection assumes that the entire stock of arrears is refinanced at a real interest rate of 5 percent and five years maturity, and that foreign borrowing for Corridor X construction is financed at a real interest rate of 5 percent and a maturity of 30 years, with no grace period. Estimates for Corridor X are based on Scenario 1. The recently approved 2009 PEPS Business Plan foresees expenditures and revenues equal to RSD66,646, with no deficit. There are two main reasons why there is no forecast deficit. (1) toll revenue projections are optimistic given the economic contraction forecast for 2009; and (2) maintenance expenditures are significantly below what has been calculated as necessary by this Report. Expropriation for Corridor X (RSD2,305 million) is to be financed from PEPS' own resources, suggestingthat ifrevenues are below what is projected, there is a genuine risk of an accumulation of arrears or the posting of a deficit in2009. 65 lower maintenance standards on regional roads by continuing their neglect, which is what has tended to occur inrecent years. Such an approach essentially would allow such roads to continue to deteriorate, ultimately raising the costs o f restoring them to acceptable conditions. 136. RaiseRevenues.An alternative would be to raise additional revenues. PEPS' non-investment expenditures are financed by fuel excise tax, toll revenues, and annual vehicle registration charges, as i s the practice in other European countries. The rate o f the excise tax on petroleum derivatives is set by the Ministry o f Finance. A percentage o f the revenues are earmarked for PEPS for the maintenance o f state roads. PEPS' share has been increasing over time, from 10 percent in 2006, to 15 percent in 2008, to 20 percent in2009. 137. A recent report on road financing in Serbia noted that increasing fuel tax revenues was not a very promising source o f new revenue. As shown in Figure 29, prices for both diesel and petrol were relatively high, compared to other Balkan countries, even before the increases after 2006." The Government's April 2009 fiscal package calls for a further increase in fuel tax revenues o f about ten percent, which implies yet another price increase. Any increases on top o f this would need careful consideration interms o f affordability. Inprinciple, PEPS' revenues could be increased by raising the share o f the fuel excise tax that i s allocated to the agency. As noted earlier, it now stands at 20 percent. Given the central government's current economic difficulties, such a move would need to be considered carefully against the other spending commitments of the Government and other austerity measures. I 1 8 0,. . . . .. ..... . .. .... .... . . ... .-... ..... -. . .... .. .____ .,.., - no f QO 1 = ; c 0 80 2. 8 60 40 20 0 I Albania Bosnia & Kosovo NR Montenegro Serbia Slov~nia Croatia Ramarua Bulgaia Hungart Greece Hemgovlna Macedonia EZZC3Diersl ..IIPetrol ----Pelmi(iux) -Dissel[Lux) Figure 29: Fuel Price Comparisons (US centsper liter, 2006) Note : November 2006 prices. Source: International Fuel Prices 2007 (GTZ 2007). 69 In addition to the expenditure assumptions cited earlier, this projection assumes that PEPS will receive 20 percent of the proposed increase in excise tax on fuel and those revenues from the excise tax, toll revenues and other own-source revenue will continue to increase at the rate o f GDP growth over subsequent years. It assumes that maintenance expenditures will remain constant in real terms and that no construction works will be financed from PEPS own source revenues. 'ODiesel (Lux) and Petrol (Lux) stand for the diesel and petrol retail prices inLuxembourg respectively, and are usually considered the benchmark for new EUaccession countries. Source: GTZ (2007). 66 138. The possibility of raising tolls also appears limited in Serbia at this time. In the first half of 2006 toll rates for domestic vehicles were raised by 20 percent and a further 18 percent in February 2008. On February 22, 2009 the government decided to equalize the price of highway tolls for local and foreign vehicles, meeting a commitment it took on nearly seven years ago. The harmonization does away with the practice of charging foreign vehicles nearly twice as much as those registered in Serbia. This could potentially lead to a decline intoll revenue in2009, although PEPS is forecasting a rise in transit traffic, and therefore a rise in toll revenue collected from foreign vehicles, from RSD 7,804 million in 2008 to RSD 8,647 million in 2009.7' However, the early indications suggest that toll revenuesare likely to be, at best, flat. 139. At first glance, it might appear that there is scope to increase overall level of tolls, but such a step needs to be viewed in terms of its affordability and in terms of the rates charged in neighboring countries that offer competitive routes. As shown in Table 16, Serbia's toll rates as of ,February 2009 are in line with those of neighboring countries with similar levels of GDP per capita. In the case of Category IV vehicles (vehicles with four or more axles) the rates appear to be on the high side, but these vehicles are the ones that do the most damage to the road network, and any change inthis rate would need careful analysis. This suggests that there is limited scope for increasing toll rates, without compromising the competitive position of Serbia as a transit country for traffic. 140. PEPS can, nevertheless, take administrative steps to reduce costs and revenue leakage at tollbooths. A recent study of electronic toll collection (ETC) systems, funded by the Public-Private Infrastructure Advisory Facility (PPIAF), reviewed the existing toll collection system and made an assessment of reform It found that ETC was little used, so that the advantages of such a system have not been exploited.73Over the medium-term, the study recommends that a private sector organization take over toll operations using the concessionaire model. Inthe short term, it recommends several more immediate measures. These include including equipping 'old-style' toll lanes with ETC and offering discounts to drivers who use them; making better use of video surveillance systems to reduce the extent of fraud; and reducing the number of vehicles eligible for exemptions. It i s estimated that these measures could increase net toll revenues by ten percent. The report also finds that there is a relatively large number o f vehicles which are exempt from paying tolls and recommends that in the short-term there should be a more detailed investigation of these exemption^.^^ 71Inthe first halfof2007 there was a significant decrease intraffic volume due to waiting times inborders and truckers decided to reroute via Romania. This underscores the fact that Serbia faces a competitive environment for transit traffic and that increases in toll rates needs to take into account the impact it could have on traffic diverting to Corridor IV. 72 Ian Catling Consulting, Review of Existing Toll Collection System and Assessment of Reform Options, Draft Final Report, January 2009. 73 Evidenceelsewhere suggests these investmentsare botheconomicallyand financially viable. 