Report No. 30741-RU Russia Fiscal Costs of Structural Reforms April 5, 2005 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank TABLEOFCONTENTS EXECUTIVE SUMMARY ..................................................................................................... .xi Chapter 1. Principles o f StructuralReforms Financing....................................................... 1 B. A. Recent Fiscal Trends. 1998-2003 .................................................................................. 1 5 C. Fiscal Costs o f Key Structural Reforms......................................................................... Costs o f Structural Reforms: Examples of Reforms inthe Civil Service, Pension .......................................................................................... D. System, and Residential Housing 8 E. The Challenge o f Financing StructuralReforms: Can Fiscal Rules Help? .................13 F. Implications for Fiscal Policy ...................................................................................... Conclusion ................................................................................................................... 18 22 Chapter 2. Fiscal Costs o f Civil Service Reform inthe Russian Federation..................... 25 A. Employment and Financing of Core Govemment Administration and the Civilian Public Sector........................................................................................................................ 25 B. 29 D. Data SourcesandKeyAssumptions............................................................................ C. A Framework for the Assessment ofReformImpact.................................................. Civil Service Reform Priorities and Progress to Date ................................................. 32 E. 35 F. Civil Service Reform Scenarios................................................................................... G. Outcomes o f Civil Service ReforminLight o f Cross-Country Comparisons .............45 50 H. 51 Overview o f RisksAssociated with ReformImplementation ..................................... Summary o f Simulations: Analysis o f Selected Reform Scenarios............................. I. 58 Conclusions and Recommendations ........................................................................... -59 Chapter 3. Fiscal Costs o f Reforms inthe Housing and Utility Sector ............................. 65 A. Background.................................................................................................................. B. 67 C. Recent trends inhousing and utility tariffs .................................................................. Govemment Involvement inFinancing Residential Housing..................................... Summary o f the Earlier Reform Efforts inthe Housing andUtility Sector.................65 -69 D. 75 E. F. Aggregated costs structure inresidential housing ....................................................... Model for Simulating Budget Implications from Increases inHousing Costs ............77 80 -83 H. G. Reform Scenarios Identified for Simulations ............................................................. Assumptions on Future Utility Costs........................................................................... 86 I. 91 J. Main Simulation Results.............................................................................................. Conclusions and Policy Recommendations ............................................................... Simulations for Phasingout Housing Privileges (lgoty) .............................................. 98 K. 102 Chapter 4. ImplicitFiscalRisksinthe RussianPensionSystem .................................... 105 A. B. 105 106 Reform Scenarios Identified for Simulations ............................................................ Background................................................................................................................ Introduction................................................................................................................ D. C. 112 Main Macroeconomic and ReformAssumptions ...................................................... 114 F. E. Demographic Assumptions........................................................................................ 122 Simulation Results ..................................................................................................... 125 iii G. Summary and Conclusions ........................................................................................ 136 References.......................................................................................................................... 139 Annexes ............................................................................................................................. 145 Annex 2.1 Definitions on Civil Service Reform ............................................................... 145 Annex 2.2 Categories o fFederalCivil Servants inthe Russian Federation..................... 149 Annex 2.4 Intemational Public Sector Pay and Employment Data, 1996-2000" ............152 Annex 2.3 Classificationof Civil Servants Rewards........................................................ Annex 2.5 Assumption on Core Government Administration Attrition Rates.................153 155 Annex 2.6 Interpretationand Analysis of Results: Impact of IndividualFactors ............157 Annex 3.1 Macroeconomic Framework Used for the Analysis ofFiscal Costs of Structural 171 Annex 3.2 MainData Sources andData Assumptions..................................................... Reforms.............................................................................................................................. 173 Annex 4.1 Sensitivity Analysis For PensionSimulations ................................................ 177 180 Annex 4.3 Main simulationresults for the pensionreformanalysis ................................ Annex 4.2 Simulation Methodology................................................................................. Annex 4.4 Summary o f pensionbenefitrules inthe current system................................ 183 199 LISTOFFIGURES Figure 1.1: Public Debt Projections.......................................................................................... 5 Figure2.1: Public Sector Employment inSelected OECD Countries and Transition Economies .......................................................................................................................... 28 Figure 2.2: Total Fiscal Costs of Selected Scenarios.............................................................. 53 Figure2.3: Fiscal Implications of Scenario 20: Core Government Administration and Civilian Public Sector .............................................................................................................. 54 Figure 2.4: Fiscal Implications of Civil Service Reform: Role of Administrative Adjustments. .......................................................................................................................... 54 Figure2.5: Total Fiscal Implications for Scenarios with the De-Linked Implementation o f Pay and Administrative Reform Components ......................................................................... 58 Figure 3.1: Housing and Utility Costs to Consumers under Different Scenarios in2006 ......-90 trillion rubles ........................................................................................................................ Figure 4.1: Trends intax bases for unified social and personal incomes taxes, 2000-03, 110 Figure 4.2: GDP Growth by Scenario, Group 1.................................................................... 115 Figure4.3: Dynamics o fthe Taxable PayrollShare inGDP, Group 1................................. 117 Figure4.4: Real GrowthinTaxable Wages, Group 1Scenarios.......................................... 120 Figure4.5: Economic Activity o fPopulation, Group 1Scenarios ....................................... 121 Figure4.6: Real InterestRate, Group 1Scenarios................................................................ 121 Figure4.7: System Dependence Ratio ................................................................................. 124 Figure4.8: PensionReplacement Rate, Group 1Scenarios ................................................. 125 Figure4.9: USTRate (Contribution Rate) Neededto Ensurethe Replacement Rate o f 30 percent, Group 1 Scenarios .................................................................................................... 126 iv Figure4.10: Ratio BetweenAverage Pension and Pensioner's Subsistenceminimum. Group 1 Scenarios ........................................................................................................................ 126 Figure 4.11: Balance inthe Base andNDC components of the Pension System, Group 1 Scenarios ........................................................................................................................ 129 Figure4.12: PensionReplacement Ration, ComparisoninScenarios Groups 1 and 2........131 Figure4.13: RatioBetweenAverage Pension andPensioner's Subsistence minimum.......132 Figure4.14: System DependencyRatio, Comparisonin Scenarios Groups 1and 3............134 Figure4.15: PensionReplacement Ratio, ComparisoninScenarios Groups 1 and.............135 FigureA2.1: Total Public Employment inthe RussianFederation ...................................... 147 FigureA2.2: Civil Servants Rewards Classification ............................................................. 152 FigureA2.3: Sensitivity of Civil Service ReformCosts to RealWage Growth...................157 FigureA2.4: Sensitivity o f Civil Service ReformCosts to RealWage Growth (for Core Government Administration Only)........................................................................................ 158 FigureA2.5: Total Fiscal Costs o f Civil Service Reform: Radical versus Moderate Pay Reform ........................................................................................................................ 159 FigureA2.6: Fiscal Costs of Civil Service Reform: Radicalversus Moderate Pay Reform 160 FigureA2.7: FiscalImplications for Federal and Sub-Federal Budgetsfor Radical and Moderate Pay Reforms .......................................................................................................... 162 FigureA2.8: InternalDecompressioninFederalExecutive HQ-BasedCivil Service: Radical versus Moderate Pay Adjustment .......................................................................................... 163 FigureA2.9: Internal DecompressioninFederalExecutive Civil Service: Radicalversus Moderate Pay Adjustment...................................................................................................... 164 Figure A2.10: Distribution o f Total Fiscal Burden o f Civil Service Reform Scenarios, Dependingupon Pay Reform Pace........................................................................................ 165 uponPay ReformPace........................................................................................................... FigureA2.11: Distribution o fFiscal Burdeno f Civil Service Reform Scenarios, Depending 166 FigureA2.12: Total Fiscal Costs ofCivil Service Reform for Different Scenarios of Administrative ReformImplementation.,.............................................................................. 167 FigureA2.13: Fiscal Costs of Civil Service ReforminCore Government Administration for Different Scenarios o f Administrative Reform Implementation ........................................... 168 FigureA2.14: Fiscal Costs of Civil Service ReforminCore Government Administration: Administrative Reform Scenarios.......................................................................................... Dynamics of Cash CompensationandNon-Wage Expenditure Increase for Different 168 FigureA2.15: Fiscal Costs o f Civil Service ReforminCore Government Administration for Different Scenarios o f Administrative Reform Implementation ........................................... 169 pensionand pensioner's subsistence, 2030............................................................................ FigureA4.1: Indifference curves for the pensionsystem: the ratio betweenthe average 178 FigureA4.2: Indifference curve for the pension system: compensatory relationship between an increase inthe payroll share and cuts inthe UST rate...................................................... 178 V FigureA4.3: Indifference curve for the pensionsystem: replacement ratio inthe PAYG system. 2030 ........................................................................................................................ 179 LISTOFTABLES Table 1.1: Expenditure Trends inRussia in 1997.2003. EnlargedGovernment Budget ..........3 Table 1.2: Public debt sustainability estimates. Debt-to-GDP ratio. ........................................ - 4 Table 1.3: Summary o f IncrementalAnnual Fiscal Costs for the Set of Analyzed Fiscal Reforms .......................................................................................................................... 13 Table 1.4: Matrix of the Proposed Fiscal Rules for Russia .................................................... 22 Employment in2002................................................................................................................ Table 2.1: Structure o f Core Government Administration and Civilian Public Sector 25 Table 2.2: Core Government Administration Employment inRussia in 1994-2002 .............26 Table 2.3: Structure o f Civil Service Employment in2001................................................... -27 Table 2.4: Cash CompensationandNon-Wage Expenditures on Core Government Administration and Civilian Public Employment in2002....................................................... 29 Table 2.5: Key Macroeconomic Assumptions ........................................................................ 37 Table 2.6: Estimated Public-Private Pay and CompensationGaps, 2002 .............................. 39 Benchmark Positions inthe Federal Civil Service .................................................................. Table 2.7: Civil Servants' Perceptions: ExpectedPay and CompensationLevels for 40 Table 2.8: Attrition Rates Expectedinthe Education Sector, 2001 ...................................... 43 Table 2.9: Assumption on Pay ReformPace, Measuredas a Share of the Overall Planned Gap Covered .......................................................................................................................... 45 Table 2.10: Possible Scope o f Pay and Administrative Reforms ........................................... 47 Table 2.11: Description o f the Civil Service Reform Scenarios IncludedinSimulations .....48 Table 2.12: Fiscal Implications o f Civil Service Reform: Summary Results(Increase in Expenditures GDP, p.p. as comparedto 2003) ........................................................................ 49 Table 2.13: Structure of ExpenditureIncrease inYear 2010 for Selected Scenarios.............55 Table 2.14: DetailedBreakdown o f the Increase inFiscal Costs o f Civil Service Reform ...56 Federal Executive Civil Service Headquarters as of 2010....................................................... Table 2.15: Fiscal Implications o f More Radical Pay Adjustmentsfor "Decision Makers" in 57 Table 3.1: Structure of Housing Financing.............................................................................. 69 Table 3.2: Financing ofthe Housing Sector ............................................................................ 70 Table 3.3: Cost Recovery and Collection Rates, 2000-02 ....................................................... 71 Table 3.4: Financial Indicators for the Housing and Utility Sector......................................... 71 Table 3-5:Average Household Expenditureon Housing and Utilities .................................. 72 Table 3.6: Structure of Budget Spending on Housing andUtility Services to the Population (main programs of direct budget support) ............................................................................... 72 vi Table 3.7: Estimates for Total Fiscal and Quasi-fiscal Support to the Housing and Utility Sector in2003 .......................................................................................................................... 75 Table 3.8: Total Volumes o f Available Financing inthe Sector, 2003 ................................... 75 Table 3.9: Inflation and Price Growth inHousing andUtilities, 1998-2000........................... 76 Table 3.10: Estimates for the Effecto f "Deferred Inflation" inthe HUS (non-energy part), 1997-2002 .......................................................................................................................... 77 Table 3.11: Cost Structure inthe Housing and Utility Sector, per 1 Square Meter o f ResidentialHousing, as of end 2002........................................................................................ 77 the Level o f Government ......................................................................................................... Table 3.12: BudgetExpenditureson Utility Services Provided to Budget Organizations, by 79 Table 3.13: Macroeconomic Assumptions: Average GDP and Real HouseholdIncome Growth for 2004-06 ................................................................................................................. 83 Table 3.14: Scenarios for Housing andUtilityReforms, Selected for Simulations ................85 Table 3.15: Real Growth inEnergy Tariffs for Households in2006 Relativeto 2002, Depending on Scenario ............................................................................................................ 86 Table 3.16: Additional Growth inResidential Tariffs to Ensurethe Eliminationo f Cross- Subsidization .......................................................................................................................... 88 Table 3.17: Effect on Costs o f Efficiency Gains inReform Scenarios .................................. 89 PeriodRelative to 2002............................................................................................................ Table 3.18: Various Factors of Growth inUnitHousing Costs, Accumulated Growth for the 90 Table 3.19: Results o f Simulations for the ThirdYear o f Reforms: Status Quo Scenarios -- Slow Reforms inHousing and Utility Tariffs, No Eliminationo f lgoty ................................. 93 Reforms inHousingand Utility Tariffs, but without Eliminationof lgoty ............................. Table 3.20: Results o f Simulations for the ThirdYear o f Reforms: Scenarios with Advanced 97 Table 3.21: Resultso f Simulations for the ThirdYear o f Reforms: Scenarios with the Advanced Reforms inHousing, Utility Tariffs, and the Eliminationo f lgoty ...................... 101 Table 4.1: Basic Parameters o f the Kussian Pension System, 1995 - 2002 .......................... 108 Table 4.2: Structure o f the total payroll................................................................................ 109 Table 4.3 : Mainpreferentialregimes for social insurance contributions, 9 months o f 2003 ..... ........................................................................................................................ 111 Table 4.4: Assumptions on Macroeconomic Parameters o f the Base Scenarios (Group I) ..115 Table 4.5: Additional Assumptions inScenarios with ReducedUST Rates. (Group 11)......117 Table 4.6: Payroll Share inGDP of Various Countries. Percent .......................................... 118 Table 4.7: Real Wage Growth inVarious CEE Countries. 1995 - 2000. Percent ................120 Table 4.8: Base Demographic Projections (scenarios from Group I-IV) ............................. 123 Table 4.9: System DependencyRatio................................................................................... 124 Table 4.10: Realpension as a percentage o f the average wage at the time of retirement.....128 Table 4.11: Some results o f sensitivity analysis for 2030: equivalence table ....................... 132 Table 4.12: Potential impact o f an increase inimmigration on the pension system..............136 vii Table 4.13: Summary o f the simulation results for replacement rate and potential fiscal costs.. ........................................................................................................................ 137 Table A2.1: FederalCivil Servants inthe RussianFederation. with the Current Legislation .................................................................................................. Categories inAccordance 149 New Legislation..................................................................................................................... Table A2.2: Tentative Classificationof Civil Service Positions inAccordance with the Draft 151 Table A2.3: Significant Administrative Reform (Attrition Rates) ....................................... 155 Table A2.4: Fair Administrative Reform (Attrition Rates) .................................................. 155 by Demographic Projections (2001- 2010) ........................................................................... Table A2.5: Assumption on Attrition Rates inCivilianPublic Sector Employment Adjusted 156 Table A2.6: Fiscal Costs of Civil Service Reform for the Selected Scenarios..................... 161 Table A2.7: ExistingPay Gap and Proposed Pay Increases inFederalExecutive HQ-based Civil Service, times................................................................................................................ 164 Table A3.1: Basic Data UsedinSimulations and RelatedAssumptions............................... 173 Table A4.1: Total affordable replacement rate...................................................................... 183 Table A4.2: Affordable average replacement rate inthe pay-as-you-go pillar ..................... 184 Table A4.3 : Affordable average replacement rate inthe fully fundedpillar......................... 185 Table A4.4: Overall affordable average replacement rate adjusted for the personal income tax ........................................................................................................................ 186 o f informal wages inthe total payroll) ................................................................................... Table A4.5 : Real overall affordable replacement rate (replacement rate adjusted for the share 187 Table A4.6: Ratio o f the overall affordable average pensionto the minimumsubsistence level ........................................................................................................................ 188 Table A4.7: Additional fundingneededto maintain the replacement rate at 30% ................189 Table A4.8: Average replacement rate inthe pay-as-you-go system ifthe pension indexation rules remain to be basedon the currently effective legislation.............................................. 190 Table A4.9: Balance of the base pension component............................................................ 191 Table A4.10: Balance o f the NDC pension component......................................................... 192 Table A4.11: Overall balance of the PAYG pension (base+NDC) ....................................... 193 Table A4.12: GDP growthrate .............................................................................................. 194 Table A4.13 : Real annual growth intaxable wage (reflecting decline inshadow wages) ....195 Table A4.14: Share of Taxable PayrollinGDP .................................................................... 196 Table A4.15: SystemDependency Ratio (ratio o fpensioners and employed) ..................... 197 Table A4.16: Tax rates for pension contributions to the NDC and fully funded components, depending on income and age, as percent of annual personal income .................................. 200 ... V l l l ACKNOWLEDGMENTS This report was prepared inresponse to a request from the Ministry o f Finance o f the RussianFederation. It was produced by the joint team ledby Lev Freinkman and included the staff of the World Bank and several Russian think tanks. Chapter 1 was prepared by Alexander Morozov (ECSPE) and Lev Freinkman (ECSPE). Lorenzo Figliuoli and Antonio Spilmbergo (both IMF) provided useful comments on the earlier draft o f the Chapter. Chapter 2 was prepared by Yelena Dobrolyubova (ECSPE) and Lev Freinkman (ECSPE). Nikolay Klishch (Consultant) prepared a background paper on the policy implications and risks associated with the implementation o f public administration reform. The team i s grateful for the comments and suggestions receivedfrom Andrey Klimenko, John Litwack (ECSPE), and Neil Parison (ECSPE). The Chapter also benefited from the discussions with Mr.Dmitriev, First DeputyMinister for Economic Development and Trade, Mr. Sharov, Head of Department for State Regulation inEconomy, Ministry for Economic Development and Trade, Mr. Petrov, Head of Civil Service Department, Ministry of Labor and Social Development, and Ms. Lamm, Head o f Financing of the Government Apparatus, Ministry of Finance, as well as other government officials. The earlier draft of the Chapter was discussed at the seminar at the Higher School o f Economics and the Moscow State University. Comments presented by John Langenbrunner, Tatiana Loginova, and Mary Canning (all ECSHD) were very useful for preparationo f the report. The team i s also grateful to Ms.Daria Kononova, Department for Macroeconomic Forecasting, Ministry for Economic Development and Trade, for her suggestions on the draft report. Chapter 3 was prepared by Lev Freinkman (ECSPE), Sergei Sivaev (Institute of Urban Economy (IUE), Moscow), Alexei Rodionov (IUE), and Irina Starodubrovskaya (Institute of Economy inTransition). The team also benefited from advice and suggestions by Alexander Puzanov (IUE), Anastasia Alexandrova (WE) and Emin Askerov (WE). Comments and advice were provided by Arvo Kuddo (ECSHD), Peter Ellis (ECSIE), Ellen Hamilton(ECSIE), Peter Thomson (ECSIE), and Ruslan Yemtsov (ECSPE). Chapter 4 was prepared by Lev Freinkman (ECSPE) and Dmitri Pomazkin (Consultant) with contributions from Saeed Batkibekov and Pavel Trounin (Institute of Economy in Transition (IET), Moscow). The backgroundpaper on eligibility rules under the existing special regimes o f pension contributions was prepared by Ekaterina Bolshakova (IET). Pavel Kadotchnikov (IET) helped with the development o f macroeconomic assumptions. The earlier draft of the Chapter was discussed at the seminar held at the IET, chaired by Dr Yegor Gaidar. The team also benefited from discussions with Mr. Karagodin, Department Director for Social Policy in the Ministry o f Economy and Trade, and his staff. Helpful comments and suggestions were provided by Olga Antimonova (ECSPE), Anton Dobronogov (SASFP), Anita Schwarz (ECSHD), Hermann Von Gersdorff (ECSHD), and Ruslan Yemtsov (ECSPE). Olga Antimonova also assisted with assembling international data. Usha Rani Khanna and Emily Evershed assisted with editing the report and Anna Bazanova, Irina Partola, and Judy Wiltshire provided excellent support during the entire preparation process. The Ministry o f Finance and the State Committee for Statistics of the RussianFederationprovided considerable assistance with the data for the report. Deborah Wetzel and Asad Alam were the Sector Managers, and Cheryl Gray was the Department Director. Kristalina Georgieva was the Country Director for Russia. Thomas Blatt Laursen (ECSPE), Goohoon Kwon (IMF), and Pedro Alba (AFC13) were the Peer Reviewers. The team is grateful to the participants of the two seminars in Moscow, held in February o f 2005 respectively in the Center for Strategic Research and Moscow office of Camegie Endowement, for comments and suggestions made on the preliminary version o f this report. xii EXECUTIVESUMMARY 1. This Report discusses the challenge of budget financing of core structural reforms within a broader framework of fiscal management reforms in Russia. It argues that explicit financing o f structural reforms i s fully justifiable because these are the investments in the institutional infrastructure with a high rate of return. Inaddition, Russia currently appears to have a fiscal room for some incremental spending. However, the number o f simultaneous reform initiatives should be kept limitedto ensure that the accumulation o f new liabilities do not undermine fiscal sustainability. In addition, the Government's commitment to explicit reform financing should be accompanied by additional steps in strengthening the fiscal management system. The adoption o f formal fiscal rules could strengthenthe government's ability to manage external shocks, as well as provide budget support for the reform process in a predictable and affordable way. 2. The Report develops estimates for fiscal costs of three key structural reforms (incivil service, housing and the pension system), reviews the feasibility of different reform options, and provides recommendations related to their planning and sequencing. For each o f the structural reforms the Report discusses various scenarios, which altogether cover a broad range o f possible options for the Government with quite different fiscal implications. The models used for the preparation o f this Report represent an easily adaptable tool that government agencies can use to develop their own fiscal cost projections for alternative reform scenarios. 3. The Report treats the key structural reforms as medium-term projects/programs, thus suggestingthat implementationof these and other structural reforms can be put inthe context o f medium-term expenditure framework (MTEF) that evolve in Russia. Thus, the proposed approach to fiscal costing o f the reforms i s in full concord with the recent Government's initiatives on introducing strategic planning, MTEF, and elements o f performance budgeting. Fiscal Costs of Civil Service Reforms 4. The report develops a general framework for costing-out the direct fiscal effects o f various reforms inthe area o f public administration. It suggests that broadreforms inthe core government administration and in the civilian public sector at large may be implemented within five to seven years but should be differentiated by the scope o f pay adjustment in various sub-sectors o f civilian employment, closely monitored for non-wage expenditure growth, and complementedby significant staffing adjustments inthe civilian public sector as well as by at least some staffreductions inthe core government administration. 5. Implementation o f such reforms would require additional budget financing as compared to the 2003 expenditure levels, but fiscal costs could be sustained within a reasonable range. Under the most realistic set o f assumptions, incremental costs for a consolidated budget would amount to 1.2-2.3 percentage points o f GDP by the end o f the reforms as compared to 2003. Moreover, about two-thirds o f this increase has already been incorporated into the 2004 budget, reflecting the pay adjustments in the core government administration and the civilian public sector introduced in late 2003. Such additional financing would provide for a major reduction inthe pay gap between the public and private sector, especially for senior government officials. 6. A more detailed analysis of the incremental fiscal costs suggests that the costs to the federal budget would amount to about 40-45 percent of the total cost increase, while the rest would become the responsibility of subnational budgets. 7. While additional spending on public administration i s necessary to ensure better quality o f policymaking and public service delivery in the country, it i s not sufficient. Successful implementation of the budget process reform, introduction o f performance budgeting,and creating incentives for better performance inthe public sector would also be needed. 8. A significant adjustment in employment levels i s critical to make the civil service reform sustainable. Even moderate pay adjustments undertaken without a cut in staffing would make the reform fiscally unaffordable. It i s expected that by 2010 average employment in the civilian public sector would decline by about 25 percent. Making the Russian government leaner also seems to be consistent with the implications o f the existing demographic trends. Fiscal Costs of Reforms in Housing and Communal Sewices 9. The report analyzes the potential fiscal and social impact o f advancing cost-recovery increases in the housing and utility sector (HUS) under different scenarios. It argues that in the current environment o f high growth inhousehold incomes, by 2006 it would be possible to attain 100 percent cost recovery in tariffs with the simultaneous elimination o f all quasi- fiscal cross-subsidization and the adjustment in domestic energy prices. Moreover, the reforms in residential housing could be made budget neutral in the medium term and they would bringconsiderable savings inthe long term. However, the high sensitivity o f results to income dynamics suggests that the Government should establish an efficient monitoring system to track the affordability o f tariff increases for the population. The Report also emphasizes the restructuring o f financing mechanisms inthe sector to ensure a higher degree of accountability of bothmunicipal governments and service providers. 10. The baseline reform scenario suggests that as a result of proposed reforms the real unit cost to households in the HUS would increase by about 90 percent relative to the prevailing 2002 level. However, given the tariff adjustments that already took place in 2003- 04, it i s expected that future cost increases would be limitedon average to about 40 percent relative to their levels at the end o f 2004. 11. An increase inenergy and utility tariffs would make the delivery of utility services to budget organizations more expensive by about 0.7-0.8 percent o f GDP per annum. Some of these costs could be compensated through increased taxation o f energy and utility providers and, later on, through rationalization o f the public sector. However, in the medium term, a fiscal gap inpublic sector financing o f 0.4 percent of GDP could emerge as a result of tariff increasesinH U S and energy. 12. The analysis suggests that elimination of housing privileges could be affordable for most o f the current lgoty recipients, while the housing allowance program would be capable of taking care o f those who face a high housing cost burden. However, given the political sensitivity of entitlementreforms, there may be a case for reforming lgoty in a more gradual way. As a starting point, the Government should monetize the lgoty to transform them into explicit subsidies and link them directly with the system o f personal social accounts. Moreover, to reduce political costs, phasing out lgoty should be coordinated with other xii structural reforms, including wage increases in the public sector, and increases in pensions and child benefits. Costs of Pension Reforms 13. Estimates o f potential fiscal costs associated with various developments in Russia's pension system are based on a comprehensive actuarial model. The report points out that such fiscal costs are likely to emerge as a result of the declining relative value of old age pensions and associated political pressures for budget support to the pension system. Without additional reforms, the existing pension system, even under the most optimistic assumptions, i s not capable of closing the growing gap between growth in wages and pensions. In the baseline "without the reforms" scenario, the average replacement rate declines from 33 percent in 2002 to 24.4-27.8 percent in 2030. Moreover, the proposed cuts in contribution rates would result in a hrther decline in the replacement rate relative to the baseline. To avoid a drastic widening in the gap between wages and pensions, a reduction in contribution rates has to be supplemented by additional reforms, including a decision on a gradual increase inthe retirementage. 14. Inthe baseline scenarios, the annual fiscal costs to the government, associated with the need to address the problems accumulated in the pension system, amount to 0.25-0.55 percent o f GDP in 2020 and to 0.55-0.90 percent of GDP in 2030. These costs are measured against a target o f maintaining the replacement rate at 30 percent. However, the potential costs would increase rapidly in all scenarios with the reduced contribution rates. In the low case, the annual costs to the budget would exceed 2 percent o f GDP in2030. 15. The analysis also suggests that trends inthe share o f the taxable payrollinGDP play a critical role in determining the future results of the pension reform. This highlights the importance o f policies aimed at stabilizing payroll and income taxation, as well as at the removal o f various administrative barriers in the economy that currently hold back the reduction o f shadow incomes and wages. Conclusion 16. The current improved economic and fiscal situation in Russia provides a unique opportunity to policymakers to undertake key structural reforms, which have a highpayoff in terms of future growth and economic and social stability. Structural reforms in public administration, housing and utility services, and pension system are affordable if planned properly and sequencedinthe context of a rule-based fiscal management framework. ... X l l l Chapter 1. PRINCIPLES OF STRUCTURALREFORMS FINANCING 1.1 This Chapter discusses recent fiscal trends and the challenge o f budget financing of core structural reforms within a broader framework of fiscal management reforms inRussia, It argues that explicit financing o f structural reforms i s fully justifiable because these are the investments in the institutional infrastructure with a high rate of return. In addition, at the moment Russia appears to have a fiscal room for some incremental spending. However, the number of simultaneous reform initiatives should be kept rather limited to ensure that the accumulation of new liabilities do not undermine fiscal sustainability. In addition, government's commitment for explicit reform financing should be accompaniedby additional steps in strengthening the fiscal management system. In this context, the Chapter suggests fiscal rules for Russia that could strengthen government ability to manage external shocks, as well as provide budget support for the reformprocess ina predictable way. Inthat respect, the fiscal rules may help in elaborating a mechanism for the medium-term financial planning stipulated, among other budget process reform measures, by the Government Resolution No.249 of May 22, 2004. The Chapter also summarizes the estimates for fiscal costs of reforms in civil service, housing and pension system, which are analyzed ina greater detail in the following chapters. 1.2 The World Bank has a long and extensive history of public expenditure analysis in Russia. It has been following the policy of preparing focused analytical pieces on specific priority topics in the area o f expenditure management rather than trying to cover all public expenditure issues in one report. Responding to changes in the macroeconomic and fiscal environment, the Bank naturally shiftedits focus from the fiscal sustainability analysis (World Bank, 1996a and 1998c) and general diagnostics of fiscal management system in Russia (World Bank, 1996b) to more narrow and more technical issues, such as quasi-fiscal subsidies and non-cash operations (World Bank, 2000) and analysis of the public investment program (World Bank, 2001). Experience showed that for the Bank this is the most productive and client-oriented way of contributing to the reforms in expenditure management in Russia. The present report continues this practice by looking at fiscal aspects o f structural reform implementation inRussia. A. RECENT FISCAL TRENDS, 1998-2003 1.3 Since the 1998 crisis, Russia's macroeconomic performance has improved considerably. The cumulative GDP growth during 1999-2003 reached 38 percent. Solid economic growth contributed to the growth in budget revenues and allowed Russia to substantially improve its fiscal performance. Starting in 2000, federal and enlarged budgets have been executed with a surplus. This brought the public debt down to 28 percent of GDP by the end of 2003 from 85 percent at the end of 1999. The problems of non-cash budget execution, pension and wage arrears inthe budget sector have been successfully solved. 1.4 Revenues o f general government recovered to the steady level o f 36.5-37.6 percent of GDP in 2000-03 after a sharp decline to 33.6 percent in 1999 from 39.3 percent in 1997. Federalbudgetrevenues grew markedly to 16.7-17.8 percent o f GDP in 2001-03 from the pre- crisis level of 12.5 percent, helped by the centralization o f tax revenues and the progressive taxation of the oil windfall. At the same time, the non-oil federal budget revenues declined to an estimated 10.5 percent o f GDP in 2003 from 12.3 percent in 2001, while the oil revenues were on the rise along with oil prices (IMF, 2003 and 2004). As a result, the federal budget's dependence on oil prices increased significantly. The share of its oil revenues grew from 45 percent o f the total in2001 to 59 percent in2003. 1.5 Indeed, a more thorough analysis reveals that boththe GDP and budget revenues were propelled largely by the high world prices of crude oil - Russia's main export commodity. It i s estimated that almost 80 percent of the incrementalincrease in general budget revenues was oil-factor driven, including spillover effects on the gas and other sectors (Kwon, 2004). Two developments played a major role here. First, with oil prices standing high, oil and gas companies increased their sales and profits and started making relatively more tax payments to the budget.Second, the tax legislationhas changed in a way that has increased the oil price elasticity o f tax payments in the sector. The most important changes have included the increased progressiveness o f oil export duties and the introduction of a linear relationship betweenoil prices and the statutory rate of the tax on the extractionof mineral resources inthe oil sector. Both taxes are paid mainly to the federal budget, making it very sensitive to changes in oil prices: In the present price range, every dollar-per-barrel change in oil prices increases or decreases the enlarged budgetrevenues by 0.45 percent o f GDP, out of which the share of federal budget i s three-quarters. 1.6 Since 1998, there have also beenconsiderable developments on the expenditure side of the budget, which point to both the large fluctuations in and the pro-cyclicality o f budget expenditures (see Table 1.1). After the initial decline to 29.0 percent o f GDP in 2000 from 43.7 percent in 1997, the enlarged budget non-interest expenditures recovered by 5.6 percentage points of GDP injust two years (2001-02) and reached 34.4 percent of GDP. The increase in total budget expenditures was less pronounced since GDP growth and the policy of budget surpluses helped to reduce both the stock o f public debt and the size o f annual interest payments. 1.7 The social expenditures of the enlarged budget (Le., expenditures on education, health, and social protection) have had a clear pro-cyclical pattem in the past. They experienced the largest reduction (by more than 50 percent in real terms) during the post-1998 fiscal consolidation, which was followed by a strong increase thereafter. Within social expenditures the after-the-crisis recovery largely financed the growth o f recurrent expenditures, such as wages, pensions, etc. Inter alia, this also included the financing o f increased staffing levels in the civil service and budget sectors (such as health) at the subnational level (World Bank, 2003b). Apparently, recurrent expenditures on wages inthe public sector atld household social benefits are those which would be the most difficult to compress politically should a fiscal adjustment be required ifand when budget revenues fall together with oil prices. 1.8 Non-social non-interest expenditures demonstrated less, but still some, pro-cyclicality. Respondingto the 1998 crisis, they fell to 14.4 percent o f GDP in 1999 from 19.0 percent in 1997. Expenditure rationalization made by the government resulted intheir further decline to 13.9 percent o f GDP in2001 before growing to 14.7 percent in2002, presumably on the wave o f the oil windfall spending and general fiscal policy relaxation. In 2003 the government managed to reduce these spending somewhat inrelative terms to 14.4 percent of GDP. 1.9 While most o f the non-interest non-social expenditures behaved pro-cyclically, some o f them did not. More specifically, expenditures on agriculture, liquidation of emergencies and natural calamities, and state industrial funds have declined inpercent o f GDP terms since 2000. In this way these expenditure cuts smoothed the pro-cyclicality o f overall non-social budgetexpenditures. 2 Table 1.1: Expenditure Trends in Russiain 1997-2003,EnlargedGovernmentBudget, (Percent of GDP) Expenditure Category 1997 1998 1999 2000 2001 2002 2003 Total Non-InterestExpenditures 43.7 36.3 29.9 29 30.7 34.4 32.1 I.TotalSocialExpenditures 24.7 10.8 15.6 14.5 16.9 19.7 17.6 1. Education 4.8 3.6 3 2.8 3.1 3.9 3.6 2. Health 3.6 3.4 2.9 2.8 2.9 3.2 3.0 3. Social Protection 11.5 10.1 7.5 6.9 8.7 10.3 9.1 4. Housingand CommunalServices 4.8 3.7 2.2 2.1 2.2 2.4 1.9 11.Total "on-Social' Expenditures 19.0 15.5 14.4 14.5 13.9 14.7 14.4 1. GeneralPublic Services 7.5 6.4 6.5 6.0 6.5 7.1 6.9 2. Public Investment/Subsidiesto 2.2 0.9 0.9 0.8 1.7 2.2 2.5 Industryand SMEs 3, TransportlRoads/Communications/ 1.o 0.7 0.5 0.5 0.7 0.5 0.3 Informatics 4. State Roadand IndustrialFunds 2.7 2.9 3.1 3.8 2.1 2.1 1.9 5. Agriculture/Fishing 1.3 0.9 0.7 0.7 0.8 0.6 0.5 6. Environment/Hydromet./Geodesy 0.2 0.1 0.1 0.1 0.1 0.2 0.1 7. Culture, Arts, and Mass-Media 0.7 0.5 0.5 0.6 0.6 0.7 0.6 8. Science 0.6 0.3 0.3 0.3 0.3 0.3 0.4 9. EmergenciesandLiquidationof 0.3 0.3 0.2 0.2 0.1 0.1 0.1 NaturalCalamities 10. Other Expenditures 2.5 2.4 1.5 1.4 0.9 0.9 1.o Sources: MoF,Rosstat, Staff calculations. 1.10 As one would expect, among the items of functional classification, budget spendingon both public investment and subsidies to industrial producers (net o f subsidies to SMEs) proved to be the most pro-cyclical expenditure category. These spendings fell more than threefold in 1997-99 -- from 2.1 percent o f GDP to 0.6 percent. They then increased to 2.5 percent o f GDP in 2003. This was the only broad expenditure category that exceeded its relative (as a percent of GDP) pre-crisis level. Medium-term fiscalprospects and debt sustainability 1.11 The fiscal surpluses of the general government recorded since 2000 have drastically improved Russia's debt profile and have practically eliminated the debt sustainability problem that was a major macroeconomic and fiscal problem for Russia in the 1990s (Figure 1.1). Recent analysis o f Russia's debt sustainability suggests that trends in the total public debt should be easily manageable under a broad set o f possible scenarios (IMF, 2003). Under the baseline scenario, it i s expected that the average primary surplus o f the general government would amount to 1 percent of GDP in 2007-09, while the average primary balance of the federal government would amount to 1.7 percent. This i s under the assumption o f average GDP growth o f 5 percent, external debt interest rate o f 5 percent, and a considerable slowdown inruble real appreciationafter 2006. 1.12 Even in the low-case scenario the debt would stabilize at a level below 30 percent o f GDP. In addition, stress tests showed that under the most gloomy macroeconomic assumptions (such as high external interest rates, a significant fall in GDP, considerable 3 depreciation of the ruble) the debt dynamics, measured by the public debt-to-GDP ratio, would deteriorate only temporary, and would still remain under control in the longer term (See Table 1.2). Moreover, since early 2004, when the most recent analysis was carried out, skyrocketing oil prices have made the public debt profile in Russia even stronger, making public debt sustainability o f even lesser concern. 1.13 However, Russia remains highly vulnerable to external shocks associated with oil prices. Should oil prices drop significantly and remain at the low level, a sizable adjustment in exchange rate and fiscal policy would be required. In addition, short-term liquidity crunches caused by adverse oil price shocks could lead to substantially lower GDP growth rates (IMF, 2004). Also the sustainability o f the country's total external debt i s o f somewhat more concern, taking into account a rapid accumulation o f foreign debts by the private sector. Significant ruble depreciationcould lift the total external debt to an unsustainable level. 1.14 Notably, the public debt sustainability does not mean that Russia does not have any problems inthe area o f debt management. Many OECD countries chose their longer term debt targets at the levels that are considerably below their debt sustainability thresholds. It i s assumed that even sustainable debt dynamics mightbe detrimental for the economic growth if the debt level i s relatively high.' Various studies estimated that the optimal debt level for individual G7 countries falls inthe range o f below 20 to 50 percent o f GDP.' Clearly, for less developed countries, such as Russia, the optimal debt level should be lower than in the G7 countries, owing to higher country riskpremiums, a higher share o f hard currency debts inthe debt structure, and the lesser efficiency o f the public sector. Tahle 1.2: Puhlic deht sustainahilitv estimates. Deht-to-GDP ratio. oercent 1998 2000 2003 2004 2006 2009 Baseline 144.4 64.8 33.3 29.2 21.7 15.7 Stress tests: 1.High interestrate(In04-05: 2 stand deviationsabovethe hist average) 37.6 37.8 30.3 2. Lower GDP growth (In 04-05: 2 standdeviationsbelow the hist average) 34.9 37.2 24.0 3. Largerprimary deficit (In 04-05: 2 stand deviationsabove the hist average) 42.2 44.7 36.6 4. Simultaneousshocks (joint shocks 1-3, using 1 stand deviation) 43.6 42.9 22.9 5. Ruble depreciation(one time 30% real depreciationin 2004) 44.6 35.5 28.2 Source: IMF (2004). 'Balls and O'Donnell (2001, pp.174-175) offer a discussionof the concept of debt optimality as an alternative 'toIbid. the conventionalview of debt sustainability. 4 Figure 1.1. PublicDebt Projections A (Inpercent of GDP) I 100, Proiections 80 n o 70 60 50 40 30 20 10 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Source: IMF (2004). B. FISCAL COSTSOF KEY STRUCTURALREFORMS 1.15 In spite of the significant progress achieved in buildingmarket economy during the 1990s and early 2000s, Russia still needs to accomplish a numbero f structural reforms, which would advance the development of its market institutions and level up the public sector efficiency. The implementation o f structural reforms usually requires budget financing and can be rather costly. Depending on the specifics of a particular reform, these costs may be of a temporary or permanent nature. From a fiscal viewpoint, launching a structural reform represents incremental, irregular spending with stronger or weaker feedback to regular expenditures. This issue i s of paramount importance for Russia, which i s committed to carrying out multiplestructural reforms simultaneously. 1.16 All structural reforms can be divided into three groups by the fiscal costs they incur. The first group consists o f those reforms that require only initial one-off financing for a limited time period or those that, with time, generate budget savings that offset their fiscal costs if the latter are o f a permanent nature. Examples include the elements o f both education reform and pension reform. Inparticular, a switch to the system o f unitary state exams (USE) for high school graduates requires investment in the institutional infrastructure for these exams, expenditures on pilots, and the like. Once it becomes operational, the USE would replace the current system of university entry exams that i s presently administered by individualuniversities. The annual costs of undertakingthe USE would be offset by savings in the costs of university admission. Inpension reform, establishing a fully funded pillar gives rise to a fiscal gap inthe pay-as-you-go part o f the pension system a few years down the road, which will disappear when the funded pillar starts to co-finance pensionbenefits. However, in the meantime, either the temporary medium-term fiscal gap inthe pension system will have to be financed from the budget, or the real value of benefits to the current pensioners will be lower inreal terms than for both the preceding and the following generations o f pensioners. 1.17 There are two important specific sub-groups o f reform initiatives that could be identifiedwithin this first group o f reforms: e Structural reforms to be implemented at the subnational level, the early implementation of which could be unattractive for regional leaders (e.g., because o f their perception of additional political risks). Under the circumstances, the federal government could be interested in establishing a special support mechanism to provide fiscal and administrative incentives to regions that are ready to pilot specific regional reforms. The recent quite successful experience with the Federal 5 Fund for Regional Fiscal Reforms (FFRFR) provides a possible model for the organization of such a federal support mechanism. It i s worth emphasizing that, even in cases where the reforms themselves may not lead to additional fiscal costs (as in the case of reforms inresidential housing), some modest costs for the federal government could arise from a need to establish a proper incentive framework for launching reform pilots in the regions. This relates primarily to the sectors that are primary responsibility of subnational governments. Reforms to promote new private sector institutions, currently missing in Russia, which could generate considerable long-term externalities. One-time public investments in such institution building could reduce the future costs of market entry for private operators. Examples o f such missing or severely under-developed institutions include entities that would provide student loans and venture fund industry. The govemment's co-financing of such institutional pilots could take the form o f either co-investments or guarantees for private investors against non- commercial risks. 