74The reportfoundthat 7,500 vehicles did not pay tolls in2008, representing930,000 journeys representsa significantnumber of toll passages and hence a significant amount o f revenue which is not collected from these exempt vehicles. Inaddition,the 12 categories of exemptionneedto be reviewed. 67 Note: Exchange ratesfrom local currencyto Euro as of February22,2009. Nominal GDP per capita data is from the IMF World Economic Outlook October 2008 and are estimates. Sources: IMF,Bank estimates. 141. The prospects for increasing annual vehicle registration fees are good. Vehicle registration fees have not been raised since 2000 and appear low compared to other countries in the region. Table 17 presents the table o f registration fees for different category o f vehicles across countries in South East Europe. Serbia's rates appear to be relatively low across the board. Such an increase, while feasible, would not generate much revenues. Evening doubling the fees would generate only RSD400 million, or about 1.7 percent o f PEPS estimated financing gap. Table 17: Annual Vehicle Registration Fees, 2007 I I I-k-kz Country I Medium 1 1 Car I Bus I truck Trailer Albania 18 44+285 45+396 60+835 70+835 Croatia 16-110 600 265 530 345 Bosnia and HerzegovindRS I 72-160 I 240 1 260 520 465 HerzegovinaFBH 12-125 420 175 415 490 Macedonia 17-93 140 280 740 400 Serbia (2009) 1-33 29 77 180 130 142. Overall, the prospects for raising significant amounts o f additional revenue from transport related taxes and fees appear to be limited. The Government i s already committed to raising the tax on fuels. Further increases in highway tolls run the risk o f diverting international freight traffic to other countries-although there i s some scope for reducing costs and revenue leakage at tollbooths. Increases in vehicle registration fees, while feasible, will not raise much revenue. This makes it all the more important for PEPS to adopt measures to improve the efficiency o f expenditure. 143. UndertakeInstitutionalReforms. The third option is to achieve better outcomes for the same amount o f money. The lack o f a professional approach to asset .management, limited use o f formal techniques o f economic appraisal in project identification and prioritization, weakness in financial planning and the under-use o f 68 performance contracting all contribute to inefficient resource use and the waste of budgetary funds. 144. The establishmentanduse of an asset managementsystem is a prerequisiteto better maintenance planning. This involves regularly collecting and computerizing data on pavement conditions and traffic counts. An economic decision model such as the Highway Design and Management Model (HDM-4) can then be usedto identifypriorities for maintenance, taking into account overall budget constraints. An ongoing World Bank project is financing improvements to the computerized road database. This is expected to be operational by the first half of 2009. PEPS should use the new database and the Highway Development and Management Model to identify those priorities for funding with the highest returns in order to ensure that only the highest priorities are supported with public money. 145. Efforts should also be made to improve project planning. PEPS' current planning and budgetary process is weak, as evidenced by the large gaps between actual and planned road works. (See Figure 30). Maintenance, rehabilitation and construction activities should be planned in advance using a medium-termdevelopment plan (MTDP) for the network. While routine maintenance activities can be easily incorporated into annual plans, the medium-term expenditures for periodic maintenance, rehabilitation, reconstruction, upgrades and new construction should be planned over a longer time- frame (e.g., 5 years. I 18000 I 2005 2006 2007 2008 Figure 30: Planned and Actual Road Capital Expenditures (RSDmn, currentprices) 146. Third, PEPS should expand the use and the scope of performance-based maintenance contracts. Such contracts, in hybrid form, have been tested on a pilot basis in two regions, MaEva and Kolubara. The two pilot contracts, involving both winter maintenance and routine maintenancefor 1,200 km of roads, were signed inAugust 2004 and ran through August 31 2008. The experience was very encouraging, with results suggesting significant reductions in costs. Unit costs for winter maintenance, for example, declined by 24 percent. This approach is now being extended to all twenty-five districts of Serbia, but there i s significant opposition from the regional maintenance companies. In addition, the opportunity to extend the original pilot schemes in the two regions, interms of the scope of activities that are subject to output basedapproaches, has 69 been missed, with PEPS management opposing testing an extension within the World Bank financedproject, that could realize further savings. 147. Finally, over the medium term, the Government should reduce the size o f the network for which PEPS i s responsible. The recent Law on Public Roads clarified the legal status and respective responsibilities for the road network. As noted earlier, PEPS i s assigned responsibility for national, magistral, and regional roads (Classes Iand 11) while local governments are responsible for municipal or local roads. However, this reclassification o f the network was introduced without a prior inventory o f the road network and the current use o f each segment. Once a detailed inventory o f the road network i s completed, some o f the lesser-used segments now under PEPS' jurisdiction could be reassigned to local governments. Box 7: Will StructuralFunds HelpFinanceSerbia's TransportInfrastructure Investments? On 29 April 2008, the EU and Serbia signed the Stabilizationand Association Agreement (SAA) and the Interim Agreement on Trade and Trade-related issues. The SAA will be submitted to parliaments for ratification and the implementationof the Interim Agreement will start as soon as the European Council decides that Serbia fully co-operates with the InternationalCriminalTribunalfor the former Yugoslavia(ICTY). Inits 2007-2013 Budget, the EU allocated 10 billion for the Instrument for Pre-Accession Assistance (IPA), to facilitate candidate and pre-candidate countries to strengthen their institutionsto utilize EU structural funds as soon as they become full members. The P A has replaced all previous instruments of support to countries with a candidate and pre-candidate EU membership status. The IPA consists of a total of five specialized components for: a) assistance in the process of transition and capacity building aimed at meeting the EU membership criteria and strengthening of administrative and legislative capacities; b) assistance in regional and cross-border cooperation; c) regionaldevelopment; d) development of humanresources; and e) rural development and agriculture.Beforeit acquires the status of a candidate, Serbia will have access only to the first two components, that is, transitionand marketreforms funding, although infrastructureexpenditure is allowed under Component 1. For this reason, for the time being, it is unlikely that the EU structural funds could help Serbia's large transport infrastructureneeds. 