1.18 The second group comprises reforms that demand a permanent increase in financing, while potential budget savings are not sufficient to cover their costs, even in the long term. Civil service reform i s one example o f such a reform. As estimated in Chapter 2, it would cost the budget 0.9-1.3 percent of GDP annually, o f which 0.5-0.7 percent would be the costs to the federal budget. These permanent additional spendings are needed to make salaries in the civil service compatible with those inthe private sector, as well as to increase non-wage costs in the civil service (such as on training and retraining, office equipment and telecommunications, etc.). Without this, it would be difficult to ensure a drastic improvement inthe efficiency ofthe civil service inRussia. 1.19 Structural reforms belonging to the third group are those that generate expenditure savings without any substantial fiscal costs for the budget. Housing reform appears to be an example of such a reform. Phasing out producer subsidies in the housing and communal services sectors and phasing in the targeted subsidies to poor families would generate substantial net budget savings, as experience has shown. In addition, potentially massive outflows from the budget to rehabilitate the communal infrastructure could be minimized. A system o f performance management contracts, leasing, and utility concessions can make it profitable and attractive for private companies to invest inthe communal infrastructure, which would remove this burdenfrom the budget. 1.20 The financing of structural reforms belonging to the first group could come from the two possible sources: (i) budget revenues, and (ii) general government borrowing. This would give rise to a temporary deficit for a budget recalculated at the long-term average oil prices, which i s discussed below in this Chapter. Structural reforms belonging to the second group should be largely financed through compensatory policies such as: (i) savings on other, less efficient, expenditures, (ii) savings achieved during the implementation of the third group of structural reforms, and (iii)revenues from additional taxes. In this case, incremental expenditures and the compensatory increase intaxes could bejustified by the expected reform gains - namely, the growing efficiency o f the public sector, including the provision o f better quality public services. However, it is likely that the financing of these structural reforms through compensatory policies will become available with a delay, because some time will be needed for the development and implementation o f such policies. Therefore, inthis case an oil price adjusted budget deficit may emerge as well, but it should gradually be phased out with the helpofcompensatory policies. 6 1.21 From the institutional viewpoint, it seems appropriate to record and monitor budget resources channeled to finance the implementation of structural reforms. For that purpose, corresponding program spendings could be analytically consolidated as structural reform funding in the federal budget. Following provisions of the Concept of the Budget Process Reform, approved by the Government in May o f 2004, such reform spendings would become a part of the `new expenditure commitments' budget. This approach would enable the government to control the overall size of such a virtual `fund' dedicated to reform financing in line with its general principles of fiscal policy (see more on this subject below in this Chapter). 1.22 It follows that the existing Federal Fundfor Regional Fiscal Reforms (FFRFR) or a similar facility should become one o f the components within such a broader category of reform spending, along with the resources raised as loans from the IFIs that aim at the acceleration o f subnational reforms. As discussed earlier, a FFRFR-type facility would provide partial financing for subnational reforms within the first group o f reforms. 1.23 The issue of costing structural reforms is not absolutely new for Russia. The government itself estimated foregone revenues or costs o f some structural reforms (e.g. tax reform and pension reform). Yet, this work has never been done systematically and put in a proper fiscal framework, which would allow comparing different reform scenarios and different structural reforms and provide proper links between the costs o f reforms and the overall budget process. Inaddition, the earlier research inthis area government (including the most advanced analytical exercise to date undertaken by the Expert Institute (Yasin, 2003)) relied mostly on rough expert estimates. The present report suggests a much more rigorous multi-factor approach and models, which allow (i)tracking expenditure changes at the disaggregated level for every specific structural refom, and (ii) presenting and estimating multiple reform scenarios. 1.24 Yasin (2003) analyzed risks associated with the unfinishedstructural reform agenda in Russia for the country's longer-term growth prospects. He argues that low gas and energy prices, together with low salaries, are the fundamental barriers for raising competitiveness in Russian manufacturing. Depressed prices for production inputs are not conducive to the development o f domestic industries as they destroy incentives to save energy and raise productivity. He develops a detailed argument to support two fundamental claims: 0 acceleration o f structural reforms in energy and housing sectors, as well as related measures in the areas o f wage policy and social security i s the core policy challenge for the Russian government; 0 there reforms are closely inter-related and should be advanced in parallel within the single financial framework, where higher domestic energy prices would bring additional taxes to pay higherpensions, social benefits and wages inthe public sector. 1.25 Yasin's paper also develops some aggregate estimates for the above mentioned financial framework within the 3-year time horizon, which are based on specific assumptions on expected increases in domestic energy prices, housing costs, as well as wage and pension levels. Inour report we intendto advance this work further, including through a generation of more accurate set o f estimates on potential costs o f particular structural reforms within this single framework under different sets of assumptions. 1.26 The specific choice o f the key structural reforms analyzed in this report-public administration, housing and communal services, and pensions-was determined by the 7 Russian Ministry o f Finance in consultation with the World Bank team. This choice reflects both the importance o f these particular reforms for the mediumterm government program, as well as perceived scale o f potential fiscal implications in case if these critical reforms are mishandled. c. COSTS OF STRUCTURAL REFORMS: EXAMPLESREFORMS INTHE CIVIL OF SERVICE, PENSIONSYSTEM, AND RESIDENTIAL HOUSING 1.27 This section summarizes the results of the simulation of the individual structural reforms that were developed inthis report for three specific sectors, such as the civil service, the pension system, and residential housing. The remaining Chapters o f the report present a full description of the respective results, including the assumptions, data and models usedin the analysis. Reforms in the civil service 1.28 The Russian government considers the implementation of the comprehensive civil service reform as one o f the key priorities in the mid-term reform agenda. The objective o f Chapter 2 o f the report i s to help the government with the planning and sequencing of its reform effort by (i) developing a general framework for costing-out the direct fiscal effects of various reforms in the area of public administration, (ii) usingthis framework for generating consolidated estimates for the incremental fiscal costs o f specific reform scenarios, and (iii) developing recommendationson the feasibility of differentreform options. 1.29 The Chapter presents more than 40 different reform scenarios as well as the potential results of their step-by-step implementation through the period 2004-1 0. The scenarios differ considerably in terms o f the pace and scope of relative pay increases in the public sector, the scope and pace o f attrition, changes in the share o f non-wage costs, etc. In addition, the Chapter analyzes the sensitivity o f potential reform costs to changes in the basic macroeconomic parameters. It identifies the future growth rate in private sector wages as the most important macroeconomic factor: the higher the wage growth inthe private sector is, the more expensive it would be for the government to close the existingpay gap. 1.30 Overall, the results of the simulations show that the reform may be implemented without a significant increase in overall fiscal costs. The Chapter argues that a fiscally affordable reform should be based on moderate pay adjustment and should assume significant staff reductions in the core government administration, and especially in the civilian public sector. Under the most realistic set of assumptions, the incremental costs for the consolidated budget would amount to 1.2-2.3 percentage points o f GDP by the end o f the reform (depending on the macroeconomic scenario) as compared to 2003. These costs do not look prohibitively highwhen considered inthe international context. Moreover, about two-thirds of this increase has already been incorporated into the 2004 budget, reflecting the pay adjustments inthe core government administration and the civilian public sector introducedin late 2003. At the same time, the remaining one-third o f the total reform costs mustbe added to ensure full financing. 1.31 The Chapter also argues that the future reform package should not be applied to the core government administration only, but should cover the entire civilian public sector. Otherwise, politically it would be quite difficult for the government to sustain a drastic increase in the compensation gap between the civil servants and the rest o f the public sector 8 employees. Moreover, our modeling suggests that the expected fiscal gains from more aggressive staff attrition in the civilian public sector would help to finance a portion of the costs related to the implementation o f civil service reform in the core government administration. At the same time, it seems desirable to synchronize primary components of the reform, such as pay and staffing adjustments. Such synchronization would allow the stabilizing o f the overall level o f employment-related expenditures on the civilian public sector. On the other hand, de-linking the reform measures might create unnecessary expenditurefluctuations, which couldbe an additionalriskto reform sustainability. 1.32 To achieve its objectives, the reform strategy has to differentiate across the sub-sectors of public employment. The pay increase should be the most significant for the headquarters- based civil servants, lower for the deconcentrated units and regional and municipal parts o f the core government administration, and lowest for the rest o f the civilian public sector. The last could be justified in part by the growing share of fee-based services provided by this sector, which would become an increasing source o f compensation for its employees. Inside the headquarters-based federal civil service, the pay increase should emphasize a larger and faster increase in compensation for decision makers. Such a focus of the pay reform would allow addressing the key constraints of the low policymaking capacity inside the government and would help retain highly qualified public officials in key positions in the federal civil service. 1.33 The results of our simulations suggest that over the period up to 2010 it would be unlikelyfor the Russian budget to be able to afford a full closure of the current public-private pay gap through a radical pay adjustment inthe entire public sector. It seems more realistic to expect that the average residual pay gap for core government administration would remain significant (in the 100 percent range of the public sector wage). However, recent surveys o f public officials reveal that such a gap may be quite acceptable, because it seems to reflect the existing expectations of the public officials. A stronger effort to close the gap (to the average level o f 50 percent) would raise the costs of the proposedreforms to about 2.2-3.3 percentage points o f GDP as compared to 2003. 1.34 Still, the proposed moderate pace o f pay reform suggests that the average wage inthe core government administration would grow at a rate that is 25 percent higher than the growth of the average wage inthe economy, while the average pay inthe civilian public sector would grow at a rate that i s 17 percent higher. Moreover, it i s recommended to complement this average increase by a more radical pay adjustment for a small number o f decision makers in the headquarters of the federal executive authorities. The recommended scenarios would result in an increase in the decompression coefficient for headquarters-based civil servants from 2.5 in2002 to 6.8-9.0 in2010. 1.35 A significant adjustment in employment levels is critical to making the reform sustainable. Evena moderate pay adjustment undertakenwithout a cut in staffing would make the reform fiscally unaffordable. It is expected that by 2010 average employment in the civilian public sector would decline by about 25 percent. Making the Russian government leaner also seems to be consistent with the implications o f the existing demographic trends. The largest reductions are expected inthe civilian public sector, where, inpart, they could be implemented through the commercialization of some of the budget sector entities. 1.36 Implementation o f the broad reform agenda in the core government administration would entail significant additional non-wage expenditures. The full-scale modernization of public service would require both significant investment costs and significant recurrent costs 9 for the operation and maintenance of computer systems and physical infrastructure, as well as the implementation of modern HRpractices (includingcompetitive recruitment, staff rotation, and training). However, the existing budget constraints suggest that the issue o f non-wage expenditure control would be quite important in the course of reform implementation. It i s expected that the non-wage expenditure share would have to go down from the current 37 percent of total costs to at least 35 percent, which would still represent a significant increase infinancing inrealterms. 1.37 A more detailedanalysis ofthe incremental fiscal costs, undertaken for several median scenarios, suggests that the costs to the federal budget would amount to about 40-45 percent o f the total cost increase, while the.rest would become a responsibility o f subnational budgets. About two-thirds o f all incremental costs would be linked to the additional funding of the core government administration, while the rest of the civilian public sector would benefit from a smaller share of the expenditure increase. 1.38 Chapter 2 also recommends that additional analysis would be desirable to overcome the important limitations of this set o f results. In particular, regional comparative pay and compensation surveys would be critical to better measuring both the current actual value of in-kind benefits and the overall current compensation gap in regions. Further costing of particular investments associated with the modernization of public service (such as the introduction of administrative operational manuals, etc.) would also be helpful to policy deliberations. 1.39 Overall, the Chapter suggests that broad reforms in the core government administration and in the civilian public sector at large may be implemented within five to seven years but should be differentiated by the scope of the pay adjustment in various sub- sectors o f civilian employment, closely monitored for non-wage expenditure growth, and complementedby significant staffing adjustments inthe civilian public sector, as well as by at least some staff reductions in the core government administration. The implementation of such reforms would require additional budget financing as compared to the 2003 expenditure levels, but fiscal costs could be sustained within a reasonable range. Sustainable public administration reform in Russia cannot be cheap, especially in the environment of rapidly growing real wages. Additional spending on public administration is the only way to ensure a better quality of policymaking and public service delivery in the country. However, this i s a necessary but not a sufficient condition. The successful implementation o f the budget process reform, the introduction o f performance budgeting, and the creation o f incentives for better performance inthe public sector would also be needed. Reforms inpayment arrangements inthe housingand utility sector 1.40 Chapter 3 of the report analyses the potential fiscal and social impacts o f advancing cost recovery in tariffs in the housing and utility sector (HUS) in Russia under the different macroeconomic and reform scenarios. The simulations are based on the disaggregated model of the full costs of operating the country's residential housing and the allocation o f the costs among the major sectoral stakeholders - households, government, and service providers. For each scenario the model estimates the expected demand inoverall budget support for the H U S that cover all types o f budget programs inthe sector. The baseline scenario suggests that as a result o f the proposed reforms the real unit cost to households inthe HUS would increase by about 90 percent relative to the prevailing 2002 level. However, given the tariff adjustments 10 that took place in 2003-04, it i s expected that future tariff increases would be limited to, on average, about 40 percent relative to their levels at the end of 2004. 1.41 This Chapter argues that, in the current environment o f high growth in household incomes, it i s affordable (both politically and financially) to attain 100 percent cost recovery intariffs with the simultaneous elimination of all quasi-fiscal financing (cross-subsidization) and the adjustment in domestic energy prices. Moreover, the reforms in residential housing could be made budget neutral inthe medium term, and they would bringconsiderable savings in the long term. However, the analysis revealed a high sensitivity of the results to income dynamics. If household income growth slows down, this may generate incremental budget costs of 0.4-0.5 percent o f GDP per annum in the medium term. The high sensitivity o f the results to income dynamics suggests that the government should establish an efficient monitoring system to track the affordability o f tariff increases for both the population in general and specific household groups. The Chapter also emphasizes the need to restructure the financing mechanisms in the sector to ensure a higher degree o f accountability for both municipal governments and service providers. 1.42 An increase in energy and utility tariffs would make the delivery o f utility services to budget organizations more expensive by about 0.7-0.8 percent o f GDP per annum. Some of these costs could be compensated through the increased taxation o f energy and utility providers and, later on, through the rationalization of the public sector. However, we estimate that in the mediumterm a fiscal gap inpublic sector financing o f 0.4 percent of GDP could emerge as a result o ftariff increases inthe HUS and energy sectors. 1.43 The analysis suggests that the elimination o f housingprivileges couldbe affordable for most o f the current lgoty recipients, while the housing allowance program would be capable of taking care of those who face too high a housing cost burden. However, given the political sensitivity o f entitlement reforms, there may be a case for reforming the lgoty in a more gradual way. As a starting point, the government should monetize the lgoty to transform them into explicit subsidies and link them directly with the system o f personal accounts. Moreover, to reduce political costs, phasing out the lgoty should be coordinated with other structural reforms, including wage increases in the public sector, and increases in pensions and child benefits. Pensionreforms 1.44 The analysis in Chapter 4 aims at estimating the potential fiscal costs associated with various developments in Russia's pension system based on the comprehensive actuarial model. It finds that such fiscal costs are likely to emerge as a result o f the decliningrelative value o f old age pensions and associated political pressures for budget to support the pension system. 1.45 The Chapter is focused on the sensitivity o f current trends in the pension system to changes in macroeconomic performance and various combinations o f potential reform measures, such as proposed cuts in contribution rates, as well as possible increases in the retirement age. The core variables analyzed in the model are (i) potential affordable the replacement rate, and (ii) the expected ratio of the pension and subsistence minimum.Most of the simulationscover a periodupto 2050. However, the analysis is focused mainly onthe first half o f the period (up to 2025-30), which i s o f larger interest to policymakers. 11 1.46 The structural nature of the pension reform in Russia relates primarily to the introduction of the 2"d fully funded pillar, which fundamentally changed the structure o f pension financing in the country. According to the estimates in this report, by 2008 contributions to the second pillar may reach 1 percent o f GDP a year, while this amount increases to 1.25 percent o f GDP by 2015. Respectively, the reform reduces the amount of funds available for paying the benefits to current pensioners through the lSt pay-as-you-go pillar, thus creating temporarily financing needs inthe pension system. 1.47 The Chapter concludes that without additional reforms, the existing pension system, even under the most optimistic assumptions, is not capable of closing the growing gap between growth in wages and pensions, which i s reflected in the declining replacement rate. Inthe baseline "without the reforms" scenario the average replacement rate declines from 33 percent in 2002 to 24.4-27.8 percent in 2030. At the same time, it provides for a gradual increase in the value o f the average pension relative to the subsistence minimum. This indicates that with economic growth the real incomes o f pensioners will also grow steadily. However, there are two concerns regarding the patterns of this future growth: (i) inthe initial period up to 2015, despite a low level o f current pensions, the growth will be rather slow, and (ii) thegrowthinpensionswillbelaggingthegrowthinrealwages,itmaybecomea because politically sensitive issue. 1.48 Inthe baseline scenarios, the annual fiscal costs to the government, associated with a needto addressthe problems accumulated inthe pension system, amount to 0.25-0.55 percent of GDP in 2020 and to 0.55-0.90 percent of GDP in2030. These costs are measured against a target of maintaining the replacement rate at 30 percent. However, the potential costs would increase rapidly in all scenarios with the reduced contribution rates. In the low case, the annual costs to the budget would exceed 2 percent o f GDP in2030. 1.49 Overall, without additional policy changes, any significant cuts in the contribution rates would result ina further decline inthe replacement rate relative to the baseline. To avoid a drastic widening o f the gap between wages and pensions, a reduction in the contribution rates has to be supplementedby additional reforms, including a decision on a gradual increase in the retirement age. The scenarios with the increased retirement age do not require any budget support to provide a replacement rate o f 30 percent. 1.50 It is quite likely that the cut inthe contribution rates of 8 percentage points under the latest government proposal would lead to an immediate and considerable decline in the Pension Fund's collections relative to the baseline. To avoid either a decline inthe real value of current pensions or an accumulation of pension arrears, such a cut would require a substantial fiscal transfer (in the magnitude of 1 percent of GDP a year) to the Pension Fund. The compensation has to be provided until considerable improvements in the revenue performance of the pension system materialize and/or policy decisions are made, which reduce the Pension Fund's financing needs. 1.51 The analysis also suggests that trends inthe share o f the taxable payroll inGDP play a critical role in determining the future results o f the pension reform. This highlights the importance of policies aimed at stabilizing payroll and income taxation, as well as at the removal of various administrative barriers in the economy that currently hold back the reduction o f shadow incomes and wages. 12 Cost summary for the analyzed set of structuralreforms 1.52 Table 1.3 summarizes our estimates of the total potential costs o f selected structural reforms analyzed in this report. We estimate that the full annual costs to the federal budget could be limited to 1.5 percent o f GDP. Given the current fiscal and macroeconomic performance of Russia, this cost level appears to be affordable. Table 1.3: Summary of IncrementalAnnual Fiscal Costs for the Set of Analyzed FiscalReforms(relative to the 2004 budgetbaseline) (YOof GDP) a. Pensionreform (effect ofthe cut in contribution rate) al. Cut in contribution rates - medium-term effect 1.o-1.1 a2. Cut in contribution rates plus long-term effect 2.2-2.5 a3. Increaseinretirement age 0 b. Housingandutility reforms bl. Residentialhousing 0 b2. Maintenance of budget organizations 0.4-0.5 olw:federalbudget 0.15 c. Civil service 0.4-0.75 olw: federal budget 0.18-0.25 Total (medium-term) for the selected sub-set of reforms 1.8-2.35 olw: federal budget 1.33-1.50 Source: Stafestimates. See Chapters 2-4 of the report. D. THECHALLENGEOFFINANCING STRUCTURAL REFORMS:CANFISCAL RULES HELP? 1.53 As mentionedabove, Russia, as a country intransition, faces the challenge of catching upwith the developed economies inmany areas simultaneously. This creates a demand for an active government structural policy aimed at the acceleration o f a broad range o f structural reforms. Suchreforms inmany cases require incremental financing, which may add too many extra claims on the budget envelope. Without financing, the implementation of many structural reforms cannot be effective. But fundingthe reforms without proper fiscal discipline and prioritization could be fiscally distractive. This presents a challenge for a fiscal policy in Russia. 1.54 Inour analysis we consider government spendingon implementation of core structural reforms to be a sort o f public investments in Russia's institutional infrastructure that ultimately have rather a high rate o f economic return. Delaying or under-financing such spending could be a sub-optimal longer term fiscal policy because it may affect longer term growth prospects o f the economy. In this sense, budget financing o f structural reforms could be treated as public investment in conventional infrastructure. As shown by Easterly and Semen (2003), under-investing ininfrastructure too often creates an illusion o f the sustainable fiscal policies. While it may help to reduce the current budget deficits, it often bringstoo high costs in terms of lower future growth, which undermines longer term fiscal sustainability. At the same time, there i s sufficient evidence that successful infrastructure investments have rates o f return that are higher than prevailing costs o f borrowing in the middle income economies, such as in Latin America. From this perspective, a certain level o f funding o f infrastructure 13 investments through public borrowing i s fully justifiable because it improves, not damages longer term sustainability o f public finance. 1.55 Therefore, we believe that earmarked well-planned and accountable budget financing o f core structural reforms i s justifiable in principle because these are the investments with high rate o f retum. In addition, Russia's strong fiscal position suggests that the government could afford some incremental spending without generating significant risks for macroeconomic stability. In other words, we do not consider an affordability argument as a reason to postpone key reforms. However, the number of simultaneous reform initiatives should be kept rather limited to ensure that the accumulation o f new commitments do not undermine fiscal sustainability. Besides, the government capacity limitations also suggest a needfor a strict prioritization of the reformprocess. 1.56 In addition, as argued in this report, government's commitment for explicit reform financing should be accompanied by additional steps in strengtheningthe fiscal management system. The challenges of structural reforms are too numerous and most of them create considerable longer-term and sometimes overlapping government commitments. This raises requirements to the quality of the fiscal framework, within which various reform strategies could be analyzed, compared and eventually prioritized. 1.57 Moreover, fiscal challenges in Russia are not limited to the financing o f structural reforms. As past Russian experience has demonstrated, world oil price volatility, not subdued by the proper macro policies, has translated into significant economic and fiscal volatility in Russia, which has brought high economic and social costs to the country. In fact, oil price volatility i s the second and largest fiscal challenge that Russia (like any large oil exporter) faces. 1.58 Additionally, the Russian economy i s subject to conventional cyclical ups and downs that are common to market economies. In Russia these changes in business conditions are closely correlated with the movements of oil prices. It i s agreed in principle and i s actually followed by some countries (for example, in the EU) that it i s desirable to make fiscal policy countercyclical (that is, expansionary during recessions and contractive during growth phases). This makes it possible to maintain the real value o f expenditures during bad times and to smooth fluctuations in economy. Yet without proper fiscal discipline, the introduction of countercyclical budgetinstrumentscouldeasily become a burden for the fiscal system. 1.59 While deciding on the implementation and financing of structural reforms, the Russian government should place these decisions in the broader context of other macroeconomic and fiscal challenges that the country faces. Otherwise, macroeconomic stability and sustainable economic development couldbejeopardized. 1.60 In order to deal with fiscal challenges and foster economic stability, it has become common in many countries to develop and apply fiscal rules (OECD, 2002; Kopits, 2004). These rules are usually fixed in legislation and represent a permanent constraint on fiscal policy. They applied to budget deficit or expenditures, or both, and may be expressed inactual or cyclically adjusted terms. Kopits and Symansky (1998) argue that the strongest case for fiscal rules i s based on political economy arguments -- the rules correct the bias of short- sighted governments to accumulate public debt at the expense o f future generations and that avoiding time-inconsistency issues results insignificant credibility gains. 14 1.61 The experience o f other countries shows that proper fiscal rules help in carrying out sound fiscal policy. They establish benchmarks for fiscal performance that help to discipline policymakers and subdue political pressures. For example, the United Kingdom uses two fiscal rules since 1997: (i) "golden rule," which stipulates that over the economic cycle, the the government will borrow only to invest and not to fund current spendings, and (ii) the `sustainable investment rule', which stipulates that the net public sector debt to GDP ratio will be held over the economic cycle at a stable and prudent level. A fiscal rule introduced in Switzerland in 2003 sets a ceiling for government expenditures, which cannot exceed cyclically-adjusted revenues (OECD, 2002). 1.62 The Russian Budget Code (para. 3, Article 92) also contains a kind o f fiscal rule stipulating that the annual budget deficit o f the federal budget cannot exceed the sum o f budget investment and interest expenditures. It means that government borrowing cannot be used to finance current outlays, except for interest expenditures. Although, at the first glance, this rule looks similar to the British `golden rule', it does not provide for sufficient fiscal rigidity. Infact, the economy could be movingto a debt trap, while be infull compliance with this rule -- new borrowing would be made to finance interest expenditures, which would increase both a stock o f debt and future interest spending, thus creating a need for even larger borrowing inthe following budget cycle. 1.63 Nonetheless, this legislated fiscal rule has been an important positive development. If it had been adopted a few years earlier, the rule could have been helpful as a tool o f fiscal adjustment duringthe period o f highbudget deficits inthe 1990s. However, inits current form the rule i s not sufficient to address the present fiscal challenges that Russia confronts. There i s a need for the further development o f fiscal rules for Russia, which would enable the government to successfully deal with the above-mentioned major fiscal challenges: oil price volatility, the financing o f structural reforms, and economic cyclicality. Each o f the three fiscal challenges requires a special risk-mitigating strategy. The adoption o f fiscal rules could be considered a part o f such mitigation arrangements. 1.64 The rest o f this section suggests three possible specific fiscal rules for Russia, which could be introduced as a part o f the overall government efforts to strengthen its budget management system. Such rules could be ultimately incorporated into the Budget Code (Article 92) and become a part o f Russia's regular budget p r o ~ e s s . ~ 1.65 Rule #I:Fiscal sustainabilitv rule. The debt stock of thefederal government may not exceed 30percent of GDP. Like the United Kingdom, Russia could benefit from having a formal fiscal rule related to the debt stock o f the federal government. For the purpose o f simplicity, the debt-to-GDP ratio couldbe suggested as an indicator preferable to more sophisticated ones that allow measuring debt on a net basis and innet-present value (NPV) terms. Given Russia's country risk profile and a highshare o f foreign currency debt inits debt structure, it appears that for sustainability Hereafter,proposals on fiscal rules are referredto Russia's federal budget. Subnationalgovernments in Russia havethere own, more binding, fiscal rules written in the Budget Code, which limit the size of both subnational debt and debt service spending. 15 reason Russia should maintain a fairly low limit for its government debt. In particular, we believe that the federal debt stock level inRussia should be limitedto 30 percent o f GDP.4 1.66 At the same time, following the best OECD practice, we believe that the rational longer term targeted debt level for Russia should be lower than i s allowed by sustainability constraints, and it may be advisable to set such a target at 20 percent o f GDP or lower. The government debt-to-GDP ratio i s already below 30 percent o f GDP and i s heading towards a 20 percent level. Thus, the current government policy i s de facto well in line with the proposed rule (i.e. in the short to medium terms this rule will not be binding for fiscal policy inRussia). However, the problem of highdebt levels may re-emerge when oil prices fall and the fiscal balance deteriorates markedly. The fiscal rule that limits future debt accumulation, adopted as a separate law or as an amendment to the Budget Code, would help the government to prevent this problem. 1.67 Rule #2: Oil price rule. An oil price adjusted Federal budget is allowed to be in deficit only temporarily to finance earmarked expenditures that are related to the implementation of specific structural reforms. The size of such a temporary deficit shall not exceed2percent of GDP. World oil prices are highly volatile. This volatility i s driven by a number o f economic and political factors on both the supply and demand sides. More or less accurate long-term forecasting o f oil prices i s virtually impossible and inaccuracies in forecasts are huge.5 However, the long-term moving average oil price, around which actual oil prices tend to fluctuate, demonstrates stability. 1.68 This makes it possible to design a fiscal rule based on the long-term average price of Such an oil price adjusted budget could be recalculated usingthe counterfactual oil price o f 20 US$/bbl for crude Russian oil. While the actual budget outcome may turn out to be in deficit or in surplus, the oil price adjusted budget (or budget in the long-term) needs to be balanced. 1.69 However, given the above-discussed needs for financing structural reforms in Russia, we believe it i s important to include a special covenant inthe fiscal rule which would permit accommodation o f reform-related fiscal costs inthe b ~ d g e t . ~ 1.70 The first group o f structural reforms, discussed above, requires just temporary budget financing for their implementation. Thus, the fiscal rule #2 would provide a binding fiscal framework for financing these reforms. Inaddition, the rule could help bridge the financing of the second group o f structural reforms that result in permanent additional government spending. Specifically, budget financing for each selected reform belonging to the second group could be provided on a decreasing basis. This financing should be program-based and provided under and within the broader budget allocation designated for new expenditure Let us note that, before the 1998 crisis, the debt-to-GDP ratio for the general government was below 50 percent, which is substantially less than the debt ratio stipulated by the Maastricht criteria. As Kopits (2004) emphasizes, markets have far lower tolerance for relatively high public debt-to-GDP ratios in emerging markets than in advanced economies. 'Pinto (1987) estimated a standard forecast deviation for off price at US$10.7 per barrel. The latter is estimated at $18-20 per barrel for the RussianUrals after accounting for the ongoing price boom. Beetsma and Debrun (2004) find -- in the context of the Euro arca's Stability and Growth Pact -- that fiscal rules sometimes may needto be relaxed for countries that are actively pursuing much-needed structural reforms. 16 commitments. For instance, it could cover 100 percent o f incremental costs during the first year o f implementation, 50 percent during the second year, and 0 percent during the third year. Such an approach would give the government time to identify and implement measures necessary to create fiscal room inthe regular budget for new liabilities that would emerge as a result o f the reforms. Controlling the overall size o f program spending on structural reforms would allow the monitoring o f whether the second fiscal rule i s adhered toS8Besides, this approach would be fully consistent with the ongoing transition to performance-based budgetingprinciples. 1.71 The proposed oil price rule should be applicable for the federal budget, as it demonstrates the much higher oil price sensitivity o f its revenues than subnational budgets. The reason behind it i s that almost all tax revenues, which are directly dependent on oil prices (oil export duties, mineral resource extraction tax on oil), accrue to the federal budget. 1.72 Rule #3: Structural fiscal balance rule. The structural budget should be kept in balance. It happens quite often that counties enjoying rapid economic growth boost their budget expenditures, which gives an additional (albeit temporary) impetus to their GDP growth. At the same time, such a pro-cyclical fiscal policy makes the public sector compete for resources with the private sector and pushes up interest rates. When the growth phase o f an economic cycle comes to an end, then it becomes very difficult to cut back the inflated government expenditures and restructure them in order to expand the counter-cyclical spending components. 1.73 This i s why, for example, the counties in the euro area have been discussing a modification of their budget rules to allow for a cyclical adj~stment.~ The budget adjusted for the impact of economic cycles is called a structural budget. The same approach, inprinciple, could be applied inRussia in order to prevent pro-cyclical budgetary expansion duringbooms (e.g., when GDP growth rate i s above 5 percent) and to allow for countercyclical budgeting during recessions (e.g., when GDP growthrate i s less than one percent)." 1.74 This rule would require that duringbooms, which inRussia are highly correlated with the periods o f high oil prices, the government should adhere to the budget surplus o f 1-2 percent o f GDP to avoid the pro-cyclicality in non-oil expenditure. Similarly, during recessions the government could allow the budget deficit o f 1-2 percent of GDP. Meanwhile, the structuralbudget would bebalancedinbothcases. 1.75 However, it appears that the formal or informal adoption o f the oil price fiscal rule precedes the adoption of the third fiscal rule. The fiscal management of oil shocks and structural reforms should be given the highest priority in the Russian case. Moreover, Russia has not yet gone through a completed economic cycle. Therefore, at this stage it appears that the establishment of rule #3 could not be done quickly because the reliable estimation of the * As discussed, the third group of structuralreforms is fiscally neutral interms of their costs. This approachwas made explicit inthe revised2001 Code of Conduct of the EuropeanUnion. loPerry (2004) argues that Latin American economies, which are subject to high macroeconomic volatility, ought to follow a rule that incorporates a countercyclical stance through a structural balance target or a stabilization fund. 17 structural budget would require additional time. Instead, a proper development of this rule should be considered as part of the longer term efforts to upgrade the fiscal management systeminRussia. E. IMPLICATIONS FORFISCAL POLICY 1.76 Inthis section we discuss someimplications oftheproposed fiscal rules, especially the oil price rule (rule #2), for fiscal policy. The oil price rule i s largely a modification o f and a potential legal basis for a de facto fiscal policy carried out by the Russian government. Indeed in drafting the 2004 budget the government used the price of 20 US$/bbl and the budget became law with the 0.5 percent of GDP surplus at this counterfactual price. Still, the government's current fiscal strategy could be strengthened ina number of aspects. 1.77 Stabilization Fund settings. The regime o f a financial reserve (before 2004) and a stabilization fund (since 2004) does not allow for saving all extra revenues associated with oil prices exceeding 20 US$/bbl. Overall, the enlarged budget has been saving only about 50 percent o f the extra oil revenues that it receives (World Bank, 2004a). Moreover, taxes on gas, which price i s strongly correlated with the oil price with a six-month lag, as well as general taxes on goods and services that also tend to grow in response to oil price increases, are not accmable to the stabilization fund in the course o f the fiscal year. Because the government saves too little o f the available oil windfall, this makes its budget policy too pro-cyclical. In particular, the government used a part of the oil windfall to expand the financing of its public investmentprogramby 1.4percent of GDP by 2002 (see Table 1.1, above). 1.78 Stabilization fund regulations need further strengtheningto broaden the revenue base for within the year transfers to the fund beyond taxes paid by oil companies, to include additional non-oil tax revenues actually received by the federal budget compared to the expected amount o f tax revenues budgetedat the counterfactual oil price of 20 US$/bbl. 1.79 This safeguard policy would sterilize the part of the revenue windfall which is an indirect result of high oil prices, and would limit spending expansion taking place in the course o f budget execution in cases when oil prices and tax revenues tum out to be higher than assumed in the budget law. Presently, the Ministry o f Finance makes quarterly projections of expected tax revenues. Therefore, it appears to be quite possible to use these projections as benchmarks for assessing the amounts o f the general tax revenuewindfall to be transferred to the stabilization fund on a quarterly basis. Such quarterly transfers should replace the current practice o f a single annual transfer to stabilization fund made in February o f the following year." 1.80 Notably, besides playing the role o f fiscal stabilizer, the stabilization fund should also perform an important function of a macroeconomic stabilizer by sterilizing the oil revenue windfall. Increasing the share o f the actual windfall accumulated in the stabilization fund would help the authorities to maintain the policy aimed at the relative stability o f the real exchange rate to preventthe development o f Dutch Disease. Technically, the amount of the quarterly transfer would be equal to the positive balancebetween the actual and counter-factualtax revenues for that quarter net of transfers of oil export duty and tax on the extraction of mineral resourcesthat have already beentransferredto the stabilization fund during that quarter. 18 1.81 Moreover, the currentcap on the size o f the stabilization fund inthe amount o f Rb 500 billion (which i s 3.8 percent o f 2003 GDP), imposed by the Budget Code, i s too small. Itmay become a serious constraint to an efficient long term fiscal policy. This cap has to be either revised or eliminated as early as 2005, when it i s expected that the accumulated reserves inthe stabilization fundwill exceed Rb 500 billion. 1.82 Expenditure control. At oil prices standing above 20 US$ibbl, the second rule would help in maintaining overall fiscal discipline, especially with respect to control over expenditure levels. During the previous years, when actual oil prices turned higher than stipulated in the budget laws, the government demonstrated little resistance to pressures on expenditure increases in the course of the budget execution. In both 2002 and 2003, amendments to the budget laws considerably increased the expenditure amounts relative to the original budgets. It means that a considerable portion o f the total oil revenue windfalls was actually spent on the basis of within-the-year decisions and without proper strategic prioritization. As a result, at a counterfactual oil price o f 20 US$/bbl the general budget would have been executed with a deficit o f 1.25 percent of GDP in 2003 (IMF, 2004). Our estimates show that in 2004, for the same reason, the budget outcome at this oil price was in deficit o f 1.7 percent o f GDP and in 2005 it is expected to well exceed 2 percent o f GDP because of unplanned spendings on benefit monetization in2005. 1.83 Thus, as follows from the analysis of recent developments, the adoption of the fiscal rules would require tightening o f the government's fiscal policy, as well as speeding up the development o f budget programs that support structural reforms. This i s also because at the moment a very little portion of government expenditures represent program spendings on structural reforms. 1.84 Tax policy. A recent government initiative to lower the social tax rates in 2005, compensating for foregone revenues o f the Pension Fund from the general revenues of the federal budget, represents a suboptimal solution. Inessence, this decision means that instead o f accumulating a higher proportion o f the oil revenue windfall in the stabilization fund, the government returns a part of the windfall to the private sector through lower taxes. To date, the private sector in Russia has demonstrated quite prudent financial behavior - it has saved almost its entire share of the windfall and invested a good chunk of it in net foreign assets stretching its use in time (World Bank, 2004b). Therefore, for the time beingthis suboptimal solution could indeed generate serious potential macroeconomic benefits. 1.85 Yet for the government the flipside of this policy is that, when oil prices eventually fall, the foregone revenues will sooner or later have to be raised. This implies that the government should have an up front commitment to raise taxes in the future, and, (either increase tax rates or broadentax bases) inthe future. Respectively, the government should not make the commitment that all of the tax cuts it suggests in times o f oil windfalls will be permanent. Given the present high level of the marginal social tax rate and the problems with administrating this tax, it would appear more appropriate to raise the rates on other taxes when time comes. 1.86 Public investment programs. Using the commodity price windfall to finance public investment has been quite a common practice for large commodity exporters around the 19 world. However, only a few countries have succeeded in doing this ina non-distorting way." A few factors led to poor results. First, the investment usually went to build structures rather than to purchase equipment, implyinga lower rate o f return (both financial and economic) on the investment. Second, the investment contributed to economic overheating by provoking higher inflation in the construction sector. This additionally reduced the efficiency o f public investment. Third, the typically weak institutional capacity o f the governments resulted in poor prioritization and in implementation inefficiencies that led to considerable waste and misuse o f public funds. 1.87 This i s another reason why Russia should pursue an opposite, countercyclical, policy in the area of public investment - being rather conservative during booms, while protecting the levels o f public investments (including reform-related spending) during "bad times" by investing a part o f the saved windfall. This would allow for better selection o f public investmentprojects and improved predictability o f their financing and it would help to smooth the fluctuations o f activities in the construction sector. Moreover, such investment policies would be consistent with the lessons learned from earlier fiscal adjustments to external shocks (Easterly and Serven, 2003). 1.88 Managing external shocks. In the oil price range o f US$16-20 per barrel, the proposed second fiscal rule would imply that the actual budget would be executed with a deficit o f up to 4 percent o f GDP. This deficit would have to be financed by a reverse transfer from the stabilization fund, which would be sufficient to maintain the level o f budget expenditures stable inreal terms. 1.89 Should oil prices temporary fall below 16 US$/bbl, the government could employ the same mechanism o f financing budget expenditures from the stabilization fund in order to maintain the constant level o f expenditures in real If the government faces an extended period of depressed oil prices, it should trigger additional adjustments in its fiscal policies by raising general taxation, cutting the least efficient expenditures, and borrowing. amount o f resources accumulated in the stabilization fund -- the more reserves that are The cut-off time for switching to such a supplemental policy package would depend on the accumulated, the later the policy would have to be launched. 1-90 Specifically, a temporary increase in taxes should (i) be accruable to the federal budget, and (ii) target taxes that have a stable and broad base and those that have the least distorting effect on the real sector, such as VAT and excises. 1.91 The cuts in expenditures should target (i) white elephant expenditures with the least transparency (and, most likely, the least efficiency), such as, the current federal targeted programs, and (ii) spending under the line "other expenditures." The real expenditure cut could also be achieved by freezing specific expenditures innominal terms. As opposed to the method o f achieving fiscal consolidation after the 1998 crisis, social expenditures should be marginally affected at the time o f future fiscal adjustments inRussia. "Inthe 1970s-80sSenegalandBotswanamanagedtoavoidinflatingrelativepricesoninvestmentgoodswhile increasingpublic investment during the periods of positive trade shocks. The reason was that in both countries investment booms were heavily concentrated in purchases of machinery and equipment, Le., tradable capital (Collier and Gunning, 1999). l3Dependingon the amount of resources accumulatedinthe stabilization fund, this policy could be sustainedfor 6-12 months. 