70 RAILWAYS 148. Subsidies to the state railway company Zeleznice Srbije (2s)75 cost the central government budget about RSDl1,12 billion in2008. Eventhis level is insufficient to cover the operating losses of the railway or to permit an adequate level of maintenance of the railway infrastructure or rolling stock. While freight traffic has been growing, passenger traffic has been declining, and the financial performance of both has continued to suffer. The Government is nevertheless planning to make a major investment in high speed rail in Corridor X. This chapter proposes a more modest and cost-effective set of initiatives 149. Overall rail traffic volumes have risen steadily since the start of the decade, after following precipitously in the late 1990s. As shown in Figure 31,the rise in traffic volume has been uneven, with a steady rise in freight traffic over the 2000-2007 period accompanying a decline in passenger traffic over the same period. In the first half o f 2008, due to the unfolding economic crises, freight traffic decreased by 5.2 pe~cent'~, driven by a 24 percent reduction indomestic freight traffic and static international traffic. Passenger traffic continued to decline (by 10.4 percent in terms of total numbers and by 6.1 percent interms of passenger km.) The limited quantity and poor quality of passenger rolling stock, combined with unreliable and low quality services and increasingly competitive road transport explains much of the decline inpassenger traffic. The average commercial speed in2007 was on 43.3 W o w . Passenger and Freight Traffic 5,000 4,000 3,000 2,000 1,000 0 2000 2001 2002 2003 2004 2005 2006 2007 Figure31: Trends in Passengersand FreightCarriedby ZS (mns ofpassengers/mn tons offreight) 150. Productivity despite improvements remains poor by EU standards. As shown in Figure 32, ZS has substantially reduced staffing levels over the past six years (by a 75 Under legislation governing the sector, public rail infrastructure (Le., trackage) is owned by the Republic and is open to all licensed rail transporters. ZS is, at present, the exclusive manager of Serbia's rail infrastructure as well as the primary provider of bothpassenger and freight services. '` Comparedto the same periodin2007. 71 total of 12,000). Incombination with the growth infreight, the ratio o f traffic units, (tons of freight, numbers of passengers) to staff has steadily improved. But as shown in Table plan, 2s had been planning to reduce its staff to 19,400 by end-2007, in line with its 18, the ratio remains considerably higher than regional averages. Through its redundancy - obligations under a loan agreementwith the EBRD, but has failed to achieve this target in part for lack of funding for severance payments." I 300,000 I 34,000 Traffic units/stad I 250,000 -- --tStaff -- 32,000 c -- 30,000 5`5200,000 -- 150,000 -- -- 26,000 S Qe -- 26,000v, 100,000 -- I- -- 24,000 50,000 -- -- 22,000 I o i I- 20,000 2000 2001 2002 2003 2004 2005 2006 2007 I Figure 32: Trends in Traffic Units per Staff and Staff Levels Source: 2s. 151. The railway also ranks poorly in terms of equipment productivity (tons of freight per wagon per day), reflecting the age and poor quality of the rolling stock and traction units, and in terms of network usage. As shown in Table 18, Serbia's passenger traffic density (passenger-kilometers per kilometer of line) is among the lowest in the region, outranking only those of Bosnia and Albania. 77 Accordingto the Business Plan 2008, the severance payment for 1,800 employees is 13.2 million; for the existing 765 staff in excess of the target, a pro-ratacalculation suggests that this figure would cost an estimated 5.6 million. This highlights the point that any reduction in wage costs will require a significant one-offbudgetary effort. 72 Table 18: Inte ational Indicators f Railway Pro( ctivity, 2007 Freight FreightWagon Locomotive Staff Traffic Passenger Productivity Productivity Productivity Density Traffic Density (ton (ton (Traffic (Ton km per (Passenger km Country kdwagodday) kdlocomotive/day) Units/Staff) km of line) per kmof line) France 3,498 80,3 16 758,302 1,439 2,825 Germany 2,608 60,405 717,545 2,685 2,205 EU 2,276 39,350 683,260 1,806 1,806 Hungary 1,895 1 1,542 227,783 1,044 783 Poland 1,612 33,722 491,034 2,243 880 Croatia 1,444 40,130 383,989 1,313 592 Czech Republic 1,417 23,821 432,001 1,788 722 Serbia 1,389 37,669 253,967 1,112 200 Bosnia- Herzegovina 1,197 21,381 196,122 1,260 53 Bulgaria 1,034 21,454 409,148 1,171 602 Romania 692 18,820 3 15,820 1,265 697 Albania 205 2,547 52,235 125 121 'ource: UIC. 152. Maintenance expenditure has been inadequate in the past, which coupled with the age o f both the rolling stock and the infrastructure, has lead to a very poor quality o f service. The existing railway infrastructure i s in poor condition. There are about 3,256 km o f railways used by i s , o f which only 38 percent are electrified, and 8 percent are d~uble-tracked.~'The railway network i s aging, with an average age o f 38 years o f age. Line speeds exceed does not exceed 60 kmkr on 52 percent o f the network, while only 2.6 percent o f the network has a line speed that exceeds 100 km/hr. Insufficient investments into maintenance has caused the instability and deformation o f tracks, eroded tracks, rotten sleepers, and to preserve safety, temporary speed restrictions have been introduced, which is actually something o f a misnomer, as limited resources meanthat the restriction stays for a lengthyperiod o f time. About 56 percent of the main lines last had a major overhaul more thanthirty five years ago. 153. The rolling stock fleet i s aged. The electric locomotive fleet amounted to a total o f 144 locomotives in the 2007. But the active fleet i s smaller than the notional inventory, with only about 60 percent o f the total inventory o f electric locomotives in operation. The current fleet o f active freight wagons, which totals approximately 3,13 8 in working order, has an average operational age o f 29 years o f age, close to the end o f their operational life. AS estimates that 66.4 percent o f its rolling stock, 57.1 percent o f passenger cars, and 49.7 percent o f freight wagons are over 30 years old and it is clear that a considerable portion o f this fleet will need to be retired from operational service in the next few years. In addition, the technical characteristics o f the current fleet are inconsistent with current and projected market demands. A modernization plan, partly funded by an EBRD loan recently negotiated, includes the delivery o f 38 self-propelled electric passenger cars. 78Total network lengthis 3,809 km, but this includes 334 km inKosovo andMetohija, 39 lanused only as factory sidings, and 180km which are out of service. 73 154. 2s has calculated required maintenance--assuming the average overhaul of tracks on wooden sleepers is 20 years and a track length o f 3808 km--amounts to 190.5 km a year--while the annual average line overhaul over 2002-2006 was only 24.5 km a year, creating an annual average backlog o f 166 km. In 2007, infrastructure expenditures, at RSD1.18 billion (US$16.4 million) was only 63.3 percent of what was planned, and in the case o f maintenance for tracks and facilities on tracks, expenditures reached only 25 percent o f planned levels. In the case of rolling stock, maintenance services reached 65.4 percent o f planned levels in 2007. Current maintenance expenditures, ceteris paribus, will lead to increased deterioration o f assets, with worsening service quality and increased risks of Table 19: Infrastructure and Rolling Stock Maintenance in 2006-2007 (RSDthousands, current prices) Source: ZS and Bank staff, 155. financial position of 2s had deteriorated and reached unsustainable levels in 2000. The Despite some improvements, 2s financial performance remains weak. The restructuring efforts and the recovery o f freight traffic have led to a subsequent improvement, but 2s commercial revenues remain insufficient to cover even labor costs. The company's operating losses are covered, in part, by Government subsidies, which now exceed its revenues from operations. The budget subsidy covers 54 percent of 79The numberofaccidents declined in2007 from 33 to 30, with one train crash, 23 train derailments, four train collisions, one clash, among others, which resulted in 61 fatalities. The poor state o fthe infrastructure means that traffic speed restrictions are imposed on large segments o f the network. 74 working costs and totaled SDNl1.126 billion (1.02 percent of consolidated central government expenditure) in 2008. As shown in Figure 33, this level of subsidies Romania, and Croatia). In addition, 2s incurs accounting losses in the form of (expressed as a percentage of GDP) is less than in neighboring countries (Bulgaria, uncompensated depreciation, debt servicing, and other nonoperating costs. The working ratio, a key financial indicator, improved from 195 percent in 2005 to 179 percent in 2007, but remains unsatisfactory.'' Cumulative financial losses totaled RSD93.7 billion in 2007. 156. One reason is that passenger tariffs are low. In 2007 the Government approved a price increase inpassenger tariffs of about 20 percent, the first such rise since 2004. Tariff levels nevertheless remain considerably below those o f neighboring countries. As shown in Table 20, in 2007, average revenue per passenger km, without adjusting for relative purchasing power, in Serbia was 9.3 percent lower than in Croatia and 37.6 percent lower thaninBulgaria. One ofthe factors affecting passengerrevenue in 2s are the number of passengers who are not paying for tickets and the low level of fines and limited fine revenue collected (see Box S), as well as the large number of individuals possessingdiscount cards. Information dating from July 2008 reveals that: (i)15 percent of passengers have a P-2 discount card, which allows free travel to railways workers from home to work; and (ii)10 percent of passengers have a P-4 card, issued to railway staff, retired persons and members of their families, who have unlimited travel at 75 to 90 percent discount. 157. Railway passengertariffs also remain lower than alternative transport modes, in particular buses. 2s calculates that bus transport prices, when compared to regular second class train prices are about 70 to 150 percent higher, although the quality of bus service is considerably higher. This suggests that, within the limits imposed by differences in reliability and quality, passenger tariffs could be raised without a significant loss in passenger volumes. (Direct comparisons between different freight tariffs are not possible, as these: (i)vary by type of commodity, wagon, and service; (ii)discounts are offered to clients; and (iii) average distance of transportation of the the structure and volumes of commodities vary from one railway to another.) *'This i s defined as the operating cost before depreciation and provisioning divided by the operating revenue, excludingbudgetsupport. 75 .-....- . ..... - . ._- -._...._ ......- . . ._.. .. . .............- ..... ........ ...- . ... Turkey Serbia Bulgaria Romania Croatia I 0.00% 0.20% 0.40% 0.60% 0.80% I.OO% Figure 33: International Comparison of Budget Support to Railways (% of GDP, 2007) Note: Includingdebt service paidby the Ministry of Financeraises the total for Croatiato 1.5 percentof GDP. Source: Bank staffestimates. Table 20: InternationalComparison of Average Revenueper Traffic Unit Passenger traffic Freighttraffic Traffic Operating Average revenue / Operating Average revenue / Country Year Units revenue Traffic Unit Traffic Units revenue Traffic Unit (million (million (million local (million currency) currency Local cent Euro NTM) currency) 1I1 local LocalEurocent currency Bulgaria 1- I 2007 I 2,423 Romania 7,417 1 2007 Croatia 336 0.25 3.37 3,305 724 0.22 2.99 I I 2007 I 1,611 371 0.23 3.14 3,870 764 0.20 2.69 1,563 1.72 2.02 3,565 7,150 2.01 2.35 Serbia I 1,626 1.94 2.45 4,232 8,817 2.08 2.63 2007 I 762 1,716 2.25 2.87 4,552 8,85 1 1.94 2.48 Source: Bank staffand ZS. 76 Box 8: Cracking Down on Scofflaws Train operators are responsible for protecting their own revenue and as such must devise a revenue protection strategy. On long-distance journeys it is usually possible for on board train staff to examine passenger tickets. However, on urbanand suburban routes, stops are fiequent and trains are busy, so it is not always possible to check every passenger's ticket between stations. Visual inspection of tickets by staff at stations before allowing access to platforms is labor intensive and can be inconvenient to passengers. The operation of an effective penalty fares scheme is one means by which train operators can tackle ticketless passengers, offering a visible deterrentto all passengers and punishing violators. The penalty fare scheme of Serbian Railways foresees that passengers who avoid ticket control or do not have the ticket and fail to report to the conductor shall pay the ticket charge and a fine equal to RSD 250 (US$3.4), while passengers who report to conductor on not having the ticket shall pay the ticket charge and RSD 120(US$1,6). Inaddition, for the suburban train, Beovoz, the fine is double the normal price. A practice used in a number of railways is to charge a specific penalty or twice the full fare as a penalty, whichever is the greater. In the case of Serbia, as the penalty is set at an absolute level, the more expensive the ticket, the cheaper it is to pay a fine. According to Serbian Railways, revenue generated from fines in 2005, 2006, 2007, and the first three quarters of 2008 amount to RSD 108,065 (US$ 1,483), which is a very low figure. The important point is that the value of penalty fares schemes is not solely in the revenue obtained from the total penalty fares charged, but also in the deterrent effect and resulting increase in normal ticket revenue. Levels of fare evasion are estimated by Serbian Railways to be between 1 to 10 percent. Given that 130 ticket offices have been closed, this increases the number of passengers who are obliged to pay on board and may have raised the number of passengers traveling without tickets. A recent initiative, No Passengers without Fare, has raised public awareness of the issue, but has not been matchedby an increase in the penalty fare or controls. It appears likely that a revision to the penalty fare scheme, a public awareness campaign and increasedcontrols, could help reduce non-payment and generate revenue. 