20 1.92 An efficient way of reactivatingpublic borrowing couldbe through usingthe deferred drawdown option (DDO) offered by IFIs. The benefit o f this option i s that it can be agreed on well ahead o f time at low cost. At the same time, borrowing commercially on capital markets in the period of low oil prices would be much more expensive for Russia, which is an unavoidable reflection o f the increasedcountry risksduringbadtimes. 1.93 When the rule #3 i s adopted, it would have an additional impact on the management of fiscal flows. Duringbooms the structural fiscal surplus would have to be transferred to the stabilization fund as another funding source. Similar to that, during recessions the reversed transfers from the stabilization fund would prevent expenditures from falling inreal terms. In addition, recessions usually bring about incremental expenditures (e.g., on benefits and retraining for the unemployed, capital investment projects and public works). This i s why runninga countercyclical policy duringrecessions usually leads to an increase ingovernment borrowing (if the stabilization fund quickly diminishes). At the same time, recessions are the best times from a political economy viewpoint for identifying and eliminating inefficient expenditures. When an economy prospers, this i s much more difficult to do. 1.94 Managing market expectations. As far as the policy of maintaining a strong fiscal balance i s not fixed inthe law, it remains unclear how strongly the government i s committed to continuing this policy in the future. Making the fiscal rule into law could be helpful in t e m s o f improvingthe perceptions of various market players, includingpotential investors. 1.95 As far as the third fiscal rule i s concerned, if and when the govemment decides to follow it, it would be very important to adopt this rule while the economy i s in a growth phase. This step would send a strong signal demonstrating government's commitment to smoothing the influence o f economic cycles on the economy. Inthis case, market participants would have stronger confidence in overall government fiscal discipline despite the govemment's expansionary, countercyclical fiscal policy duringa period o f recession. 1.96 The rolefor the IFIS.The fiscal framework for structural reforms, which i s analyzed inthisChapter, suggestsconsiderable variation inbudget expenditures (Le. their fluctuation in response to various challenges and shocks). However, to ensure that these expenditure interventions are timely and efficient, there should be appropriate institutional capacity within the public sector for strategic management of financing o f both public investment programs and structural reforms. In this area the World Bank and other donors could offer their expertise and project management experience. For instance, joint investment projects with the World Bank in support o f various structural refoms and infrastructure upgrades could be designed in such a way that the Bank would provide only the start-upand pilot financing and would help to set up project implementation mechanisms. Scaling-up the projects would become entirely the govemment's responsibility. For this purpose, the govemment could use the same project preparation and implementation mechanisms that would be jointly establishedunder the initial WorldBank projects. Such a project framework may significantly increase the efficiency o f government spending on investment and structural refoms if relatively large volumes o f government co-financing would leverage the World Bank funds. In addition, the Bank and other IFIs, and the donor community in general, might provide direct technical assistance to public sector institutions to address the binding capacity constraints. 1.97 The govemment could keep the amount o f tied borrowing from IFIs stable, but at the same time could vary its contribution to joint investment projects by reducing co-financing of structural reforms during recessions. Thus, the government's share in IF1projects could be lower during recessions and higher during high growth phases. This would help when 21 necessary to free the money for maintaining the stability of vital current expenditures in real terms. Operationally, this could be done, for example, by varying the number o f regions in which originalpilot projects are replicated. F. CONCLUSION 1.98 The outcomes of the proposed second and third fiscal rules can be summarized in Table 1.4. Table 1.4: Matrix of the ProposedFiscalRulesfor Russia oilprice, US$/bbl Budget balances under differentfiscal rules, % of GDP 20+ 16-20 12-16 1. Oil price adjustedbudgetbalance balanced balanced Balanced 2. Oil price adjustedbudgetbalance with the ceiling for structural reforms costs (2ndfiscal rule), not more than -1-2 -1-2 -1-2 3. Savings/expendituresfrom counter-cyclical policy (3rdfiscal rule), not morethan +1+2 0 -1-2 4. Baselinebudgetbalanceestimatedat actualoil price (excluding spendingon both structuralreforms and counter-cyclical policy), % of surplus -0-2 -2-4 GDP Overall actualbudget outcome (=2+3+4) surplus -1-4 -1-8 Source: WorldBank staffestimates. 1.99 It follows from the last column of Table 1.4 that the combination of all three fiscal rules may give rise to an actual budget deficit o f up to 8 percent of GDP ifthe period o f very low oil prices and output fall coincides with the implementation o f structural reforms. Specifically, fiscal sustainability rule would generate a budget deficit of up to 4 percent o f GDP. Oil price rule would add to the deficit up to 2 percent o f GDP, and structural balance rule would add another 2 percent o f GDP. However, inthe terms of structural budget balance, the deficit would not exceed 4 percent o f GDP - 2 percentage points stemming from financing structural reforms under oil price rule and another 2 percentage points stemming from countercyclical policy (structural balance rule). It should be recognized though that duringthe periods of low oil prices the authorities could draw from the oil stabilization fund and, inthis sense, a portion o f respective deficits would be self-financed. Followingthe same fiscal rules during a period o f high oil prices exceeding 24 US$/bbl requires that the federal budgetis executed with a s~rplus.'~ 1.100 In terms o f sequencing, the first two fiscal rules address the most crucial fiscal risks that the Russian fiscal policy faces, and therefore they have to be adopted first. The third fiscal rule i s somewhat supplementary and couldbe adopted thereafter. 1.101 Overall, following the proposed fiscal rules would allow the government to maintain the fiscal stability and yet adequately address the emerging structural and cyclical challenges l4At oil prices inthe range of 20-24 US$/bbl the overall budget outcome i s ambiguous. It would depend on the extentto which the govemment would finance structural reforms and/or conduct countercyclical policy. 22 through the embedded flexibility. This fiscal framework would make the budget policy more predictable and sustainable, thereby facilitating the development of the private sector. 1,102 Inaddition, the adoption o f the proposedfiscal rules would contribute to the successful attainment o f objectives o f the government's Concept of Budget Reform intwo program areas -- introduction o f medium-term budget planning and division of budget spending into the existingcommitments and new expenditure commitments. 1.103 The World Bank as well as other international financial institutions may help the government to overcome its institutional weaknesses in managing the implementation of reform and public investmentprograms. This appears to be the way to increase the efficiency o f public spending, so that it fully complies with the proposed fiscal rules. 23 Chapter 2. FISCAL COSTSOF CIVIL SERVICEREFORM 2.1 This Chapter provides recommendations on the planning and sequencing of civil service reform in Russia based on (i) estimates for direct fiscal effects o f various reforms in the area o f public administration, and (ii) analysis o f feasibility of different reform options. The Chapter suggests that broad reforms in the core government administration and in the civilian public sector at large may be implemented within five to seven years but should be differentiated by the scope of pay adjustment in various sub-sectors of civilian employment, closely monitored for non-wage expenditure growth, and complemented by significant staffing adjustments in the civilian public sector as well as by at least some staff reductions in the core govemment administration. Implementation o f such reforms would require additional budget financing as compared to 2003 expenditure levels, but fiscal costs could be sustained within a reasonable range. A. EMPLOYMENT FINANCINGOF COREGOVERNMENT AND ADMINISTRATION AND THE CIVILIAN PUBLIC SECTOR 2.2 Overall core government administration and civilian public sector employment accounts for a significant part of the Russianpopulation: in2002, employment inthese sectors was about 11 million people or about 7.7 percent o f the country's total population. The structure of civil employment inRussia i s presented inTable 2.1. Table 2.1: Structureof Core Government Administrationand CivilianPublic Sector Employment in 2002 Sector Employment Employment (`000 persons) (YOof total) Total: Core govemment administration and civilian 11,062.2 100.0 public sector (A+B) A. Core government administration employment 1,252.3 11.3 Civil service and local self-govemment 919.5 8.3 -Federalcivilservice 463.7 4.2 -Subnationalcivilservice(incl.localself- 455.8 4.1 government) Other core govemment administrationemployment 332.8 3.0 B. Civilian public sector employment 9,809.9 88.7 Health, physical culture and sports, social services 3,356.1 30.3 Education 5,282.3 47.8 Culture and Arts 915.4 8.3 Researchand Development* 256.1 2.3 *Data are for 2001. Source: Rosstat. 2.3 In 2002, core govemment administration employment accounted for 1.25 million people (or roughly 0.9 percent o f the total population) with about 47 percent employed by the federal authorities and 53 percent employed at the regional and municipal levels o f the govemment. Employment in executive authorities (including local self-government with executive functions) accounts for more than 86 percent of core government administration employment. 2.4 Over the past eight years, the core government administration employment has increasedby almost a quarter with the most drastic increases inlegislative, judiciary and other government bodies. Employment growth in the executive branch o f core government administration was slower, but still since 1994 there have been significant increases in absolute staffing levels (Table 2.2). Table 2.2: Core GovernmentAdministration Employmentin Russiain 1994-2002 (in thousands) 1994 1995 1996 1997 1998, 1999 2000 2001 2002 Total core government 1004.3 1061.8 1093.0 1108.9 1102.8 1133.7 1163.3 1140.6 1252.3 administration employment as % of the 1994 employment 100 105.7 108.8 110.4 109.8 112.9 115.8 113.6 124.7 A. Legislativeauthorities 7.1 8.8 10.5 11.1 11.0 14.5 15.5 19.1 20.4 as%in 1994 100 123.9 147.9 156.3 154.9 204.2 218.3 269.0 287.3 B.Executive authorities 894.4 945.1 971.3 984.4 983.9 1006.5 1029.5 983.7 1072.6 as % in 1994 100 105.7 108.6 110.1 110.0 112.5 115.1 110.0 119.9 B l . Federal executive 379.9 416.2 433.0 442.5 409.9 412.8 404.7 377.1 446.8 as YOin 1994 100 109.6 114.0 116.5 107.9 108.7 106.5 99.3 117.6 -HQ-based 33.8 33.8 31.2 30.4 28.9 30.9 30.3 28.8 28.8 as % in 1994 100 100 92.3 89.9 85.5 91.4 89.6 85.2 85.2 -Deconcentrated 346.1 382.4 401.8 412.1 381.0 381.9 374.4 348.3 418.0 as % in 1994 100 110.5 116.1 119.1 110.1 110.3 108.2 100.6 120.8 B2. Subnational executive 514.5 529.0 538.3 542.0 574.0 593.4 624.8 606.6 625.8 (incl. Local self-government) as % in 1994 100 102.8 104.6 105.3 111.6 115.3 121.4 117.9 121.6 C. Judiciary authorities 102.7 107.3 110.1 111.5 105.6 110.1 115.2 134.3 153.9 as % in 1994 100 104.5 107.2 108.6 102.8 107.2 112.2 130.8 149.9 D.Other authorities 0.1 0.6 1.2 1.8 2.3 2.6 3.1 3.5 5.5 as %in 1994 100 600 1200 1800 2300 2600 3100 3500 5500 Source: Gimpelson (2002) and Rosstat (2002). 2.5 It is notable, that the only category of core govemment administration employment that has been declining over this period was employment in HQ-based federal executive authorities, while the employment in deconcentrated federal executive bodies and particularly inthe subnational executivebranchhasbeen steadily growing. This disproportional growthin part may be explained by the changes in government functions from central planning to regulation of economic activities during the period o f transition. These functional changes led to growing numbers o f core government administration employees working for federal deconcentrated units (treasury, tax administration, customs, employment service, anti-trust agencies and property management units). Decentralization and transfer o f some o f the government functions from the federal to sub-federal levels may also explain part of the employment growth at the subnational level (see Gimpelson, 2002), while some o f the increases may have resulted from a lack o f control o f staffing levels in the regional administrations and local self-government, Moreover, the recent World Bank report points to excessive employment generation in the public sector at the regional level due to insufficient job creation in the private sector. As a result, budget employment started to function as a quasi social safety net (see World Bank, 2004b). 26 2.6 Federal and subnational civil sewice in 2001 accounted for about 75 percent of total core government administration employment. The structure o f the federal and subnational civil service i s presentedinTable 2.3. Table 2.3: Structure of Civil ServiceEmploymentin 2001 Federalcivil servants Subnational civil servants Category Legislative Executive Judiciary Others Total Subjects Munici- Total Total ofRF pal Total core 4.5 377.1 122.0 1.3 504.9 N/A N/A 635.8 1140.6 government administration employment, (`000 persons) O/w total civil 3.8 322.5 95.2 1.3 422.8 153.3 283.6 436.9 859.7 service employment, (`000 persons) As % of total civil service employment Total civil 100 100 100 100 100 100 100 100 100 service Category A 11.7 0 20.4 2.2 4.7 3.6 4.3 4 4.3 Category B 26 0 1.7 3.2 0.6 3.7 4 3.9 2.3 Category C 62.3 99.9 46.6 94.6 87.5 92.7 91.7 92.1 89.8 Top 1.9 0.4 0.1 8.4 0.4 2.2 3.3 3 1.7 Chief 16.3 1.4 0.2 28.4 1.3 9.7 12.4 11.4 6.4 Lead 27.6 24.9 2.1 45.1 19.9 23.2 14.9 17.8 18.8 Senior 11.7 53.6 10.5 10.7 43.4 35.9 30.9 32.6 37.9 Junior 4.8 19.7 33.8 2 22.7 21.7 30.2 27.2 25.0 Prosecutors 31.3 7 3.4 Source:Rosstat (2002). 2.7 The remaining 25 percent of the core government administration personnel do not have a status o f civil servants and their employment contracts are regulated by the Labor Code. Most o f these employees perform supporting or logistical functions (secretaries, drivers, etc.). 2.8 Comparison of cross-country data on public sector employment (see Figure2.1. below and Annex 2.4 to this Chapter for more information) confirms: first, that the total employment in core government administration and the civilian public sector in Russia is quite large as compared to other countries; but, second, the composition o f this employment i s rather distorted. It i s characterized by fairly tiny federal HQ-based government authorities, as opposed to much better staffed federal deconcentrated units, to even more numerous regional and municipal administrations, and to a clearly overstaffed civilian public sector. Some o f these imbalances were inherited from the central planning era, while others (especially an expansion in subnational administrative employment) have emerged over the years of transition. 2.9 The core government administration in Russia is rather small as compared to high- income OECD economies as well as many transition economies. In fact, only Poland and 27 Kazakhstan have smaller core government administrations, while all federative countries presented in the sample, especially those with comparable population numbers, have significantly higher core government administration employment, particularly at the sub- national level, as illustrated by the Figure 2.1 below. At the same time, the number o f education and health employees inRussia i s significantly higher than in all other countries in the sample, including transition economies that historically also had high employment in the health and education sectors. Figure 2.1: PublicSector Employmentin SelectedOECD Countries and Transition Economies (as a percentageof population) PCivilian Central Government W SubnationalGovernment 0EducationandHeaRhEmployees Note: For some countries not all types of employment data are available, hence there are some gaps inthe figure above. Source: WB Database. 2.10 Overall, public sector employment i s rather high, and its financing i s not adequate vis- a-vis the employment level. This coexistence o f highemployment and low fundingrepresents at the moment a major disproportion inthe Russianpublic sector. 2.11 Based on the 2002 baseline data, the current status of financing core government administration and civilian public sector employment could be summarized as the following. As Table 2.4 shows, in 2002 the total budget expenditures on cash compensation in core government administration and the civilian public sector accounted for about 6.6 percent o f GDP. Cash compensation incore government administration was only 1.7 percent o f GDP out o f which only about 0.7 percent o f GDP was spent on the federal government's wage bill (or 1.4 percent o f GDP including cash compensation o f the armed forces). This i s quite a low ratio by international standards. In the 1990s, the central government wage bill (including armed forces) in large developed economies with a federal structure varied from 1percent to GDP in Germany to 2.1 percent in Australia. Some sources quote OECD averages for the early 1990s as being equal to 4.5 percent o f GDP, while Latin America and the Caribbean (LAC) as 4.9 percent (see Annex 2.4). Inaddition, there i s a view that while the expenditures on core government administration (specifically, on federal executive civil service) in Russia are lower than in other countries, the functions of the federal executive authorities in Russia are broader than in comparable countries with the federal structure of the government (see HigherSchool ofEconomics (HSE),2004, for more information). 28 Cash Compensation, YOof 2002 GDP Non-WageExpenditures,YOof 2002 GDP* Total Federal Sub-federal Total Federal Sub-federal Budget Budgets Budget Budgets Core Government Administration 1.7 0.7 1.o 1.o 0.4 0.6 CivilianPublic Employment 4.9 1.1 3.8 3.2 0.2** 3.0** Total 6.6 1.8 4.8 4.2 0.6 3.6 ***Estimated. Non-wage expendituresincivilian public sector employmentdo notincludebudgettransfers to extra-budgetaryfunds and government transfers to population. Source: Rosstat, Ministry of Finance, and staff calculations. 2.12 The 2002 ratio o f the ,average cash compensation in the federal core government administration to per capita GDP accounted for 1.21, which was lower than in other middle- income countries as well as in Europe and Central Asia (ECA) and OECD economies. For example, the ECA average i s 1.3, OECD average i s 1.6 while in L A C countries the indicator accounts for 2.5. 2.13 Hence, although core government administration pay and employment figures vary from country to country and are highly dependent on the structure and functions of the government as well as on the methodology used for national public sector statistics, the cross- country comparison shows that interms of spending Russia's core government administration i s quite small. Both wage bill inthe core government administration (as a percentage of GDP) and the average cash compensation (as a ratio to GDP per capita) are considerably lower than i s common internationally. B. CIVILSERVICE REFORMPRIORITIES AND PROGRESS TO DATE 2.14 The need to increase productivity and transparency of the public administration and improve quality o f public services has been widely recognized in Russia both inside and outside the Government as a necessary condition for ensuring implementation of other reforms and sustainable development o f the country. Hence, implementation o f the civil service reform (including administrative reform) i s one of the Government's key priorities in the mid-termreform agenda. 2.15 The priorities for civil service reform were defined in the Presidential Decree No. 1336, dated November 19, 2002: On Federal Program, "Reform of the Civil Service of the Russian Federation (2003-2005) ", containing a set of measures aimed at improving legal, organizational, financial, and methodological framework o f the public service" inthe Russian Federation. The main areas of reform include the following: 0 creating a comprehensive normative and legal framework governing the public service o f the Russian Federation; 0 developing efficient mechanisms for implementing a personnel policy for the purpose o f optimizing the structure o f public service personnel; The term "public service" i s usedinthis section to denote civil service, police service and armed forces. 29 developing measures aimed at improving compensation o f public servants and financial, economic, and logistic support o f the public service o f the Russian Federation, and at using advanced information technology in the public service system in a rational manner; implementing staff training programs for the public service o f the Russian Federation and professionaldevelopment programs for public servants; and forming a system o f management o f the public service o fthe RussianFederation. The specific objectives are described inthe FederalProgram as following: creating conditions for optimal organizational and legal support o f the public service; defining roles, powers, and responsibilities o f public servants on the basis o f job (service) descriptions; implementing new techniques for planning, financing, stimulating, and assessing public servants' activities, and using public service system resources in a rational manner; ensuring openness o f the public service for the benefit o f civic society development and strengthening o f the state; applying efficient methods for selecting qualified personnel for the public service and for assessing professional performance o f public servants, as well as creating conditions for their job (service) promotion; implementing staff training programs for the public service and professional development programs for public servants; implementing mechanisms for identifying and solving public service-related conflicts o f interest, as well as introducing a legal regulation of professional ethics of public servants; creating an optimal material and technical environment for efficient functioning o f the federal public service and for performance by public servants o f their official (service) duties; and ensuring the development o f a public service administration system. During2002-2003, significant progress has been achieved inconceptualization o f the reform approaches and preparation o f the legislation that i s to provide a basis for reform measures implementation. A framework Federal L a w No. 58-FZ dated M a y 28, 2003 "On State Service System of the Russian Federation ", established a two-level system o f public service in Russia: federal public service (administered by the Federal Govemment); and civilian public service o f a subject o f federation. Federal public service i s subdivided into civilian, military and law enforcement service. Rule o f law, pre-eminence o f human and civil rights and freedoms, equal access to public service for all citizens, integrity o f the public service, interconnection o f civil and municipal service, openness and accountability, professionalism and protection from undue interference in public service activity by private individuals or legal entities are declared as guidingprinciples for public service system inthe country. The law introduces a notion o f a consolidated register o f public servants positions as well as class, diplomatic, military, and special ranks that are to substitute the professional grades that existed under the previous legislation. 30 2.18 The framework law requires separate laws to be enacted for eachtype of public service (civilian, military, and law enforcement). To this end, a draft Federal Law No. 79-FZ dated July 27, 2004, "On State Civil Sewice of the Russian Federation, was enacted in February " 2005. This law defines rights and obligations o f civil servants; limitations and prohibitions o f the state service requirement to the conduct of the civil servants; regulation of the conflict of interests; and obligations to annually submit information on income, property, and related obligations. The law provides for a classification of civil service positions inaccordance with categories and groups o f appointments (see Table A2.2 in Annex 2.2). The law contains a number o f important innovations, including competitive recruitment for civil service positions; introduction o f job descriptions that should, inter alia, contain performance indicators for each position; and changes in remuneration related to introduction of performance-oriented incentives.l6 2.19 In the framework of the civil service reform based on functional reviews several experiments related to introduction o f performance budgeting and performance management have been launchedinfive pilot regions participating inthe Program. 2.20 Significant attention i s also given to the administrative reform agenda that, in the narrow sense, includes inventory of functions of the federal executive bodies and restructuring and right-sizing o f the government. Priority areas for implementation o f administrative reforms for 2003-2004, defined in the Presidential Decree No. 824 dated July 23, 2003, "On Measures of Implementation of Administrative Reform in 2003-2004, I`are the following: 0 restriction of the state interference in economic activity o f entrepreneurs, including restriction of state over-regulation of the business; 0 elimination of duplication o f functions and authorities of organs o f the Federal Executive powers; 0 development of the system o f self-regulating organizations inthe economy; 0 organizational separation o f functions related to the regulation of economic activity, control and supervision, management o f state property and provision o f service to the citizens and legal entities by the state structures; and 0 completion of the process o f separation of functions between structures o f executive power at federal and regional levels, optimization of the activity o f the territorial structures of the federal executive power. 2.21 In February 2004 a specially created Government Commission chaired by the Vice Prime Minister completed the review o f all 5,600 functions o f the federal executive bodies and it was expected that about half o f the functions would be revised or abandoned. This would require significant changes inthe legal andregulatory framework to be implemented by the end of 2004. The results o f this inventory o f function^'^ became the basis for a major l6Accordingto some analysis,however, the provisionsfor linkages betweenpay andperformanceinthe law are still not sufficient since about 84 percent of paywould not depend on performance(HSE, 2003). l7It shouldbenotedthat the review lookedat the functions as they are set up inthe existinglegal andregulatory framework, without any analysis of whether these functions have been actually implemented or whether the functions have been actually performed but are not reflected in the regulatory framework. More detailed agency-specificreviews would be needed later to increase the efficiency of executive bodies. These future reviews should be also linked with the introduction of administrative operations manuals and automation of governmentbusinessprocesses(see HSE, 2004). 31 reform inthe federal executive government structure launched by the President inearly March 2004. As opposed to the previous structure o f the federal government, the new one introduces a clear mechanism for division o f authority and responsibility between the three tiers o f the executive power: federal ministries responsible for policy development, federal services for exercising control and supervisory functions, and federal agencies in charge o f public service delivery. The number o f federal ministries was drastically cut from 23 to 14 (later the number o f ministries was increased to 16). 2.22 As o f March2004 some work on a number o f issues relatedto administrative and civil service reform, including development o f concepts and legal framework providing for introduction o f administrative operations manuals, quality, and accessibility standards for public services, e-government architecture, and deregulation issues as well as access to information and development o f the pre-court appeals system, had been started. Some achievements have been made in introducing performance-oriented budgeting at the federal level" and initial steps have been taken to strengthen anticomption activities. This demonstrates the comprehensive nature o f public sector reform that i s being prepared in Russia with most o f the implementation still planned for the future. c. A FRAMEWORK FOR THE ASSESSMENTREFORMIMPACT OF 2.23 As described in Section B civil service reform includes a variety o f measures that, for the purpose o f estimating the fiscal implications o f the reform, could be grouped into the following key components. B Pay reform intended to reduce the existing gap between the compensation inthe public and private sectors (these efforts refer specifically to salary adjustments in core government administration employment). Implementation o f this component o f reform would have direct fiscal implications caused by an increased wage bill. '% HR management reforminvolving the changes inrecruitment, promotion, rotation o f staff, training, performance appraisal, disciplinary procedures, career paths, retirement procedures, etc. Implementation o f this component would lead to increased expenditures on humanresource management and training.'' b Modernizationof public service that would entail development and introduction o f new business processes and procedures based o n modem information technologies and e-government approaches. The direct fiscal implications o f this component include additional spending on equipment, use o f infrastructure and support services (such as telecommunications), and other overhead costs o f core '*In accordancewith the Concept for Reforming the Budget Process (2004-2006) approvedby the Govemment of the Russian Federation in May 2004, as a part of 2005 budget cycle all ministries and agencies reporting directly to the Government of the Russian Federation were required to prepare Reports on Results and Main Areas of Activity, which have to contain objectives, tasks, and performance targets for the relevant ministries (including their subordinate agencies) and present a budget breakdown by programs and tasks to be implementedby the ministries. l9Another possible fiscal implication of the HR reform would be an increase in intensity of inter-regional staff rotation. Currently the scope of rotation inside the civil service is limited because of the fiscal constraints (Le. provision of housing, relocation benefits, etc.), and usually it is HQ-directed. Since to the knowledge of the authors, there have been no estimates yet for the desirable intensity of rotation, this aspect of reforms is not takeninto account for the purpose of the Note. 32 government administration. The costs o f modernizing the public service would entail both significant investments and recurrent expenditures on operation and maintenance o f the investments made. Estimating the requirements for investments needed for public service modernization i s a separate significant task that itself would suggest a number of reform scenarios. Hence, in this Chapter we will be focusing exclusively on current expenditures." For the purposes o f this analysis, it i s assumed that though automation of some functions within the government may lead to reduction in staffing, this effect i s not analyzed separately for this reform component. All expected reductions instaff levels inthe model are associated with administrative changes and are accounted for under the public administration reform component'. % Public administration reformz2includes decreasing the scope o f government interference in economic processes, eliminating duplication in functions and authorities of different federal executive bodies, development o f self-regulating organizations, differentiation between regulatory, supervisory, property management, and service provision functions o f the government, as well as separation of responsibilities between federal and regional executive bodies. It i s expected that the reform process would involve restructuring and right-sizing of the core government administration, better allocation o f functions between various government agencies, out-sourcing of non-essential functions to the private sector, etc. This component o f reforms would mostly affect the executive branch o f the core government administrationz3. The reform would result inbudget savings from the reduced number o f core government administration employees, as well as in significant efficiency gainsz4. 2.24 The ongoing debates on civil service reform are mostly focused on salary increases and other improvements in the civil service (both federal and subnational), especially in the executive branch of the federal civil servicez5. However, the political economy of the reform suggests that a significant pay adjustment inthe civil service i s likely to become a trigger for corresponding increases in compensation in the whole core government administrationz6,as ' O Fiscal costs of implementation of some HR management reform components (for example, introducing competitive recruitment for civil service positions) have beenestimatedby HSE (2004). 21 Some altemative estimates of the costs of public service modernization related specifically to implementation of electronic administrative manualsare presentedinthe recent HSE (2004) report. ''Formally, the administrative reform in Russia is not seen as apart of the civil service reform. It is considered to be a complementaryreform with the priorities defined in Presidential Decree No. 824 of July 23, 2003 On Measures of Implementation of Administrative Reform in 2003 - 2004. 23 Civil servants employed by other branches of the core govemment administration may be affected by other plannedon-going reform efforts (such as a judicial reform). However, for the purposes of this Chapter it is assumedthat these other branches remain unaffectedby the administration reform. 24 Inthe framework ofpublic administrativereform, it is envisagedthat some ofthe functions ofthe government entities would be abolished, while some of these functions would be transferredto "quasi-government" entities. For the purpose of this Note, by "a reducednumber of core government employees" we mean absolute reduction inanumberof civil servants/publicemployeesas aresult of abolishing some of the existing functions. 25 The most recent discussionsrelated to possible scenarios of the pay reform are presentedin the HSE (2004) report. 26 Rationalefor inclusion of non-civil servants into the analysis of fiscal implications of the pay reform may be illustrated by the example of the State Customs Committee that currently has three types of personnel: (i) civil servants; (ii)uniformed personnel (that includesstaff performing intelligence functions, as well as all staff at the positions of Division Chief and above); and (iii)staff with employment regulated by the Labor Code. In this 33 well as inthe public sector at large (inthe general government), even if the current legislation provides no formal linksbetweenthe wages levels of civil servants' and those o f other civilian public sector employees. Otherwise, it would be quite difficult politically for the Government to sustain a drastic increase in the salary gap between the civil servants and the rest o f the public sector employees. The importance of this link was highlighted once again by the recent government decision regarding salary increases in the public sector, including a 30 percent pay increase in the entire public sector inOctober 2003 and the proposed similar all- inclusive 50 percent increase for 2004. Hence, the approach followed in this Chapter i s that the fiscal implications o f the proposed reforms in civil service should be simultaneously assessed intwo different frameworks: 0 ``narrow approach" covering fiscal costs of reforms in core government administration; and 0 "broad approach " assessing fiscal implications of a comprehensive reform package and salary adjustment inthe entire civilian public sector employment. 2.25 It is recognized that in fact the proposed "broad approach" may be expanded even further to include the fiscal costs associated with the increased pay in police service and armed forces. These spheres were not included in this Chapter because of the lack of data readily available for simulations. Moreover, inclusion of the armed forces into the analysis would call for additional analysis of military reform options, which i s a large task in itsele7. Also, it i s recognized that implementation o f civil service reform would create additional pressures on the pension system, since keepinga decent proportion between the pensions and wages i s another important policy constraint. In particular, reducing positions that are to become vacant due to staff retirementpresents a relatively soft option for adjusting staffing levels. Inaccordance with Rosstat data, in 2003 19.3 percent o f civil servants and 29.2 percent of municipal servants were over 50 years old. However, the increase inpension age proposed as part of pension reform in Chapter 3 reduces potential benefits of such scenario. Proper consideration o f this area would require separate research and would have to be analyzed in the context of the pensionreform. Fiscal implications o f the pension reform are presented ina companion Chapter preparedjointly by the Bank and the Institute for Economy inTransition. 2.26 It should be noted that the implementation of civil service reforms would bringabout not only fiscal costs but also result innoticeable fiscal and economic benefits associated with a stronger productivity growth, better resource allocation (achieved through granting more autonomy to government agencies in reallocation of budgeted funds, adoption o f more transparent procurement practices and wider application o f competitive procurement methods, etc.), automation of business processes, increased efficiency inthe use o f public property, and better quality o f public services. However, it is difficult to estimate most o f these benefits in quantitative terms and it will take time for most o f these benefits to materialize. Hence, they have not beentaken into account for the purposes o f this analysis. 2.27 The earlier attempts to estimate fiscal implications o f the on-going civil service reform inRussiahave beenlimitedand were focused onparticular components ofthe overall reform costs. In2003-04, the Higher School o f Economics publishedsome estimates related to fiscal case, should it be decidedto adjust the pay scale for civil servants, it clearly has to be adjusted at least for the category(ii),but also probably for the category (iii). 27Some analysis of possible military reform options is presentedinthe IET (2002) report. 34 effects o f .targeted pay increase for federal headquarter-based civil servants, as well as investment costs associated with the introduction of new technologies in the federal civil service. The later were further refined in the framework o f drafting a Federal Program Administrative Reform (2005-1 0). The Presidential Program Reforming the Civil Service (2003-05) also presented some estimates for investment costs related to drafting new civil service legislation and developing new methodologies and instructions needed to carry out both pay and HR management reforms. However, there have been neither attempts to estimate the full fiscal costs o f all the above components, nor any assessments of a systemic fiscal effect o f the on-going reform on the entire costs of financing the civilian public sector. D. DATA SOURCES AND KEYASSUMPTIONS 2.28 Datu Sources. The data used for calculating the fiscal implications o f civil service reform includes October 2002 public sector statistics publishedby Rosstat in 2002; data on pay and employment incore government administration for 2001 and 2002 (Rosstat 2002 and 2003); results o f pay and public officials surveys conducted in 2002-2003 and sponsored by DFID (HSE, 2003, forthcoming); international statistics on the education and health sectors (including WHO and Bank data); international pay and employment database maintained by the Bank'(Wor1d Bank database); as well as data obtained from other sources (a complete list o f sources i s presentedat the end of the Chapter). 2.29 Because o f the lack of up-to-date comprehensive pay and employment data inRussia, the following assumptions were appliedto calculations ofthe pay andemployment baseline: > For Core Government Administration 0 Because o f the lack o f data on employment breakdown by groups/categories for 2002, we assumed that the 2001 employment structure remained unchanged and applied the 2001shares to the 2002 employment levels. Because of the lack o f data on compensation by groups/civil service categories for 2002, we applied the actual average compensation increases in relevant branches and levels o f the government to all 2001 categories/groups proportionally assuming that the compression ratio remained unchanged. > For Civilian Public Sector Employment 0 We assumed that the compensation level as o f October 2002 i s equal to the average compensation level in2002. 0 Rosstat (2002) data contained average compensation levels with breakdown per four major civilian public sector branches (see Figure A2.1 in Annex 2.1) and per level of government (Le,, federal, regional and municipal). These compensation levels were applied to all sub-sectors within each branch. 0 It was assumed that organizations classified as being federal property are financed from the federal budget, while the entities in regional and municipal property are financed from sub-federal budgets. 0 Rosstat (2002) data cover all organizations in the civilian public sector that are partly or fully financed from the budget. For the purposes o f this Chapter it i s assumed that the reported wage bill and social charges o f civilian public sector employees were fully financed from the respective budgets. 35 2.30 Key Macroeconomic Assumptions. Macroeconomic assumptions for our simulations were developed jointly with the experts from the Institute o f Economy in Transition. The basic principles o f the macroeconomic framework used in our work for costing out various structural reforms are presented in Annex 3.1, In sum, we took the Government's baseline macroeconomic projections for the period 2004-2006 and used them as a basis to build a set o f four macroeconomic scenarios, each of which reflects a specific combination o f two primary determinants o f Russia's future macroeconomic performance-average oil price and expected speed of structural reforms. Our initial plan was to use the four macroeconomic scenarios as a basis on which we would design and elaborate further more detailed sub- scenarios o f specific structuralreforms2*. 2.31 However, in the case o f reforms in public administration we came to the conclusion that it would be appropriate to modify our original plan and follow a simplified set of macroeconomic assumptions. The reason for such simplification derived from our early observation that the simulation results (presented in GDP p.p. 29) are not sensitive to GDP growth rate. The simulations are also not sensitive to inflation, since the model assumes full annual indexation o f wages in both core government employment and the civilian public sector. Therefore, we decided that it would be appropriate to develop cost estimates within the single macroeconomic scenario. 2.32 What significant for the outcomes of our costing exercise i s the differences ingrowth rates, primarily the difference between the expected growth rate in private sector wages and GDP growth rates. The higher future wage growth inthe private sector, the more expensive i s the government's task of closing the existing pay gap. Overall, we concluded that the key macroeconomic parameter in the model i s the ratio between real wage growth rate and real GDP growth rate. Currently, the share of payroll (including hidden wages) in the Russian GDP i s quite low, but in the medium term we expect that in response to structural reforms it will be growing quite rapidly to approach the levels prevailing in other economies in transition. This would occur owing to increased competition, reduced share of the informal sector, and therefore improved wage/profit ratio in the economy. However, the costs of the proposedreform incivil service depend fundamentally upon the speed o f this future catch up. 2.33 Based on the above considerations, we decided to base our simulations on the single macroeconomic scenario, but to consider three sub-scenarios that have different growth rates for the payroll share inGDP (Table 2.5). From our original four macroeconomic scenarios we selected the most optimistic one as our single macroeconomic framework. This scenario i s described as a scenario with the advanced reforms andhighoil prices inAnnex 3.1. 28 This initial plan with multiple macroeconomic scenarios was fully implemented, however, for analysis of fiscal implications for two other structuralreforms- inthe pensionsystem and inhousinghtilities. The results for these two reformsare reflectedin Chapters3 and 4 of this paper. 29 Costs reflectedinabsolute values, e.g., inconstant 2002 rubles,wouldbe different. 36 Table 2.5: Key MacroeconomicAssumptions (in constant 2002 rubles) 2003 2004 2005 2006 2007 2008 2009 2010 GDP growth, % 7.3 5.5 5.0 4.5 4.0 3.5 3.0 2.5 GDP, trln. RUR 10.0 10.5 11.1 11.6 12.0 12.4 12.8 13.1 GDP per capita, `000 RUR 69.9 74.1 78.2 82.0 85.7 89.1 92.2 94.9 Inflation, annual, % 10.0 10.0 10.0 10.0 10.0 10.0 10.0 Realinvestment growth, % 9 7.5 6.5 5.5 4.5 4 4 Real householdincomesgrowth, % 7.5 7 6.5 6 5.5 5 4.5 Population (million people) 142.8 142.1 141.5 140.9 140.3 139.7 139.0 138.4 Scenario 1(real wage growth rate exceeds GDP growth rateby 2 p.p.) RealWage Growth, % 8.1 7.5 7.0 6.5 6.0 5.5 5.0 4.5 Scenario 2 (real wage growth rate exceeds GDP growth rateby 1p.p.) RealWage Growth,% 8.1 6.5 6.0 5.5 5.O 4.5 4.0 3.5 Scenario 3 (real wage growth rate is equalto GDP growth rate) RealWage Growth,% 8.1 5.5 5.0 4.5 4.0 3.5 3.0 2.5 Source: Rosstat, MoEDT, and staff calculations. 2.34 We selected the most optimistic scenario in order to highlightboth the costs and risks of implementingthe proposed reform package. At the same time it i s quite clear that in the scenarios with low oil prices and respectively with much higher fiscal pressures, the same reform task would become much more challenging politically. Inthe case o f a tight budget it would be much more difficult for the government to justify the need for a radical pay increase in public administration and convince society that a considerable budget restructuring that favors govemment employees i s indeed urgently needed. In the discussion below we do not consider these additional and non-quantifiable policy risks could be derived from the overall deterioration in the macroeconomic situation, and which most likely would initiate additional pressures towards slower rates of reforms. 2.35 The selected macroeconomic framework is based on the assumption of an average growth rate o f 5 percent for 2004-0630 and 4 percent for 2004-2010. It i s worth noting that these growth rates are lower than those assumed inthe corresponding government projections. Overall, we expect that growth rates will decline after 2006. Better utilization o f existing reserves inthe economy, which was a critical growth factor in 1999-2003, cannot support the future growth inthe same way as before because the reserves are to a large extent exhausted. At the same time, we admit that this may be considered as a "conservatively-optimistic" scenario. 2.36 Private Sector Benchmarks for Net Cash Compensation Increase. The primary objective of the pay reform i s to narrow the existing compensation gap between the core government administration and the private sector. However, so far the overall scope of the civil service pay reform remains undecided. Although some proposals on increasing the compensation only to a limitednumber o f senior civil servants (or those holding "strategically 30 The current consensus estimate for Russia's growth in 2004-05 is rather favorable and is close to about 5.2 percenta year. 37 important positions") have been formulated (HSE, 2003 and 2004) and have recently been implemented through a Presidential Decree, drastically increasing the cash compensation for some HQ-based federal civil servants3', the sustainability o f such an approach remains unclear and no further policy decisions related to deconcentrated units of federal bodies, as well as to core government administration in general, let alone the civilian public sector, have been taken so far. 2.37 For the purpose o f this Chapter, the data from a comparative pay and benefits survey of the public and private sector (HSE, 2003) were utilized as a basis for the estimates for HQ- based federal authorities employees, and then was adjusted to account for the regional wage differential, which was estimated based on the actual 2001 information about the pay differences between Moscow and other regions (Denisova, 2001). The specific figures applied are presentedinTable 2.6. 2.38 These available data have important shortcomings related to a limited sample of the participating Russian private sector companies in Moscow. Also, while the employees in the public sector especially in the top grades have access to a considerable amount of in-kind benefits, the available information on such benefits i s still limited and could be not statistically representative. Also, the survey did not allow for measuring the existing gap in remuneration for the top employment categories (Categories A and B) o f the civil service. For the purposes o f our analysis, it was assumedthat the pay o f Category B employees should be at least 20 percent higher than that for Category C-top, and for Category A-20 percent higher than that for Category B, while estimated non-cash benefits are assumed to be the same for all top civil servants (Category C-top and Categories A and B) and their private sector comparators. 2.39 To estimate the increase in net compensation of the core government administration employees that do not have the status o f civil/municipal servants (ie., core government administration employees that are regulatedby the Labor Code), an assumption was used that the compensation increase for these employees is equal to a simple average increase for Category C Senior and Junior positions in the appropriate brancwlevel o f the government administration. This assumption was usedbecause most o f these employees are technical and support staff. 2.40 The survey showed, inter alia, significant variance in both the pay and total compensation gap, especially for the top grades. Hence, to estimate the existing pay gap correctly, some in-kind benefits provided to civil servants are assumed significant and proposed to be retained (medical benefits, car, and driver), while others (sanatoria, apartments, dachas) are proposed to be discontinued (but their value i s to be reflected indirectly through a proposed pay increase). Due to a limited number o f recipients of the latter type of benefits, the direct fiscal savings from their discontinuation are negligible- estimated savings for the HQ-based federal executive service in2002 accounted for about 550 million rubles or about 0.01 percent o f 2002 GDP. However, incase the government chooses to commercialize all related infrastructure, potential one-off revenues from disposal o f these assets at market prices would be much more significant. It should also be noted that the available data on benefits cover only federal H Q executive civil service, hence, the adoption 3 1Presidential Decree No. 5 19 On Improving the Compensationto Public Servants Holding Specific Positions of the Russian Federation and Specific Positions in the Federal Civil Service of the Russian Federation dated April 10, 2004. 38 o f a similar policy for federal deconcentrated and subnational core government administration employees would probably yield additional fiscal savings. 2.41 Our results suggest that in 2002 the public-private gap in cash compensation varied between 2.4 and 8.6 times (Column 1, Table 2.6). However, the estimated gap intotal (gross) compensation (Column 2) while still significant i s much smaller - between 1.7 and 4.3 times. This is because of considerable amount of in-kind benefits i s available to the top public servants, which in these employment groups exceeds their cash compensation by 3-5 times. The proposed reform strategy suggests elimination o f most of these benefits, which would require a major increase in their cash salaries to compensate for lost benefits, make their overall compensation competitive and facilitate the establishment o f the environment supportive of the government anti-corruption efforts. This increase would be partly accommodated by decreasedfinancing of the in-kindbenefits for civil servants. 