77 DIRECTIONS FORREFORM 158. Both freight and passenger services will continue to play a key role in Serbia's transport system. Several measures are required to ensure that it does so at a cost affordable to the Government. 159. Reduce staffing levels. To reduce operating unnecessary costs, 2s should proceed with its target for reducing staff to no more than 19,400 and should ensure that there i s adequate funding for severance payments. Once a network rationalization program is developed and implemented (see below) staffing levels will need to be reviewed again. 160. Abandon underused track. 2s will also need to define and implement a network rationalization program. This will enable the company to reduce its excess railway track and concentrate on the segments o f the network where rail performs the most useful transport role. This rationalization i s expected to bring its rail traffic density, currently at 1,3 11,63O/TU/l km o f network, some way towards levels inthe EU countries which average (3,600,000/TU/ 1 km o f network). This will also reduce the maintenance backlog and costs considerably. 161. Rationalize passenger services and subsidies 2s will have to reduce non- profitable passenger services, as definedinthe Plan o f Activities Regarding Reduction o f Non-Profitable Services. The scope o f these cuts will depend upon the Government's desire to continue to provide subsidies for socially necessary services. The legal framework for passenger subsidies i s specified in the new Law on Railways, which became effective on March 1 2005, and which i s consistent with the structure envisaged with the relevant European Commission Directives (Dir. 1991/440, 2001/12/13/14.) The Law created a legal framework for the introduction o f a public service obligation (PSO). This would require the Government (a term which could include local governments) to provide explicit subsidies to ZS to cover losses on passenger services undertakenin the public interest. Production o f the methodology for setting transport prices was finalized by a consulting firm in 2008, but the Government still needs to adopt regulations governing the content o f PSO contracts, the method o f calculating the amount o f the PSO and the procedures for paying it." Delays to date mean that the introduction of a new contractual relationship between the Government and 2s has not taken place, and that subsidies are not tied to the performance o f specific services. On the basis of the methodology as well as the projections set out in the Strategic Plan for 2006-2010, it was estimated that RSD3 billion (US$42 million) would be needed to finance the PSO for the second half of 2008, funded through local and central governments. If these projections are accurate, on annual basis the PSO would be about RSD6 billion, considerably below the current subsidy, which exceedsRSD I O billion. 78 Box 9: The New EBRD Loan:ReformConditions InMay ofthis year (2009), the Government signed a100 loanagreementwithEBRD to finance the purchase self-propelled suburban passenger cars. As a condition of effectiveness, the loan requires the Ministry of Infrastructure to submit the methodology for calculatingthe amount of the Government's public service obligation. This is being prepared by consultants and is expected to be completed by end May, 2009. The Ministry of Infrastructureis then requiredto issue a by-law establishingthe PSO by December31 2010. The loan further requires:(1) the adoptionof separate business plans for each of ZS's businessunits (by December31 2010); (2) the adoptionof cost accountingfor each sub-unit of the railroad's passenger services-international, local, urban, sub-urban--(by December 31 2010); (3) the preparation by the Ministry of Infrastructure of a by-law introducingtrack access charges and the publicationof network statement on AS website (by December 31 2010); and (4) an increase in the proportion of passenger service costs to be covered by tariff revenues to 30 percent (by 2013). If successfully implemented, these will support the proposedreforms inthe financingofthe railway sector. 162. The Transportation Institute (CIP) has analyzed the data on revenues and costs for each segment of the network, to calculate the viability o f passenger services. This analysis has been updated to reflect the new accounting system introduced in 2s. The Government should finalize process for prioritizing which loss making services to supportthrough the PSO and analyzing whether socially necessary passenger services are most cost effectively provided by rail. 163. Increase passenger tariffs. At the same time, ZS should increase passenger tariffs on lines where subsidies are not justified for social reasons. As noted above, existing passenger tariff levels are low, by regional standards. ZS should also take steps, suchas increasingpenalty fares, to reduceevasion. 164. self-sustaining company with its existing stock of debt. As of the end of 2008, 2s had Restructure debt. ZS has little prospect of emerging as a financially long-term debt and liabilities to international financial institutions of 199.2 million (excluding debt owed to the World Bank.) The Government is now servicing the debt on the railroad's behalf. ZS is nevertheless required to reimburse the Government for debt Article 23 of the Decree on the Establishment of the 2s specified the need for a Financial service paid on IF1 loans contracted before 2000, except those from the World Bank. Consolidation (Le., debt restructuring) Program. A Financial Consolidation Plan was preparedby the Committee appointed by the then- Ministryof Capital Investments-now the Ministry of Infrastructure-and was submitted inJuly 2007 to the relevant ministry.It has not yet been adopted. Due to data constraints, it is not clear that ZS is, infact, paying any of the debt service it owes to the Government. If it is, the cost to the budget of forgiving this obligation could be considerable. Inany event, the Government should not do so prior to a confirmed reformplan for ZS has beenagreed upon and until it is assured that that ZS is in a position to cover its operating costs and remaining financial 79 obligations in the future. Premature debt forgiveness could reduce management's incentive to makethe necessary reforms. 165. ReconsiderdesignspeedinCorridorX. Despite the precarious physical and financial condition o f the existing network, the Government plans to begin a major modernization and reconstruction o f the Corridor X railways. The stated objective i s to raise speeds in this corridor to 160 km/hr, which will require among other things, electrification and the construction o f a second railway track on a number o f sections.82 The total value of the necessary investment is estimated at about 1.7 billion to 2 billion, for total track length 1,O 16 km.83 166. The Government and 2s should review the technical specifications for Corridor X, and in particular, the need to raise design speeds to 160 W.Project costs could be considerably reduced if speeds were raised to only 120 km/hr. This would be more in line with the composition and level o f current and projected traffic. Freight traffic constitutes over 80 percent o f total traffic and revenue, traffic which does not need high speeds. It would also be consistent with the design speed o f the existing infrastructure. All electrical-technical plants on all lines on Corridor X, with the exception o f the NiS-Dimitrovgrad line allow for speeds o f 120 kmihr and 55.4 percent o f considerably cheaper. According to 2s estimates, total superstructure and substructure lines were designed with route elements for speeds o f 120 km/hr. and would be costs are estimated at 880,000 per track km for a speed o f 120 kmihr,but rise sharply to 2.12 million per track km in the case o f 160 Given the cost differential o f upgrading infrastructure from 120 km/hr versus 160 M h r and the important maintenance backlog on other part o f the railway network which will require upgrading over the medium to long-term, alternative options should be c~nsidered.~~ An agreement has been signed with GermanRailways to prepare a master plan for the modernization o f Corridor X. This could review alternative design speeds. This master planwill be usedas the basis to access EUfunds for future investments inthe sector. 167. Address alternativeconstraintson railtraffic flow. Another option, which may even be preferable in the short to medium term and likely to have higher economic returns, would be to defer large scale investments in upgrading the rail 82 The National Road and Rail Infrastructure Development Planfor the Republic of Serbiafor 2008-2012 sets out the road and rail infrastructureprojects that are considerednational priority objectives. 83 The Government's Action Plan 2009 for the Construction of Rail Corridor X was presented on 30/01/2009. It calls for work to start on three rail sections for a total cost of Euro 136.3 million. All three sections are to be financed by EIB credits in the amount of 80 million, with the remainderto be financed from domestic resources. 84 A comparison with the costs quoted by a railway company operating in the region suggests that these unit costs are on the high side, and may be applicable only to those parts of the infrastructure which are in the poorest condition. These alternative costs for scheduling a railway line overhaul for 1 km of track include 210,000 for materials, which is half of the cost quoted by AS, and an additional 110,000 of labor costs. " Inthisregard,theargumentechoesthatintheWorldBank,"Options fortheDevelopmentoftheRoad and Rail Infrastructure on Corridor X", August 2008. 80 infrastructure on Corridor X, and focus instead on making only those investment necessary to address current speed restrictions and reduce the physical and institutional impedimentsat the border crossings. The requiredmeasures include: (i) relocation of the change of locomotives for freight trains and the related train technical checks (brake testing) from border crossing points to the nearest marshalling yard; (ii) implementation of information technology solutions to facilitate advance processing by railways and border agencies; (iii) promotion of joint processing of freight trains by Customs administrations at marshalling yards; (iv) improving of scheduling to build on the first three points. The improvement intrade facilitation at the border i s likely to have a greater impact on the operating speed of transit traffic for modest cost, than significant investments in upgrading at this time. A program of investment to address all the speed restrictions, investing to return to the current design standard of 100 km/hr and 120km/hr, and improving border crossing times, would appear to be the more logical choice inthe circumstances. 168. PostponeWorks. The time frame for the investments of Corridor X also need to be reviewed, given the financial costs of finalizing the restructuring process of i s , the investments on the Corridor X highway, and the Government's overall fiscal constraints. The Government should consider delaying the railway Corridor X rehabilitation until the road works in Corridor X are completed. In the interim an independentfeasibility study should be commissioned to evaluate costs andbenefits of all the proposedimprovement options. 81 Box 10: InlandWaterways Given its geographical position, Serbia has the potential for intensive river transportation as well as its canal network. The Serbian waterways network comprises approximately 1,600 km and includes, among others: (i)the Danube river over a total length of 588 km; (ii)the Sava river, whichjoins the Danube at Belgrade is currently restricted in use, but normally provides some 207 kilometersof navigablewaterway leadinginto Bosniaand Herzegovina,Croatiaand Slovenia; and (iii)the Tisa, which contributes 164 kilometers of navigablewaterway leading to the Hungarianborder.During the 1990s, inlandwaterway transport fell into poor condition due to a number of factors.Navigation is also hinderedat several locationsdue to unexploded ordnance, the poor state of navigation locks, the state of existing ports, among other factors. While waterborne traffic increased in EU countries by 12 percent during 1970-1998, river transport in Serbia decreased in 1990-98 by 40 percent. Regarding annual transport volume and capacity, the most important ports are Belgrade, PanCevo, Smederevo and Prahovo.Most of the ports are connected with main railway lines and roads or are very close, which has strategic and logistic importance.Only ports Belgradeand PanEevo have container terminals, althoughthe containertransportvolume is very low. Inlandwaterways are an importanttransportmode in Serbia. They are the second most widely used form of freight transport after railroads. Inorder to help define the government's strategy in the sector, a Master Plan and Feasibility Study for Inland Waterway Transport for Serbia was preparedin 2005, with the aim of presentingthe current situation and preparingfeasibility studies for: (i)restoring unhindered navigation; (ii)rehabilitation the Serbian waterway transport network; and (iii)preparing a port development plan. The Master Plan makes an assessment of priority projects and provides a cost estimate for proposed river improvement works to solve bottlenecks. Initial construction costs for these improvements have been estimated at 203 million for the Danube, 58 million for the Sava, and 20 million for the Tisa. The Directoratefor Inland Waterways, PLOVPUT, is responsiblefor public works in relation to maintaining navigability, marking the waterways, developing and maintaining river information systems, and developing international and interstate waterways. Central government allocations to PLOVPUT are small, at about US$3 million in 2007-2008. The budget allocation for PLOVPUT in 2009, at RSD 205.6 million (US$2.8 million) remains inadequate to finance needed rehabilitation works. In addition, PLOVPUT has a little over US$400,000 of own revenues, but the total volume of funds allocated means that larger investments need to be financed by international financial institutions. Serbia's transport strategy recognizes the importance of inland waterways as a transport mode, while acknowledging that the infrastructure is inadequate. However, budgetary allocations are heavily focused on the road network and railways and have not provided the necessary funding for upgradinginfrastructure. 