2.42 Columns 3 and 4 in Table 2.6 present our estimates for the public-private net to be retained. The net compensation gap i s a critical parameter in our model - it defines the compensation gap, which i s the gap incurrent cash compensation adjusted for in-kindbenefits benchmark for necessary pay adjustment in the public sector to make salaries there fully comparable with the pay inthe private sector. At the moment, the net compensation gap varies between 2.3 and 7.8 times for HQ-based staff and 1.2-3.9 times for federal deconcentrated and subnational service. However, based on the data available, we cannot estimate variances inthe compensationgap amongindividual positions and specialties. Table 2.6: EstimatedPublic-Private Pay and CompensationGaps, 2002 Measured as a Ratio of Pay (Compensation) inthe PrivateSectorto Pay (Compensation)in Federaland SubnationalCivil Service FederalCivil Service SubnationalCivil Service Public-Private Public-Private Category/ Public-Private Gross Net Public-Private Group Pay Gap, Compensation Compensation Net CompensationGap, time& Gap, timesb/ Gap, timesg times 1 2 3 4 Category A 8.6 2.2 7.8 3.9 Category B 7.1 1.9 6.3 3.2 Category C TOP 8.5 1.7 7.8 3.9 Chief 5.8 4.2 5.7 2.8 Lead 5.2 4.3 5.1 2.5 Senior 3.7 3.4 3.6 1.8 Junior 2.4 2.2 2.3 1.2 Notes: The estimated gaps are not adjusted for 30 percent compensation increase in core government -administrationestimated gap incash compensation only. and civilian public sector implementedin October 2003. a/reflects the b' reflects estimated differences in cash compensation combined with value o f all in-kindbenefits currently received. /reflects estimated differences in cash compensation combined with value o f in-kindbenefits to be retained. Source: HSE (2003) and staff calculations. 2.43 Based on the understanding that some o f the in-kindbenefits will be discontinued, for the purpose o f this Chapter it i s assumed that the total amount o f expenditures on in-kind 39 benefits in core government administration would not exceed the current level. Hence, the expenditures associated with in-kind benefits are not included in net increases o f fiscal expenditures on core government administration. 2.44 Given the large current compensation gap with the private sector it i s highly unlikely that in the medium-term Russia could close most o f it and reach the level o f OECD countries where the common remuneration gap with the private sector i s on average o f about 10 percent32,while it varies for particular positions/prQfessions. Thus, even in the medium-term, the proposed scenarios aim only at a partial closure o f the gap. In our simulations below even the "radical" scenario o f the pay reform assumes a relatively highresidual compensation gap o f 50 percent. 2.45 The results o f the survey o f the public servants' pay expectations (see Table 2.7) undertaken in2002 (INDEM) also seem to confirm the feasibility o f a relatively highresidual compensation gap for federal civil service. The comparison o f pay expectations and the existing compensation gap shows that the surveyed officials either underestimate their existing pay differences with the private sector or they are prepared to accept lower pay levels (at least at Department/Division Chief levels). While the estimated pay gap accounts for a 5.0-7.5 increase, public officials would consider a 2.54.0 pay increase to be adequate. This difference between the actual pay gap and pay expectations could be explained by a number o f factors, including: (i) insufficient awareness o f public servants o f the comparative pay levels inthe private sector; (ii) the extent to which the public servants surveyed have access to and value various non-cash benefits provided; and (iii) higher job security and lower intensity associated with a public sector employment as compared to the private sector positions. The results o f the public officials surveys may also reveal the fact that some o f the civil servants in Russia may not perceive their official pay as the only (and the primary) source o f income. This argument i s confirmed by the recent research on ownership structure in Russia. In fact, there i s little, if any, separation between the authorities and business community in the country33. Table 2.7: Civil Servants'Perceptions:ExpectedPay and CompensationLevels for Benchmark Positionsin the FederalCivil Service (based on Public Officials Survey) ExpectedPay Pay Increase ExpectedPay (in (with all in- (with case all in-kind Pay Increase Benchmark Category kindbenefits retained in- benefitsare (without in- Position C retained), kind discontinued), kindbenefits), Group US% benefits), US$ times times Deputy Minister top 2,182 4.0 3,733 6.9 DepartmentHead top 864 1.9 1,223 2.6 Division Head lead 672 2.2 887 2.9 Chief Specialist senior 448 2.6 634 3.7, Sources: INDEM (2002), HSE (2003), and staffcalculations. 32Schiavo-Campoet al. (1997), p.10. 33 For more information on this issue, see the recent Russia Country Economic Memorandum (World Bank, 2004). 40 2.46 The analysis of both Tables 2.6 and 2.7 points out that both the existing net public- private compensation gap (Table 2.6) and the pay increases, which would meet the current expectations of the civil servants (Table 2.7), vary greatly between the different grades and positions and tend to be lower for more junior servants and higher for more senior public officials. This suggests that the compression ratio in the core government administration i s significantly lower than the one in the private sector. Indeed, in 2002 the compression ratio for federal executive HQ-based civil service accounted for only 2.5. A compression ratio for federal executive civil service was slightly higher (3.7), but still quite low by international standards. Internationally, the compression ratio varies widely from highs of 30:1 to lows of 2:l with the mode o f around 6 or 7 to What follows from this analysis i s that the decompression o f the existing pay scale in Russia by at least 2.0-2.5 times would be quite desirable inthe mediumterm. 2.47 Low compression ratios in Russia suggest that the pay increases in core government administration should be focused more on the managerial (higher level staff) than on junior staff. It should be noted, however, that the proposed reform strategy to close the same percentage of the existing pay gap for all employee categories would automatically generate considerable pay decompression (because the net compensation gap i s so much higher today for the higher grades). 2.48 At the same time, it is worth notingthat selective increases that do not cover all of the core government administration employees, at least to some extent, may create additional distortions in the staff grading process. Recent public officials surveys pointed out that there are quite a number o f federal and subnational civil servants that though formally employed at senior positions, in fact perform the responsibilities of junior staff (see HSE, 2003, for more information). To improve the incentives for proper grading in the core government administration and sustain the employment structure under the conditions o f significant pay adjustments for higher level staff, a "minimum wage increase for all grades rule should be " made a part of any pay adjustment strategy. In our simulations such minimum adjustment amounts to a 20-30 percent increase in real wage relative to its 2003 level depending on the reform scenario. 2.49 Substantial cash compensation increases in core government administration, if not expanded to other categories o f general government employment, would significantly affect the balance between remuneration of the core government administration and the rest of the public sector. It is assumed that this imbalance would be politically unsustainable and the Government would ultimately have to adjust pay in the entire public sector regardless of the fact whether there i s or not at the moment a legally explicit link between compensation of these two categories. For the purpose o f this Chapter, an average cash compensation increase in core govemment administration is used as a trigger for estimating a corresponding pay adjustment in the civilian government employment. Given the current number o f civilian employees, it is unlikely that it would be possible to sustain the current cash compensation ratio between the core government administration and the civilian public sector. It i s also expectedthat the ratio ofnon-budgetary sources o f civilianpublic sector employees' wage bill would be increasing owing to the growing share o f commercial services provided by the entities in this sector for additional pay. Hence, in our scenarios we assumed that the pay 34Schiavo-Campoet a]. (1997), p. 44. 41 adjustment in civilian public service would be equal to 70 percent of the average increase in cash compensationinthe core government administration. 2.50 Payroll Taxes. All the estimates of cash compensation include payroll taxes- personal income tax (PIT) and Unified Social Tax (UST)-Le., we estimated gross compensation aggregates. The existing regressive scale o f UST has been explicitly taken into account with tax brackets adjusted for real wages growth. No possible future changes in the UST tax rates have been accounted for35.Increase in PIT collection because of higher cash compensation in core government administration and civilian public sector employment was accounted for as a fiscal gain to the sub-federal budget. 2.51 Verification. Although there i s no "best international practice" related to the relative pay levels in the public sector, a ratio of average federal government administration wage to GDP per capita could be used as a monitoring indicator to verify the viability of the analyzed reform scenarios. In OECD countries, this indicator on average i s 1.636,while in Russia in 2002 it was 1.2, which i s quite low, especially when compared to either the global average (3.0) or an average for L A C countries (2.5) that could be used as a proxy for middle-income countries 2.52 Relative Pay Adjustment: HQ-based Employees and Deconcentrated Staff. The estimated compensation gap i s higher for federal HQ-based core government administration employees than for deconcentrated federal civil servants and for subnational civil servants. This is partially a reflection o f the existing imbalance inremuneration among the federal HQ- based civil servants, civil servants working in deconcentrated units o f federal bodies, and subnational civil servants. Currently, in 60 percent o f regions, salaries of regional servants are higherthan those o f federal deconcentrated units; in45 percent o f regions, salaries o f regional servants are higher than the salaries of HQ-based federal civil servants (see HSE, 2004, for more information). Hence, the ratio of the compensation between these categories (current1 favoring subnational employees) should be reversed to reflect relative labor market prices3r. T h i s policy may also be considered as part o f the proposed decompression agenda, since it suggests that as a result o f the reform the compression ratio in the federal executive branch (including deconcentrated units) would become significantly higher. 35The on-goingdebates on tax reform in Russia also includesome proposals on decreasingthe rate for Unified Social Tax up to 10 p.p. but no final decision had been taken on this issue by the time simulations were performed. 36The relativelylow levels of this ratio inthe OECD countries are due to smaller skill differentialsbetweenthe central governmentemployees and the rest of working population.Hence, for Russiaa higher ratio than the one for OECD (probably the one approachingthe level of middle-incomeLAC countries) could be expected. Cross country comparisonsshould, however, be usedwith caution since e.g. in many countriesparamilitarypersonnel is considered as a part of the central government, while this report separates police service from the core government administration. 37 Source: Schiavo-Campo et al. (1997). An International Statistical Survey of Government Employment and Wages, p. 7; staff calculations for Russia(based on the Rosstat data). 38This assumption suggests a completely different approach to compensation for the deconcentratedfederal executive bodiesthat in effect creates a significantpay gap between HQ-basedstaff and those employed in the territorial branches of the same government entity. Implementation of this approach would call for (i) continuous monitoring of regional labor market prices throughout the country, and (ii)imposing strict limitations on employment growthinthe Headquarters of federal executivebodies. 42 2.53 It is worth noting that the future nominal levels of cash compensation inthe civilian public sector should vary from region to region, depending on the local price level as well as on some other factors, while our scenarios suggest only the national average estimates for a potential cash compensation for different employment categories. 2.54 Attrition Scenarios entailing implementationof the administrative reform component assume some reduction in both core government administration employment, as well as in employment inthe civilian public sector. 2.55 As far as the core government administration employment is concerned, for the purpose o f this Chapter it i s assumed that attrition would take place only in the executive branch o f the core government administration. Since the detailed structure o f management bodies within the core government administration has not been determinedyet, it i s especially difficult to make projections for a possible attrition rate. Moreover, the Wagner Law claiming that public administration employment tends to grow with economic growth also may play a role in determining future trends in staffing levels in the core government administration in Russia. The attrition rates proposed in the Chapter are more conservative than the latest announcement made by the top government officials on the expected 30 percent cuts in the number o f civil servants because it i s not clear whether these expected reductions refer to HQ- based federal executive civil servants or to the whole federal executive civil service. Moreover, giventhe low current staffing levels inthe HQpart o f the executive service, we are concerned that a 30 percent cut for this component o f the public administration may be excessive, Our analysis of the dynamics o f core government administration employment in 1994-2002 (see Section A) also suggests that subnational executive civil service and local self-government may accommodate higher attrition than federal executive authorities. These attrition rates are used here only as preliminary benchmarks for the analysis. More accurate forecasting o f future attrition rates would require much more detailed research o f the current employment rates, costing out the separate executive functions and analyzing broad statistical data onregional variations inactual employment. 2.56 The specific attrition rates that have been used for simulations are presented inAnnex 2.5 and briefly summarized below. Employment adjustments for the education sector were calculated based on the difference between the studentdteacher ratio inRussia and inmiddle- income OECD countries. The target level was either estimated as an average between middle- income and OECD countries levels or as the one approaching the OECD countries level, depending on the current indicators for Russia and the feasibility o f the quick changes inthe staffing levels. Depending on the sub-sector, possible attrition rates in the education sector were estimated in the range o f 10.5-26.9 percent o f the 2001 employment levels (see Table 2.8). For later years, demographic projections on the number o f students were also taken into account. Table 2.8: Attrition Rates Expectedin the EducationSector, 2001 Sub-Sector" Current ratio of Benchmarkratio of Expected Attrition studentdteacher studentdteachers Rate (YO) HigherEducation 11.4 15.6 26.9 General Education 11.5 15 23.3 Vocational Education 10.2 11.4 10.5 * For other sub-sectors ofthe education sector a general 10percentreductionrate usedfor all "other civilian public sector employment" was applied. Source: Poletaev et al., (2003), and staffcalculations. 43 2.57 Futureemployment adjustments for the health sectorwere estimated basedonthe ratio o f doctors and paramedical staff per 100,000 of population inRussia as compared to the same indicator in OECD countries. Given considerable cross-country variance o f this indicator as well as the comparatively highratio inRussia, it i s assumed that inthe medium to long term the number of health sector employees can be reducedby 20 percent. 2.58 There is little if any information on a potential for staff reduction in the rest of the civilian public service. For the purposes of this Chapter it was assumed that this reduction would account for 10percent. 2.59 At the same time, for the scenarios that do not assume any administrative reform actions, it was assumed that the actual employment incore government administrationmay in fact grow, given that in2001 about 5.6 percent of positions inthe government administration were vacant and some of these are likely to be filled in after the salary increase. In2001 the difference between the share o f vacant positions in subnational service (which i s considered relatively more attractive given local market opportunities) and federal civil service accounts for about 2 percentage points. At the same time, in the process of reforms the requirement of public servants' qualifications will increase significantly, and as a result some of the positions may remain vacant. Hence, for the purpose o f simulations, it was assumed that the salary increases would lead to an increase in employment of 2 percent, if not accompanied by a separate effort to cut staffing. The same assumption was applied to the civilian public sector employment. 2.60 Expenditures Related to Staff Attrition. For scenarios that assume employment cuts, additional expenditures related to one-time separation costs in core government administration have been calculated based on the current legislation (i-e., Labor Code, legislation on civil service). These costs were estimated under the assumption that the redundant civil and municipal servants are paid an equivalent of their four-month pay, while the other civilian sector employees are paid an equivalent of their two-month pay. 2.61 Verification.Even a superficial cross-country comparison shows significant variation of the employment numbers in different layers of the core government administration and civilian public sector at large. Although there are significant methodological limitations to the actual application o f international comparative data, it was helpful to use cross-country comparisons of various employment ratios, such as shares o f core government administration employment, civilian public sector employment, and employment in health and education sectors inthe total population, as monitoringindicators for reformprogress. 2.62 Non-Wage Expenditureson Core Government Administration. Modemization of public service and implementation o f the full-scale HR management reform would call for increased financing o f non-wage expenditures on core government administration, especially those on equipment, communication and other services, and training. The UKdata for running costs (used as a benchmark) shows that the average ratio o f non-wage expenditures accounts for 33-40 percent o f total expenditures on government administration, while inRussia in2001 this ratio (for federal budget only) accounted for about 38 percent (36 percent in2003). This shows that the current share o f non-wage expenditures in Russia appears to be in line with international practice. It i s anticipated that a full-scale implementation o f the comprehensive civil service reform package would sustain this ratio at the level o f 25-36 percent (depending on the reform scenario). In other words, we assume that the share o f non-wage costs in total budget expenditures on core govemment administration would be either sustained or would somewhat decline primarily owing to an opportunity for considerable savings associated with 44 more efficient use of existing office space (and associated decline in maintenance and utility costs). Inaddition, the possibility o f reallocating some of the core government administration staff and subsequent renting of government real estate could bring about additional fiscal gains. 2.63 It should be noted that non-wage expenditures in civilian public sector employment are highly sector-specific. Identification of appropriate benchmarks would require a separate research o f the current status of civilian public sector financing and the need for further development. This research goes beyond the scope of this analysis although we acknowledge the importance of non-wage expenditures inthe civilian public sector as a parameter of civil service reform inRussia. E. CIVILSERVICEREFORMSCENARIOS 2.64 Fiscal implications o f the civil service reform depend upon a large number of factors. For the purpose of this Chapter, the following parameters were used to identify the possible reform scenarios and estimate their fiscal implications: k Macroeconomicscenarios (three sets ofindicatorsbasedonthe change inthe share o f real wages inGDP); k Extent of pay reform implementation, measured as the amount of the residual public-private compensationgap: 0 Radical pay reform (by the end of the reform period, Le., by 2010, the residual public-private compensation gap would not exceed 50 percent, Le., average compensation in core government administration would amount to two thirds (66 percent) of the average compensation in the private sector. Minimum pay increase incore government administration i s 30 percent. 0 Moderate pay reform (by the end of the reform, the residual public-private compensation gap would not exceed 100 percent, Le., average compensation in core government administration would amount to half (50 percent) of the average compensation in the private sector). Minimum pay increase in core government administrationi s 20 percent. k Payreformpace. Three variants were usedfor simulation:high,medium, andlow. The differences in pace across pay reform scenarios are presented in Table 2.9 below. For consistency, it was assumed that, the pace of administrative reform (measured as the share o f number o f staff cuts in the total attrition planned within the specific scenario) i s identical to the pace o f pay reform (measured as the share o f the public-private compensation gap closed since the beginning o f the reform in the total gap to be closedwithin the specific scenario). Table 2.9: Assumption on Pay ReformPace, Measured as a Share of the OverallPlanned Gap Covered, YO 2004 2005 2006 2007 2008 2009 2010 . Low-pacedreforms(1) 15 30 45 60 75 90 100 Medium-pacedreforms(2) 20 40 60 80 100 100 100 High-pacedreforms (3) 35 70 100 100 100 100 100 45 > Administrativereform implementation and the extent of results inattrition as keys o f fiscally measurable impact. Specific attrition rates for core government administration are presented in Annex 2.5. Given the political sensitivity o f the issue, as well as the comparatively low share o f core government administration in the total population o f the country, the following scenarios have beenproposed: 0 N o administrativereform, which would lead to increased employment incore government administration and the civilian public sector at large. This scenario assumes significant expenditure growth without any fiscal gains from staff attrition. As a result, it is likely to lead to fiscal pressures, which would prevent any substantial progress in implementation o f both HR management and public service modernization reform components. Thus, the share o f non-wage expenditures in total expenditures on core government administration i s assumed to be at the lower end (25 percent) o f a broader interval. 0 Fair administrative reform with moderate achievements in both HR management and public service modernization components. This would provide for cuts in staffing by 7-10 percent in federal authorities and 17-20 percent in regionaladministrations and local self-government and bring the share of non- wage expenditures in total expenditures on core government administration to about 32 percent. 0 Significant administrative reform with significant progress in HR management and public service modernization. This would provide for cuts in staffing by 15-20 percent in federal authorities and 25-30 percent at the regional and municipal levels and sustain the share o f non-wage expenditures in total expenditures on core government administrationat the level of 36 percent. 2.65 Possible combinations of scenarios with differentpay and administrative refom scope are summarized inTable 2.10. 2.66 Based on the factors listed above, we developed a matrix containing 54 potential scenarios (all possible combinations of selected variables). After a brief analysis of the matrix, the scenarios that have inherent inconsistencies and are unlikely to be implemented (Le., radical pay reforms implemented at a low pace, moderate pay reforms implemented at high speed, slow-paced reforms combined with significant administrative changes) have been removed from the list. The matrix with the description o f 36 remaining scenarios that were selected for actual costing-out and analysis i s presented inTable 2.11. The basic results (cost estimates) o f the simulations for these 36 scenarios are summarized inTable 2.12. 46 Table 2.10: Possible Scope of Pay and Administrative Reforms No AdministrativeReform Fair Administrative Reform SignificantAdministrative + Reform Employmentincore + Employment inexecutive + Employmentin executive governmentadministration branchof core government branch of core government and civilianpublic sector is administrationandcivilian administrationand civilian increasedby 2 percent public sector is cut by 7-20 public sector is cut by 15-35 + Share of non-wage percent percent expenditures intotal + Attrition ratesfor civilian + Attritionratesfor civilian expenditures on core public sector are appliedin public sector are appliedin govemment administrationis accordancewith sectoral accordancewith sectoral 25 percent assumptions assumptions + Residualpublic-private + Share of non-wage + Share of non-wage compensationgap is 50 expendituresintotal expenditures intotal percent expenditureson core expenditures on core + Minimumpay increaseis 30% government administration government administrationis is 32 percent 36 percent Note: this group of scenarios was found unrealistic and was not + Residualpublic-private + Residualpublic-private included in simulations compensation gap is 50 compensationgap is 50% percent + + Minimumpayincreaseis 30 Minimumpay increaseis 30 percent percent + Employmentincore + Employmentinexecutive + Employmentinexecutive government administration branchof core government branch of core govemment andcivilianpublic sector is administration and civi1ian administrationand civilian increasedby2 percent public sector is cut by 7-20 public sector is cut by 15-35 + Share of non-wage percent; percent; expendituresintotal + Attritionratesfor civilian + Attritionrates for civilian expenditureson core public sector are appliedin public sector are appliedin govemment administrationis accordance with sectoral accordancewith sectoral 25 percent assumptions assumptions + Residualpublic-private + Share of non-wage + Share of non-wage compensation gap is 100 expendituresintotal expenditures intotal percent expenditures on core expenditures on core + Minimumpay increaseis 20 governmentadministration government administrationis percent i s 32 percent 36 percent + Residualpublic-private ,+ Residualpublic-private compensationgap is 100 compensation gap is 100 percent percent + Minimumpayincreaseis 20 + Minimumpayincrease is 20 percent percent 47 No. Scenario Macro (shareof Pay Reform Pay Reform AdminReform Code wages in GDP) Extent Pace 48 8 L CI 0 m I: 50 2.67 The mainfindings from the base set of simulations couldbe summarized as follows: 0 Potential fiscal costs of civil service reform are quite sensitive to relative growth rate of private sector wages. A decrease in real wages growth by 1 p.p. leads to about 0.5-0.7 p.p.o f GDP inannual budget saving. 0 The radical pay reformaimed at a 50 percent residualpublic-private compensation gap i s on average about 1.5 times as expensive as the moderate pay reform that allows for 100 percent public-private compensation gap at the end o f the reform process. In addition, the more ambitious the pay reform, the more volatile are its fiscal costs, Le., they are more sensitive to the changes inother factors. 0 Sub-federal budgets would have to carry a much larger share o f the overall fiscal burden, associated with the reforms. About two thirds of the total incremental costs would become responsibility o f subnational authorities. 0 The scenarios with bothradical and moderate pay adjustments result insignificant pay decompression in HQ-based civil service, from the current ratio between average wages inTop and Junior Groups of 2.5 to 5.2-8.3 in 2006 and to 6.8-8.3 in2010. 0 The sensitivity of fiscal costs to the pace o f pay reform implementation is relatively low. 0 Both radical and moderate pay reforms would not be affordable for the budget system ifthey are not complementedby administrative adjustments. F. OUTCOMESOFCIVILSERVICEREFORMINLIGHTOFCROSS-COUNTRYCOMPARISONS 2.68 Inthis section we intendto validate various reform options using several monitoring indicators, which allow placing some expected reform outcomes in a broader comparative framework. 2.69 Section D o f this Chapter suggested several indicators based on international public sector statistics that may be used as tentative guidance in analyzing and evaluating the results o f our estimates. One o f these indicators was a ratio of average federal government administration wage to GDPper capita. The results of simulations suggest that this indicator falls within the interval between 1.8 (moderate pay reform in case of unchanged share o f real wages in GDP-scenarios 29-36) and 2.4 (radical pay reform in case the real wages growth rate i s 2 p.p. higher than the GDP growth rate-scenarios 1-4). These are more or less inline with the indicators for LAC (2.5), lower than the world average (3.0), and higher than the average ratio for OECD (1.6), and ECA (1.3) (see table in Annex 2.4). The results suggest significant increase inthis indicator as comparedto the baseline of 2002 (1.2). 2.70 Another relevant indicator i s the number of core government administration employees as apercentage topopulation, which inour simulations varies from 0.8 percent to 0.9 percent, which i s quite close to the baseline value (0.87 percent of population in2002). This indicator i s in line with the numbers for civilian employment in central and subnational government in Poland (0.7 percent), Bulgaria (0.8 percent), Ukraine (1.1 percent), and India (0.9 percent), but it is significantly lower than inMexico (1.4 percent) and Brazil (1.6 percent), let alone the high-income OECD countries where government employment varies between 2.9 percent of population in Australia to 6.9 percent in the United States. The only country in our sample where the ratio i s significantly lower than inRussia i s Kazakhstan(0.5 percent). 50 2.71 A similar ratio for the number of health and education employees suggests that by the end of the proposedreforms the number of these employees may vary between 6.0 percent o f population (in case no administrative reform i s implemented) or 4.2 percent (in case the administrative reform takes place), while the 2002 baseline i s 5.6 percent. The cross-country data vary greatly, but it i s clear that implementation o f administrative reform will bringRussia closer to OECD averages (3.4 percent) with the closest comparators in transition economies group being Hungary (4.5 percent), Kazakhstan (5.1 percent), and Ukraine (3.9 percent). By international standards, employment in health and education sectors will still remain quite high,but, onthe other hand, the specifics of Russia &e,, vast territory most of which has low density o f population) mayjustify a higheremployment inpublic services. Indeed, inCanada the indicator appears to be 5.0percent, inAustralia andthe US-3.8 percent. 2.72 Generally, the brief comparison of the simulation results with international statistics suggests that the reform scenarios implying administrative reform would bring about considerable general improvements to the structural characteristics of both Russian civilian public sector and core government administration employment as compared with the current status. G. SUMMARY OFSIMULATIONS: ANALYSISOFSELECTEDREFORMSCENARIOS 2.73 The analysis presented inSectionFandAnnex 2.6 allows narrowing down the number of reform scenarios that could be potentially implemented within the middle to long-term time frame. Inthis Section we would: first, discuss these potential policy constraints, and second, concentrate on the analysis o f the relatively viable reform options. 2.74 Our simulations show that it is unlikely that by 2010 the budget would be able to afford a radical decrease in the public-private compensation gap for the entire core government administration employment and civilian public sector. In case when the residual public-private gap amounts to 50 percent, the additional budget expenditures caused by the reform would account for about 2.1 to 3.3 pop.o f GDP in annual extra costs, as compared to 2003 expenditure levels, depending upon a future wage growth rate in the private sector. These figures, although not very high in the international context, are still unlikely to be affordable for the Russian budget, especially for sub-federal budgets that would have to accommodate most o f the necessary increase inbudget expenditures. In fact, these estimates do not capture other important costs, such as the need for a parallel and similar inmagnitude increase in financing o f the police and armed forces, which will result in even higher fiscal pressures. Hence, in our view, the scenarios assuming radical pay adjustment for the entire civilian public sector employment (scenarios 1-4, 13-16, and 25-28) should be excluded from further consideration. 2.75 Secondly, implementation o f the administrative reform component and, more specifically, employment adjustment in the civilian public sector i s critical to the success of the broad civil service reform agenda. Pay adjustments not complemented by administrative reform efforts, even in case o f moderate pay adjustment, would result inunaffordable growth o f budget expenditures (by 2.8-4.2 p.p.GDP as compared to the 2002 baseline). This leads us to exclude scenarios with no administrative reform actions (scenarios 5, 7, 10, 17, 19, 22, 29, 31, and 34). 2.76 The results presented in Annex 2.6 also highlight a need to control for non-wage expenditure growth while implementing the administrative reform and modernizing public service in the country. Although the administrative reforms and associated attrition calls for an increase in non-wage expenditures, Russia i s unlikely to sustain the current share of non- 51 52 wage expenditures intotal core govemment administration costs (37 percent). It seems that this share would have to go down to at least 35 percent, which would represent a significant increase infinancing inreal terms. 2.77 It is notable that the conclusions made above significantly limit the range of civil service reform scenarios that appear affordable. In fact, only five groups o f reform scenarios that satisfy the constraints described above as the following: (i)moderate pay reformimplementedat a lowpacewith a fair administrative reform effort (scenarios 6, 18, and 30); (ii)moderatepayreformimplementedatamediumpacewithafairadministrative reform effort (scenarios 8,20, and 32); (iii)moderate pay reform implemented at a medium pace with a significant administrative reform effort (adjusted scenarios 9,2 1, and 33); (iv) moderate pay reform implemented at a high pace with a fair administrative reform effort (scenarios 11,23, and 35); and (v) moderate pay reform implemented at a high pace with a significant administrative reform effort (scenarios 12, 24, and 36). 2.78 In summary, a realistic reform strategy should provide for a moderate pay reform targets and be supported either by significant or fair administrative reform effort. 2.79 As shown in Table 2.12, the total fiscal costs of these reform scenarios in 2010 vary between 1.2 p.p. GDP and 2.3 p.p. GDP. Figure 2.2 illustrates the range o f fiscal implications simulated for the selected groups of scenarios. To simplify further analysis and because the variation in pace o f pay reform does not have a major impact on fiscal costs o f reforms (as discussed in Annex 2.6), hereinafter we concentrate on analyzing the two groups of above mentioned scenarios, namely, (ii) and (iii) represent a median of the reform options that that we consider practical. To reflect the earlier finding with respect to a need for limiting the share of non-wage costs in the total core govemment administration costs, we undertook further adjustment of scenarios inthe group (iii) (scenarios 9, 21, and 33) to reduce a share o f non-wage expenditures to 35 percent. 52 Figure 2.2: Total Fiscal Costs of Selected Scenarios (p.p. of GDP as compared to 2003)39 2.5 Scenario8 2.o -Scenario 9 (adj) 1.5 0Scenario20 O S c e n a r i o 21 (adj) I.o -Scenario 32 (adj) 0.5 -Scenario3321(adj): Scenario trendline 0.o 2004 2005 2006 2008 2010 Source: Staffestimates. 2.80 As shown inFigure 2.2, the fiscal implications of the selected reform scenarios highly depend on the share of real wages in GDP. In fact, if the real wage growth is 1 pap.higher than GDP growth (as in the scenarios 20 and 21), the fiscal costs o f the reform implementation are conveniently stable and account for about 1.6-1.7 pap.o f GDP in additional spending as compared to 2003. This stability i s partly related to the substantial increase in cash compensation implemented in October 2003 for the entire civilian public sector4'. The fiscal implications o f this policy measure have already been incorporated into the 2004 budget, and hence in fact a significant portion (more than two thirds) o f the overall reform-related additional costs has already been accommodated by the government. Overall, the analysis suggests that the reform strategy that provides for a moderate pay increase in core government administration combined with signi9cant attrition (though differentiated by employment groups) and considerable increase in financing of non-wage costs could be implemented within (or almost within) the currently established fiscal constraints, i.e. relatively little of additional financing may be needed compared to 2004 budget. However, some redistribution of funding between the various expenditure categories within the consolidated budget would be required. 2.81 Suchan outcome of low reform costs is possible, however, only ifa broad approachto civil service reform i s followed. Figure 2.3 presents a breakdown o f the total reform costs between two parts o f the public sector employment. It shows that in fact even moderate pay increase in core government administration (complemented by some attrition) would bring about an increase in budget expenditures on this part o f public employment. However, this increase may be compensated by a decline in overall costs o f financing the civilian public 39See the text of this section for the description o fthe selectedscenarios. 40It i s noteworthy that significant attrition may also cause additional expenses, since some of the benefits provided to civil servants do not cease after the termination of their employment. The situation may already change with the adoption o f the new laws On State Civilian Service and On Municipal Service. Assessing the impact of such additional costs would require substantial analysis related to age profile o f the civilimunicipal service and evaluating the opportunity costs for running parallel medical care structures, etc. Such analysis is beyondthe scope ofthis Report. 53 54 sector, which becomes possible because o f lower salary increases and higher attrition rates proposed for this sector as compared to core government administration. Figure 2.3: Fiscal Implicationsof Scenario 204':Core GovernmentAdministrationand Civilian Public Sector (p.p. of GDP as compared to 2003) 4 4 1.20 0 0 0 N 1.oo E L0 0.80 fI 0.60 0.40 u 0.00 i - 0.20 f 2004 2005 2006 2008 2010 RCore GovernmentAdministrationC&s CivilianPublicSector Costs Source: Staffestimates. 2.82 This pattern, however, changes drastically if the civil service reforms are not complemented by any administrative effort to reduce overstaffing (especially in the civilian public sector). Should moderate pay reforms be implemented without any employment reductions, the fiscal costs o f reforms would be almost twice higher and reach about 3.5 p.p. o f GDP by 2010 (in case o f medium growth o f the real wages). Figure 2.4 illustrates this drastic expenditure growth. Figure 2.4: FiscalImplicationsof Civil Service Reform: Role of AdministrativeAdjustments (pep.of GDP as comparedto 2003) ' 4.0 6 3 5 0 mderatepay -$ t%' 3 0 adjustmnt wlth 2 5 admnistrabvereform (Scenario 20) G 2 0 1 5 mderate pay adjustmnt w lthout w i 3- 1 0 admnistrabvereform 0 5 (Scenario 19) U 0 0 2004 2005 2006 2008 2010 Source: Staff estimates. 2.83 The detailed structure o f additional fiscal costs resulting from the reform implementation (for the year 2010) i s illustrated in Table 2.13, for which we selected two scenarios with a medium growth rate o f wages, namely 20 and adjusted 21. As demonstrated inTable 2.13, the broad structure o f expenditure increase under the selected scenarios is very similar. 41Scenario20: Moderatepayreformimplementedat a mediumpacewith a fair administrativereformeffort and mediumgrowthinprivatewages. 54 Table 2.13: Structure of Expenditure Increase inYear 2010 for Selected Scenarios (in p.p. of GDP) Medium-PacedModerate PayMedium-PacedModerate Pay Reform with Fair Reform with Significant AdministrativeAdjustment AdministrativeAdjustment (Scenario 20) (Scenario 21, adjusted)* A. Core Government Administration 1.13 1.11 o/w: costs to Federal Budget 0.58 0.58 Costs toSub-Federal Budgets 0.55 0.53 a. Cash Compensation 0.77 0.64 Federal Budget 0.38 0.33 Sub-Federal Budgets 0.39 0.31 b. Non-WageExpenditures 0.44 0.53 Federal Budget 0.20 0.25 Sub-Federal Budgets 0.23 0.28 c. StaffAttrition 0.00 0.00 FederalBudget 0.00 0.00 Sub-Federal Budgets 0.00 0.00 Increased Collection of PIT (sub- -0.08 -0.07 federal budgets) B. Civilian Public Sector 0.60 0.60 Employment o/w: costs to Federal Budget 0.23 0.23 Costs to Sub-Federal Budgets 0.37 0.37 a. Cash Compensation 0.63 0.63 Federal Budget 0.23 0.23 Sub-Federal Budgets 0.42 0.42 b. StaffA ttrition 0.01 0.01 FederalBudget 0.00 0.00 Sub-FederalBudgets 0.01 0.01 c. Increased Collection of PIT -0.06 -0.06 (sub-federal budgets) Total ExpenditureIncrease (A+B) 1.71 1.69 o/w: coststo Federal Budget 0.82 0.81 Costs to Sub-Federal Budgets 0.91 0.89 * Assumption related to the share of non-wage expenditures in total costs of core government administration was adjustedto 35 percentas comparedwith the original share of 36 percent. Source: Staff estimates. 2.84 As discussed above, significant portion o f these fiscal costs has already been incorporated into the consolidated budgets. Based on our simulations, fiscal implications o f civil service reform in core government administration account for almost two thirds o f the total additional costs and require increased financing o f about 1.1 p.p. o f GDP, distributed almost equally between the federal budget and sub-federal budgets. Most o f these additional costs would be required for financing the executive branch of the core government administration. 55 56 Table 2.14: Detailed Breakdownof the Increasein Fiscal Costs of Civil Service Reform in Core GovernmentAdministration as of 2010 (in constant 2002 billionrubles) Medium-PacedModerate Medium-Paced Moderate Pay Net ExpenditureIncrease Pay Reformwith Fair Reform with Significant Administrative Adjustment AdministrativeAdjustment (Scenario 20) (Scenario 21, adjusted)* A. Cash Compensation 156.59 139.06 AI. Federal Authorities 71.88 65.50 legislative authorities 3.09 3.09 executive authorities 43.64 37.27 o/w HQ-based 6.62 5.71 judiciary authorities 24.26 24.26 other authorities 0.88 0.88 A2. Regional and Municipal Authorities 84.71 73.56 legislative authorities 3.74 3.74 executive authorities 77.74 66.59 judiciary authorities 2.24 2.24 other authorities 0.99 0.99 B. Non-Wage Expenditures 57.92 70.63 B1. Federal Authorities 27.10 33.11 legislative authorities 1.37 1.64 executive authorities 15.82 18.39 judiciary authorities 9.52 12.61 other authorities 0.39 0.47 B2. Regional and Municipal Authorities 30.82 37.52 legislative authorities 1.47 1.95 executive authorities 28.09 33.89 judiciary authorities 0.87 1.16 other authorities 0.39 0.52 C. IncreasedPIT Collection (15.91) (14.17) Total Costs (A+B+C) 198.60 195.52 olw: costs to FederalBudget 98.98 98.61 costs to Sub-FederalBudgets 99.62 96.91 Total Costs (as a % of GDP) 1.13 1.11 Note: Figures may not sum up due to rounding. *Assumption related to the share of non-wage expenditures in total costs of core government administration was adjusted to 35percent as compared with the original share of 36percent. Source: Staffestimates. 2.85 Table 2.14 presents additional breakdown o f the expected expenditure increase under the scenarios 20 and 21 (adjusted) if the reforms are limited only to core government administration. It shows that most o f the expected fiscal gains from more significant administrative actions in scenario 21 are likely to be spent on additional non-wage expenditures. 2.86 The recent debates on the pay reform in Russia suggest that the pay adjustments should have a stronger focus on the middle and top management o f the civil service ("decision-makers") employed inheadquarters o f the federal executive authorities. This would allow retaining the most qualified managers within the executive bodies and would result in better government performance, including improved policy making (for more deliberations on this scenario see HSE,2004, report). 2.87 To take into account this policy option, we introduced additional modifications to the selected moderate pay reform scenarios, namely: scenario 20, assuming fair administrative changes; and scenario 21, assuming significant administrative reform. These modifications 56 decreasedthe residual public-private pay gap for HQ-based federal executive civil servants o f Categories A, B, and part of the Category C (Top and Chief Groups) from 100 percent (default value for all scenarios with moderate pay adjustment, such as e.g. scenarios 20 and 21) to 50 percent. As o f 2002, the share o f these civil servants in the total number o f civil servants employed in H Q of federal executive bodies accounted for about 14.1 percent. It is expected that administrative reform efforts would not significantly affect this share (in2010, the share of officials in these Categories would account for 14.2 percent of HQ-based executive civil service incase o f fair administrative reform and for about 13.9 percent in case o f significant attrition, respectively). Table 2.15: FiscalImplications of More RadicalPay Adjustments for "Decision Makers" in Federal Executive Civil Service Headquartersas of 2010 Cash Compensationfor Federal ExecutiveHQ-Based Officials Medium-PacedModeratePay Medium-PacedModeratePay Reformwith Fair Administrative Reformwith Significant Adjustment(Scenario20) AdministrativeAdjustment (Scenario 21 adjusted) Constant 2002 % of 2010 Constant 2002 % of 2010 billionrubles GDP billionrubles GDP Costs of the base case, with 10.66 0.080 9.76 0.073 No additionalmeasures Costs of additionalpay 11.30 0.085 10.34 0.078 adjustmentfor "decision Source: Staff estimates. 2.88 Table 2.15 shows that the extra pay increase for the "decision making" group of HQ- based federal executive authorities would not bring about significant additional fiscal pressures: the costs o f such an extra step would account for about 0.6 billion rubles in 2002 prices (less than 0.005 percent o f 2010 GDP). Such a policy measure would result in both better external competitiveness o f the civil service in the labor market and in greater internal decompression: by the end o f the reform, the compression ratio for HQ-based federal executive civil service would increase to 9.0, which i s 0.8 higher than in the base case with moderate pay reform. 2.89 Another suggestion often made in the debates on implementation o f civil service and administrative reforms relates to a possible separation o f administrative reform from the pay reform incore government employment and from the corresponding adjustment inthe civilian public sector at large. To simulate potential effects of this reform strategy, we made further modifications to the above scenarios 20 and 21 based on the following additional assumptions: 0 Administrative reform measures in core government administration and civilian public sector would be implemented within 2004-2006, i.e., under the compressed schedule; 0 Pay adjustment in core government administration would be implemented in 2004- 2008; and 0 Pay adjustment inthe civilian public sector would be implemented in2004-2010, Le., at a slower pace. 57 58 Figure2.5: Total FiscalImplicationsfor Scenarios4*with the De-LinkedImplementationof Pay and Administrative ReformComponents (p.p. of GDP) L 2.50 li 1 4n2.00 -fg150 1 0 0 6 3a & 050 2 Ly z 0.00 2004 2005 2006 2008 2010 0Scenario20 Scenario21 Source: Staffestimates. 2.90 Although de-linkingthe reform measures may be more attractive to the budget during the first three reformyears (see Figure 2.5), a consequent rapid growth of public expenditures in 2006-2010 may make it difficult to sustain the pace of the reform agenda, which may in turn compromise the reform achievements. Overall, it seems as a less desirable strategy than the one that assumes a more steady reform effort and should be used with caution since it yields a less predictable fiscal implications pattern. H. OVERVIEWOFRISKSASSOCIATEDWITHREFORMIMPLEMENTATION 2.91 Implementation of the broad civil service reform that would affect more than a million of employees in core government administration and about ten million o f those employed in the civilian public sector, let alone the implications of the reform innon-civilianparts o f the general government, i s by definition a complex and risky task. Detailed analysis o f all possible risks associated with the reform implementation process calls for a self-standing piece o f research. Therefore, in this section we try to briefly overview only some key risks that should be taken into account during the reform implementation and that are directly linkedto the fiscal implications of the reformprocess. 2.92 Firstly, a balanced relationship between different components o f the public sector reform, especially between the reforms in core government administration and the rest o f the public sector has to be established. The absence of such a relationship may lead to accumulation of disproportions in the public sector. In particular, it could trigger a non- sustainable wage differential between the core government administration and the public sector at large, which could cause either a slowdown in wage adjustment in the public administration or may result in excessive in creases in budget costs because o f the need to adjust the remuneration inthe highly overstaffed civilian public sector. As confirmed by our simulations, attempt to increase the real wages for the current level of employment in the civilian public sector would require additional Jinancing of about two percent of GDP by 42Scenario 20: Moderate pay reform with a fair administrative reform effort and medium growth in private wages. Scenario 2l(adj): Moderate pay reform with a significant administrative reform effort and medium growth inprivatewages (share of non-wage costs i s 35 percent). 58 2010. Analysis of the current government plans for advancing the civil service reform suggests that this riski s not taken into account yet inthe existingpay reform design. 2.93 Secondly, successful implementation of public administration reforms calls for a close coordination between the civil service and budget reforms. Slow delegation o f authority for budget spending to budget units would undermine the overall efficiency o f the reforms and would reduce incentives for implementing signiJicant staf cuts both in core government administration and in the civilian public sector. Introduction o f performance budgeting mechanisms that require a combination o f more flexible rules for budget spending and higher degree of accountability for the results i s a necessary precondition for implementation of many components of civil service reform, includingHR management reforms, modernization o f public service, as well as administrative reform. The failure to ensure such coordination may result in top-down mechanical cuts in staffing levels that would not be sufficiently concerned that adjustment in staffing was driven the considerations about quality o f public service delivery. The recently approved Conceptfor Reforming the Budget Process (2004-06) provides a basis for integrating performance budgeting principles and the performance management agenda advocated inthe framework of civil service reform. However, the risk of possible disconnect between the two should be fully accounted for and controlled throughout the reformprocess. 