82 SUMMINGUP 169. The economic crisis will force Serbia to cut expenditures. The downward turnineconomy and the dwindling of international credit would require the Government to reduce spending even if deficits were macro economically sustainable. And they are not. Inthe short term, the Government's best option is to pursue the measures specified in its agreement with the IMF: across-the-board cuts in the major economic categories o f spending, particularly wages and capital works. Cuts inpensions would also be desirable ifthey were politically feasible. But these are not long term solutions. Over time, they would lead to a continuing deterioration o f services. They do not address structural sources o f inefficiency. The Government therefore needs to start on more fundamental reforms, even ifthe payoff, infiscal terms, i s not immediate. 170. The proposals discussed in this report have different fiscal impacts over different time horizons and differing degrees of impact on the efficiency o f public services. In some cases, such as changes in pension parameters, the timing and scale o f fiscal impacts can be determined with some accuracy. In other cases (such as the introduction o f new payment mechanisms for health care) it cannot. This final chapter o f the report summarizes the report's recommendations, estimates their fiscal impacts where this i s feasible, and suggests immediate steps to be taken in conjunction with the 2010 budget. 171. Pensions.In terms o f overall fiscal impact, the most important reforms will be in pensions. Extending the nominal freeze in pension benefits through 2010 would yield savings equal to about 3.5 percent o f consolidated central government expenditure. Subsequently reverting to inflation-only indexation until such time as the replacement rate i s more in keeping with levels in EU countries; limiting early retirement costs by reducing the number o f years a worker can retire early and reducing pension benefits for such workers regardless o f their years o f contribution; and raising the retirement age for women to match that o f men will have significant long term impacts on the Government's pension obligations. Their short term impacts will be smaller. Because changes cannot be imposed retroactively, limits on early retirement and increases in the retirement age for women will only affect new retirees. The impact o f indexing to inflation (as opposed to wages) will also appear inthe longer term, as the nominal growth in wages outpaces the rate of inflation. (In the short term, real wages may, in fact, decline.) 83 Table 21: Summarv of Measures and Fiscal Conseauences Total Spending Annual (RSD bn.) Saving* 2007 2008 Pensions 340.16 386.4 II 0 freeze nominal pension benefits thru 2010, then index to inflation II II II3*5% II 0 Dhase in minimumage and reductions for earlv retirement -- schools in the same municipality Social assistance 54.8 I 58.9 I increase MOP and child allowance (1-0%) Agriculture 25.0 31.2 means-test area-based payment 0.9% Economy 25.3 32.1 0 close, privatize mines 0.1% 0 scale back softhnrecoverable loans to social enterprises 0.3% Roads 44.0 61.O 172. Health and education. Both the health and education sectors suffer from overstaffing and the inefficient use of material inputs. Inthe short term, efficiency gains can be achieved by administrative fiat. In the health sector, the HIF should close 86 Figure for 2007 is sum of former Ministry of Education and Sport and current Ministry o f Education. 84 underused primary clinics and reduce staffing in both primary clinics and hospitals. It should reexamine its benefits package and its use o f high-technology. But it should also begin to address the overall incentives confronting primary care physicians and hospital directors. To institutionalize incentives for efficiency, it should abandon input-based budgeting and shift to capitation-based financing for primary care, and DRG-based financing for hospital care. This fiscal impact o f introducing capitation based financing for primary care cannot be determined with any accuracy. A rough idea o f the potential impact o f introducing DRG's for hospital payments is provided by Hungary's experience, where the introduction o f DRGs reduced spending on acute care by 14 percent. If Serbia were able to cut total hospital spending by a similar proportion, the savings would total about RSD12 billion or about one percent o f consolidated central government spending. 173. Inprimary education, the Ministry of Education, acting through its regional offices, could institute the consolidation o f under-enrolled primary classrooms within individual schools over the 2009 summer school break. The savings would be fairly modest: about 0.3 percent o f consolidated government expenditure. At the same time, the MOEcould initiate the closure o funder-enrolledschools andthe transfer o f their students to other schools within a given municipality. To this end, the MOE could proceed with its proposed centrally-driven rationalization plan. Alternatively, it could enlist local governments inthis effort, by shifting to an enrollment based payment system and giving municipalities greater leeway in deciding which schools to close and where students should be transferred. Closing schools i s likely to have considerably more impact that consolidating classrooms within schools alone. On an annual basis, the impact i s calculated to be about one percent o f consolidated central government expenditure. Taken together, the two measures would reduce consolidated central government expenditure by about 1.3 percent per year. 174. Social assistance. Inprinciple, there is a case for introducing means- testing into the Government's two most expensive social assistance programs: maternity and veterans benefits. If this succeeded in cutting the costs by half, it would save the equivalent o f 1.2 percent o f consolidated central government expenditure. But it i s not clear that such a step would be worth the political cost-or that means testing would actually produce this result. A more modest proposal would be to lower the cap on maternity benefits and reduce its duration to no more than one year. 175. The Government should instead focus on increasing spending on programs that are already means tested: the MOP and the child allowance. The recession is likely to further reduce the incomes o fthe poor and addto their numbers. The Government should consider increasing level o f MOP and child allowance benefits. And it should prepare to fund an increasing number o f beneficiaries. This would not be particularly expensive. Doubling funding for both programs would represent an increase o f only one percent in consolidated Government spending. 