2.94 Thirdly, the ultimate success of civil service reform depends largely on the incentives o f subnational governments to pursue a similar agenda across the country. This i s especially crucial because most o f the public service delivery i s taking place at the subnational level, hence the failure to demonstrate some tangible improvements in these parts of the public sector would undermine overall credibility of the reform process. Hence, the reform agenda should encompass both the activities related to building the capacity to implement public administration reforms at the regional and local levels (these issues are partly covered by the Presidential Program Civil Service Reform 2003-2005 that provides financing to undertake pilots and reformexperiments inthe regions) and, more importantly, buildingthe incentives of the regional and local administrations to engage in the reform process. These incentives are associated with both getting additional authority in resource allocation and increased transparency and accountability o f subnational administrations. One of the relevant policy measures that could strengthen the subnational incentive framework i s currently proposed abolishment of a Unified Salary Scale in the public sector, which would be accompanied by the delegating to the regions the authority to determine the pay levels inthe civilian branches of the public sector, taken into account the regional fiscal situation and local labor market prices. 2.95 Finally, successful implementation o f the reforms in deconcentrated units of the federal civil service would also require a proper accounting for cross-regional variation in staffing needs and remuneration levels i s also crucial for. Insufficiently differentiated pay adjustments for HQ-based staff and deconcentrated employees would create major disproportions inthe regional labor markets and may also trigger significant additional budget expenditures (up to about 0.5 percent of GDP in 2010). Hence, identification and establishment of the pay levels for federal deconcentrated employees that are based on local market realities i s one o f the key preconditions for successful and sustainable pay adjustment incore government administrationinRussia. I.CONCLUSIONSANDRECOMMENDATIONS 2.96 Increasing efficiency of the government machinery and improving the quality o f governance and service delivery inthe country are prerequisites for successful implementation 59 60 of the structural reforms inRussia. At the same time, implementation of these reforms would create additional fiscal pressures on the budget system of the country. Fiscal implications of public administration reforms depend on a large number o f variables and are sensitive to changes inmacroeconomic conditions. The model described in this Chapter does not capture all variables involved. However, the analysis of the incremental costs of civil service reform estimated for more than 40 reform scenarios allows for some preliminary conclusions. 2.97 Clearly, the reform approaches to different sub-sectors o f the civilian part o f general government would need to vary both in terms of pay adjustment and in terms o f attrition measures. To achieve the reform objectives, thepay adjustment should be the most significant for the HQ-based portion of the core government administration, lower for deconcentrated units and regional and municipal parts of the core government administration, and the lowest for the civilian public sector. The latter could be justified inpart by the growing share o f fee- based services provided by this sector, which would become a growing source o f compensation o f its employees. Inside the federal HQ-based executive service, the pay adjustment should be focused on managerial levels, which currently are affected by the highest pay gap with the private sector. Such a focus o f the pay reform would allow addressing the key issues of low policymaking capacity inside the government and would help retain highly qualified public officials at keypositions inthe federal civil service. 2.98 Our simulations suggest that over the period till 2010 the budget i s unlikely to accommodate a full closure o f the current pay gap through a radical pay adjustment in the entire core government administration and civilian public sector. As a result, the average residual net public-private pay gap for core government administration would still remain significant (inthe 100 percent range). However, the recent surveys o f public officials reveal that such a gap may be quite acceptable and in general it reflects the existing expectations of the public officials. 2.99 In fact, implementation of moderate pay reform suggests that average wages in core government administration would grow 1.25 times faster than the average wages in the economy, while the average pay in civilian public sector would grow 1.17 times faster than the average wages inthe economy respectively. Our simulations also show that by 2010 the average real wages incore government administration would be 1.7 times higher than current wages (average real wages in the civilian public sector would be respectively 1.6 times higher). 2.100 If such pay adjustment i s complemented by more radical adjustments for a small number of "decision makers" in the headquarters of federal executive authorities, it is likely that the pay reform would achieve its objectives through both increasing external competitiveness of employment in the public sector (especially in core government administration) and significant internal decompression that would facilitate the introduction of a performance oriented system in the Russian civil service. The selected scenarios assume a substantial decompression: the compression ratio for the federal HQ-based executive civil service would increase from 2.5 in 2002 to 6.8-9.0 in 2010, while the compression ratio for the federal executive branch as a whole would by 2010 grow up to 9.9-13.2 from 3.7 in2002. The proposed pay reform would allow retaining a highly qualified cadre, which ultimately would result inbetter quality of policy making and public service delivery, as well as support the government anti-corruption effort. 2.101 The results o f our estimates confirm that implementation o f the administrative reform component and, more specifically, employment adjustment in core government administration and civilian public sector i s critical to the success of the broad civil service reform efforts. 60 Evenmoderate pay adjustments not complemented by significant employment changes would result in unjustified growth of annual consolidated budget expenditures (2.8 to 4.2 p.p. o f GDP in 2010). The expected employment dynamics in the core government administration and civilian public sector, as well as cross-country comparisons justify higher rates of attrition inthe civilian public sector and in sub-national core government administration, but smaller cuts for the federal portion of core government employment. It i s expected that by 2010 employment in the civilian public sector would decline by about 25 percent on average as compared to the 2002 baseline, while the overall employment cuts in core government administration would account for about 9-1 3 percent and attrition in the executive branch reaching 15 percent on average (with 13 percent attrition rate inthe federal executive branch and 17 percent in subnational executive bodies). Similar adjustments would also be required in the non-civilian public sector that is reportedly overstaffed and for which cash remuneration i s directly linked to the pay levels inthe civil service. 2.102 Implementation o f the broad reform agenda in the core government administration would entail significant additional non-wage expenditures. Full-scale modemization o f public service would require both significant investment and significant recurrent costs for operation and maintenance of computer systems and physical infrastructure, as well as implementation o f modern HR practices (including competitive recruitment, staff rotation, and training). However, our simulations show that Russia is unlikely to sustain the current share o f non-wage expenditures in total costs of supporting the core government administration (37 percent). It i s more likely that this share would have to go down to at least 35 percent, which would represent a significant increase in financing in real terms. Similar issues o f non-wage expenditure control are also likely to arise in the civilian public sector at large, but specific estimates related to fiscal implications o f growing non-wage expenditures in this sector go beyond the scope of this Chapter because these are highly dependable on specific situations ineach particular sector (Le., education, health, etc.). 2.103 Overall, the results o f our simulations show that successful implementation o f civil service reform calls for a broad approach to the reform, i,e. simultaneous reform measures in core government administration and civilian public sector at large44.Political economy o f reform also suggests that civil service reform should not be applied to core government administration only. Infact, a significant pay adjustment incore government administration i s likely to become a trigger for corresponding increases in civilian public sector and general government at large. Otherwise, it would be quite difficult for the government to sustain a drastic increase in the compensation gap between those in core government administration and the rest o f the public sector employees. 2.104 Moreover, significant adjustments inthe civilian public sector employment within five to seven year timeframe would yield the savings needed to keep overall fiscal costs o f public administration reform more or less within the current limits. Although the fiscal implications of a moderate pay reform complemented by administrative efforts would account for about 1.2-2.3 p.p. o f GDP as compared to 2003 levels, it i s assumed that a substantial part o f these costs (about two thirds) has been already included into the 2004 budget as a result .of the October 2003 salary increase. Implementation of civil service reform complemented by 43 OECD/SIGMA(1997), p.52. 44 Infact, these reforms should be closely linkedwith parallel adjustments inmilitaryand law enforcement parts of public service. Such integral approach seems to be sharedby the government: representatives of both armed forces and law enforcement agencies have been involved in the preparation of civil service reform, and administrative restructuring of all non-civilianparts of public service has been recently announced by the RussianPresident. 61 62 significant reform efforts in civilian public sector would call for some redistribution of funds across expenditure categories within the consolidated budget: apportion of savings from staff attrition in the civilian public sector may be used for providing additional funds to support core government administration. pay adjustment and administrative changes - to be synchronized. Although de-linking the 2.105 Our simulations also suggest that it i s desirable for both components o f the reform - reform components with employment adjustment implemented before the completion o f pay increase would cut the expenditures associated with the reform inthe earlier years, completion o f the pay adjustment would call for an accelerated increase in consolidated budget expenditures that may be politically and fiscally difficult to afford. As a result, the sustainability of the reformachievements would be at risk. 2.106 It should be noted that, although the estimated budget costs of the civil service reform implementation are very significant if compared to the 2002-03 levels o f financing of respective expenditures, the expected fiscal outcomes do not look prohibitively high when considered in the international context. This Chapter argues once again that public administration reform in Russia cannot be cheap, especially in the environment o f a rapidly growing economy and real wages. Otherwise the objectives of reform could not be achieved and sustained. 2.107 The estimated fiscal costs depend significantly on the private sector real wage dynamics, and more specifically on the change o f the share of real wages inGDP. The higher i s the growthrate o f real wages as compared to GDP growth, the higher i s the additional fiscal pressure on the budget system. This introduces a significant factor o f uncertainty in the costing out future reforms, because future growth inprivate sector wages i s largely outside o f the government control. Moreover, giventhe low share of wages inGDP at the moment, there are reasons to believe that in the medium term growth in real wages would be higher than GDP, which would pushthe reforms costs towards the higher end of the estimated range. 2.108 Because of the magnitude of core government administration and civilian public sector restructuring required, this Chapter argues that such reforms are not likely to be implemented within a mid-term time frame (Le., 2004-06) and would requirea longer period of time - five to seven years. This is especially important given the fact that the success of the civil service reform would significantly depend on its coordination with budget management reforms aimed at introduction o f performance budgeting principles, including delegation o f greater expenditure autonomy to spendingunits and significant increase intheir accountability for the results. Thus, advancing budget management reforms would create additional incentives for reforms inthe public administration. 2.109 Implementation o f civil service reform i s a complex task entailing multiple risks which, ifnot controlled for, may result insignificant fiscal implications. For instance, a failure to synchronize civil service reform incore government administration with similar reforms in the civilian public sector may result in significant growth o f budget expenditures (about 2 percent o f GDP a year in 2010). A failure to account properly for regional labor market conditions when determining future pay increases for federal deconcentrated employees may also cause significant extra costs (up to 0.5 percent o f GDP in2010). 2.110 The results o f our simulations should be interpreted and applied with caution. For instance, the estimates of net public-private compensation gap used for the simulations are quite rough and could be improved based on the results o f regional comparative pay and compensation surveys. These surveys would also allow to better measure the actual value o f 62 in-kindbenefits receivedby the core government employees of federal deconcentrated units, as well as regional and municipal servants. 2.111 It should be also noted that some of the financial resources needed for reform implementation could be derived from other reform actions that have not been cost .out in our simulations. Given the market prices for real estate in the largest Russian cities, more efficient use of federal, regional and municipal property may provide additional fundin8, (however, it i s unlikely that these gains would be sufficient to cover the total reform costs)4 There may also be important productivity gains from the introduction o f more efficient business processes and automation of routine procedures, adoption o f more transparent procurement practices, etc. 2.112 Overall, the above arguments suggest that broad reforms in the core government administration and in the civilian public sector at large may be implemented within five to seven years but should be differentiatedby the scope o f pay adjustment invarious sub-sectors o f civilian employment, closely monitored for non-wage expenditure growth, and complemented by significant employment adjustments inthe civilian public sector as well as by at least some staff reductions inthe core government administration. However, evenunder the most optimistic assumptions, the residual pay gap betweenpublic and private sectors may remain considerable. 4sWe agree that, basedon introductionof new budgetingmechanisms and improvedincentives, a more efficient utilization of publicly-ownedreal estate indeedcouldbring substantial fiscal benefits. However, it is difficult to estimate the magnitude of these savings in advance. Moreover, it seems that most of these savings are likely to be concentratedin Moscow (with some of them representingone-time gains). As a result, these savings could be considerable if comparedto the initial costs of reformingthe federal executive part of the core government administration, but it is unlikely that those could be a longer-termsource of funding for reforms in the entire system of the core govemment administration. Overall, we do not find it appropriateto reflect these savings in the model in additionto the above mentioned effect of a possibly lower share of non-wage expenditures inthe total expenditureson core governmentadministration. 63 Chapter 3. FISCALCOSTSOFREFORMSINTHEHOUSING UTILITY SECTOR AND 3.1 This Chapter analyses potential fiscal and social impact of advancing housing and utility sector reforms in Russia under the different scenarios. It argues that in the current environment of high growth in household incomes, it i s affordable by 2006 to attain 100 percent cost recovery in tariff with simultaneous elimination o f all quasi-fiscal cross- subsidization and the adjustment in domestic energy prices. Moreover, the reforms in residential housing could be made budget neutral inthe medium term, and they would bring considerable savings in the long term. However, high sensitivity o f results to income dynamics suggests that the government should establish an efficient monitoring system to track the affordability of tariff increases for population. At the same time, it i s estimated that adjustment in domestic energy and utility prices could trigger a fiscal gap o f 0.4 percent of GDP due to increased utilitycosts inthe public sector. A. BACKGROUND 3.2 The Russian government has been pursuing reforms in the housing and utility sector (HUS) since the early 1990s. To date, these reforms have been only partially successful, and the sector remains one o f the major parts of the economy that still operates on predominantly non-market principles. The share o f the government's involvement in both ownership and management of the sectoral entities i s excessive, while the institutional framework to support market-based development i s rather weak. As a result, the housing stock i s deteriorating, the quality of utility services i s low, and the sector i s the single largest recipient o f government subsidies. 3.3 In2003 the government made a new commitment to accelerate reforms inthe housing and utility sector aimed at its market transformation, the phasing out o f budget subsidies, and the mitigation of social risks. This policy priority has been driven by the multiplicity of problems that have accumulated in the sector as well as by the demands coming from parallel reform efforts in public administration and the energy sector that call for both the rationalization of budget spendingand an increase indomestic energy prices. 3.4 There are several channels through which reforms in the energy and utility sectors, which are expected to take place as a part o f the government's mediumterm reform strategy, would have a major impact on the costs of operating the housing stock and thus would affect both householdand government budgets. These channels include the following: 0 A direct increase inutility and housing maintenance tariffs, aimed at increasing cost recovery under the existing cost patterns 0 A substantial increase in domestic energy prices, which would affect the costs of services both directly (direct electricity and gas consumption) and indirectly (use of electricity and gas for heat generation and water supply) 0 An additional increase inutilitytariffs to incorporate a capital repair component 0 The phasing-out o fthe existing cross-subsidization inutilitytariffs. 3.5 Overall, the reforms are expected to change considerably the current proportions inthe financing of the existing housing stock: the share o f the consolidated budget i s expected to decline, the share o f the population would increase, while the current financing by the enterprise sector through cross-subsidization i s intended to be fully phased out. In addition, 65 the budgets should be able to collect additional taxes from the energy sector, primarily the gas industry. However, while budget subsidies and total government spending on housing would decline, budgetexpenditures on social support to the poor are likely to increase, at least inthe initialperiod that would follow the tariff increases. Objectivesof the Chapter 3.6 The overall objective o f the Chapter is to estimate the potential fiscal impact of the expected changes in energy prices and utility tariffs, based on the series o f simulations of different reform scenarios. The reform scenarios aim to reflect changes in the government's fiscal and social policies intended to reduce budget subsidies to the sector, increase the overall efficiency o f government expenditures inhousing and utilities, and, at the same time, help to mitigate the impact o f tariff increases on vulnerable groups o f households. The reform analysis follows the approach suggestedby Yasin (2003), which emphasizes a reform strategy that closely links future tariff increases in the energy and housing sectors with structural reforms inthese sectors, as well as with adjustments inbothpublic sector wages and pensions. 3.7 Important factors to be reflected in simulations o f the various reform scenarios include: 0 Existing cross-regional variations inthe main underlying factors, such as actual utility costs, access to utility services, andhouseholdincomes 0 Projections for future dynamics in domestic energy prices and their likely impact on utilityprices 0 Expectedtrends inthe real dynamics o f householdincomes 0 Elimination(phasing out) of cross-subsidization inthe utility sector 0 Elimination(phasing out) o f privileges (lgoty) inhousing and utilities 0 Actual participation rate o f households eligible for housing allowances. 3.8 More specific objectives o f the Chapter include: a. Costing out potential medium term reform scenarios in the housing and utility sector. b. Estimating an overall demand for budget financing related to the reforms in the sector that would cover (i) residual subsidies during the transitional stage (for the periods when the full cost recovery will remain below 100 percent), (ii) costs the of social mitigation programs, and (iii) additional utility costs o f budget sector entities (inhealth, education, etc.), causedby the increased utilitytariffs. c. Developing estimates o f budget revenue impact associated with an increase in housing and utility tariffs and improved tax yield. 3.9 The projections developed in this Chapter are based on a flexible imitational model, which could be used for the simulation of different reform scenarios inRussia's housing and utility sector. An improved government capability to analyze and forecast the consequences of the respective sectoral reforms i s critical to designing adequate mitigation policies, reducing the risks o f social tensions, and avoiding an excessive budget burdeninthe course of reforms. To become a practical tool for policy analysis, the model explicitly takes into account several features o f the housing situation that are Russia-specific, such as the 66 considerable cross-regional variation in the main parameters and specific characteristics o f Russia's housing allowance program. B. SUMMARY OF THE EARLIER REFORMEFFORTSTHE HOUSING UTILITY IN AND 3.10 The reforms in the housing and utility services (HUS) were identified as a major structural reform challenge fairly early inRussia's transition. Despite this, the H U S were not a part o f the original Government o f Russia's economic liberalization program o f early 1992. Instead, the Government's decisions o f 1992-93 provided for the gradual attainment o f full cost recovery in the sector (supported by considerable market transformation) by 1998. However, later these reform targets were adjusted and made considerably less ambitious. In particular, in 1999 the Duma adopted an amendment to the Law on the Foundations of Federal Housing policy which set 2008 as a target for attaining full cost recovery inhousing. 3.11 Overall, reforms inthe HUS to date have produced three main achievements: 0 Divestiture of enterprise housing (which in the early 1990s accounted for about 40 percent of the housing stock) to municipalities i s practically completed. 0 Cost recovery in tariffs has been radically increased, with the current federal benchmark set at the level o f 90 percent, while collections remainhigh. 0 A new system of targeted means-tested social assistance - housing allowances - has been established by almost all municipalities and has proved to be an efficient tool for protecting vulnerable households. 3.12 At the same time, the reforms to date have not succeeded in resolving the main problem - transforming the sector from an administratively managed one into a market driven one. The major remainingproblems can be summarized as follows: 0 The sector's operations remain heavily subsidized. 0 Consumers inthe HUS have quite limited opportunities to influence the performance o f service providers. 0 Competition in housing maintenance i s highly restricted, and the sector i s still dominated by municipally owned monopolies and quasi-monopolies. 0 Most operators in the sector do not have real incentives to improve the efficiency o f their service delivery. 0 The sector does not have a market-based mechanism for investment financing. At the same time, it i s affected by years o f under-maintenance and under-investment, which undermines the quality and reliability o f service delivery. For the time being, government budgets are the only regular source o f investment financing inthe HUS. 0 Regulatory policies and practices in the sector are affected by administrative interventions and are neither sufficiently transparent nor stable. This makes the H U S quite unattractive/risky for the newprivate sector. 0 The incidence o f new private arrangements (such as condominiums) for managing multi-unithousing is low. While more than half o f the housing units are privately owned,, practically all urban housing is still de facto managed by municipal 46This sectionis basedprimarilyon Starodubrovskaya(2003) andIUE(2003b). 67 administrations. By late 2003, less than 6,000 condominiums were established in Russia. 3.13 The political economy analysis o f developments inthe sector suggests that the existing incentives o f municipal governments are likely to be a key barrier to reform acceleration in the HUS. Municipal governments are seen as the main potential losers in future market transformation in the sector. The reforms would drastically reduce the sphere o f their administrative control, while their access to a considerable cash flow in the sector would be lost. Moreover, government control over utilities at the moment creates a major political advantage for administrations during the election campaigns, becauseit allows for easy access to and communicationwith localvoters. 3.14 It is not surprising that under the circumstances most municipal governments have never been keen on supporting market reforms inthe HUS.Instead, considerable efforts were made to imitate market transformations but to keep the essence of administrative regulation intact. This lack of reform incentives at the municipal level was further aggravated by additional factors. First, there has been little external pressure for change -- neither from the private sector (tenants or business) nor from the federal authorities. The federal government was not consistent in its political signal to expedite HUS reform, which, owing to its political sensitivity, has never topped the list of federal policy priorities. Second, the system of inter- budgetary relations has been unstable and generally has not provided sufficient incentives for budget savings, and thus for reforms that could lead to a reduction in municipal budget subsidies. Massive redistribution o f funds by regional administrations in general tends to benefit those municipalities that are not too proactive in reforming the H U S and thus could easilyjustify their needs for additional budget transfers. 3.15 The government adopted its newHousing and Communal Services ReformProgram in November 2001. The program called for significant reforms in the sector to be implemented in steps over 2002-10. The mainpriorities of the government strategy include: increased cost recovery intariffs, which should be achieved parallel with strengthening the safety net for the poor; improvements in the quality of services and in reducing the costs o f services; and the expansion of private sector participation. However, the implementation progress has been modest to date. The acceleration o f cost recovery increases after 2000 has been the most visible aspect of the change. While progress remains highly uneven across the regions, the reported average cost recovery in tariffs exceeded 70 percent in 2003, which i s a major step forward when compared to about 30 percent in 1997-99. The Government resolution of August 26,2004 set upan objective of attaining 100percent cost recovery intariffs in2005. 3.16 The Bank analyzed the reform implementation constraints in the HUS, faced by subnational governments and other stakeholders, in the recent Policy Note, "Housing and Communal Services in Russia: Completing the Transition to a Market Economy" (World Bank, 2003a). 3.17 Based on the lessons from the earlier analytical work (see World Bank 1998 and 2003; IUE, 2003a; Starodubrovskaya, 2003a), this Chapter argues that the reform accelerationinthe H U S could be sustained only through simultaneous government actions in several important directions: 0 Advancing reforms in inter-budgetarv relations, in particular providing a more transparent regulatory framework for fiscal redistribution among municipalities within the regional budgets and expanding a local tax base for municipalities. Recent 68 amendments to the Tax and Budget Codes were an important step in this direction. A future priority relates to the introduction of a local real estate tax. In addition, more active political pressure and monitoring on the part o f the federal government would be requiredto change prevailingmunicipal government incentives. Reforming housing financing: changing disbursement mechanisms by transferring most of the budget support funds directly to customers o f the H U S instead o f service providers. Improving the regulatory framework, especially for tariff setting, as a tool to make the sector attractive to large private investors. While day-to-day regulation would remain decentralized, the federal government has to strengthen a unifiednational approach to HUS regulation which would define clear rules of the game for subnational regulators. Creating conditions for established private investors to enter the HUS. Under the circumstances, large private firms, with their ability to withstand potential local political pressures, could become a real driving force for market transformation inthe sector. This would require, inter alia providing a legal framework that supports longer-term contracts between private operators and municipalities (such as a good Law on Concessions inlocal utility networks) and strengthening the judiciary's ability to enforce such contracts. c. GOVERNMENT INVOLVEMENT INFINANCINGRESIDENTIAL HOUSING Recenttrends inhousingand utilityfinancing 3.18 At the moment, the population is still paying about halfof the total housing and utility costs, with the rest coming from budgets, enterprises, accumulation o f arrears and degradation of the housing stock andutilities (Table 3.1). 3.19 The total costs of government programs that explicitly relate to the financing of the H U S exceeded 1.75 percent o f GDP in 2003 (about US$8 billion, see Table 3.2), while the large cities historically have been spending about a third o f their budgets on housing and communal services.47Direct subsidization of heating and hot water services make up about 40 percent o f the total budget support inthe system. Table 3.1: Structure of Housing Financing(YO) 2001 2002 2003 Population 47.4 46.8 52.7 Budget (only direct budget support) 35.8 41.4 37.7 Enterprisesector and under-financing(residual) 16.8 11.8 9.5 Source: Staff estimates based on the datafrom Rosstat and Gosstroi. 47This includesspendingon operational subsidies, housingallowances, and investmentsinrehabilitation. 69 Table3.2: Financingof the HousingSector (YOof GDP) COSTS/EXPECTED ACTUAL FINANCING FINANCING 2001 2002 2003 2001 2002 2003 Total costs of services, estimatedonthe basis ofthe reportedtariffs 4.03 4.73 4.67 3.35 4.17 4.22 olw: 1. Households 2.01 2.46 2.68 1.91 2.21 2.46 2. Budget: 2.01 2.26 1.99 1.44 1.96 1.76 -- compensationof direct subsidiesto cover tariff gaps 1.42 1.66 1.34 1.21 1.61 1.26 tariff benefits(lgoty) 0.54 0.46 0.42 0.19 0.24 0.30 - housingallowances 0.06 0.14 0.23 0.04 0.11 0.20 MEMO: 1.Enterprisesector, residual 0.68 0.56 0.44 2. GDP, bn 8,944 10,834, 13,285 Source: IUE. 3.20 The average cost recovery intariffs inresidential housingreached 69 percent in2002, a major improvement relative to 53.5 percent in2000 (Table 3.3) and 33 percent in 1993. The collection level remains relatively high, at about 90 percent. However, the actual billings are almost 20 percent lower than those that would correspond to the reported cost recovery levels because of considerable benefits (lgoty) in the system. As a result, the actual household payments amount tojust 50 percent o f the current housing costs. 3.21 By the end of the third quarter of 2003, the average cost recovery intariffs increased further to 73 percent, and cost recovery in payments increased to 55 percent. However, it should be emphasized that the cost recovery levels reported in Table 3.3 are considerably upwards biased because they are estimated relative to the prevailing average level o f domestic prices for energy resources, but not as a ratio to the long-term marginal costs (LRMC) o f energy production and delivery. This discrepancy i s especially important for gas and electricity supply. In the gas sector, for instance, assuming the average RLMC level i s betweenUS$35 and US$40 per 1000 cub m, the 2002 average gas tariffs in Russia were at about 40 percent o f the RLMC. Because gas tariffs for households were below the average tariff, while the costs o f gas delivery to households were above the average, the properly estimated cost recovery intariffs for population at the time was closer to 30 percent. 70 Table 3.3: Cost Recovery and CollectionRates,2000-02 (YO) Total Housing maintenance 2000 2001 2002 2000 2001 2002 Cost recovery intariffs 53.5% 58.2% 69.1% 50.9% 50.0% 67.5% Billing coeff. 11 86.8% 84.2% 80.6% 78.5% 79.1% 76.8% Collection ratio 86.9% 87.8% 89.8% 89.4% 89.4% 91.8% Cost recovery by payments 40.3% 43.0% 50.0% 35.7% 35.3% 47.6% Water Heatingandhot water 2000 2001 2002 2000 2001 2002 Cost recovery intariffs 53.6% 61.9% 68.8% 43.5% 52.5% 65.8% Billing coeff. 11 87.9% 84.1% 79.9% 84.6% 83.4% 84.1% Collection ratio 84.2% 86.1% 88.6% 87.0% 85.2% 87.4% Cost recovery by payments 39.6% 44.8% 48.7% 32.0% 37.3% 48.4% Power Network gas 2000 2001 2002 2000 2001 2002 Cost recovery intariffs 76.8% 81.8% 81.5% 82.5% 88.6% 85.7% Billing coeff. 11 85.7% 83.2% 78.3% 100.0% 90.0% 80.0% Collection ratio 87.4% 89.6% 91.1% 82.6% 92.8% 90.9% Cost recovery by payments 57.5% 61.0% 58.2% 68.2% 74.1% 62.3% 11Actualbillings adjustedfor discounts and Igoty. Source: Staffestimates based on the datafrom Rosstat and Gosstroi. 3.22 The financial performance of the housing and utility sector remains weak, which is a result of both the existing government tariff policy and the poor operational efficiency o f the service providers, which primarily remain unreformed municipally owned entities. Total losses in the sector amounted to 0.8 percent of GDP in 2002 (Table 3.4), while the stock of total payables (mostly for energy) was at the level o f 2.7 percent o f GDP in early 2003. Table3.4: FinancialIndicatorsfor the Housingand Utility Sector (% of GDP) TOTAL o/w Heating & Hot Water 2000 2001 2002 2000 2001 2002 Total losses inthe sector -1.03 -0.99 -0.79 -0.51 -0.49 -0.36 o/w: from servicesto population -1.21 -1.26 -0.95 -0.66 -0.61 -0.41 Total receivables,stock 3.17 2.54 2.22 1.02 0.83 0.76 olw: from budgets -- from 0.95 0.66 0.59 0.41 0.30 0.25 population 0.44 0.44 0.52 0.13 0.14 0.18 Total payables, stock 3.25 2.58 2.69 1.20 1.02 0.95 olw: - to budgets 0.39 0.37 0.41 0.11 0.12 0.13 Source: Staffestimates based on the datafrom Rosstat and Gosstroi. 3.23 In 2000-01, average spending on housing and utilities amounted to only about 6-7 percent of household budget expenditures. In 2002, housing and utility prices rose by 50 percent inreal terms, while real incomes increased by about 20 percent. As a result, the share o(Table f housing s ending exceeded 8 percent, which historically i s a relatively high level in Russia 3.5).4! 48This share amountedto about 2 percent in 1993 (World Bank, 1998a). In comparison, households in Poland and Hungary typically spend over 20 percent of their income on HUS (World Bank, 2003). Even in Belarus, 71 Table 3.5: Average HouseholdExpenditureon Housingand Utilities (YOof total householdexpenditures) 1997 1998 1999 2000 2001 2002 Average 6.3 6.7 6.0 6.1 7.0 8.5 Median 4.6 4.6 4.3 4.5 5.2 6.5 Source: Staff estimates based on the HBS. 3.24 It is worth mentioning, however, that even after accounting for the lgoty, the existing statistics on utility tariffs for households continue to overestimate considerably the actual level o f cost recovery. This i s due to two primary factors: 0 Services to the population are still cross-subsidized through higher tariffs from commercial consumers. The total annual amount o f cross-subsidization may be close to 1percent of GDP (see Table 3.7 below). 0 The tariffs heavily underestimate the investment component. It is estimated that the amount of annual under-investments inrehabilitation and repair o f local utilities may reach another 1percent o f GDP. 3.25 Therefore, the actual financing needs of the sector, at the current level of its efficiency, are much higher than the cost estimates based on the reported information on tariffs and cost recovery, and exceed 6.5 percent o f GDP (see also Table 3.8 below). At the moment, annual household housing payments make up about 2.5 percent of GDP (i.e. they cover less than 40 percent o f the total amount). Structure of budget support inthe housingandutilitysector 3.26 Table 3.6 presents the structure of the mainbudget programs inthe sector. There have been four main types o f budget channels that provide regular financing to the HUS.Most o f this financing is provided by the subnational budgets. The role of the federal government in HUSfinancing is limitedto: (i) providing special central budget transfers to regions related to the financing of housing recently divested from local enterprises; and ii)providing indirect budget support to the sector by financing non-regular programs o f the emergency type (Table 3.7). Table 3.6: Structureof Budget Spendingon Housing and Utility Servicesto the Population(main programs of directbudget support) (YO) ExpectedFinancing Actual Financing 2001 2002 2003 2001 2002 2003 Totalbudget spending: 100 100 100 100 100 100 - direct subsidiesto cover tariff gaps 70.4 73.5 67.4 83.7 82.1 71.4 -- housingallowances compensationof tariff benefits(lgoty) 26.7 20.4 21.2 13.2 12.3 17.1 2.9 6.1 11.4 3.1 5.7 11.5 Memo:Federaltransfers to regionsto compensatefor housingdivestiture, as % 1.3 2.5 n.a. of totalbudget spending Source: Staffestimates based on the datafrom Rosstat and Gosstroi. ~~ budgeton HUS. where structuralreformsin general have beenlagging, in 2003 households spent on average 11 percent of their 72 3.27 Mainbudget programs inthe sector could be summarized as follows: Budget subsidies to service providers. This i s the main budget program in the sector, which accounts for more than 70 percent of the total direct budget support. Its preservation i s a direct reflection of the slow pace o f reforms in the HUS. Prevailing tariffs are below actual costs and respective govemments have to compensate local utilities for these tariff deficiencies. This i s a traditional example o f "bad subsidies," which are highly regressive: wealthier households tend to live in larger apartments, consume more services and receive more subsidies. This subsidization also undermines incentives to improve efficiency inthe sector. Housing benefits &OW). Recently, more than one-third o f Russian households were eligible for various discounts from the established HUS tariffs, usually as high as 50 percent. The total annual cost of this program amounted to 0.45 percent of GDP in 2002-03. Most o f these programs are occupation-based, but not income-based, and many are relicts o f social program priorities o f the Soviet era. The analysis o f their beneficiaries clearly suggests that non-poor households received most o f the benefits. In addition, actual budget financing of benefits never exceeded 70 percent. In the summer of 2004, the Government of Russia adopted a radical program of reforming the lgoty starting from 2005. So far, it has announced a limitedmonetization o f lgoty through a replacement of specific programs in health and public transportation. See a more detailed analysis of lgoty ina separate section below. 0 Housing allowances. The introduction o f the housing allowance program represents one o f the major successes o f housing reform during the 1990s. It i s the first Russian program o f targeted social assistance, and as such it has a major impact on definingan overall approach to the reform of social protection in the country. The program has been established practically in every municipality, and its evaluation suggests that it has been rather effective. As of September 2003, 13.8 percent of Russian households participated in the program. In several Russian regions more than 30 percent of households became recipients of housing allowances. It i s expected that in the course o f the reforms the share o f participants would increase further, but it i s believed that the existing infrastructure has the capacity to handle more applicants. A replacement of other current programs o f housing financing with additional funding of housing allowances would add a major efficiency gain to the system, because it would provide for the channeling o f most o f the budget expenditure in the H U S for the support o f low-income households and more broadly for a reallocation o f subsidies from producers to consumers. 3.28 Before 2004, the current legislation on housing allowances provided for two parallel eligibility criteria as well as a formula to determine the benefit size: a) General. Household spending on housing and utility services (within the established limits of housing size) should not exceed the threshold share in overall household income; this threshold i s established by the decisions o f regional govemments and in most regions equals 22 percent o f family income. b) For low income households. Households with per capita incomes that are below the subsistence minimum are eligible for a housing allowance if their total housing and utilityspending exceeds 50 percent o fthe current minimumwage. 73 3.29 It was believedthat the second criterion was too generous, and that it ledto excessive subsidization of low income households and distorted the entire process o f housing assistance. This is especially problematic, given the quite low value of the Russian minimumwage. In 2004, the Govemment adopted the decisionto eliminate the 2"dcriterion. 3.30 The elimination of the second criterion, as estimated by the WE, other factors being intact, would reduce the number of eligible participants in the program by about 75 percent, bringingthe number to 7.5 percent o f households. Inour simulations we also assumedthat the second criterion would be eliminated. Respectively, the number 7.5 percent i s used as a benchmark for our analysis o f potential changes in the number o f allowance recipients as a result o f the proposedtariff adjustments inthe HUS. 3.3 1 Inthe course ofrecent reforms inthe division ofpowersbetweendifferent government levels in Russia, it was decided that all functions related to social policy should be concentrated at the regional level. As a result, the responsibility for financing housing allowances was moved from the municipal to the regional budgets. While this change expands the fundingbase for housingallowances, it also creates a potential incentive problem. Having no responsibility for financing this program themselves, municipalities may relax control over program administration, which could erode the efficiency of its targeting. The government's instrument to mitigate this risk is through an introduction of the unified system of regional housing standards that would encourage municipalities to follow a single region-specific set of eligibility criteria for housing allowances. 3.32 The earlier analysis identified two essential weaknesses in the administration of housing allowances (Hamilton, 2004). First, their availability shows too high a variation by region, which by far exceeds the variation in housing affordability. Second, the program remains predominantly urban-based, with rural households beingmostly left without accessto it4'. 3.33 Indirect budget ,Drograms (investment support, expenditures for liquidation o f emergencies and deliverv o f fuel to the North). This i s the least transparent and least efficient part o f public financing in the HUS, and i s provided mostly on an ad hoc basis. The funds spent through this channel are not reported in the conventional budget reports as spent on housing sector support. Preservation o f these programs reflects primarily the failure o f sectoral reforms to date to create an institutional framework for sustainable investment finaricing that would include efficient mechanisms for tariff regulation, competitive mechanisms for service delivery, and providers that are creditworthy and have access to regular commercial credit. The overall amount of budget funding through this channel i s estimated by the IUEto amount to about 0.6 percent of GDP a year, or more thanone-third of fundingunderthe conventionalbudgetprograms inthe H U S (Table 3.7). 3.34 When indirect budget financing i s included, total annual government spending on the HUS is estimated to amount to 2.4 percent of GDP. If,inaddition, the amount of quasi-fiscal financing through cross-subsidization i s reflected, the total amount of fiscal and quasi-fiscal fundinginthe sector reaches 3.3 percent o f GDP (Table 3.7). 49 It i s worth noting, however, that most rural households in Russia continue to benefit from much lower energy tariffs. 74 Table 3.7: Estimatesfor Total Fiscal and Quasi-fiscal Support to the Housing and Utility Sector in 2003 bnrbl %ofGDP Direct budget support provided 234 1.76 Indirect budget support --- Emergencyrehabilitation 82 0.62 Preparation for the winter 55 0.41 Fueldelivery to the North 16 0.12 11 0.08 Quasi-fiscalfinancing: Cross-subsidies (*) --- electricity 125 0.94 heating 20 0.15 50 0.38 gas 55 0.41 Total fiscal and quasi-fiscalfinancing 441 3.32 Costs of annual under-investments (missing rehab) 135 1.02 (*) Estimatesrelate to servicesprovidedto bothpopulationandbudget organizations. Source: IUE 3.35 Table 3.8 presents an estimate of the total volume of actual financing in the Russian housing and utility sector through all available channels. It amounts to 5.8 percent of GDP, which i s quite high, given the predominantly low quality o f the existing housing. Moreover, given the available estimates of annual under-financing in proper maintenance and rehabilitation, the total annual sector needs in financing under the existing institutional arrangements may amount to as much as 6.8 percent o f GDP. The latter number should be considered primarily as an indicator o f the sector's inefficiencies, and not as a benchmark for future growth in tariffs and subsidies. Many sectoral experts believe that under the proper incentive regime the total costs o f operation inthe sector could be reduced by at least 20%. At the same time, as shown below, the sector is facing some additional cost increases associated with growth in domestic energy prices and with a need to incorporate capital costs into the tariffs. Table 3.8: Total Volumes of Available Financinginthe Sector, 2003 (% of GDP) I,Directfinancingofhousingandutilityservices 4.2 o/w: Households 2.5 Budget 1.8 2. Indirectbudget financing 0.6 3. Total cashfinancing, (1)+(2) 4.8 4. Quasi-fiscalfinancing 0.9 5. Total availablefinancing, (3)+(4) 5.8 6. Under-financing 1.o 7. Total financingneeds under the existinginstitutionalarrangements, 6.8 (5)+(6) Source: Staffestimates, based on the data collected and estimated by the IUE. D. RECENTTRENDSINHOUSINGAND UTILITYTARIFFS 3.36 Inthe periodthat followed the 1998 Russia crisis, the government tariff policy inthe housing and utility sector was inconsistent. In 1998-2001 the tariff growth in housing and utility services lagged behind general price growth (Table 3.9). The largest gap between the 75 two occurred in 1998: HUS prices grew by only 12 percent?', while annual CPI inflation was 84 percent. This led to a considerable deterioration in the financial situation in the sector. In 2002, as a reaction to this extended period of under-financing, the annual tariff increase inthe H U S for the first time surpassedthe rate of CPI inflation. Table 3.9: Inflation and Price Growth inHousing and Utilities, 1998-2000 Implicitgrowth of Consumer Annual growth of electricity growth Annual growth tariffs for anon- Price Index of HUStariffs tariffs of gas tariffs energy part of HUS services 1998 184.5 112.0 102.2 108.6 113.8 1999 136.5 125.0 119.7 113.6 128.7 2000 120.2 111.3 141.7 162.0 90.9 2001 118.6 110.9 137.0 129.0 99.1 2002 115.7 134.5 125.0 130.0 138.0 Source: WE, Rosstat. 3.37 It is worth noting that in 1998-99 the tariff growth inthe electricity and gas sector was even slower than in the HUS. However, the period of depressed tariffs was much shorter there. Starting from 2000, growth in electricity and gas tariffs surpassed the CPI rates and provided enterprises inthese sectors with a more adequate level of cost recovery. 3.38 Overall, the HUS was used as a major "shock-absorber" during the period o f high inflation and economic instability of 1998-99. By keeping housing and utility services costs low, the Government of Russia was trying to maintain social and budgetary stability at the cost of the deteriorating financial position o f the HUS. 3.39 The last column inTable 3-9provides estimates of growth intariffs that corresponds to a non-energy component o f H U S services. These indices were estimated by de-composing the total costs o f housing and utility service delivery in energy (electricity and gas) and non- energy component^.^' As can be seen from Table 3.9, in2000-01, when energy tariffs started to grow relatively rapidly, but growth in HUS tariffs remained depressed, the implicit tariffs for a non-energy part o f the HUSwere declining even innominal terms. 3.40 Since the growth in H U S tariffs in the post-1998 period was kept below general inflation, the accumulated under-financing has become quite significant. In this report it i s assumedthat to provide the sector with conditions for financial rehabilitation at.least a portion o f this deferred inflation effect has to be compensated through future accelerated tariff growth. 3.41 InTable 3.10 we estimated the accumulated size of this under-financing in the non- energy part of the H U S for the period o f 1997-2002 as o f 126 percent relative to the CPI inflationrate. It i s worth noting that this price gap declined considerably in2002 after a major adjustment in HUS tariffs. Still, this gap indicates that the real level of non-energy housing tariffs in 2002 was less than half o f their 1997 level. In our simulations, we assumed that the future growth in housing tariffs would compensate for about 25 percent of this effect of 50The index of HUStariffs is estimatedbasedon the data for a federal standard of HUS costs, annually adopted by aresolution ofthe RussianGovernmentas apart of the budgetpreparationprocess. Both direct and indirect consumption of electricity and gas were taken into account. See the next section for more details. 76 deferred inflation. This means that in the model we assumed that for the first three years o f reforms the non-energy part o f the H U S costs would grow at a rate about 10 percentage points higher than the expected CPI rate, after accounting for all other factors that will influence the future H U S tariff dynamics. Table 3.10: Estimatesfor the Effect of "Deferred Inflation" inthe HUS (non-energy part), 1997-2002 CPI, Implicit growth of tariffs for a Deferred inflation effect (a gap 1997=100 non-energy part of HUS between CPI and non-energy services, 1997=100 HUS tariff index) 1998 184.5 113.8 1.62 1999 251.8 146.4 1.72 2000 302.7 133.1 2.27 2001 359.0 131.8 2.72 2002 411.8 182.0 2.26 Source: IUE, Rosstat. E. AGGREGATEDCOSTSSTRUCTUREINRESIDENTIAL.HOUSING 3.42 Inthis section we review the role of energy in overall housing and utility costs. This analysis i s important for further simulations of the link between expected growth in domestic prices for gas and electricity and a future increase in housing costs. For the purposes of such an analysis, it i s important to account for full gas and electricity consumption in the sector, that is, both their direct and their indirect (as input for production o f other utility services) consumption. 3.43 The first column in Table 3.11 presents the conventional cost structure in Russia's housing sector. It i s worth noting that the share of housing maintenance intotal costs i s quite low, less than one-fourth. In developed economies this share often amounts to 50 percent of the total. Non-maintenance costs are much higher in Russia because of several factors: (i) climate, (ii) energy inefficiencies in the sector, and (iii) under-estimation o f actual major the maintenance costs in the existing maintenance feeskariffs, which cover operating costs (though not in the full amount) but exclude provisions for rehabilitation and major repair costs, land fees, and insurance payments. The implementation of tariff reforms would require a major increase inhousing maintenance fees. Table 3.11: Cost Structure in the Housingand Utility Sector, per 1 Square Meter of ResidentialHousing, as of end 2002 (YO) Cost structure that accounts for Conventional cost structure indirect consumption of electricity and gas Electricity 9.4 15.8 Water and sewerage 13.6 9.5 Heating andhot water 48.6 34.9 Gas 5.8 19.2 Housingmaintenance 22.6 20.3 Total 100 100 Source: IUE. 3.44 The second column inTable 3.11presents the estimates for the full electricity and gas intensity o f the Russian housing sector. While the costs o f directly consumed electricity and gas make up only 15 percent of total, the costs o f full consumption o f these energy inputs 77 amount to 35 percent. This i s because all main utility services are quite energy intensive. Electricity costs account for nearly 30 percent o f the total costs in water and sewerage. The share o f heat generated in gas-operating boilers reaches 60 percent. Gas accounts for 35 percent o f the total heat costs in such boilers. Respectively, 50 percent o f electricity i s generated by gas-operating power plants and gas accounts for 35 percent o f the generation costs o f these plants. 3.45 The estimates in Table 3.11 were used directly as weights in our simulations o f future housing costs, which were based on the utilization o f three primary independent variables: growth in gas and in electricity prices, and growth in tariffs in the rest o f the sector. The last variable reflects a need for tariff adjustment to ensure proper maintenance and rehabilitation o f the housing stock and utility networks, but it i s unrelatedto the costs o f energy inputs. Budget spendingon utility services providedto budget organizations 3.46 The Russian budget statistics do not generate consolidatedestimates for federal budget expenditures o n utility services consumed within the public sector. This i s because utility payments are budgeted as parts o f the budgets o f individual ministries and agencies, and the Ministry o f Finance does not provide for across-the-board consolidation of such spending. However, the budget reporting at the subnational level provides sufficiently adequate data on ' government utility spending. 3.47 The consolidated estimates o f the budget costs o f utility payments were developed by the IUE based on the detailed analysis o f the budget execution for 2001. Respective expenditures made by both regional and municipal govemments were extracted from individual regional consolidated reports on the annual budget execution. Respectively, estimates for the 2001 federal budget spending on utility services were developed using several assumptions on the unit costs o f such services in the sectors, for which direct reports are unavailable. The latter includes expenditures under the Ministries o f Defense and Interior. Annex 3.2 presents a description o fthe various assumptions made. 3.48 Consolidated budget expenditures on utility services for 2002 and 2003 were generated based on the simplified procedure, using the base estimates for 2001 and general trends in budget spending at all levels o f the government in 2002-03, as well as the dynamics o f utilityprices. 3.49 Total budget expenditures on utility services have been slightly larger than 1 percent o f GDP in recent years (Table 3.12). H a l f o f these expenditures are made by municipal budgets. The analysis also suggests that at the subnational level (both regional and municipal) the largest component of the total utility spending relates to the costs o f operating budget institutions inthe education and health sectors, which account for about 70 percent o f the total utilityexpenditures. 78 Table 3.12: Budget Expenditureson Utility Services Providedto BudgetOrganizations,by the Level of Government(billion RUR and YO) Inbl Rbl In % of GDP 2001 2002 2003 2001 2002 2003 Total expenditures 89.9 130.8 152.3 1.o 1.2 1.1 - Federalbudget - Consolidatedregionalbudget 29.1 43 46.9 0.3 0.4 0.4 60.8 87.8 105.4 0.7 0.8 0.8 - regional including: budgets 18 28.1 33.8 0.2 0.3 0.3 - municipalbudgets 42.8 59.7 71.6 0.5 0.6 0.5 Memo:GDP 8.944 10,834 13.285 Sources: IUE, MOF, Rosstat. 3.50 At the federal level, utility spendingon education andhealthis considerably lower and accounts for less than 40 percent o f the respective total. This i s because o f the completely different structure of government functions at the federal level. The specific federal itemsthat have significant utility costs relate to defense and law enforcement, includingthe operation of the penal and penitentiary system. This group of government functions accounts for nearly one-third o f all federal budget spending on utility services. 3.51 It is worth noting that, owing to a considerable improvement in budget discipline in Russia since the late 1 9 9 0 ~actual government payments for utility services have improved ~ considerably. The level o f current non-payments has become negligible. The stock of government arrears for utility payments declined to below 0.6 percent o f GDP by the end of 2002 (Table 3.7, above). Most o f the remaining quasi-fiscal financing takes places through cross-subsidization (Table 3.6 above): in many regions budget organizations have benefited from the same low utility tariffs as households. 3.52 As is shown below, we expect that under the reform scenarios the unit costs o f H U S would increase by about 90 percent by 2006 relative to the 2002 Even adjusting for the expected modest efficiency gains inthe sector, our estimates suggest that full adjustments in energy and utility prices would result in the total HUS costs to the government reaching 1.9-2.0 percent o f GDP, with half to be incurred by municipalities. This amounts to 0.8-0.9 percent of GDP in additional expenditures for Russia's consolidatedbudget.We estimate that about half of these incremental costs could be compensated through additional taxes paid by energy firms and utility providers that are the primary beneficiaries o f the proposed tariff reform. This leaves a residual fiscal gap o f about 0.4-0.5 percent o f GDP. In the longer term, we expect that most o f this gap would disappear as a result of the expected rationalization and consolidation of the budget sector, first of all in health and education. But public sector rationalization could become a relatively lengthy process. Meanwhile, the government has to findways to finance the gap. 52 This cost increase reflects some compensation for under-investments in the previous period. See below for more details on cost assumptions. 79 F. MODEL FOR SIMULATING BUDGET IMPLICATIONS FROM INCREASES INHOUSING COSTS Lessonsfromthe earlier simulationsof housingreformsinRussia 3.53 Given the long history o f the government's attempts to accelerate housing reforms, which would include reforms in financing through higher cost recovery by tenants and reduced subsidies, it i s not surprising that there has been some experience of quantitative analysis o f the potential impact of the proposed tariff increases.53Most o f this work has been undertaken initially by the staff o f the Urban Institute, as part of the housing reform program sponsored by USAID. 3.54 The primary focus of the earlier work was related to the introduction of the housing allowances programs inparticular regions and municipalities, as well as to the analysis o f the actual efficiency o f suchprograms. This requireddeveloping the procedures for: (i) estimating the variation in future reform impacts across particular types o f housing and specific household groups, (ii) forecasting changes in the demand for housing allowances, and (iii) makingprojections for trends inthe total requirementsfor budget financing, including costs of both conventional budget subsidies and social assistance programs, such as housing allowances. These studies have beenbased on various types of the survey data on household incomes, expenditures, and housing conditions, which have helped provide an important understandingregarding the comparative advantageso f specific statistical sources. 3.55 The main lessons from the earlier simulations o f housing reforms, which were fully incorporated inthis paper, couldbe summarized as follows: 0 Estimates of the current level ,ofhousing costs to households have to be imuuted, and they should not be directly based on expenditure values self-reported inthe household surveys. The conventional household surveys inRussia, includingboth the RLMS and the regular Household Budgetary Survey (HBS), bring an unacceptable level o f distortions when they deal with the issues related to housing and utility spending. Apparently the basic question, "How much did you spend on housing and utilities last month?" was interpreted quite differently by respondents. This i s because of the multiplicity o f available discounts and benefits (lgoty) to households, the high incidence of arrears and late payments, and the still relatively low share of housing- related payments intotal spending for many households. 0 Accounting for cross-regional variations incosts and incomes i s important. There i s a striking cross-regional variation in all key parameters that determine a potential reform impact on both the population and the fiscal system. In particular, regional differences are highwith respect to average housing costs, average income levels, and income inequality. As a result, as was shown, for example, in World Bank (1998b), while on the average the reform impact could be modest, some regions (particularly Siberia and the Far East), could be badly affected by the shift to full cost recovery in housing. 0 Regional income ,distribution i s a key determinant of a demand for housing allowances. Relatively simple models that are based on i)forecasting o f average regional housing costs, and ii)the aggregated income distribution by main income groups, proved to be sufficiently accurate in generating projections for both number 53 See specifically Housing Allowance Program (1996), Kolodeznikova and Stmyk (1997), World Bank (1998b), IUE (2003b). 80 o f future applicants and average size o f their allowance. Switching to more detailed (household-level) information on incomes and housing conditions greatly complicates the analysis and does not bring significant improvements in the accuracy of projections. Model: EstimatingfutureHUScosts" 3.56 The base model for estimating future average costs o f operating the residential housing stock hadthe following structure: HUS (t)= (H(t)*x l +E(t)* x2 + G(t)*x3) * Sav(t), (1) where HUS (t)- a housing cost index that reflects an increase infull average costs o f operating 1 square meter (sq. m) of the housing stock inyear t relative to the base year (2002), H(t)-anincrease inunitnon-energy costs ofoperatinghousinginyear t relative to the base E(t) and G(t)-- increases inunit costs of electricity and gas, respectively, inyear t relative to year, the base year, x l,x2, and x3 - shares of maincomponents of housing costs (non-energy, electricity and gas, respectively - see section 3 for additional information) inthe total costs, Sav(t)-parameter o f cost savings inyear t, which reflects the expected efficiency gains (primarily energy savings) as a result of reforms, as a percent of the base 2002 costs. 3.57 All costs are estimated in constant 2002 prices. Regional average costs HUS(r, t) are determined based on actual cross-regional variations inunit costs inthe base year 2002: HUS(Y,t)=HUS(t)*cost(r) Where cost (Y) - the coefficient of regional costs that reflects the 2002 ratio between unit costs inregionr and average costs for Russia. 3.58 Future growth inelectricity and gas tariffs, E(t)and G(t),i s determinedby two factors: -- el/gl expected average real growth indomestic energy prices, - e2/g2 - expected effect o f elimination of cross-subsidization that would provide for a higher growth inresidential tariffs relative to their average growth. 3.59 Futuregrowth innon-energy unitcosts H(t)is determinedby: - h1- expected real growth intariffs for non-energy services inthe sector, mostly relatedto a partial compensation for earlier under-financing in the sector; hl was selected at the level o f 1.32 (i.e., for the period 2004-06 non-energy costs in the sector would be growing at a rate that is 32 percentage points higher than general inflation), which would compensate for about a quarter of earlier deferred inflation. - h2 -- expected effect o f the elimination o f cross-subsidization in non-energy services (primarily water) that would provide for a higher growth inresidential tariffs relative to their average growth. Therefore: 54 A detaileddescription ofthe modeli s providedinIUE(2003a). 81 E(t)= el(t) G(t>=81@) g2(0 H(t)=hl(t) h2(t) ***e2(t) 3.60 Projections for costs of housing and utility services for budget organizations are also based on the above HUS(t) index (ie., it i s expected that future costs o f HUS for budget institutions would grow at a similar rate as those for households). Estimating the demandfor housingallowances 3.61 Estimates for a potential demand for housing allowances were developed on a region- by-region basis, using region-specific information on (i)average unit housing costs, (ii) average household incomes, and (iii) differentiation among eight main income groups. income 3.62 The core variable for this bloc of the model is the future average housing costs per capita inregion r, CC(r, t). It was estimated usingthe data on average regional housing costs HUS (r, t) and the aggregated distribution of the Russian population by available housing space. A standard assumption on housing distribution was used: 10 percent o f the population use 33 sq mper capita, 15 percent use 21 sq myand 75 percent use 18 sq m. It i s worth noting that hausing space in excess of these amounts i s not eligible for subsidizing through the housing allowance program. 3.63 For each region and income group, average household incomes I (r, j, t) were compared with the regional per capita unit housing costs CC(r, t) to determine if this income group i s eligible for budget support through housing allowances. allowances, where q - i s a maximum share of household income that could be spent on 3.64 IfI(r,j , t) * q > CC(r,t),then membersof income groupj are ineligible for housing housing. 3.65 Only household housing costs that exceed q are eligible for full budget compensation through housing allowances. According to the current legislation, in most regions q=22 percent. Thus we assume that subnational governments would largely follow federal guidelines on the eligibility threshold. In our simulations, we also considered alternative values o f parameter q such as 10 percent and 15 percent. 3.66 The average allowance for a member of income groupj inregion r inyear t, Allow (r, j , t) is estimated as: Allow (r,j, t) = CC(', t) - I (T, j, t)*q (2) 3.67 Aggregation by income groups and regions allows for estimating the total number of recipients of housing allowances and the total costs of this program for each scenario. 3.68 Inthe scenarios, which did not provide for the elimination of lgoty, the size of each income group was proportionally reduced to reflect the number o f lgoty beneficiaries inthis group. This would reduce an overall demand for housing allowances. 3.69 It is assumed that the future participation rate in the housing allowance program will be 75 percent (Le., three-quarters o f eligible households would apply for a benefit).The IUE field research suggests that at the moment the participation rate in the program i s about 60 82 percent. It i s expected to increase in the medium term in response to the growth in real housing costs. G. REFORM SCENARIOSIDENTIFIEDFORSIMULATIONS Macroeconomic scenarios 3.70 Macroeconomic assumptions for our simulations were developed jointly with the experts from the Institute o f Economy in Transition. The basic principles and parameters o f the macroeconomic framework usedfor costingout various structural reforms arepresentedin Annex 3.1. In sum, we took the government's baseline macroeconomic projections for the period 2004-06 and used them as a basis to build a set of four longer-term macroeconomic scenarios, each o f which reflects a specific combination of two primary determinants of Russia's future macroeconomic performance - the average world market oil price and the expected speed o f structural reforms in the country (Table 3.13). Then we used these four macroeconomic scenarios as a basis on which we have designed and elaborated further, more detailed, sub-scenarios that reflect specific reformpackages inthe housing and utility sector. Table 3.13: MacroeconomicAssumptions:Average GDP and RealHousehold Income Growthfor 2004- 06 (Yo) Slow reforms Advanced reforms Moderateoil prices ($18.5), low growth Scenario 1 Scenario2 GDP growth- 2.0 GDP growth - 1.5 Income growth - 8.4 Income growth - 6.4 High oil prices($22.5 and higher), high growth Scenario 3 Scenario4 GDP growth -4.0 GDP growth - 3.5 Income growth - 8.4 Income growth - 6.4 3.71 Overall, we based the analysis on a rather conservative macroeconomic framework. It i s worth noting that GDP growth rates assumed in our scenarios for the period 2004-06 are lower than those assumed in the corresponding government projections. We believe that without advancing reforms growth rates will decline: better utilization of existing reserves in the economy, which was a critical growth factor in 1999-2003, cannot support future growth inthe same way as before because the reserves are to a large extent exhausted. At the same time, the "advance reforms" scenario implies that reforms are likely to temporarily slow down GDP growth compared to the "no reform" scenario, other things being equal. Therefore, in this case, growth rates are also likely to be lower for the next few years than those assumedby the government. 3.72 The four baseline scenarios couldbe summarizedas follows. I.Moderate oil prices and slow reforms. This scenario presents the most difficult macroeconomic environment, with low budget revenues, a relatively high expenditure burden, and a deteriorating balance of payments that triggers inflation and exchange rate pressures and damages investment expectations. This i s the scenario with low growth inboth investments and household incomes. Lack o f reforms would mean that cross-subsidization of households would remain intact. 11. Moderate oil prices and advanced reforms. Under this scenario the macroeconomic environment for reforms remains rather difficult, with low government revenues and depressed earnings in the real sector. However, real devaluation of the Russian ruble impliessome protection o f domestic producers from international competition. This in combination with the reform momentum could be favorable for investments and 83 growth in sectors with a highdegree o f processing. The reform o f natural monopolies and tariff reforms would lead to a gradual increase in electricity and gas tariffs for commercial consumers and a sharp increase intariffs for households. 111.Highoil prices and slow reforms. This is aninertial scenario, under which the existing structure of the economy remains basically intact with the energy and other primary sectors generating most of the investments, taxes, and exports. As with the first scenario, cross-subsidization of households i s preserved. IV.Highoil prices and advanced reforms. This is the most optimistic scenario. It implies a deep economic restructuring and would be more sustainable in a longer term economic growth backed by vigorous structural and institutional reforms. This scenario provides for the highest annual growth rates of GDP, investments, and household incomes. However, given the required level of restructuring in the real sector, unemployment would remain relatively high, which would preserve a strong demand for govenunent spending on social protection. 3.73 One o f the core variables inthe model relates to the real growth inhousehold incomes. Inour base case simulations, we assume that the current highgrowth inreal incomes would continue, while the income differentiation would remain unchanged. Inparticular, inline with the latest projections of the Ministry of Economy, the real income growth for 2005-06 in all scenarios i s assumed to be 8 percent per annum. This i s equivalent to an average growth o f 10 percent a year for the entire period 2003-06. At the same time, we did a sensitivity analysis with respect to income growth: an alternative set of estimates was obtained for the scenario, in which real incomes grow at a rate o f 5 percent a year for the period2003-06. Scenarios for the housing and utility sector 3.74 Two baseline scenarios for reforming payment arrangements the housing and utility sector were developed for 2004-06 along the similar lines: 0 Slow reforms, mostly inertial development 0 Implementation of the reformpackage that would include, inter alia, the elimination of the cross-subsidization in tariffs, additional growth in housing tariffs to compensate for earlier under-financing, and encouragement o f investments inenergy efficiency 3.75 The first scenario implies that housing and utility tariffs would grow only in line with the changes in the overall level of inflation, and electricity and gas tariffs. In other words, there i s no autonomous growth in utility/housing tariffs to compensate for earlier under- financing ("depressed inflation") and to create more favorable conditions for rehabilitation in the sector. 3.76 The second scenario provides for higher rates o f tariff growth in the sector, through which it would be compensated for a de facto tariff freeze during the period o f high inflation in 1998-99.We assume that about a quarter oflost revenues duringthat tariff freeze wouldbe compensated through an additional growth inhousing and utility tariffs in2004-06. Moreover, cross-subsidization would be eliminated through the introduction of a single tariff for all categories of consumers in all relevant sectors (power, gas, and local utilities). In addition, some efficiency gains are expected inthe utility sector, first of all through a reduction inheat and water losses. (See the next section for additional details on potential efficiency effects). 84 Scenarios for the tariff policy 3.77 For each of the four baseline scenarios, we have also developed a set of possible policy options with respect to future changes in tariffs and cost recovery. These options for the tariffpolicy derive from differentcombinations ofthe two keypolicy parameters: 1) Level o f cost recovery intariffs for household-three altematives were considered: 0 Tariffs are set to cover 90 percent of the total running housing and utility costs, while the remaining 10 percent i s still financed from the budgetas a direct subsidy to providers o f housing and utility services 0 Tariffs cover 100percent of the total running housing and utility costs 0 Tariffs cover 100 percent o f the total running housing and utility costs, and, in addition, expenditures on major repairshehabilitation o f the housing stock are also included intariffs (these rehab-related costs are estimated to amount to 20 percent o f the current runningcosts). 2) Maximum share o f housing and utility expenditures in household incomes, above which households would be eligible for social assistance through the housing allowance - three altematives were considered: 10 percent, 15 percent, and 22 percent. 3.78 Ultimately, our simulations covered 36 scenarios (4 initial baseline scenarios options for the level of cost recovery in tariffs * * 3 3 options for the share of housing expenditures inhouseholdincome). Table 3.14 summarizes our sectoral reform scenarios. Table 3.14: Scenariosfor Housing and Utility Reforms, Selected for Simulations Parametersof the tariff policy Moderateoil prices High oil prices The maximum Cost recovery in Slow reforms Accelerated Slow reforms Accelerated share of housing tariffs reforms reforms expendituresin household income 10% 90% Option 1-1-1 Option 1-2-1 Option 2-1-1 Option 2-2-1 100% Option 1-1-2 Option 1-2-2 Option 2-1-2 Option 2-2-2 100%+ costs of Option 1-1-3 Option 1-2-3 Option 2-1-3 Option 2-2-3 capital repairs 15% 90% Option 1-1-4 Option 1-2-4 Option 2-1-4 Option 2-2-4 100% Option 1-1-5 Option 1-2-5 Option 2-1-5 Option 2-2-5 100%+ costs of Option 1-1-6 Option 1-2-6 Option 2-1-6 Option 2-2-6 capital repairs 22% 90% Option 1-1-7 Option 1-2-7 Option 2-1-7 Option 2-2-7 100% Option 1-1-8 Option 1-2-8 Option 2-1-8 Option 2-2-8 100%+ costs of Option 1-1-9 Option 1-2-9 Option 2-1-9 Option 2-2-9 capital repairs 3.79 In the process of the simulations of each scenario, the following parameters were estimated for each administrative region o f Russia for the period 2004-06: 85 1. Total costs o fhousing and utility services suppliedto households 2. Numbero f households eligible for housing allowances 3. Share o f allowance recipients inthe population 4. Budget expenditures on financing housing allowances 5. Budget expenditures on financing housingprivileges (lgoty) 6. Budget expenditures on financing housing subsidies to cover costs not covered by housing tariffs (under the options with 90 percent cost recovery intariffs) 7. Budgetexpenditures on financing the repair andrehabilitation o f the housing stock 8. Total budget expenditures related to operations on residential housing 9. Share o f budget spending in financing the total costs o f housing and utility services suppliedto households 10. Average share o f housing and utility expenditures inhousehold incomes 11. Share o f housing allowances inhousehold incomes inthe lowest income groups. 3.80 In addition, for a limited set of selected scenarios, we undertook complementary simulations to estimate the impact of elimination o f housingprivileges. H. ASSUMPTIONS FUTURE ON UTILITY COSTS Adjustment in domesticenergy prices 3.81 The scenarios for reforms in Russian domestic energy pricing remain at the center o f heated political debates. Still, in the gas sector there i s a growing consensus that the longer- term marginal costs of Russian gas are in the interval o f US$36-40, and the domestic gas price, which on average amounted to US$22.7 per 1,000 cubic meters in 2003, should ultimately reach this level. The latest draft of the government reform program suggests a ' US$36 target (VAT excluded) in2006 (through both annual price hikes and an expected ruble appreciation), but similar plans inthe past have failed to be fully implemented. 3.82 Inthe power sector, however, serious disagreements remain even about the direction for a future trend inthe electricity price. A number o f analysts believe that, given the existing over-capacity, liberalization in the sector i s likely to lead to a decline in the average real electricity price. The current draft reform program envisions an increase o f 16-21 percent in real ruble terms in 2004-06, and no need for additional increases in later years when the effects of sectoral reforms are fully utilized. 3.83 Table 3.15 summarizes the assumptions inthis Chapter on the expected tariff growth for gas and electricity. In the most radical scenario, the real gas tariff would be 25 percent higher than in 2002, while the electricity tariff would be increased by one-third. It is worth noting that our macro assumptions provide for a considerable real ruble appreciation (in excess of 40 percent) for the period 2003-04. This means that dollar energy tariffs would almost double. Inparticular, the gas tariff would be higher than US$33 per 1,000 cub m, while in2002 it was less thanUS$19. Table 3.15: Real Growth in Energy Tariffs for Householdsin 2006 Relativeto 2002, Depending on Scenario Inrblterms InUS$ terms 86 Power gas power gas Scenario 1- slow reforms, low oil prices 1.02 1.10 1.45 1.57 Scenario 2 - advanced reforms, low oil prices 1.19 1.16 1.70 1.65 Scenario 3 - slow reforms, high oil prices 1.05 1.16 1.49 1.65 Scenario 4 - advancedreforms, high oil prices 1.34 1.25 1.91 1.78 Assumptions on the elimination of cross-subsidization 3.84 Given the considerable volume o f cross-subsidization, merely increasing the average H U S tariffs to achieve full cost recovery would be insufficient. In addition, a considerable tariff rebalancing i s needed to ensure that tariffs reflect the costs o f service delivery for particular consumer groups. Such tariff restructuring would result in a considerable additional growth in residential tariffs relative to their average expected growth for the economy (see Table 3.16). 87 Table 3.16: AdditionalGrowth in ResidentialTariffs to Ensurethe Elimination of Cross-Subsidization Additional growth in residential tariffs Electricity 1.23 Gas 1.14 Water 1.45 Heating 1.13 Housingmaintenance 1.oo Average for the sector 1.15 Sources: IUE, staff estimates. Assumptions on potentialefficiency gains 3.85 The Russian housing stock i s quite inefficient and there i s considerable room for energy savings. For instance, the water consumption of Russian urbanhouseholds i s estimated to be more than 70 percent above the WesternEuropeanlevel (World Bank, 2003a). However, the realization o f most of these efficiency gains requires fundamental institutional changes in the sector, which could not be achieved in the short to medium term.55 Inthe period up to 2006, under the most optimistic assumptions about the pace of sectoral reforms, only a small portion o f these gains could be realized. This i s due to the existing institutional and technological constraints that would hamper the introduction of energy and resource saving technologies, especially at the household level. The situations regarding heating, water, and electricity are discussed immediately below. 3.86 Heating. Considerable savings inheating are technically possible but would require a major change in how Russian multi-apartment housing units are managed. In short, most savings couldbe achieved at the buildinglevelbut not at the apartment-level, To realize these savings, apartment owners must (i) transform the current means o f housing management into condominiums or similar, (ii) be capable of contracting and supervising efficient managing companies for their multi-unitbuildings, and (iii) invest in the proper metering of delivered heat. The experience o f the past 10 years suggests that the introduction o f new management arrangements has beenquite slow. 3.87 Water. The situation with water metering, which i s a pre-condition for incentives to save water, i s even less advanced than that for heating. And, because of technical constraints, water metering at the apartment level is not cost efficient. The efficient solution for metering has to be at the buildinglevel, which has the same limitationrelated to the need to introduce a collective form of housing management such as condominiums. In addition, water utilities in most cases are not interested inthe installation o f meters, because the current billing i s based on norms that usually exceed actual household consumption. The introduction of meters would reduce the revenues o f water utilities. 3.88 Electricitv. In contrast to water and heating, households have better opportunities for savings in electricity. Most housing units have power meters and it i s much easier for households to save electricity than to save heat or water. Thus, the price elasticity o f .''See also Eismont et al. (2003). 88 electricity demand i s higher than that for other utilities. Still, at the household level we do not expect substantial electricity savings for the period up to 2006. This i s due to the following: 0 Expected growth in the ownership of household appliances, driven by the continuation o f a fairly strong growth inreal household incomes 0 Growth inthe installation o f electrical ovens, especially innew apartments 0 Improvedquality o f the maintenance of the housing stock (better lighting, etc.) 3.89 Overall, a combination o f two opposite factors - energy savings due to increased tariffs and growth in demand due to higher real incomes - would keep the absolute level of electricity consumption inthe residential sector basically constant. 3.90 Therefore, it i s assumed in the scenarios with advanced reforms that most efficiency gains in the HUS would be realized not within the household sector but by service providers inheating and water supply. Inparticular, it is expected that by 2006 it would be realistic to achieve the following main improvements: 0 As a result o f investment in network modernization and improved management, the current level o f water losses could be reduced by 25 percent 0 Investments in energy savings could reduce unit electricity consumption in water supplyby 12percent 0 The reductioninheat losses inheatingdistribution couldreach 10percent 0 Improved energy efficiency in heat generation (mostly through replacement/modemization o f boilers) could amount to 10 percent 3.91 Overall, we assumed that owing to efficiency gains in the reform scenarios, by 2006 the total unit costs in the HUS could be reduced by 10-12 percent. In the non-reform scenarios, however, it i s assumed that no reduction inunit costs would take place because of the efficiency gains. Table 3.17: Effecton Costs of Efficiency Gains in ReformScenarios Percent in decline inreal unit costs due to energy savings Scenario 2 - advancedreforms, low oil prices 10 Scenario 4 - advancedreforms, high oil prices 12 Source: IUE. Assumptionson the effect of delayed inflation(under-financing) 3.92 Inour simulations we also assume that the non-energypart ofhousing costs should be adjusted upwards additionally to compensate for the period o f the de facto tariff freeze inthe late 1990s. It i s assumed that this compensation would amount to a quarter o f the gap between CPI growthand HUS tariff growth in 1998-2002. Summary ofthe assumptionsfor H U S costs 3.93 Based on the above, we quantified four scenarios for the different cost dynamics o f service delivery in the Russian housing sector, reflected as changes in the real costs of maintaining and servicing an average 1 sq. m o f residential housing. Figure 3.1 presents a 89 comparison of housing and utility cost increasesunder the different scenarios. It suggests that by 2006 in the most advanced scenario--scenario 4 n o s t s to residents would increase by almost 50 percent. It i s worth noting that even in the inertial scenarios without reforms (scenarios 1 and 3), housing costs would also be growing in real terms, reflecting the continuation of the (although slow) domestic energy price adjustment. Figure 3.1: Housing and Utility Coststo Consumers under Different Scenarios in 2006 (real growth r- relativeto 2002) Growthin costsrelative to 2002 4 3 I 2 1.40 1 1 0.0 0.5 1.0 1.5 2.0 1- slow reforms, low oil prices Scenario2 - advancedreforms, low oil prices Scenario 3 - slow reforms, high oil prices Scenario 4 - advancedreforms, high oil prices 3.94 This expected growth in US costs is a combined effect of the following four core factors (Table 3.18): (i) an increase in domestic energy prices (gas and electricity), (ii) an increase in non-energy costs related to the need to compensate the sector for a period o f suppressed information, (iii) the elimination of cross-subsidization, and (iv) the realization of some efficiency gains. Table 3.18: Various Factorsof Growth inUnit HousingCosts, AccumulatedGrowth for the Period Relativeto 2002 Scenarios Total costs Power Gas Non-energy Cross- subsidization Savings 1 1.02 1.oo 1.03 1.oo 1.oo 1.oo 2 1.40 1.17 1.08 1.32 1.15 0.90 3 1.04 1.03 1.08 1.00 1.00 1.00 4 1.47 1.32 1.17 1.32 1.15 0.88 Source: Staffestimates. 3.95 In addition to accounting for these four factors, the implementation of housing and utility reforms implies that the existing tariff structure should be further adjusted to incorporate an adequate level o f investment costs in the sector. Based on the analysis undertaken by the IUE, this would require an additional increase in average tariffs by 30 90 percent. This would increase our estimate for the overall future growth in tariffs to cover all economic H U S costs inscenario 4 to 90 percent. 3.96 The latter estimate suggests that if the full reform package is implemented the total costs to consumers o f the HUS (including budgets as a source of residual subsidies) of the H U S would increase from 4.8 percent of GDP in 2002 to about 9 percent o f GDP in 2006. Such a change would produce not only a major tariff increase but also a considerable change inthe cost structure inthe sector, inwhichmaintenance fees and capital charges would play a more prominent role than they now play. 3.97 The important caveat to the above estimates derives from the fact that they do not reflect actual changes in H U S tariffs that have taken place in Russia since 2002. Meanwhile, according to Rosstat, in the last two years actual tariff adjustment has been considerable -- real growth inHUS tariffs was 23 percent in2003 and i s estimated to reach at least 13 percent in2004. Overall, the tariff growthin2003-04 amountedto almost 40 percent (i.e. itmade up half o f the tariff adjustment that i s claimed necessary to complete the reforms inthe sector in 2006). 3.98 While the recent pace of tariff adjustments appears to be consistent with the medium term objectives of H U S reforms, there i s a risk that recent tariff growth was not accompanied by adequate institutional changes inthe sector. Without complementary institutional reforms, tariff and cost increases in the sector could become excessive. In this respect, the current weaknesses of municipal regulators are of special concern. Ifnot addressed, these weaknesses may lead to a non-justified growth inthe tariffs o f local utilities, especially water and district heating providers. In addition, without institutional reforms, it would be difficult to expect that sectoral players may become capable o f realizing any non-trivial efficiency gains. I. MAINSIMULATION RESULTS Scenarios for the continuation of the current policies 3.99 Table 3.19 presents the results of the simulations of the group o f status quo scenarios that do not provide for any significant acceleration o f tariff reforms inthe housing and energy sectors. In particular, they assume the preserving of cross-subsidization in tariffs and a relatively low pace of adjusting domestic energy prices. 3.100 Among 12 different scenarios, the one in Column 10 i s the closest approximation of the current government policies and ongoing macroeconomic trends: this i s the scenario with highrates of economic growth and with an unchanged level of cost recovery intariffs of 90 percent. 3.101 Under such assumptions, the total explicit costs (excluding cross-subsidization) of HUS will grow by about 30 percent, to 6.2 percent of GDP from 4.8 percent in 2002. Households will have to absorb practically all o f these incremental costs. As a result, in 2006 households would pay 60 percent more inreal terms for H U S (about 4 percent o f GDP), while in 2002 they paid about 2.5 percent of GDP. Given the high growth of household real incomes in 2002-06 (45 percent for the period), this would increase the average share o f housing costs in total population spending only slightly, from 8.5 percent to 11 percent. Overall, the number o f recipients o f housing allowances would decline considerably, from the 7.5 percentbenchmark to 3.7 percent. 91 3.102 Budget spendingon H U S would basically remain at the 2002 level and would amount to 2.3 percent o f GDP. However, the share of the budgetintotal H U S financing would decline from almost half in 2002 to less than 40 percent. The largest portion of these costs would be related to the continuation of budget investment support (capital rehabilitation). 3,103 However, the above estimates hide considerable longer-term implicit budget costs and liabilities that the sector would continue to accumulate in such a no-reform scenario. First, the volumes o f quasi-fiscal financing (cross-subsidization) would remain significant (1-25percent of GDP). Second, the incomplete tariffreform would delay changes inincentives inthe sector, which would prevent the reduction of current inefficiencies, while the government would continue to accumulate considerable contingent liabilities associated with the deterioration of the municipalhousing stock and the urbaninfrastructure. 3.104 In this group o f scenarios, a further increase in cost recovery in tariffs to 100 percent (column 11) would shift additional costs in the amount of 0.5 percent o f GDP from the government to the population. The share of budget spending in total sector financing would fall below 30 percent. The share of households that receive allowances would grow by 30 percent but would still remain low (less than 5 percent). 3,105 Incorporating the capital repair charges into tariffs (column 12) would produce a major additional reduction in the budget financing o f the sector: total budget costs would decline to 1 percent o f GDP or to 16 percent of the total financing. Inthis case, the population would be expected to pay more than 5 percent of GDP inH U S charges (i.e. more than twice as much as in 2002). The number of housing allowance recipients would increase to 7.4 percent of the population. 3.106 The main conclusions drawn from simulating the continuation o f the current policies could be summarized as follows: The continuation of the current policies provides for only a gradual shift of responsibility for housing financing from the government to households, while keeping budget liabilities in the sector at the current level o f about 2.3 percent of GDP. This fact, however, considerably underestimates the longer term costs of the current policies related to the preservation of non-transparency in financing and be distorted incentives o f sector players. Such inertial policies seem fully affordable to the population, and they do, for example, a reduction inthe eligibility threshold for housing allowances). The sensitivity o f these results i s high relative to the modest changes in the level of residential tariffs (e.g., all other factors being the same, the transition from 90 to 100 percent in cost recovery could increase the number o f recipients of housing allowances by 30 percent. The sensitivity of the results is low relative to the variation in rates of economic growth: in our scenarios with lower economic growth (columns 1-6), we assumed a lower level of domestic energy prices and housing costs, but this effect i s largely compensated by lower household incomes inthese scenarios. 92 3 0 0 W c 10 W 3 0 0 0 0 2 Y) W z Scenarios for the acceleration of housing and utility reforms 3.107 Table 3.20 presents the results for simulations of the group of scenarios that provide for a significant acceleration o f tariff reforms inthe housing and energy sectors, including the elimination of cross-subsidization and a much more aggressive pace o f adjustment in domestic energy prices. However, these scenarios do not assume an elimination of the housing lgoty (privileges). 3.108 We again use as a base case the scenario reflected in Column 10, which assumes that the key current parameters o f housing policy remain unchanged: an eligibility threshold for allowances of 22 percent and a 90 percent cost recovery in tariffs. This simulation suggests that the overall costs o f the housing sector's operations would increase to 9 percent o f GDP (Le., a growth o f almost 90 percent relative to 2002). However, these are the full costs that assume the full elimination o f quasi-fiscal financing. 3.109 Inthis scenario, the budget would be responsible for 37 percent of the total costs (3.3 percent of GDP), while households would have to pay the remaining 5.7 percent of GDP, which i s an increase of 130 percent relative to 2002. This would bring the share o f housing spending in household budgets to 16 percent. About 9 percent of households would become recipients o f housing allowances. Inonly 1 out o f 88 regions the share of allowance recipients would exceed 25 percent. 3.110 Inthis scenario the total costs o f housing lgoty, which grow inline with growth inunit housing costs, would reach 0.9 percent of GDP. This would amount to 30 percent of total budget spending inthe sector. 3.111 A further shift o f the housing cost burdento the population (as reflected inColumn 11) seems to be fully affordable as well. Under the 100 percent cost recovery, the share of the government intotal housing financing declines to 30 percent (2.7 percent o f GDP, namely, 0.3 percent more than in2002). The share of allowance recipients reaches 11 percent, which does not sound as prohibitively high. 3.112 In the most advanced scenario (column 12), which assumes the full incorporation o f capital repair charges intariffs, the share o f the government would fall to 18 percent of total financing (1.65 percent o f GDP, namely, 30 percent below the 2002 level). Households would face out-of-pocket housing expenditures o f 7.35 percent of GDP (an increase o f almost 300 percent relative to 2002). More than 15 percent o f the population would become recipients of allowances, while in 10 regions the share of recipients would exceed 25 percent. The average share o f total housing spendinginhousehold budgets would be close to 20 percent. This i s 2.5 times higher than the 2002.level, but still somewhat below the levels currently common in CEE countries. 3.113 Thus, in the most advanced scenario, the households' out-of-pocket expenditures on HUS would be by almost five percent of GDP above its 2002 level. This is a very large increase indeed. How affordable i s such a burden? Inthe view of the authors o f this report, it could be broadly affordable, assuming the latest trends o f high real income growth are sustained. As discussed in Chapter 4, it i s expected that inthe medium term the real average wage growth inRussia will remain above the rate of GDP growth. This inpart will be driven, as shown in Chapter 2, by the reforms in civil sector employment that should result in a considerable increase in public sector wages. Annual budget spending on (non-military) wages may increase by 1 percent o f GDP under the advanced reform scenarios. Overall, preliminary estimates suggest that about 60 percent of the future out-of-pocket housing 94 spending could be funded through a parallel growth inreal household incomes. More accurate estimates could be obtained when the full information for 2003-04 i s available. Those were the years, during which real housing costs have beenincreasing at a relatively high for Russia pace, but this was mitigatedby a highand broad-based income growth. 3.114 Our further interpretation o f these results i s that a practical policy option for the government for the next two to three years would be some combination o f scenarios reflected in columns 11 and 12 of Table 3.20. That is, regions with higher incomes should follow scenario 12 (with the incorporation of capital repairs charges in tariffs), while those with lower incomes may follow scenario 11 (100 percent cost recovery, but no household responsibility for capital repairs). Such a combination o f policies would keep the government spending in the sector at roughly its current level o f about 2.2 percent o f GDP. On average, households would be spendingabout 18-19 percent o f their budget on HUS. 3.115 An important conclusion from our analysis i s that the acceleration o f energy and housing reforms in the current macroeconomic conditions should not bring any incremental costs to the consolidated government budget. Inthe worst case, the reforms should be budget neutral inthe medium term, and they shouldbringconsiderable savings inthe long term. 3.116 At the same time, it i s worth noting that in the analyzed scenarios while the total budget outlays on the sector would decline, their structure o f budget spending would remain rather inefficient. This i s because, without the elimination o f lgoty, their costs would expand drastically in line with progress on the tariff reforms. Inscenario 12, the costs o f lgoty would amount to 1.2 percent o f GDP, and they amount to three-fourths of the total budget spending inthe sector. This means that phasing out lgoty is critical for any significant improvement in the efficiency ofbudgetary spending inthe sector. 3.117 The low income regions, where the population cannot afford capital repair charges as yet, should be partially supported by the federal government. It i s recommended that the federal government should develop a separate program for investment grants to support the rehabilitation of the housing stock in such regions on a co-financing basis. Funding for such a program could be obtained by phasing out the existing nontransparent programs of housing sector financing as described above. 3.118 The analysis identified two groups o f regions that are likely to be most affected by the proposed cost increases inthe HUS, as follows: a Regions located in the south o f Eastern Siberia and in the Far East. These are the regions with high unit housing costs due to their remoteness and climate conditions. At the same time, these are relatively poor regions, which are significantly dependent upon federal budget transfers. a Autonomous (internal national) republics in various parts o f the country. These are predominantly the rural, least developed regions with a relatively low share o f urban housing that has full access to subsidized and expensive network utilities. Thus, these regions spend relatively little on housing, but their policy in the sector has been traditionally the least reformed and they have had among the lowest levels of cost recovery in tariffs. As a result, the future costs o f residual adjustment to reach full cost recovery in these regions could be quite high. Moreover, these regions are the most transfer-dependent entities in Russia, and this reduces their fiscal room for adjustment to additional policy and price shocks. 95 Additional sensitivity analysis 3.119 Additional analysis was undertaken to check the sensitivity o f the above results to changes in what are seen as the most important parameters in the model: i)the dynamic o f household incomes, and ii)the level ofhousing costs. 3.120 Sensitivitv to changes in incomes. A set of simulations was undertaken with significantly lower rates of real income growth, which would result in 2006 household incomes being 20 percent below the level assumed in the base case scenario above. The simulation showed rather high sensitivity: in the scenario with 100 percent cost recovery (column 11) a 20 percent decline in incomes increases the number of recipients on housing allowances by about 80 percent (to 20 percent of the population). The total budget spending on housing allowances would increase by 140 percent (to 0.57 percent o f GDP). In such an environment an aggressive policy o f tariff adjustments may be much more difficult to implement. 3.121 The high sensitivity of the results to income dynamics suggests that the government should establish an efficient monitoring system to track the affordability of tariff increases for the population in general and for specific household groups. While within the current favorable macroeconomic environment the government should make a strong pushtoward the necessary adjustments in domestic energy prices, the elimination of cross-subsidization, and the achievement of full cost recovery intariffs, thispolicy maybe modifiedincases where the economy faces a major slowdown in household income growth. Some low income regions could be allowed to move with the reforms at a slower pace than the rest of the country. 3.122 If, in the environment with high growth in household incomes, the advancement o f housing reforms could be easily made budget neutral, this i s not case when income growth slows down. The latter may generate incremental budget costs o f 0.4-0.5 percent o f GDP per annum in the medium term, which would be mostly related to the additional financing of housing allowances. 3.123 Sensitivity to changes in housing. costs. Additional simulations assumed are relatively high sensitivity (although lower than in the case of household incomes) - the approximately 20 percent higher level of housing costs. These simulations revealed a number of recipients of housingallowances increasesby 27-31percent depending on the level of cost recovery. 96 2 0 5 o msz z 3 3 0; J. SIMULATIONS FORPHASINGOUT HOUSING PRIVILEGES (LGOTY) Incidence of lgoty 3,124 Various non-cashhousing privileges (lgoty) represent one of the major deficiencies of Russia's housing policy. Lgoty provide their beneficiaries with considerable discounts (usually 50 percent) against their housing and utility bills. Lgoty are category-based benefits, and as such they are an inefficient policy instrumentof social assistance that tends to channel most support to middle-income and high-income groups (World Bank, 2004~). 