176. Subsidies. There are potential savings in subsidies to agriculture and enterprises. The principal agricultural subsidy (the area based payment) i s now costing 85 the Government about 1.9 percent of consolidated spending per year. If means-testing reduced that by half, the savings would be equal to 0.9 percent o f consolidated spending. Closing and or privatizing the remaining state owned mines would save about 0.15 percent o f consolidated expenditure. Terminating subsidies to socially owned enterprises would save 0.3 percent. 177. From a fiscal standpoint, the more immediate issue i s the cost of the Government's proposed program of soft loans to private firms and consumers. While the scale o f the proposed investment in the FIAT/ZASTAVA venture has been reduced, it has been replaced by an increase incredit subsidies. This may bejustified in light o f the present economic crisis. But fr.om a longer term perspective the Government should reconsider the level o f subsidies that it intends to provide (in the form o f equity, subsidized or unrecovered loans, and tax breaks) to attract future private investors. At this point the numberis not large. Butit couldbecome large. 178. Roads. In the roads sector, the time required to complete land acquisition and other preconstruction activities on Corridor X i s likely to reduce the level o f Government contributions to this project in2009 and 2010. Adjusting the timetable for road works in Corridor X (from scenario 3 to scenario 1) would reduce costs to the Government budgetg7by about 0.5 percent o f consolidated government expenditure in 2009. This would, o f course, merely shift these costs into future years. This could nevertheless be wise, given the Government's immediate fiscal constraints. 179. With respect to expenditures on road maintenance, there appear to be few immediate options for savings, beyond stretching out the timetable for addressing the existing maintenance backlog. The company also confronts the prospect of paying off a large stock o f arrears, although the proposed commercial loan will allow these costs to be spread out over several years. The Government's planto increase fuel taxes will help pay for some o f this, as would improvements in toll administration. (Raising the fuel tax alone i s expected to generate RSDl billion, assuming PEPS continues to receive 20 percent o f the revenues of the tax.) But estimates o f even a minimal program o f road maintenance and rehabilitation still imply a budget subsidy to PEPS for road maintenance equal to about 1.8 percent o f consolidated central government expenditures.8g Over longer term, institutional reforms in PEPS-better identification of maintenance and rehabilitation priorities, timely payment o f contractors (which will reduce future bid prices) and wider use ofperformance contacts reduce the needfor subsidies. 180. Railways. Inthe rail sector, immediate savings could be achieved by postponing major works in Corridor X. Postponing planned expenditures in 2009 by one year would save RSD4.4 billion. or 0.4 percent o f consolidated e~penditure.~'Again, this Excluding costs financed by foreign loans. 88 Of this about one-third would be required to pay the estimated debt service on PEPS' current stock of arrears. This figure refers to savings in the Government's own contribution and excludes expenditures that would be financed from foreign loans. 86 would delay, rather than eliminate these costs. Permanent savings could be achieved by reducing the proposed design speed o f the proposed works from 160 km/hr to 120 kmihr. This would reduce the costs o fthe project by 60 percent, or RSD86 billion. 181. In terms of recurrent expenditures, there are potential savings in reducing railway subsidies.These are now equal to 0.8 percent o f consolidated central government expenditure. How far subsidies can be reduced will depend, in part, on the impact o f the proposed cost-cutting measures--further reducing staff, abandoning under-used track, terminating underused passenger services-and on the Government's willingness to raise tariffs. It will also depend on the costs upgrading what remains and the Government's interest in retaining certain passenger services for social reasons. Cutting subsidies in half, would reduce expenditures by 0.4 percent o f consolidated central government expenditure. 182. In total, the immediate savings yielded by these proposals are fairly significant. Summing only the savings that can be roughly quantified yields a total equal to about eight percent o f consolidated central government expenditure. The proposed increase in the MOP and child benefits, along with the start o f debt service on PEPS arrears, would reduce that figure to 6.7 percent. Although the fiscal impact o f structural reform will become more evident over the medium term, their most important impact will be on the efficiency o f public service delivery. These reforms will stand Serbia in good stead even after economic growth resumes. 87 MAP SECTION IBRD 34847 19°E To 20°E 21°E Kiskoros To Szeged SERBIA H U N G A R Y SELECTED CITIES AND TOWNS Subotica Kanjiza To 46°N Arad OKRUG (DISTRICT) CAPITALS Bajmok Senta POKRAJINE (PROVINCE) CAPITALS Backa Kikinda To Topola Ada Timisoara Sombor NATIONAL CAPITAL Apatin RIVERS Sivac Veliki Becej Kula ROMANIA MAIN ROADS V O J VTisaO D IElemirA N RAILROADS CROATIA Temerin To OKRUG (DISTRICT) BOUNDARIES Timisoara Backa Zrenjanin POKRAJNE (PROVINCE) BOUNDARIES To Palanka Vinkovci INTERNATIONAL BOUNDARIES Novi Sad To Sid VrsacVrsac Vinkovci Fruska Gora Alibunar This map was produced by the Map Design Unit of The World Bank. 45°N Ruma Indija The boundaries, colors, denominations and any other information shown on this map do not imply, on the part of The World Bank Group, any judgment on the legal status of any territory, or any Pancevo Bela Crkva endorsement or acceptance of such boundaries. Sremska To Carbunari Sabac Mitrovica To To Zeleznik BELGRADE Kovin Bijeljina Sava Gura Vail Zminjak Danube Prnjavor Pozarevac Vlasic Obrenovac Smederevo Golubac Kladovo Loznica Planina Jovan Deli Mladenovac Drina Velika- Velika Homoljske Majdanpek Petrovac P lanina Negotin Ljubovija PValjevo Plana To S E R B I A Zagubica Sarajevo ovlen Maljen oraMva To BOSNIA Gornji Beljanica Bor Vidin 44°N Milanovac Kragujevac 44°N AND Rogacica Ravni Gaj Cuprija ´ To Cacak Vidin HERZEGOVINA Uzice Jagodina Zajecar Pozega (Svetozarevo) Paracin ´ Rtan To Gorazde Zlatibor Soko Kraljevo ZapadnaJuhor Banja Morava Knjazevac Zaglavak Ivanjica Midzor (2168 m) Jastr Velebac iki Prijepolje Javor Cemerno Aleksinac Krusevac Usce Zelgin Nis Bela Balkan Sjenica Raska Prokuplje Palanka Golija Su Beloljin JuznaMorava Planivna a Pirot Mts. Novi Pazar Sit Radan 43°N Rogozna Kopaonik cian Prepolak Leskovac 43°N To Kosovska Vlasotince To Podgorica Mitrovica Sofia To Vucitrn Sofia MONTENEGRO Pec´ Pristina Priboj Surdulica BULGARIA Daravica B (2656 m) ile K O S O V O Vranje mirD Gnjilane Bosilegrad To Orahovac Bujanovac Pernik Dakovica Urosevac Ljuboten Presevo (2486 m) SERBIA Lake Scutari Prizren To To Sar nina To Kumanovo Kukes Pla Tetovo To Skopje 0 25 50 75 Kilometers ALBANIA FYR 0 25 50 Miles MACEDONIA 19°E 20°E 21°E 22°E AUGUST 2006