3.125 Moreover, lgoty i s a fairly non-transparent policy instrument. At the federal govemment level, there i s no single regulatory and legal framework to govern its allocation and financing. The existing system o f lgoty i s quite complicated because of the dozens of government resolutions .(some of them more than 60 years old) that have introduced different eligibility criteria for potential benefi~iaries.~~a result, there are no reliable statistics on the As actual budget and overall costs o f providing lgoty. The most commonly used source of information on lgoty (deriving from Rosstat's Form 26-ZhKH) i s believed to be somewhat unreliable because it i s filed not by lgoty recipients but by providers of H U S (see Annex 3.2 on other issues relatedto the availability o f the data). 3.126 On the basis o f the 2000 Rosstat's data for six regions Misikhina (2002) estimated that about 40 percent o f Russian households benefit from various housing payment privileges. Pensioners and budget sector employees have been the primary beneficiaries. Posarac and Rashid (2002), on the basis o f the 2000 RLMS data, also show that low income households receive a much smaller share than higher income households of having privileges. 3.127 The conventional Rosstatreport, based on Form 26-ZhKH, suggests that in2003 about 43.4 millionpersons (29.9 percent o f the entire population) benefited from different housing payment privileges. This i s a 13 percent decline since 2002, which was a result o f the decision made in May 2002 to monetize housing lgoty for military servicemen. About half of all current beneficiaries receive their lgoty on the basis of the Law on Veterans, which grants considerable non-cash benefits to most Russian individuals with an extended employment history. An additional 30 percent of beneficiaries claim their lgoty on the basis of the Law on Social Protection of People with Disabilities. 3.128 As reported by Rosstat, the total costs o f lgoty in 2003 amounted to about 0.4 percent of GDP (see Table 3.2, above). The average per capita value o f the monthly benefit amounted to Rbl 93.8 or about US$3. However, as was shown inthe earlier section, traditionally, lgoty have been under-financed. In2002-03 the average level of their actual financing was about 70 percent. Thus, the lgoty operate as an implicit tax on sector providers (World Bank, 2003a). 3.129 This Chapter uses newly available information on lgoty from the NOBUS survey, undertaken inthe spring o f 2003. NOBUS i s believed to be a more reliable source o f both the incidence and the costs o f lgoty (see Annex 3.2 on different sources of data). According to NOBUS the share of individuals benefiting from lgoty was 33.3 percent of the population (Le., their overall number was 11 percent higher than inRosstat reports). Moreover, the share of households that benefit from lgoty i s even higher - 41 percent. This i s because the number of beneficiariesi s higher among smaller families, such as pensioners. The analysis by the IUE identifiedthe list of govemment decisions that had established various housing privileges,which is morethan 30 pageslong. 98 3.130 More important, the region-by-region comparison of the results from the two sources shows a systemic bias in the Rosstat data set: this set gives much higher shares of lgoty recipients for poor regions (suchas various Russian national republics) and underestimates the number o f recipients in wealthier regions (regions of the European center). This is an important bias from the perspective of the impact analysis o f reforming the system to phase out the lgoty and provide cash compensation to vulnerable households: simulations that are based on the Rosstat data would produce a much higher demand for cash compensation for the existing lgoty recipients. This i s the reason why this report employs the NOBUS data for the simulationofthe fiscal andsocialeffects relatedto the eliminationoflgoty. 3.131 The Russian Government has initiated reforms inthe lgoty system in 2004. It intends to define more precisely the sources o f financing o f the privileges, and it intends to shift the responsibility for financing a relatively large part o f them to the regional administrations, Moreover, the regional govemments would be allowed to terminate specific privileges if they could not afford their funding. However, to date the govemment has made only those decisions that provide for the phasing out and partial monetization of privileges in public transportation and medical services. There i s no agreed timetable as yet for the reforms o f existing lgoty inthe HUS. Reformscenarios that assumethe eliminationof non-cashhousingprivileges5' 3.132 Table 3.21 presents the results o f simulations of reform scenarios that combine policies reflected in the scenarios in Table 3.20 with the elimination o f housing lgoty (privileges). The results suggest that the elimination o f lgoty would have a very modest impact on the vulnerable group o f households. This reflects the fact that lgoty are concentrated inmiddle-income and high-income groups, and that, therefore, phasing out lgoty would not make housing unaffordable for too many o f their current beneficiaries. 3.133 In the base case (column lo), the elimination o f lgoty increases the number o f allowance recipients by one-third, bringing their total number to 12 percent o f the population. Inother words, only every tenthbeneficiary out ofthe current lgoty beneficiaries (thenumber of which i s close to 30 percent) would become eligible for housing assistance through allowances. This would provide the budget with considerable savings through the elimination o f Igoty: the costs o f additional housing allowances (0.06 percent o f GDP) amount to only 8 percent o f the costs of eliminated lgoty. 3,134 Inthe case o f 100percent cost recovery (column 1l), the above-mentionedproportions hold: the elimination o f lgoty increases both the number o f allowance recipients and the total cost o f the allowance program by about one-third, while generating considerable budget savings. Overall, the scenario for the 100 percent cost recovery and fully removed housing lgoty in 2006 sounds quite affordable: the share o f the population that would receive " The alternative set of simulations to assess the potential impact of the elimination of privileges was undertaken for the Russia Poverty Assessment (World Bank, 2004). Our modeling reflects two important differences relative to this earlier work. First, it uses 2002 (not 2000) as a base year for the analysis. As a result, our estimates reflect a major recovery in householdincomes that took place in 2001-02. Second, we stimulate the phased-in reforms with tariff increases being expanded through three years, which provides household incomes with additional room to catch up with incremental costs. Overall, our estimates suggest that the elimination of the lgoty would generate an even smaller increaseinthe demandfor housing allowances than was reportedinthe PovertyAssessment. 99 allowances i s only 15 percent, and the average household pays about 14.5 percent of its income for HUS. 3.135 Our interpretation o f the latest set o f simulations i s as follows: if the current trend of high growth in household incomes continues, then the elimination of housing lgoty in about 2006 should not be a social/political problem. This i s even more so if, as currently planned, the government would succeed in advancing other structural reforms, including those in public administration, pensions, and the social services, which would result in an additional growth inpublic sector wages, old age pensions and childbenefi@. 3.136 With respect to sequencing the necessary reforms, this report argues that, given the sensitivity o f the lgoty issue, the entire set o f reforms could be dividedinto two main groups: 1. 2004-06: Accelerate tariff and financing reforms to increase tariffs, make the financial flows in the sector more transparent, and increase the accountability o f sectoral players, including municipalities and service providers. As a part o f this process, the Government should make a major effort to monetize the lgoty to transform them into explicit budget subsides and link them directly with the new system o f individual social accounts that would track all benefits and entitlements for which households are eligible. 2. 2006-07: Eliminate lgoty (probably with a partial cash compensation to particular categories o f recipients). The specific strategy for phasing out lgoty would depend upon the actual progress in housing and other reforms, the basic proportions between the prevailing levels of household incomes and housing costs, and the capabilities o f the social assistance system. Additional modeling of housing affordability would be justifiable at that stage. '*Yasin (2003) provides a detailed discussionof linkages between changes in public sector wages and benefits and affordability of tariff increasesinthe HUS. 100 b y o 0 W 0 N N q o o f 0 2m 2 0 v , s 10 e.0 0 m c b 0 y o 0 W 0 v, Nv, " y o d 0 0 d N y o 0 M 0 o\ croo 3 3 0 O W r0?0 O ? b + 9 0 0 00 e W 9 0 0 In N 0 In x b W y W m K. CONCLUSIONS ANDPOLICYRECOMMENDATIONS 3.137 Despite more than 10 years of reform efforts, the housing and utility sector remains among the least reformed segments of the Russian economy. It i s increasingly becoming a bottleneck in the country's overall development strategy. The HUS i s too dependent on government support, too costly to the budget, and too politicized and non-transparent, while the quality o f the provided services remains low. 3.138 At the same time, when compared with the previous decade, the current economic environment inRussia i s much favorable for advancing the H U S reforms. This i s the case for the following reasons: - Russia's macroeconomic performance i s much stronger and it provides all major players with an additional income space for addressing the potential short-term costs o f reforms. - Recent improvements in cost recovery indicate that a significant portion o f the entire task of phasing out inefficient housing subsidies has already been accomplished; what remains i s a much more manageable tariff increase relative to what was a major challenge in2000. - Russia's federal government has a much stronger influence over subnational developments and could be more proactive in setting up proper incentives for regions and municipalities to advance the HUSreforms. - Russia's private sector has shown for the first time a real interest in entering the housing sector. 3.139 Our analysis suggests that the acceleration o f cost recovery increases and other H U S reforms in the current macroeconomic environment i s not supposed to bring any significant incremental fiscal costs. From the fiscal perspective, the main immediate challenge o f the reforms in housing is not about changing the current level o f budget financing o f the sector but is about the radical restmcturing of the existing financing mechanisms inthe direction of the following: - A reduction ingovernment involvement inthe HUS operations and financing - Improved transparency inthe residual government commitments inthe sector - A strengthening of the accountability of the government for the full financing of its commitments - Improved transparency o f the financial flows inthe sector - A strengthening of the accountability of both municipalities and service providers for - the use ofbudgetfunds and the quality of services Improved targeting of budget support to vulnerable households and depressedregions 3.140 Our estimates suggest that, as a result o f tariff adjustments and other reforms, the real unitcost inthe HUS would increase by about 90 percentrelative to the prevailing 2002 level. This would bringthe annual costs of operatingresidentialhousing to 9 percent of GDP. At the same time, given considerable tariff adjustments that already took place in 2003-04, much a smaller additional growth in tariffs (of about 40 percent) i s needed to reach the levels that cover full economic costs inthe sector. 3.141 This Chapter argues that in the current environment o f high growth in household incomes, it i s affordable (both politically and financially) to attain, by 2006, a policy target o f 102 100 percent cost recovery in tariffs with the simultaneous elimination o f all quasi-fiscal financing (cross-subsidization). In addition, in the wealthier regions it could be possible to incorporate capital repair charges intariffs. Insuch a scenario, households are expected to pay 6.5 percent o f GDP, a major increase relative to 2.5 percent in 2002. Government budgets would pay the residual 2.5 percent of GDP (which i s roughly the same level o f budget spending as in2003). 3.142 The simulations also suggest that, in the current environment, with high growth in householdincomes, advancing reforms inthe residential housing couldbe easily made budget neutral in the medium term, and this would bring considerable savings in the long term. However, if household income growth slows down, this may generate incremental budget costs o f 0.4-0.5 percent o f GDP per annum in the medium term, which would be mostly relatedto the additional financing o f housing allowances. 3.143 Overall, the analysis revealed a high sensitivity o f the results to income dynamics. This suggests that the government should establish an efficient monitoring system to track the affordability of tariff increases for both the population in general and specific household groups. A policy of aggressive push toward more justifiable domestic energy prices and H U S tariffs may be modified if the economy faces a major slowdown inhousehold income growth. Some low income regions could be allowed to move with the reforms at a slower pace than the rest ofthe country. 3.144 The analysis suggests that inmost regions attaining the full cost recovery intariffs and phasing out cross-subsidization in three years would be possible without a drastic expansion in the share of housing allowance recipients. The average share of households that would become allowance recipients might reach 11 percent in the reform scenarios with the 100 percent cost recovery in tariffs. Inthe regions with high income differentiation, an additional mitigation measure could include a reduction in eligibility requirements for housing allowances, e.g., from the current 22 percent o f household income that i s spent on HUS to 15- 18 percent. 3.145 From the financing perspective, the following steps constitute the core policy measures necessaryto support the accelerationo f HUSreforms. At the subnational level: 0 An increase intariffs to reachfull cost recovery 0 The phasingout o f direct subsidization of service providers 0 The expansion ofthe program o fhousing allowances At the federal level: 0 The phasingout o f the federal programs o f indirect financing o f the HUS 0 The establishment o f a federal program o f investment grants to support subnational rehabilitation projects inutilities, to be funded onthe basis o f co-financing 0 The expansion o fthe scale o f federal budget assistanceto regions inthe Northand Far East (e.g., by establishing a respective federal investment program) 3,146 An increase in energy and utility tariffs would make the delivery o f utility services to budget organizations more expensive. Marginal annual costs are conservatively estimated to amount to 0.7-0.8 percent o f GDP. These costs would be mitigated by the ongoing reforms in the public sector, which, inter alia, would bring about some savings (e.g., through the consolidation o f the existing budget institutions inhealth and education. Inaddition, the tariff increases would increase the amount o f taxes to be paid by energy and utility firms. However, we estimate that inthe medium term a fiscal gap inpublic sector financing o f 0.4 percent o f 103 GDP could emerge as a result of tariffincreasesinHUS and energy. Inaddition, to realize the expected tax gains, further strengthening of the tax administration may be required, in particular inthe energy sector. 3.147 The analysis suggests that the elimination of housing privileges could also be affordable for most of the current lgoty recipients, while the housing allowance program would be capable o f taking care of those who face too high a housing cost burden. This i s primarily due to the fact that lgoty are concentrated in the middle income and high income groups o f the population. However, given the political sensitivity o f the entitlement reforms, there may be a case for postponing drastic reforms in lgoty untilmore basic H U S reforms are more advanced. To reduce political costs, phasing out lgoty should be coordinated with other reforms, including wage increases inthe public sector, and increases inreal old age pensions and child benefits. To better understand the interlinkages between the reforms, additional modeling may be needed. In parallel, the government has to strengthen the accountability mechanism for the financing of the remaining lgoty. 3.148 In addition, to support the above reforms in housing financing, additional structural reforms should be facilitated. This reform package is fairly well known and i s described elsewhere. Its important components include, but are not limitedto the following: 0 Additional reforms ininter-governmental fiscal relations 0 An improvedenvironment for private sector participation inthe HUS 0 The depolitization and increasedtransparency ofthe tariff settingprocess 0 The strengthening of federal monitoring over reformprogress at the subnational level 0 The disseminationof the best regional and municipal practices for H U S reforms. 3.149 It also seems appropriate to provide additional fiscal incentives for subnational governments that are ready to accelerate the H U S reforms. The experience o f the Fund for Regional Fiscal Reforms, which provided federal budget transfers to regions that were leaders in reforming their fiscal systems, could be replicated to facilitate US reforms at the subnational level. 104 Chapter 4. IMPLICIT FISCALRISKS INTHE RUSSIAN PENSIONSYSTEM 4.1 This Chapter estimates potential fiscal costs associated with various developments in Russia's pension system based on the comprehensive actuarial model. It finds that such fiscal costs are likely to emerge as a result o f the declining relative value o f old age pension and associated political pressures for budget support to the pension system. Without additional reforms, the existing pension system, even under the most optimistic assumptions, i s not capable o f closing the growing gap between growth in wages and pensions. In the baseline "without the reforms" scenario the average replacement rate declines from 33 percent in2002 to 24.4-27.8 percent in 2030. Moreover, the proposed cuts in the contribution rates would result in a further decline in the replacement rate relative to the baseline. To avoid a drastic widening in the gap between wages and pensions, a reduction in contribution rates has to be supplemented by additional reforms, including a decision on a gradual increase in the retirement age. A. INTRODUCTION 4.2 This Chapter is based on the findings of the earlier World Bank Report (2003~) "Pension Reform in Russia: Structure and Implementation" and it focuses on a financial trends in Russia's pension system, including the pay-as-you-go (PAYG) and funded pillars, within the framework o fthe reform o funifiedsocial tax. 4.3 The Chapter develops a multifactoral analysis of potential fiscal costs associated with the future developments o f Russia's pension system. The analysis i s based on a set o f alternative formulations for the future pension reforms. It utilizes the comprehensive actuarial model as a tool of simulations of the pension system trends.59The analysis makes an emphasis on the sensitivity o f the current trends in the pension system to changes in macroeconomic performance, to the changes in demographic trends, as well as to various combinations o f potential reform measures. It provides an analysis o f the potential impact o f proposed cuts in the unifiedsocial tax (UST) rates and increases inthe retirement age. The Chapter highlights both medium- and long-term trends inthe pension replacement rate, as well as on the changes inthe purchasingpower ofthe averagepensionbenefit. 4.4 The Chapter is using the level of the replacement rate (ratio of average pension to average salary) as a core policy variable for the pension reform analysis. This i s because such a ratio reflects a basic proportion between the pension and pre-retirement income o f pensioners, i.e., it characterizes income losses associated with retirement. In the OECD countries, the pension system replacement rates vary, ranging from 36 percent in the USA to 49 percent in Finland (World Bank, 2003~).However, sustaining the replacement rate above 40 percent has increasingly become a policy target in the developed countries. The World Bank experience generally indicates that in the mandatory pension system, for a typical full career worker to maintain a subsistence income in retirement, an initial target for a replacement rate i s likely to be around 40 percent (Holtzmann, 2004, p. 33). 4.5 In addition to the replacement rate, the analysis in Chapter 4 pays a considerable attention to the dynamics o f the ratio between average pension and subsistence minimum.It is 59 Andrews (2001) and Anusic and Petrina (2003) provide the examples of broadly similar analysis of sustainability ofthe pensionsystemsunder different reform options for other CIS countries. 105 justified by the fact that, when the average pension i s either below or close to the pensioner's living subsistence level, as it is in Russia, the value o f the replacement rate can not be an informative indicator o f changes inpensioners' real incomes. 4.6 The fiscal cost analysis in this Chapter i s undertaken for an expanded set o f macroeconomic and reform scenarios. The basic group o f scenarios reviews the trends in the current pension system and analyzes their sensitivity to the core macroeconomic and demographic factors. Additional scenarios focused more specifically on the potential impact o f such changes as cuts inthe contribution rate, increase inthe retirement age, and changes in certain pension rules, such as a reduction inthe existing USTbenefits for the self-employed. 4.7 This Chapter does not discuss either the issues related to the policies on social pensions, which at the moment are received by 1.5 million pensioners (3.8 percent o f the total number o fpensioners) who are not eligible for a regular old-age pension. Social pension i s the parameter o f the government's social protection policy and it has to be dealt with separately from the issues related to social insurance and reform in the old-age pension system. Respectively, financing o f social pensions should be de-linked from payroll taxation and instead be included in the overall government budget expenditure on social protection policies. Overall, we agree with the view that the govemment has to adopt a strategy for a gradual increase in the social pension from its current low level o f 55 percent o f the minimum subsistence to the level o f pensioners' subsistence minimum(Smirnov and Isaev, 2003). B. BACKGROUND 4.8 Currently the Russianpension system faces the challenge o f establishing a sustainable longer-term framework for a gradual increase in the level o f pension benefits. To this end, as part o f the comprehensive pension reform, the govemment introduced a multi-pillar system in early 2002. The move to a multi-pillar system in Russia follows recent similar reforms in several other economies intransition, including Hungary, Kazakhstan, Latvia, and Poland.60 4.9 Two out o f three main components o f the reformed pension benefit in Russia -- notional defined contribution (NDC) and mandatory funded -- directly depend o n the size o f actual pension contributions (a portion of the unified social tax). Inaddition, the reforms that have been introduced made a major step toward simplification o f the benefit formula and more transparent eligibility criteria. This i s expected to improve incentives for compliance, cut down the unreported and untaxed share o f the payroll, and therefore improve the revenue performance of the pension system. However, so far the informal part o f the payroll has been shrinking only slowly. This is believed to be largely a result o f rather a high rate o f established contributions of 28 percent. Inaddition, the incentives for proper payroll reporting were further weakened because only half o f the paid pension contributions (14 percent o f payroll) were taken into account to assess the future size o f benefits, while another half was reserved for financing the basic pension, which i s the same for all pensionersn6lAnnex 4.4 provides some details on the main benefit rules inthe current pension system. 4.10 At the same time, the Russian Government set up a strategic objective of easing the tax burden on the economy, which would include reduction in U S T rates. It i s expected that 6oHoltzman et al. (2004) presents a summary of the lessons from the early pension reforms in ECA. See also Rutkowski (2002, 2004). 6'This was changed by the decision made earlier in 2004 (and which will become effective in 2005). These recent changes are discussedin detail below. 106 the tax rate reductionwould helpbroaden the tax base for pensioncontributions, thus partially compensating for the direct loss from the UST rate cut. Under the circumstances, there i s significant interest inthe analysis of potential fiscal implications of both the current trends in the pension system, as well as the changes that would be generated by various reform proposals, such as the proposedcuts inUST rates. 4.11 The Russian Government has been considering improvements in pensioners' welfare through a growth in real pension benefits as its core longer term policy objective. It also considers a comprehensive pension reform, which provides for an introduction o f the fully funded pillar, as a primary instrument o f achieving this objective. At the same time, as a part of its macroeconomic and growth policy, the Government intends to reduce the average UST rate, which i s likely to bring about a decline in pension contributions, at least in the short term. 4.12 Despite recent steps to reform the social insurance system, the current situation with the old-age pensions inRussia, similar to other pre-reform pension systems incountries of the former Soviet Union, i s characterized by the low retirement age (60 for men and 55 for women), highcontribution rates, high system dependency ratio, almost universal coverage for the current retirees, and the flat structure of benefits that are not linked to the contributions.62 There i s some evidence, however, that participation rates have been declining among employees, but in Russia this trend has been weaker so far than in some other countries in tran~ition~~. There is also a highnumber (almost 30 percent o f the regular old-age pensioners) of pensioners who benefit from various early retirement schemes, determinedby special laws inherited from the Soviet era which have so far remained largely ~ n r e f o r m e d ~ ~ . 4.13 Basic parameters of the current pension system inRussia are presented inTable 4.1. It shows a rapid growth in the system dependency ratio owing to the ageing population. Moreover, in the 90s the ratio between the number o f pensioners and employees (60 percent in 2002) had been growing even faster than the share of pensioners inthe entire population (27 percent in2002). This i s owing to an increased incidence o f both unemployment and non- participation in the labor force65. Annual contributions to the pension system amount to approximately 6 percent o f GDP. 4.14 At the same time, it is worth noting that in the short term the broader demographic structure o f the Russianpopulationhas a less unfavorable impact on pension financing. This i s because the recent increase o f older people's share in the population was accompanied by a drop inthe birthrate, and therefore the overall share o f people o f non-labor age (children and retirees combined) in the total population did not increase. According to the Rosstat, in 2002 60.8 percent o f Russians were o f labor age, while this share was 57.0 percent in 1989 and 57.3 percent in 1997. This means that inthe short term sustaining a pension system inRussia may not necessarily require an immediate increase in the overall tax burden on the working population. However, some restructuring of the overall public taxation and spending may be needed to reflect the ongoing demographic changes. Still, in the longer run Russia's pension 62De Castello-Branco(1998) provides a general overview of the main features of pension systems in the FSU economies as they were inherited from the plannedeconomies. 63InRomania,for instance, the number of contributorsto the pension system declined from 8.2 million in 1990 to 4.4 million in2003 (Plevko 2004). 64The new draft Law on professionalpensionschemes is currently under considerationinthe Duma. 65It shouldbe noted, however, that despite the growing pressures, the Russianpension system i s still in a more favorable position relative to one e.g., in Ukraine, where the ratio of pensionersto employees reached0.92 and overall pensionspendinghavebeenapproaching 10percent of GDP (Anusic and Petrina, 2003). 107 system will have to deal with the same set o f fiscal problems associated with the ageing population as most developed and middle income countries. Table 4.1: BasicParametersofthe RussianPensionSystem, 1995 2002 - 1995 2000 2001 2002 Number of pensioners(end year), million 37.08 38.41 38.63 38.43 olw: old-age pensioners 29.01 28.81 28.99 29.70 Share of pensionersinpopulation, % 25.1 26.5 26.8 26.9 Pensionsystem load (ratio of pensioners and employed) 0.56 0.60 0.60 0.60 Average monthly pension,including compensations (1995 - 188.1 694.3 1,024.1 1,378.5 `000 rubles; after 1998-rubles) O/w: old-age pension 258.5 894.0 1,254.9 Minimumpension,rubles 89.6 427.8 474.1 522.4 (*) Replacementrate (ratio of averagepensionto average 40 31 32 32 monthly salary), % Ratio of averagepensionto the pensioner'sliving 101 76 90 100 subsistenceminimum, % Ratio of averagepensionto the minimum pension, % 210 162 216 264 Employed inthe economy, million 66.44 64.33 64.70 65.40 Economically active, million 70.86 71.46 70.97 71.90 Economic activity rate, YO 48.0 49.4 49.3 50.2 Average (reported) monthly salary(1995 `000 rubles; after 472.4 2,223 3,240 4,360 1998 rubles) - - Taxable payroll amount, estimated(billion rubles) 377 1,716 2,516 3,422 - taxable payroll amount, as % of GDP 24.5 23.5 28.1 31.6 Total payroll amount, incl. informal wages but excluding n.a. 2,304 3,141 n.a. payrolltaxes (billion rubles) - totalpayroll amount, as % of GDP n.a. 31.5 35.1 n.a. Total collections by the PensionFund, as % oftaxable 19.6 22.1 19.7 18.7 payroll - collections as % of GDP 4.8 5.2 5.5 5.9 Pensionexpenditurebythe PensionFund, as % of GDP 4.4 5.2 5.9 PensionFundbalance, as % of GDP -0.03 1.24 0.19 -0.83 Memo: Total population (endyear), million 147.60 144.80 144.00 143.10 GDP, billion rubles 1,541 7,306 8,944 10,834 (*) -Base componentof the oldagebenefit. Source: Rosstat, Pension Fund. 4.15 The current pension system provides for a relatively highnominal replacement rate of about 32 percent66.However, the purchasing power o f the average pension i s low, because it merely exceeds the pensioner's living subsistence minimum.According to the Pension Fund, due to a real growth in pensions in 2002-03, by August o f 2003 the average pension reached 108 percent o f the subsistence minimum6768. This contradiction between the highreplacement 66Here and everywhere in this Chapter the replacementrate i s estimated(according to the establishedtradition) on the basis ofthe average gross wage. A more accurate way to estimatethe replacementratewould be based on the net wages (wages after personal incometax). This would result inthe replacementrate estimates that are 13 percent higher. 67Kabani and Sajaia(2004) argue, however, that the current official subsistenceminimum inRussiais inflated and in the case o f pensionersexceedsby about 25 percent the true cost of the consumer basketthat covers both minimum calorierequirementsandbasic non-food needs. 68 Monetarization of some non-cashbenefits, introduced in early 2003, resulted in some additional increase of the replacementrate, but did not change muchthe ratio betweenaveragepensionand subsistenceminimum. 108 rate and low pension purchasing power i s explained by a large proportion o f informal wages inthe economy that are not taken into account incalculations of the average reported salary size. Informal wages are not reflected inthe conventional estimates o f the current replacement rate that are significantly upwards biased and somewhat misleading (especially for those in higher income brackets). 4.16 Under the circumstances, inaddition to the values of a nominal replacement rate, it i s worth looking at the estimates o f a real replacement rate (ratio of the average pension to average real salary, i.e., salary that includes both formal and informal nontaxable wages). Based on the Rosstat publications of national accounts, we estimated that the share o f informal wages in 2001-02 amounted to about one-third6' o f the total labor earnings and to about a half of the taxable payroll amount (Table 4.2). Respectively, the real average replacementrate was one-third lower than the nominal one and amounted to only 22 percent. Table 4.2: Structureof the totalpayroll, as YOof GDP ~ 1997 1998 1999 2000 2001 2002 Total payroll, as % of GDP 50.0 47.2 40.6 40.2 43.2 -- payroll taxes 11.0 10.6 8.3 8.7 8.1 8.0 all wages 39.0 36.5 32.2 31.5 35.1 olw: taxable wages 27.3 26.4 21.2 20.4 23.9 24.5 hiddenwages 11.7 10.1 11.0 11.1 11.2 Share oftaxablewages inall wages, % 70.0 72.3 65.8 64.8 68.0 Other non-transferincomes,as % of GDP 14.8 14.9 14.8 12.6 12.1 Source: Staff estimates based on Rosstat 'spublications on National Accounts. 4.17 The prevailing statutory contribution rate is 28 percent, which is highrelative to most OECD countries but much lower than in the transition economies of Central Europe (IMF, 2002). The 2002 effective UST rate has been 18.7 percent of the reported (Le., without informal wages) payroll amount, and it shows a steep decline relative to 22.1 percent in2000. This i s owing to two primary factors: (i) introduction of lower UST rates for higher wages the in200170;and(ii) growing incidence ofUSTbenefits for those employedby SMEs. the 4.18 The high UST rates are usually considered among the primary reasons for a high incidence of informal wages in the economy. Indeed, World Bank experience indicates that mandated contributions in excess o f 20 percent are likely to be quite detrimental for compliance inmiddle and high-income countries (Holtzmann, 2004, p. 34). 4.19 Since the introduction o f the flat PIT rate o f 13 percent in Russia in 2001, there has been a striking difference in trends between UST and PIT tax bases. Nominal growth in the PIT tax base was twice as high in 2000-03 than the one for UST - 106 versus 59 percent (Figure 1). The response to a cut inaverage income tax rate was quite strong, and a noticeable share o f personal incomes was legalized, especially in2002-03. While the marginal UST rates were also reduced in 2001, this cut was much smaller and apparently had a weaker incentive effect. As a result, the growth inthe PIT tax base occurred primarily due to the legalization o f 69 It i s worth noting, however, that the share of informal wages declined somewhat since the middle of the 90s as aresult ofthe post-1998 fiscal stabilization and phasingout non-cashsettlementsinthe economy. 'O Since the beginning of 2001 the portions of the wages that exceed 100,000 Rbl a year (about $3,500) have beentaxedat lower (steeply declining) rates. See also Annex 4.4. 109 non-wage incomes, especially among those in top income deciles for whom a substantial share o f total income comes from non-wage sources (Sinelnikov et al., 2003). This observation provides indirect support in favor o f further cutting the existing U S T rates, and may give additional backing for an assumption that future cuts in U S T rates could be accompaniedby some expansion inthe tax base. 4.20 One o f the problems with the existing UST regime relates to its fragmentation, with a considerable proportion o f the labor force being eligible for significantly reduced UST rates. As can be observed from Table 4.3, in2003 the share o f such employees amounted to at least 16 percent of the total number o f UST taxpayers, and on average they paid three times less in their pension contributions than the "regular" employees. The actual number o f beneficiaries o f the special U S T regimes i s even higher, inparticular because the data in Table 4.3 do not cover those who are eligible for a single agricultural tax. The Pension Fund staff estimates that the overall share o f those who benefit from preferential U S T rates amounts to 19 percent. The main beneficiaries o f these tax privileges are farmers and employees in small firms, who are eligible for a 50 percent reduced UST rate (Box 4.1). These employment groups do not make any contribution to their base pension, and therefore they are implicitly cross-subsidized by regular UST payers. Figure4.1: Trends in tax bases for unifiedsocialand personalincomestaxes, 2000-03, trillion rubles SST and PIT tax bases, trln rbl 1 2000 2001 2002 2003 Box 4.1: PreferentialUST regimesfor employees of smallbusiness According to the RussianTax Code (Article 18), there are three specific tax regimesthat de facto provide considerabletax benefitsfor specific employment categories: e Simplified taxation for SMEs e Singletax on imputed incomes for businessesengaged inspecific types of economic activities e Single agricultural tax for agricultural producers Inall these cases, the eligible businessesbenefit from the overall lower taxation rateand simplifiedtaxation structure, under which severalmajor taxes are replacedby a singletax on revenues or imputed incomes.As a part ofthis benefit package, SMEs are eligible for lower USTrates, which include: (i)a 50 percentdiscount on USTtax on employeeearnings(fullexemptionfor contributions to the base pensionbenefit); and (ii)avery low fixed paymentof 150ruble a month for employers and self-employed.Inaddition, agricultural producersbenefit from further reducedUST rates (10.3 percent versus 14 percentfor other SMEs and28 percent for the rest ofthe enterprise sector). Source: Bolshakova(2004). 110 4.21 Another peculiar feature of the pension system relates to the currently effective indexation rules for both the base and NDC parts of the pension benefit. Currently, the indexation rules are broadly linked to the actual inflation rate, but the rules are insufficiently formal and provide the government with considerable discretion powers with respect to both timing and scale of indexation. In addition, the pension legislation (Federal Law No. 173, Article 17) envisions complementary indexation o f NDC components incases when growth in wages exceeds the size of basic pension indexation. However, the scale o f such additional indexation can not exceed the growth rate o f actual Pension Fund revenues per one current beneficiary". This de facto makes the NDC component partially based on wage growth indexation, adjusted for real growth incollectionrate and growth ina number o f beneficiaries. 4.22 As o f January 1,2002, when the Russia's pensionreformwas launched, the size o f the monthly base pension was 450 rubles, the minimum pension - 600 rubles, and the overall average labor pension amounted to 1241 rubles. Subsequent indexation to compensate for the inflation so far has been broadly in line with the provisions of the law. Moreover, indexation in both 2002 and 2003 of the average pension exceeded the inflation rate. The latest indexation took place in April 2004. According to the Pension Fund, at that time the base pension reached 621 rubles, the minimum pension reached 944 rubles, and the average pension - 1892 rubles per month. Table 4.3: Main preferentialregimesfor social insurancecontributions, 9 months of 2003 Number of contributors Number of Average monthly to the PensionFund firms contribution, ruble 1. ReducedUST tax for particular employment groups ---- individual agriculture 4,052,947 154.8 entrepreneurs 2,030,569 141.4 2. Simplified tax regime for small business(*) -- firms -- individual 1,088,541 120,949 347.9 entrepreneurs 475,706 317,137 109.4 3. Presumptivetax on small business (*) -- firms 1,485,162 165,018 174.7 -- individual entrepreneurs 2,373,903 1,582,602 89.9 Total for taxpayersunderthe preferential schemes (**) -- share in total number of UST 11,506,828 158.0 the taxpayers 16.3% -- averageratio betweena preferential and regular contribution 32.3% Memo: regular UST taxation 59,270,991 489.8 (*)-Number of employeesis estimatedbasedonthe average size of respectivebusinesses. (**) - Total does not include the number of taxpayers that pay single agricultural tax, for which no data are available. Source: Estimates by the IET based on the datafrom the Taxation Ministry. " The inflation-basedindexation rules provide for the real values of pensionsfor current pensionersto be maintained at their end-2001 levels. However, historically this was a period with rather low real pensions. The real value of average old age pensionsin Russia declined dramatically as a result of the 1998 crisis, and at the endof 2001it still was below the 1997level (Smirnov and Isaev,2003, p.19). 111 4.23 The earlier World Bank Report (2003) "Pension Reform in Russia: Structure and Implementation" undertook a comprehensive review of the Russian pension reform strategy. The report identified major risks that the strategy has been facing and which relate to the following issues: (i)fiscal sustainability, (ii)deficiency in the system's design that may weaken workers' incentives to contribute to the system and work longer, (iii) weakness of the financial sector that may constrain the development of the funded pillar, and (iv) limited implementation capacity inthe government. 4.24 The results from the earlier World Bank Report (2003) also demonstrate that the currently effective indexation rules are a critical determinant for the fiscal prospects of the Russian pension system. Under the current rules, if they are fully enforced, the NDC portion o f the system i s expected to run considerable deficits, while the base component o f pension benefit would be generating surpluses that in principle would be sufficient to finance the deficits of the NDC portion and keep the entire system fiscally sustainable. However, to base the pension system's fiscal sustainability on such permanent redistribution o f funds between its two components could be a risky strategy in the longer term. Given the rather informal nature of the existing indexation rules, there i s a considerable risk for pressure to mount that would push toward using the surpluses in the base pension component either for additional indexation of the base pension or for other social spending. c. REFORM SCENARIOSIDENTIFIED FOR SIMULATIONS 4.25 Our simulations were primarily focused on the trends in the following performance characteristics of the pension system: (i)potential (fiscally affordable and actuariallyjustified) average replacement rate, (ii) o f the average pension and pensioner's living subsistence ratio minimum,(iii) pension system balance at the preset rules of pension indexation, and (iv) fiscal gap, estimated as a value o f additional public funding needed to provide for a desirable replacement rate. 4.26 While some of the reform elements are still being defined by the government, five groups of scenarios were identified and simulated for the purposes o f the current analysis. These are based on the expanded actuarialmodel o f the Russian pension system. 4.27 Group I.Base Case: N o reforms inthe unified social tax (UST) - unchanged pension contribution rates. Scenario 1(11). Low growth and absence o f structuralreforms, includingabsence of USTreforms Scenario 2 (12). Low growth and advanced structuralreforms, but without UST reforms Scenario 3 (13). Highgrowth and absence o f structural reforms, includingabsence of USTreforms Scenario 4 (14). Highgrowth and structural reforms, but without USTreforms 4.28 This group o f scenarios primarily reflects the status quo (i.e. continuation of policies as of early 2004) and its main objectives are to (i) provide estimates for the base line, Le., what would happen to the pension system if no reforms are undertaken, and (ii) estimate the impact of main macroeconomic factors, such as variation in GDP growth rates. In these scenarios the main parameters o f the system, including the retirement age, USTrates, number 112 o f contributors, etc., remained unchanged. However, it i s worth noting that our base case scenario fully reflects the changes envisioned in the draft Law on Professional Pension Schemes that provides for a considerable reduction in implicit subsidies to early retirees. It i s assumed that this Law would become effective inthe near future inits present draft form. 4.29 Scenarios of Groups 11-Vconsider various potential reforms inthe pension system.For the purposes of consistency, we assume that the reforms in the pension system will be undertaken only if there i s a broader across-the-sector acceleration o f structural reforms in Russia, i.e., there i s considerable progress in structural reforms as reflectedin Scenarios 2 and 4 o f Group I. Thus, in each of the Groups 11-V we consider only two scenarios (both with advanced reforms) instead o f four scenarios inGroup I. 4.30 Group 11. With the UST reform - cuts in the pension contribution rates by 4 and 8 percentage points Scenario 5 (22). Low growth and advanced structuralreforms, includingUSTreforms Scenario 6 (24). High growth and advanced structural reforms, including UST reforms 4.3 1 Group 111.UST reforms as in Group I1supplemented by an increase in the retirement age by 5 years for menand 10years for women Scenario 7 (32). Low growth and advanced reforms, including UST reforms, and increase inthe retirement age Scenario 8 (34). High growth and advanced reforms, including UST reforms, and increase inthe retirementage 4.32 Group IV.UST reforms as in Group I1supplemented by an increase in the retirement age and other changes and further reduction o f current UST privileges: (i)increase in contributions by self-employed and other beneficiaries from the existing tax regimes with lower USTrates; and (ii) reductioninthe total number o f such beneficiaries. Scenario 9 (42). Low growth and advanced reforms, includingUST reforms, increase inthe retirement age andadditionalcuts inUSTprivileges Scenario 10 (44). High growth and advanced reforms, including UST reforms, increase inthe retirement age and additional cuts inUSTprivileges 4.33 Group V. These are the scenarios with the active migration policy, in which we looked at a potential impact o f a considerable migration inflow to Russia. The question we have been asking inthis part o f the analysis was: to what extent immigration on its own could ease Russia's pension system problem and as such be considered as a substitute to the comprehensive pension reform? We consider two altemative options for such an extra migration inflow (relative to the baseline reflected inGroup I). Scenario 11 (52). Annual number o f migrants increases by 300,000 a year Scenario 12 (54). Annual number o f migrants increases by 500,000 a year 4.34 The UST reforms in Group I1were modeled through cuts in the pension contribution rate for the base portion o f the old-age pension. Within the Group I1 we consider two alternatives for a potential effect o f such a cut in the contribution rate on the tax base and collections: (i)base case -the rate cut does not lead to any expansion inthe UST tax base and bring about a proportional decline in collections; and (ii) optimistic case, which assumes a 113 considerable expansion in the UST tax base (a share of taxable payroll in GDP)72.Moreover, inthe latter case, it was assumedthat the growth rate of taxable payroll exceeds the pace of decline in the contribution rate. These two cases represent quite extreme altematives, which together help identify a range (low and high ends for collections), within which an actual future dynamic of the pensionsystem i s likely evolve. 4.35 The scenarios inGroup I11simulate the impact of increase inthe retirement age. These scenarios assume a gradual increase in the retirement age for men and women up to 65 starting in2006. These changes are fully inline with the global and regional trends - increases inretirement age have beencommon for the recent pensionreforms worldwide (Schwarz and Demirguc-Kunt, 1999). Most transition economies managed to start similar increases over the last ten years (Holtzmann et al., 2004, p. 85). There i s also a global tendency for closing the gap in retirement age between men and women. As was shown in analysis o f similar multi- pillar pension systems in other countries, such as Poland, preserving an earlier retirement for women would result insignificantly lower pensionbenefits for them because the new systems deliberately reward longer carriers and later retirement (Balcerzhak-Paradovska et al., 2003). 4.36 It was assumed in simulations that transition to the new retirement age would be completed for men by 2016 and for women by 2026. In case of the retirement age increase, the GDP growth rate was recalculated based on the assumption of maintaining the labor productivity growth rate at levels similar to the respective scenarios of Group I1(with UST reforms), but with a highernumbero f the employed. 4.37 Additional simulations in the scenarios o f Group IV analyzed somewhat more comprehensive reform strategies that assumed simultaneous undertaking o f cuts inUST rates, increase inthe retirement age, and a considerable reduction inthe existingprivileges for those who are eligible for the reduced UST rates. Inparticular, it was assumed that (i) the share of those who benefit from preferential UST regimes would decline from 15 to 10 percent o f the employed, and (ii) contribution rates for those who remain beneficiaries o f the reduced UST rates would increase by 100percent, from 150 to 300 ruble a month in2002 rubles. 4.38 Most o f the simulations in this Chapter cover a period up to 2050. However, the analysis i s mainly focused on the first half o f the period (up to 2025-30), which i s of greater interest for policymakers. The main simulation results are presented in Annex Tables A4.1- A4.15, and they are discussed in Sections F and G. Annex 4.2 presents the main features o f the modelthat was usedinour simulations. D. MAIN MACROECONOMIC AND REFORMASSUMPTIONS 4.39 Macroeconomic assumptions for our simulations were developed jointly with experts from the Institute o f Economy in Transition. The basic principles of the macroeconomic framework used for costing out various structural reforms are presentedinAnnex 3.1. Insum, we took the government's baseline macroeconomic projections for the period o f 2004-06, and used them as a basis to build a set o f four longer-term macroeconomic scenarios, each o f which reflects a specific combination o f two primary determinants o f Russia's future macroeconomic performance - average world market oil price and expected speed o f o f four macroeconomic scenarios - scenarios with advanced reforms, 12 and 14 -- as a basis structural reforms in the country. As described in the previous section, then we used two out ''Dmitriev at al. (1999) used a similar approach to the analysis of the potential impact of UST rate cuts on Russianpensionsystem. 114 on which we have designed and elaborated further more detailed sub-scenarios that reflect specific reformpackages inthe pension system. Noreforms Advanced reforms Moderate oil prices (18.5), low growth Scenario 11 Scenario 12 High oil prices (22.5 and higher), highgrowth Scenario 13 Scenario 14 4.40 Overall, we based the analysis on a rather conservative macroeconomic framework. It i s worth noting that GDP growth rates assumed in our scenarios for the period 2004-06 are lower than those assumed in the corresponding government projections. We believe that without advancing reforms growth rates will decline: better utilization o f existing reserves in the economy, which was a critical growth factor in 1999-2003, cannot support future growth inthe same way as before because the reserves are to a large extent exhausted. At the same time, the "advance reforms" scenario implies that reforms are likely to temporarily slow down GDP growth compared to the "no reform scenario", other things being equal. Therefore, in this case, growth rates are also likely to be lower for the next few years than the ones assumed by the government. 4.41 The key macroeconomic parameters in the model include real GDP growth rate, payroll share in GDP, population growth, economic activity, unemployment level, inflation, and interest (yield o f pension savings) rate. The payroll share in GDP i s an important parameter that inthe model determines the dynamics of the UST tax base.73 4.42 Macroeconomic parameters of the base scenarios (Group I) are provided in Table 4.4. Projections for GDP growth used for all scenario groups are also presented inFigure4.2. Figure 4.2: GDP Growth by Scenario, Group 1 7 T GDP growth by scenario, Group 1 2004 2009 2014 2019 2024 2029 2034 2039 2044 2049 ~O L Table 4.4: Assumptions on MacroeconomicParametersof the Base Scenarios (Group I) 2004-2010 2011-2015 2016-2020 2021-2025 2026-2050 Real GDP growth (annual rate, YO) 73 Roik (2003) also emphasizes a linkbetween the growth intaxable wages in GDP and the financial position of Russia's pension system. 115 2004-2010 2011-2015 2016-2020 2021-2025 2026-2050 Scenario 1 (11) 2% 2% 2.5% 2.5% 2.0% Scenario 2 (12) 1.5% 3% 4.5% 5.5% 5.0-3.5% Scenario 3 (13) 4% 4% 3.5% 3% 2.5% Scenario 4 (14) 3.5% 5% 6% 6.5% 6.0-4.5% Inflation (annual average, YO) Scenario 1 (11) 7% 6.5% 6% 5.5% 5% Scenario 2 (12) 10% 8% 6% 4% 3yo Scenario 3 (13) 10% 9% 8% 7% 6% Scenario 4 (14) 20-9% 9-7% 7-5% 5-3% 3% Taxable payroll, share in GDP Scenario 1(11) 26-28% 28-29% 29-30% 30% 30-31% Scenario 2 (12) 27-30% 30-31% 31% 31% 31% Scenario 3 (13) 26-28% 28-29% 29-30% 30% 31% Scenario 4 (14) 27-30% 30-31% 31% 31% 31% Real interestrate/yield Scenario 1 (11) 3% 3% 3% 3yo 3% Scenario 2 (12) 5% 5% wager -1% wager -1% wager -1% Scenario 3 (13) 3.5% 3.5% 3.5% 3.5% 3.5% Scenario 4 (14) 6% 6% wager -1% wager -1% wager -1% Unemploymentlevel(YO) Scenario 1 (11) 9% 9% 9-8% 8Yo 8% Scenario 2 (12) 8-7% 7-6% 6-5% 5?`o 5yo Scenario 3 (13) 9% 9-8% 8-7% 7% 7% Scenario 4 (14) 7% 7-5.5% 5.5-4% 4% 4% Memo items that correspondto the above presentedassumptions: Taxable payroll, growth rate (YO) Scenario 1 (11) 3.4% 3.5% 3.8% 3.8% 3.4% Scenario 2 (12) 3.7% 3.3% 5.2% 6.0% 5.5% Scenario 3 (13) 6.2% 5.2% 4.8% 4.3% 3.8% Scenario 4 (14) 6.2% 5.9% 6.6% 7.3% 6.4% Labor productivity,growth rate (YO) Scenario 1 (11) 1.8% 2.6% 3.3% 3.4% 3.3% Scenario 2 (12) 0.5% 2.8% 5% 6% 5.5% Scenario 3 (13) 3.8% 4.3% 4.4% 3.7% 3.7% Scenario 4 (14) 2:4% 4.2% 6.3% 7.2% 6.4% Notes: a) wage r stands for annual real growth rate in average wage in the economy; b) projected productivity growth rate i s higher than GDP growth rate due to expected decline inboth population and employment. 4.43 Our simulations assume that wage growth i s proportional to labor productivity growth. At the same time, it is accepted that the taxable wage may grow faster than productivity through a decline inthe share of informal wages and respective growth o f the taxable payroll share in GDP (i.e., due to growth in formally registeredwages and the expansion inthe share o f the formal sector facilitated by institutional reforms). Thus, one o f the particular features of our projections is an introduction of an additional factor in the model that reflects faster growth of taxable wages (and respectively, the UST tax base) relative to labor productivity growth. Under such an approach, the share o f the taxable payroll in GDP i s an exogenous parameter, for which we consider a set o f altemative trends depending on a scenario. The dynamics of the payroll share in GDP for the Group I("No reforms") scenarios i s shown in Figure4.3. 4.44 Inthe Scenarios of Group I1("Reduced UST rates"), it is assumed that the cut inthe UST rate will become effective on January 1, 2005. Parameters o f the PAYG pillar have a linear dependency on the effective contribution rate. Therefore,' in the base case, the direct impact of changes inthe contribution rate on the pension system couldbe estimated by simple recalculation of the base case results. 116 Figure 4.3: Dynamicsof the Taxable payroll share in GDP, Group 1 Dynamics o f the taxable payroll share in GDP, Group 1 35 33 _ _ _ -Scenario 1 -Scenario --Scenario3 2 I I 2004 2009 2014 2019 2024 2029 2034 2039 2044 2049 2 5 + Table 4.5: AdditionalAssumptions in Scenarioswith ReducedUST Rates, (Group 11) 2004 2005-10 2011-15 2016-20 2021-25 2026-50 Share of taxable payroll in GDP (optimistic case) Scenario 5 (22) 27% 29-33% 35% 35% 35% 35% Scenario 6 (24) 27% 29-43% 35% 35% 35% 35% Contribution rate to finance the base pension Scenario 5 (22) 14% 10% 10% 10% 10% 10% Scenario 6 (24) 14% 10% 10% 10% 10% 10% 4.45 In the optimistic case of Group 11, the payroll share in GDP is expected to expand (Table 4.5). Quantified estimates for such an increase were developed based on the assumption o f unchanged contributors' net incomes in cases when their tax rate i s reduced74. This assumption is similar to a hypothesis on a proportional link between the relative size of the USTrate cut and an associated additional increase inthe share of taxable payroll inGDP. Under such assumptions, as a result of UST rate cuts, absolute amounts of both pension contributions and pensions remain unchanged, while taxable wages would increase (due to a declined share o f shadow wages), which means a decline inthe nominal average replacement rate7'. 4.46 The cross-country data suggest that indeedthere is a room for Russia to increase the share of taxable payroll, but given high stability o f this indicator in most counties, such changes could happen only gradually. In addition, considerable institutional reforms will be 74The idea behind this assumption is as follows: in response to the tax rate reduction, taxpayers will hide less o f their incomes, and the share o f their taxable wages will increase. It is also assumed that the size o f such an increase could be conservatively estimated based on the assumption of constant net income: taxpayers, who benefit from an opportunity to reduce their hidden incomes (and, therefore reduce the size o f potential penalties), will not seek additional financial benefits and will keep unchanged both the amount o f paid tax and the size o f their actual total incomes after taxes. 75 Note that the real replacement rate does not change in this case, because the total payroll (including the informal part) remains intact. 117 neededto support this trendtoward higher formalization of the economy. In2002, the share o f taxable payroll (including payroll taxes) in Russia was estimated at about 32.5 percent o f GDP, which i s similar to the share in Mexico and somewhat higher than the one in Turkey. Taxable payroll net of payroll taxes amounted to about 25 percent of GDP (Table 4.2 above). Table 4.6 presents the comparative dynamics of this indicator across the OECD countries. Table 4.6: PayrollShare in GDP of Various Countries,Percent 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Australia 48.44 48.09 47.69 47.38 47.62 47.95 48.68 47.91 48.42 48.10 48.07 47.48 Austria 53.25 53.86 54.15 54.72 54.11 54.29 53.14 52.91 52.50 52.60 51.87 51.80 51.56 Belgium 51.52 52.92 52.92 53.17 52.40 52.03 51.72 51.31 51.17 51.48 51.26 52.45 53.16 Canada 55.06 56.19 56.26 55.17 53.34 52.46 52.04 52.14 52.79 52.07 51.94 52.89 Czech Republic 44.82 45.32 46.48 47.10 48.28 47.84 46.10 46.18 46.43 45.60 46.36 Denmark 55.54 55.22 54.92 54.68 52.80 53.02 52.93 52.92 53.91 54.22 53.35 54.33 54.60 Finland 55.55 58.36 56.54 52.77 50.68 49.85 50.15 48.64 48.03 48.88 47.86 48.93 48.96 France 52.49 52.73 52.89 52.96 52.07 52.09 52.15 51.91 51.51 51.86 51.98 52.49 52.76 Germany 56.04 56.18 56.74 56.67 55.37 55.34 54.87 53.98 53.46 53.51 54.18 54.12 53.63 Greece 35.16 32.57 31.54 31.44 31.32 32.25 31.92 32.77 33.24 33.35 33.09 33.14 32.95 Hungary 46.94 45.66 44.95 44.79 43.15 45.30 Iceland 48.77 51.66 52.36 50.58 49.49 51.04 52.56 52.25 54.24 56.91 58.47 58.19 59.26 Ireland 46.28 47.40 47.90 47.76 47.25 45.30 44.39 42.52 41.67 40.63 40.63 41.28 Italy 46.14 46.29 46.19 45.85 44.27 42.56 42.52 42.66 40.58 40.74 40.64 40.91 41.24 Japan 52.16 52.93 53.26 53.96 54.36 54.58 54.04 54.25 54.40 54.31 54.23 54.61 Korea 45.87 47.01 46.90 46.70 46.61 47.65 48.85 47.20 45.23 43.59 43.99 45.50 44.99 Luxemburg 53.26 53.05 55.02 53.66 53.31 53.44 53.10 50.53 49.04 48.23 47.40 50.42 53.04 Mexico 29.53 30.88 32.88 34.75 35.34 31.08 28.86 29.63 30.60 31.23 31.25 Netherlands 51.61 52.00 52.90 52.96 51.61 50.86 50.59 50.35 50.82 51.42 51.05 51.37 52.26 New Zealand 45.06 44.71 44.26 42.75 42.58 42.57 43.26 43.45 43.62 42.47 41.84 41.98 Norway 49.15 48.58 49.16 47.96 47.92 47.36 46.37 46.48 50.15 49.16 43.57 44.67 47.16 Poland 41.90 47.05 44.17 42.66 42.92 44.33 45.15 45.22 44.93 43.75 Portugal 43.83 46.12 47.05 46.86 44.76 47.71 47.97 47.93 47.81 48.22 Slovak Republic 43.25 42.30 42.25 42.46 43.45 43.23 41.34 Spain 49.86 51.14 51.70 52.19 50.71 49.91 49.76 49.78 49.93 50.04 50.13 50.13 49.91 Sweden 57.51 56.99 56.70 55.20 54.29 52.66 54.66 54.28 54.20 53.22 55.23 57.38 57.94 Switzerland 61.98 63.31 63.68 62.80 62.00 62.59 62.59 62.88 62.61 62.35 62.45 Turkey 27.21 31.86 31.67 30.88 25.54 22.18 23.93 25.80 25.46 30.68 29.19 28.34 Great Britain 56.72 57.27 56.92 55.68 54.25 53.77 53.20 53.38 54.12 54.60 55.23 56.16 55.74 USA 58.31 58.30 58.26 58.00 57.49 57.33 ,56.77 56.39 ,57.28 57.69 ,58.69 58.69 57.63 Source: World Bank. 4.47 Inour projections for the scenarios in Group I("No UST reforms"), it was assumed that without UST rate cuts the share of taxable payroll (net o f payroll taxes) would increase gradually from the current 25 percent o f GDP to 31 percent o f GDP by 202576.It i s worth noting that after such an increase, in 2025 the payroll share in Russia will still be lower than the actual payroll share observed in such countries as Poland and the Slovak Republic in 2000. This indicates a possibility for a somewhat stronger expansion in payroll share under more favorable conditions associated with an additional reform effort and lower payroWincome taxation. 76Or 39-40 percent of GDP, if payroll taxes are included. 118 4.48 At the same time, in the optimistic case in scenarios of Group I1 ("Reduced UST rates") it was assumed that the payroll share may grow considerably faster than in the scenarios o f Group I,and the payroll growth rate would exceed the pace of decline in contribution rates by about 20 percent.77Under such assumptions, the growth in the taxable payroll share would be more than 50 percent higher than the growth in similar scenarios of Group I,and the overall increase in the payroll share within the period would amount to 9.5- 10 percentage points (p.p.) o f GDP instead of the original 5.5-6 percent. Consequently, starting from 2020, the taxable payroll share in GDP (net o f payroll taxes) in the optimistic case o f Group I1(Table 4.5) amounts to 35 percent (25+6-14/10*1,q7'. Such payroll shares are comparable to the actual current levels of this indicator inthe most advanced reformers among transition economies inthe CEE. Inother words, under the conditions of aggressive structural reforms and with lower UST rates, within approximately 15 years additional expansion o f the UST tax base could amount to 4-4.5 p.p. of GDP, from 30-31 to 35-35.5 percent. It is worth emphasizing that we consider the above assumptions to be really optimistic and we do not see much room for a further growth inthe payroll share. Inother words, it seems unlikely that the payroll share in Russia may expand above this 35 percent of GDP threshold, even if more radical cuts inthe UST rates (inexcess of 4 percentage points) are introduced. 4.49 International experience, however, suggests that there i s no strong correlationbetween reductions in the payroll tax rates and expansion in the tax base. To attain a desired policy outcome, cuts in the UST rate in Russia have to be complemented by other policies aimed toward formalization o f the labor market and improvements in expectations of market participant^.^^ 4.50 Inaddition to a closer look at two quite differentcases ("base" and "optimistic") for a potential dynamics of the tax base, we develop a broader framework for a sensitivity analysis of our simulation results (Annex 4.1). Ithelps to address questions such as "What may happen to the pension system if the key parameters such as UST rate and payroll share inGDP would reachparticular levels?" 4.51 Our payroll growth rate assumptions are equivalent to projecting relatively high growth rates for real taxable wages. Inparticular, it i s assumed that in the optimistic case o f Group 11, in2004-10 the real average (taxable) wage will grow at an annual rate of 4.4 percent inScenario 2, and 6.3 percent -- inScenario 4. Andthis growthwill remainrather highfor the period 2011-20 (Figure 4.4). This i s 0.5 to 1.5 percentage points higher than the growth in 77 Recent study of the Institute of Economy in Transition (Sinelnikov et al., 2003) has found that simultaneous reduction of income tax and UST rates in 2000 had rather a substantial effect on growth in tax bases of these taxes in the period of 2001-02. Given all existing uncertainty, these results cannot be directly applied for developing precise quantitative estimates for elasticity of the payroll tax base. However, they seem to confirm the plausibility of our assumptions. 78 Note that we linkthe potential incentive effect exclusively with the reduction of the base pensioncontribution rate (from 14 to 10 percent), but not with the change in the total pension contribution rate (from 28 to 24 percent). This i s a reflection of the current situation where the contributors are least interestedin contributing to the basic portion of pensions, because this part does not have any effect on the size of their future pension benefits.Respectively, incentives for participantsseem to be the most sensitiveto adjustmentsinthis component ''ofIn the overall contribution rate. Argentina, for instance,the payroll tax base did not expand despite the considerable cuts in tax rates inthe course of pension reforms in the 90s (Rofman, 2002). This is believed to be related to other non-tax impediments to the formalization of the labor market (such as hiring and firing restrictions), as well as low credibility of reforms and respectively strong expectations of the employers that the tax cuts had been temporary. At the same time, several Eastern European countries, including Poland, the Slovak Republic and Czech Republic, have a relatively highpayroll tax base despitepreserving one of the highest payroll tax rates in the world, well above 40 percent(IMF, 2002). 119 labor productivity in the same period. Given that a substantial portion of this wage growth will be associated with a reductioninshadow incomes, such growth rates seem quite feasible. Incomparison, the real wage grew at an annual rate of 5 percent and more in a number of CEE countries during the second half of the 90s (Table 4.7) despite the fact that the share of shadow wages inthese countries at that time was significantly lower. Figure4.4: RealGrowthinTaxable Wages,Group 1Scenarios 10 9 1 -~cenano2 I 8 7 6 5 4 3 2 1 0 I 2004 2009 2014 2019 2024 2029 2034 2039 2044 2049 4.52 Overall, the assumptions inthe optimistic case provide for the share o f shadow wages to decline to 20 percent of taxable wages (as compared with the current 45-50 percent). This indicates that total labor incomes (combined taxable and nontaxable wages) will be growing at a rate that i s about 20 percent lower than indicated inTable 4.4. Table4.7: RealWage GrowthinVariousCEE Countries,1995 2000, Percent - Croatia 34.5 Estonia 23.2 Lithuania 38.6 Slovak Republic 62.0 Poland 24.7 Source: World Bank. 4.53 Inaddition, it is assumed inthe reform-advanced Scenarios 2 and 4 in all groups that the economic activity o f the population would be higher, which reflects an assumption on expansion in labor demand under the conditions of sustainable growth and a favorable investment environment created by structural reforms. In Scenario 2, it i s assumed that the economic activity parameter would grow at an annual rate of 0.5-1 percentage point up to 2010, and it would remain unchanged after this. In Scenario 4, economic activity would grow at a rate of 1percentage point a year up to 2010, and additionally 0.5 percentage points a year during 2011-15. Consequently, the average level of economic activity during the period increases by 5-8 percentage points (Figure4.5). 4.54 The share of non-payers in the pension system (delinquent contributors) was assumed unchanged at a level of 5 percent throughout the entire period. This corresponds to the assumption that most workers would continue to participate inthe pension system, while the primary form of tax evasionwould remain to be a underreporting o f wages, but not a complete 120 withdrawal from the system. This relates to two specific features o f Russia's pension system, which strengthen employees' incentives to participate such as, (i) eligibility requirements low for the base pension, which requiresjust 5 years of service and no minimumcontribution, and (ii)availability of generous preferential tax rates for self-employed and other categories (such as farmers) that have a higher propensity to drop out of the system. At the same time, we This elasticity has a simple linear character -- a decline inparticipation in the pension system estimated the elasticity o f the average pensionwith respect to changes inthe participation rate. by 5 percent wouldreduce the averagereplacementrateby about 5 percent. Figure4.5: EconomicActivity of Population, Group 1Scenario o.8, Economic activityofpopulation,Group lscenarios, YO ~ 0.5 -- -Scenario 1 0 1 -- Scenario4 0 0 2004 2009 2014 2019 2024 2029 2034 2039 2044 2049 4.55 InGroup IV ("Reforms to Reduce Tax Privileges") we assume a considerable change inthe structure ofcontributors. While inallthe scenarios ofGroups1-111the structure ofthose employed inthe economy was: 80 percent - regular employees, 15 percent - the self-employed and others eligible for lower UST rates, and 5 percent are UST delinquents, inthe scenarios in Group IV, the share o f regular employees i s gradually increasing to 85 percent, while the share of those included in the privileged schemes i s declining to 10 percent. It i s argued that the current eligibility rules for reducedpension contributions are too soft and, moreover, they are often abused by employers who register their staff as self-employed to reduce payroll costs. In addition, it i s assumed that for those who keep their UST privileges, the fixed monthly pension contribution would be doubled inreal terms. 4.56 No changes are envisioned in technical parameters of the pension system, such as collection and exemption rates, and coefficient o f regressivity, which respectively remain at the base levels of 0.96, 0.94, and 0.94. 4.57 The real interest rate (pension savings yield) i s higher for reform-oriented scenarios 2 and 4 than in the scenarios 1 and 3. For the period till 2015, the rate amounts to 5-6 percent per annum inthe former case, while it i s 3-3.5 percent inthe latte. For the period after 2015, inthe scenarios with reforms the yield is fixed at the level of one percentage point below the real growth rate inaverage wages o f the respective year (Figure4.6). 121 Figure 4.6: Real Interest Rate, Group 1,Scenarios YO 1 Real Interest rate, Group 1,scenarios % ~7.0 6 0 5 0 4 0 I 2.04 I - - - - Scenarioi14 -Scenario 1 0 - -- 2 Scenario3 ' Scenario4 4.58 Another important assumption relates to the set o f used indexation rules. In our simulations, all pension benefits are fully indexed for an actual inflation rate, and thus inflation does not have an impact on real values of pension benefits and respectively replacement rates. Moreover, in addition to full indexation for inflation, we used two alternative sets o f indexation rules to assess trends in real values of the average pension benefit: (i) The rule that corresponds to the existing legislation - the base pension i s indexed with the average inflation, while the NDC portion o f the benefit i s indexedbased on the nominal growth rate of the Pension Fundcollections per beneficiary. (ii) The rule that corresponds to estimates for the maximum affordable pension - indexation o f the base pension remains the same (inflation-based), but the NDC portion i s additionally indexed to distribute all available surpluses in the PAYG pillar. 4.59 Inaddition, a separate indexation rule is used for the subsistence minimum, which is indexed for inflation plus additional indexation equal to 30 percent o f the real growth in average taxable wages.*' Justification for the latter indexation rule derives from the fact that the baseline scenario assumes relatively high growth in real wages for the entire period forecast. If the subsistence minimum i s not adjusted for this growth in labor incomes, it quickly erodes and becomes a meaningless concept. Substantively, the real growth in subsistence minimum could be interpreted as expansion in the set o f minimum social needs with the growth inaverage income (Chen andRavallion, 2000). E. DEMOGRAPHIC ASSUMPTIONS 4.60 Basic demographic assumptions applied in the model are presented in Table 4.8. The key features o f the baseline demographic projections for Russia provide for both a drastic 80 For the cross-country data Chen and Ravallion (2000) estimate that elasticity of the poverty line by consumption level equals one third. 122 reduction inthe general population (particularly after 2025)81and a monotonous growth o f the system dependency ratio (ratio o f pensioners and contributors): from the current 0.6 to 0.65 - 0.75 in2025 and 0.9 - 1.0 in2050 (Table 4.9). 4.61 For the simulations, we have used the demographic parameters that are mostly identical to the existing base demographic projections for Russia (Population o f Russia, 2002). However, inthe "advanced reform'' Scenarios 2 and 4, we increased the birthrate from the conventionally projected 1.40 to 1.45 (Scenario 2) and to 1.50 (Scenario 4), Le., as compared with the base case projections, the maximum increase o f the birthrate inthe model does not exceed 7 percent. Table 4.8: Base DemographicProjections(scenarios fromGroup I-IV) 2003 2004 2005 2006 2010 2015 2020 2030 2040 2050 Birth rate* 1.31 1.33 1.35 1.37 1.40 1.40 1.40 1.40 1.40 1.40 Life expectancy 66.6 66.8 67.0 67.1 67.2 67.7 68.3 69.5 70.6 71.8 - Men 60.3 60.7 60.9 61.1 61.3 61.9 62.5 63.8 65.0 66.3 -Women 73 73 73 73 73 74 74 76 77 78 General population 143 143 142 141 139 136 132 121 110 99 -- Men 67 67 66 66 65 63 61 55 50 45 Women 76 76 76 76 75 73 71 66 60 54 Elderlyloadcoefficient for general o,33 population(60/55) 0.32 0.32 0.32 0.35 0.40 0.46 0.49 0.58 0.71 - Men(above 60) - Women(above 0.20 0.19 0.18 0.18 0.19 0.22 0.26 0.29 0.32 0.43 55) 0.46 0.46 0.47 0.48 0.52 0.60 0.67 0.71 0.86 1.03 * Birth rates presentedare those for the most conservative options that correspondto the Scenarios 1 and 3 of Group I. Source: Demographic projections used in this Chapter were developed by Ye. M. Andreyev, the Centre for Human Demography and Ecology. See Population of Russia (2002). 4.62 Inall the scenarios of Groups I-IV, net immigrationwas set at a conservative level of 60-100,000 persons per year. InScenarios 1and 3 o f Group I, the net migrationis close to the baseline demographic projections of 60,000 persons a year, while in Scenarios 2 and 4 it was increased approximately by 50 percent. However, this growth in immigration has an insignificant impact on the results. This i s primarily owing to the fact that, inour view, the net migration inflows are (unrealistically) low in the conventional demographic projections. In a long-term perspective with a continuing population decline, one should expect that the annual net immigration to Russia will expand substantially. This i s the reason why inthe scenarios o f Group V ("Additional immigration"), the net migration i s additionally increased to 300,000 and 500,000 persons per year through the extended periodg2. *' However, it is worthmentioning,that such sharp populationreductionmayleadto amuch stronger expansion innet immigration,which is to some extentreflectedin scenarios of GroupV. 82 Accordingto the data providedby the Instituteof Economic Forecasting, over the last 15 years the maximum level of net migration of 800,000 persons per year was observedin 1994. Since then the flow of net migration declinedsignificantlyandin2001 it made up only 50,000 persons. 123 F. SIMULATION RESULTS Group I.No changesinthe unifiedsocial tax, no changesinthe retirementage Estimating baseline trends 4.64 As noted above, the main output indicators for the pension system in our scenarios include the overall affordable replacement rate (Figure 4.8), the contribution rate required to ensure the targeted replacement rate (Figure 4.9), and the ratio between the average pension benefitand the pensioner's living subsistence (Figure4.10). 4.65 Inthe scenarios of Group I,the variation inbaseline macroeconomic parameters does not lead to any tangible variation inthe overall replacement rate (Figure 4.8 and Annex Table A4.3). That is, our results suggest that acceleration o f economic growth, lower inflation, and reduced unemployment all have rather a modest effect on the future average size of the pension benefit (relative to the average wage). While differences in annual growth rates between Scenarios 11 and 14 are significant and amount to 3-4 p.p., they produce a modest variation in the replacement rates that in 2030 stays in the interval of 24.4-27.8 percent.83 Even under the most favorable macroeconomic assumptions, by 2025 the replacement rate drops considerably relative to its current level and then it stays basically unchanged. It i s expected that the rate o f decline would be the most pronounced between 2016-25, when it would lose on average about 3.5 p.p. Figure 4.8: PensionReplacement Rate, Group 1Scenarios 40 I]"/ PensionReplacementRate,Group 1 scenarios, YO I ~ -- I 30 25 I "1 , -Scenario -- 2 Scenario3 2004 2009 2014 2019 2024 2029 2034 2039 2044 2049 _____ 83The contribution of the fully funded pillar to the overall replacement rate amounts to about 10 percent in 2030, but it reachesa quarter in 2050. 125 Figure4.9: SST Rate(Contribution Rate) Neededto Ensure the ReplacementRate of so%, Group 1 Scenarios SST rate (Contributionrate) neededto ensurethe 40 replacementrateofl-roup 1scenarios, %-, 35 30 25 I -Scenario --Scenario3 2 10 0 I 2004 2009 2014 2019 2024 2029 2034 2039 2044 2049 Figure4.10: Ratio BetweenAverage Pensionand Pensioner's SubsistenceMinimum, Group 1Scenarios Ratiobetweenaveragepensionandpensioner's subsistenceminimum, Group lscenarios - - - - Scenano 1 -Scenario2 --Scenanod 4 0 Scenano 4 4 4.66 None o f the scenarios in Group Icould bring the replacement rate to the level o f 30 percent. While higher economic growth results in much higher real pensions as well as in a higher ratio between average pension and subsistence minimum (Annex Table A4.6 and Figure 4-10), it does not help to close the gap between growth in wages and pensions. It appears the existing pension system, even under the most optimistic assumptions, does not provide pensioners with a fair share of gains from economic growth (Box 4.2). 4.67 These results confirm that the current pension system does not have internal reserves to prevent a broadening of the income gap between employees and pensioners, and as such they justify the need for additional reforms in the pension system. Without the reforms, the broadening of the gap may trigger a pressure for support o f the pension system through 126 additional budget transfers. W e estimated potential fiscal costs of such budget support, based on the assumption that the government would be forced to keep the replacement rate at the level o f 30 percent. In this case, annual fiscal costs of supporting non-reformed pension system would amount to 0.5-1.5 percent of GDP a year in2030 and it increases to the level of 1.5-2.5 percent o f GDP by 2050 (Table A4.7). The pension system would b e able to generate the same amount of funding internally only through a considerable additional increase in the average contribution rate, from 28 percent to on average 32 percent in 2025 and to about 35- 36 percent in2050 (Figure 4.9). Box 4.2: Drivers of the declinein the replacement rate Separating the PAYG and fully funded pillars o f the system helps to understand better the main factors behind the declining trends in the overall replacement rate. Without significant reforms inthe current pension system, in the longer run the replacement rate in the P A Y G pillar will inevitably decline to approximately 17-20 percent (Table A4.2). This is because the current pension system, despite all its recent positive innovations, has largely preserved its re-distributional natures4,in which the replacement rate is inversely proportional to the system dependency ratio. With the system dependency ratio approaching 1 (under the current retirement age), the average replacement rate inthe PAYGpillar will be to a growing extent determined by the average UST rate. Taking into account also that the actual collections o f the Pension Fund are lower than statutory collections due to such parameters as incidence o f non-payments, exemptions, and the regressive nature o f the tax rates5, in the long run the average replacement rate in the P A Y G pillar could be maintained only at a level o f about 19 percent (22*0.94.0.94.0.96=18.7 percent). However, taking into account that the self-employed and other special categories pay their P A Y G contributions at a much lower rate (on average approximately only 3 percent), while their number comprises 15 percent o f contributors, the longer-term ceiling for the replacement rate inthe P A Y G system will be even lower - - 16.3 percent ((0.94~0.94~0.96)~(0.85.22+0.15~3) =16.3). For the fully funded pillar (FFP), a rough estimate for the potential longer-term replacement rate (Table A3) may be obtained in a similar manner. Assumingthat the wage growth rate is equal to the interest rate (yield) in the economys6, the FFP replacement rate as o f the date o f retirement is proportional to the contribution rate and the number o f years o f service and inversely proportional to the average life expectancy after retirements7.Given the current maximum contribution rate for the funded pillar o f 6 percent, prevailing values o f both collection and regression factors, and working lives o f 30-35 years, the FFP replacement rate will be about 8.5-10 percent (6*0.94-0.96(30-35)/19)88. As a result, with the dependency ratio equal to 1, the combined (PAYG and FFP) longer-term potential replacement rate for the existingsystem would stay within the interval o f 24-25.5 percent. 4.68 Table 4.10 provides a compressed summary of the expected trends in the replacement rate under the Group Iscenarios. It suggests the following observations: 84The current legislation provides that only 21 percent of pension contributions (6 percent out of 28 percent of payroll) will be directedto the fundedpillar. 85As mentioned above, the model uses the following estimates for these three parameters: 0.94, 0.94, and 0.96 respectively. 86 This is a standard assumption for this kind of model: if the interest rate is below the wage growth, the incentivesfor participation in a pensionsystemweakennotably. 87 Currently, annuity has not yet been fixed for the funded pillar, and the model is using variable "life expectancy" with the same values as for the PAYG pillar (12-19 years dependingon the year of projections). 88This is an upward biased estimatebecause it does not account for several factors, such as e.g., administrative costs inthe FFP. 127 0 Each cohort of the pensioners starts its retirementwith quite a low real pension, and this relativevalue ofpensionat retirementis declining with time, i.e. eachcohort is better off at the moment of retirement than those that follow. 0 The real value o fpensionfor each cohort is growing with time, however this growth i s concentrated inthe secondpart o f their retirement "career". 0 Practically inall cases duringthe first part of retirement the average pension i s less than one third o f the average wage at retirement. Table 4.10: Realpensionas a percentage of the averagewage at the time of retirement. Year of 2002 2010 2020 2030 2040 retirement Scenario2 (12): Low growth 2002 31 34 41 - 2010 - 26 31' 48 - 2020 - 20 31 49 Scenario 4 (14): Highgrowth 2002 31 37 50 - 2010 - 25 34 57 - 2020 - 19 32 56 4.69 The dynamics of real replacement rate (Table A4.5), i.e. the replacement rate estimated as a ratio o f the average pension to the full labor income (but not just to its taxable portion), shows much less variation than the conventional (nominal) replacement rate (Table A4.1). It suggests that, when one accounts for an expected decline in the share of informal wages inthe total payroll, the overall trend inthe relative size of the average pension i s quite different. It shows modest growth for the period between 2005 and 2020 within the range o f 24-27 percent, and after this it declines much more gradually that the conventional (nominal) replacement rate. Still, innone o f the scenarios in Group I,the real replacement rate exceeds 26 percent in2030. 4.70 At the same time, all scenarios in Group Ishow a gradual increase in the value o f average pension relative to the subsistence minimum (Figure 4.10). The latter ratio i s a better measure of absolute changes in pensioners' standard o f living than the replacement rate. It indicates that, even within the unrefomed pension system, with economic growth the real incomes of pensioners will also grow steadily. However, there are two concerns with the patterns o f this future growth: a) in the initial period till 2015, despite a low level o f current pensions, this growth will be rather slow, and b) as reflected in the falling replacement rate, the growth inpensions will be lagging growth inreal wages, which may become a politically sensitive issue. 4.71 Incontrast to the replacementrate, the ratio between average pension and subsistence minimum shows a considerable variability depending on the growth rate inthe economy, as well as growth inreal average wages. By 2030, the difference inincreases o f this ratio across the scenarios i s about 3 times, from 50 percent in Scenario 11 to 165 percent in Scenario 14 (which i s the scenario with the highest GDP growth rate). 4.72 The analysis also suggests that the existing pension indexation rules could be somewhat relaxed without compromising the stability o f the pension system. In the expected environment o f the high wage growth the existing indexation rules, which are primarily inflation based, are too restrictive and do not allow for utilization o f the entire pension 128 contributions. Ifthe PAYG pensions are adjusted only for price inflation, and do not reflect real wage growth, the future average pensions would be much lower (Table A4.8) than those that the system could afford, and the pension system would generate considerable surpluses that would reach 1.8-2.7 percent of GDP by 2030 (Table A4.11). Comparison between the average pension, estimated for the "inflation-only" indexation rules, with the average pension that could be afforded by the pension system (ie., with the zero balance o f the PAYG pillar, (Table A4.1) shows that by 2020 the difference,(measured by the replacement rate) amounts to 50 percent and it reaches 100 percent by 2050. 4.73 While under the existing indexation rules overall the PAYG system would generate considerable surpluses, they would be generated entirely by the base pension component (Table A4.9). The NDC component would have considerable deficits (Table A4.10 and Figure 4.11). Figure4.11: Balance in the Base and NDC Componentsof the Pension System, Group 1 Scenarios,% of GDP Balanceinthe baseandNDCcomponents ofthe pension 3.5 system, Group-1.__..._-..I... scenarios, YOofGDP " .- - - --Scenario 1 3.0 -Scenano 2 2.5 2.0 1.5 1.o 0.5 I 4.74 The inflation-only indexation rules would have a major effect on the real value of the base pension and could lead to a quick erosion o f its value. If the real wages grow at 6 percent, the replacement rate for the base pension will decline by factor of approximately 3 by 2025, and by factor o f 18 by 2050. The share o f the base pension inthe total average pension benefit will decline from 30 percent in 2003, to 20 percent in 2025 and to only 5 percent in 2050. 4.75 Under the circumstances, it would make sense (probably after 2030) to convert the base pension into an instrumento f social assistance, link its size to the minimum subsistence benefit, and respectively move the responsibility for its funding from the PensionFundto the federal budget. This would be consistent with the nature o f the basic pension, which aims to provide basic income security for the elderly (Hotzmann et al., 2004, pp. 10-1l).*' This would cause an annual increase in government budget expenditures o f about 1 percent of GDP in A similar suggestion was made by the Russian Union o f Industrialists and Entrepreneurs (Concept of the Pension Reform. Employers' Position, 2001). Hungary and Poland abolished their minimum pension and providea minimumincome support for the elderly from outside of the pension system(IMF, 2002, p.30). 129 2030, but this amount would be declining quickly to less than 0.4 percent o f GDP in 2050. Accordingly, the same amount o f pension fund resources could be freed and may be used for an increase inthe PAYGpensions. Impact of other specificfactors on the replacement rate 4.76 The sensitivity of the results for future average pensions to changes inthe interest rate proved to be quite high.An increase inyield by 1p.p. leads to a 15-20 percent increase inthe replacement rate for the fully funded component o f the pension benefit. For 2050 this could bring about an additional 1.5 pap.to the overall replacement rate inthe system. However, for the periodtill 2030 the share of the fully funded component inthe overall benefitremains low, and therefore changes inthe interest rate have only a marginal impact on the real value of the overall pension benefit. Our baseline estimates assumed a rather conservative assumption regarding the interest rate, which i s kept at 1 p.p.below the real wage growth, while usually these parameters are assumed to be equal in the longer term. Our choice of a lower interest rate relates to the fact that the baseline scenarios assume quite high real wage growth, and equalizing interest and wage growth under the circumstances may produce unrealistically high yields. 4.77 While the pension system may have potential financial reserves associated with an increase in the economic activity (labor market participation) o f the population, as well as with decliningunemplovment, the possible impact of these two parameters seems to be low. It appears inour assumptions usedfor Scenario 14 (highgrowth with advanced reforms) that we have already accounted for most o f the realistic reserves inthese two areas. 4.78 An increase in the birth rate along with the effects of declining unemployment, growing economic activity, and higher net migration inflows inthe base scenarios o f Group I helpto reduce the system dependency ratio by about 10 percent and, respectively, to increase the average pensionbenefit by 10percent. 4.79 The above estimates for baseline trends inthe replacement rate assume a considerable compression of the early retirement schemes as proposed by the Draft Law on Professional Pension Systems. Thus, there are no significant unaccounted reserves associated with this factor. We estimate that if the current draft Law becomes effective, this would reduce the number of early ensioners from about 3 million in 2002 to 2.25 million in 2030, and to 1 million in 2050. 83 Given that the total incremental annual payments to early retirees, as estimated, amount to about 4 percent o f total PensionFundspendingon pension benefits, total savings due to the proposedreduction inthis benefit would be modest - from 1percent o f the total pensionpayments in2030 to 2.7 percent in2050. 4.80 The existing proposals for UST reforms also suggest considerable changes to existing tax brackets, and also introduce a regular indexation of the tax brackets for the future. Ifimplemented, this would change both the average effective tax rate and the coefficient of regressivity (currently equal to 0.94) for the pension system. However, it i s difficult to predict the magnitude o f possible changes in the latter because we do not have data on actual distribution of wages within each group of contributors. More importantly, it is quite difficult to make longer-term projections for future changes inwage differentiation. We have therefore kept the coefficient of regressivity intact in our simulations and focused primarily on the impact o f changes inthe average contribution rates. 90However,the number of early pensioners would increase till 2012 dueto the grandfatheringclauses. 130 Group 11. USTreform:cuts inthe pensioncontributionrates Modest rate cut of 4percentagepoints 4.81 The pension system understandably i s very sensitive to even modest cuts in the contribution rates. Scenarios o f Group I1were used to undertake a series o f simulations that help evaluate the possible effects o f cuts inthe average USTrate on the overage benefit under the assumption that such a cut is not complimented by other reform efforts. Inthe base case, when there is no increase inthe UST tax base relative to the baseline of 31 percent of GDP, the proposed rate cut i s expected to cause by 2030 an additional decline in the average replacement rate by 6.5-7.5 pap.This would bring the replacement rate to the level about 20 percent, i.e. one quarter below the replacement rate that i s likely to be generated inthe system without the reforms (and 40 percent below the current replacement rate). This would represent a major compression o f the real value of the pensionbenefit. 4.82 The ratio of the average pension and subsistence minimum in the latter scenario is about 20 percent below those in the Group Iscenarios. It i s worth noting, however, that it remains considerably higher than 1 (by 60-120 percent in 2030), which i s a major improvement relative to the current situation. 4.83 Inthe optimistic case, despite an assumedconsiderable expansion inthe tax base (See Section D), the average replacement rate would decline relative to the baseline inthe Group I scenarios. By 2030 the decline would amount to 3.5-4 percentage points, which bring the replacement rate to the range o f 22.7-23.8 percent (Figure 4.12). Along with this, potential fiscal costs o f closing the gap in the pension system (measured against the replacement rate target o f 30 percent) would increase from approximately 0.6-0.9 to 1.7-2.1 percent of GDP. Figure4.12: PensionReplacement Ratio, Comparisons of Scenariosin Group 1 and Group 2 PensionReplacementRatio,compariosons of scenarios 400,- - _inGroup 1andGroup2, % _ - - __ _ _ - I_ - - --Scenario 2 (Grl)- -Scenario 100 - --Scenano4(Grl)2 (GR) - Scenano 4 (GR) 5 0 - 0 0 r 4.84 Due to a linear dependency among the parameters in the PAYG system, the above results may be used for generating rough estimates for potential alternative plans for cuts in the contribution rates inthe PAYG pillar. On average 1 percentage point inreduction of the UST rate could be compensated for by an increase inthe share o f taxable payroll in GDP by 1.0-1.3 p.p. At the same time, a 1 p.p. cut inthe UST tax rate could be compensated by 0.25- 0.3 percent of GDP in additional external (budget) transfers to the pension system (Table 131 4.11). Annex 4.1 presents additional results o f sensitivity analysis: it helps to get the idea of the variation inreal value o f average pension for different pairs o f UST contribution rate and payroll share inGDP. Table 4.11: Some resultsof sensitivity analysisfor 2030: equivalencetable 1 p.p, in USTtax rate cut Could be compensated by an expansion in the UST tax base of 1.O -1.3 P.P. 1p.p. inUSTtax rate cut Could be compensatedby an additional budget transfer of 0.25- 0.30% of GDP 4.85 An important conclusion from thispart o fthe analysis is that without an increase inthe retirement age even a rather substantial growth o f the tax base does not allow halting the decline in the replacement rate. At the same time, as shown below, with the retirement age increase (as in Group I11scenarios) and particularly when the retirement age increases along with cuts in UST benefits (as in Group IV scenarios), the effect of a declining replacement rate i s fully compensated. 4.86 The basic distinctions between the scenarios within Group I1 (Figure 4.12) are associated with the differences inGDP growth rate, as well as with differences in the shares o f payroll inGDP. The results also suggest that despite a considerable decline inthe replacement rate, the ratio between the pension benefit and the subsistence minimum i s less sensitive to cuts in the U S T tax rate (Figure 4.13). The latter ratio did not change much relative to the results insimilar scenarios of Group I. Figure 4.13: Ratio BetweenAverage Pensionand Pensioner's SubsistenceMinimum, Comparisons of Scenarios i,nGroup 1 and Group 2 Ratio betweenaveragepensionandpensioner's subsistenceminimum, compariosonsofscenarios in I 2004 2009 2014 2019 2024 2029 2034 2039 2044 2049 " ' I Latest governmentproposal: rate cut of 8percentage points 4.87 Earlier in 2004, the Government announced a plan to introduce a more drastic cut in the UST rate compare to what has been discussed in this Chapter. Specifically, the latest Government's proposal calls for a rate cut o f 8 percentage points, from 28 to 20 percent, 132 which will be allocated among base (cut by 6 pep.)and fully funded (cut by 2 p.p.) components of the pension benefit. The proposal includes the following compensatory measures: a) substantial reduction in participation in the filly funded pillar, which, as we estimate, would reduce the average replacement ratio in this pillar by 0.8-1.0 p.p. by2030; b) utilization of the currently expected surpluses in the base component to accommodate a decline in the contribution rate to this component from 14 to 8 percent; and c) compensatory budget transfers to the Pension Fund funded either from regular budget revenues or from the StabilizationFund." 4.88 We estimate that the current plan, if implemented, would utilize about a half of the expected surplus in the base pension component, and it would reduce the affordable replacement rate in the PAYG system by 4-5 p.p. Thus, if there i s no compensation for the pension system from the budget, the total reduction in the replacement ratio (including FFP) in2030 couldreach 5-6 p.p. relative to the baseline scenario inGroup Iwithout the rate cut. This i s equivalent to a 20-25 percent decline in real average pensions. This would raise potential (longer term) fiscal costs of supporting the replacement ratio o f 30 percent to above 2 percent o f GDP a year. 4.89 Moreover, the short term negative impact o f the proposed could be also significant. The PAYG component may lose more than 25 percent (6/22) o f its regular incomes, while the affordable replacement rate would decline correspondingly. Based on the estimates obtained inthis Chapter, we would recommend that the government should be prepared to compensate most o f the lost revenues in the PAYG component to avoid either decline in real value o f current pensions or accumulation of pension arrears. Such compensation may amount to an equivalent of not less than 4 pap.o f the UST rate and be funded from regular budget revenues. It is estimated that the total annual budget costs associated with such compensation would amount to 1.0-1.1 percent of GDP. The compensation has to be provided until either considerable improvements in the revenue performance o f the pension system materialize, driven either by strong recovery in real wages or by a decline inthe share o f informal wages or by both, or policy decisions are made, which reduce Pension's Fund's financing needs. Sensitivity to the high rates of economic growth 4.90 To reflect the fact that our base growth assumptions are quite conservative, we also undertook an alternative set o f simulations based on the assumption o f the much higher (6 percent average) growth for the period 2004-09. However, in this case we found rather an insignificant change to our base results. This i s because higher GDP growth i s expected to be accompanied by higher growth in real wages, which limits opportunities for additional improvements inthe replacement rate. 91The government indicatedthat in 2005 it may spend as much as 0.4 percent of GDP from the Stabilization Fundto cover the Pension Fund deficit that would derive from the introducedthe UCT rate cut (Prime-TASS, August 24, 2004). 133 Group 111.UST reformssupplementedby an increaseinthe retirementageby 5 years for men and 10 years for women 4.91 The growth inthe retirement age by 5 years for men and by 10 years for women inthe scenarios o f Group I11leads to a considerable decline in the system dependency rate. This decline amounts to approximately 40 percent by 2050 (Figure 4.14), and, therefore, to an increase in the average P A Y G pension benefit by about 40 percent. Due to the increase of the contribution period, the affordable overall replacement rate in the optimistic case (with a higher tax base) increases to 33-35 percent in 2030 (Figure 4.15), Le., by 20-30 percent as compared with the similar scenarios o f Group I.Even in the base case, i.e. without the expansion o f the tax base, the replacement rate still remains above 30 percent. This considerably reduces the risks o f budget support for the pension system. 4.92 The ratio between the average pension and the subsistence minimum increases in the long term by approximately 50 percent as compared with the scenarios o f both Groups Iand 11.Inabsolute terms, the latterratio mayreach 2.5-3.5 in2025-30. 4.93 Our sensitivity analysis suggests that, without a mandatory increase inretirement age, a voluntary delay in retirement by 5 years under the current conditions and respective extension o f the contribution period would have a noticeable impact on the individual pension. The replacement rate would be higher by 8-25 percent for both NDC and fully funded components, depending upon a particular combination o f values for real wage growth and real interest rate. However, the size o f this increase may be insufficient to create strong incentives for participants to retire later. The Government may consider amendments to the benefit formula to strengthen the link between the retiree's length o f service and hidher benefit size. The latter step would be o f special importance if there i s a delay inresolving the issue o f mandatory increase inthe retirement age. Figure4.14: SystemDependency Ratio,Comparisons of Scenarios in Group 1and Group3 1 , Systemdependencyratio,compariosonsofscenariosin 1.0 Group 1andGroup3, YO ~ I 0.94 0 7 0 8 0 5 0 4 0 3 0 2 0 1 0 0 2004 2009 2014 2019 2024 2029 2034 2039 2044 2049 134 Figure4.15: PensionReplacement Ratio, Comparisonsof Scenarios in Group 1and Group 3 PensionReplacementRatio,compariosonsofscenarios 45.0 ___l__ll____ inGroup1andGroup3,--__- YO 40.0 35 0 30 0 125 0 20.0 1 1 - - --Scenario 2 ( G r l ) l dI -Scenario --Scenario4 2 (Gr3) (Grl) 100 - 5 0 - 0 0 7 Group IV. Additionalreformsaimed at the reductionof currentUSTprivileges 4.94 The results of the scenarios o f Group IV show a modest additional growth o f the affordable overall replacement rate by about 1 percentage point relative to the similar scenarios in Group 111. This additional growth reflects primarily the following changes: (i) a one time twofold increase inthe rates of pension contributions paid by the self-employed and indexation o f these rates in accordance with the growth rate o f average wages, and (ii)a reduction in the number o f people who are eligible for reduced UST rates from 15 to 10 percent o f the employed. The latter decline may be achieved through a tighter definition o f eligibility rules and their stricter enforcement. 4.95 Even under the most optimistic reform and macro assumptions as in Group IV, the replacement rate remains below 36 percent in2030. Thus, if the policy target i s reaching a 40 percent replacement rate for the average pension, this objective can not be achieved without an expansion in additional private pensions, based on additional voluntary contributions. Encouraging the development o f private pensions i s a large, separate task that would require the introduction o f supportive regulatory, institutional and tax frameworks. GroupV. Effectsof expandedimmigration 4.96 Inthis group of scenarios we simulated a potential impact of a considerable migration inflow to Russia on the trends in the pension system. As a base case, we used a scenario 12 from the Group 1, and compared it with the two additional scenarios that differ by both the level and structure o f migration inflow. Inthese additional scenarios, the annual net number o f migrants increases from about 100,000 to respectively 400,000 and 600,000 a year. In addition, it i s assumed that the immigration inflow would have a higher share o f people o f working age (and thus, a higher share of contributors to the pension system) than the average share of working age people inthe Russianpopulation. The latter assumption i s reflected ina higher share of labor force participants, which i s increased by 4-7 p.p. 135 4.97 While an increase in migration would have a drastic impact on the overall demographic trends in Russia, it has rather a modest influence on parameters o f the pension system. Table 4.12 presents the main results of these simulations, which do not change much our previous conclusions - the system dependency ratio continue to growth rapidly, while the replacement rate still declines below 30 percent. Even under the boldest assumption on immigration level (which by 2050 bring a considerable increase of 30 percent in the total population), the replacement rate increases by less than 15 percent. In no way, the active migrationpolicy inRussia could become a substitute for a comprehensive pensionreform. Table 4.12: Potentialimpact of an increaseinimmigrationon the pensionsystem Russia's population, System dependency Replacementrate: total, million ratio percent 2030 2050 2030 2050 2030 2050 Base scenario 12, Group I,low growth 122 102 0.72 0.93 26.4 24.6 The same scenariowith additional immigration: a) of 300,000 a year 132 119 0.68 0.84 28.0 26.9 b) of 500,000 ayear 138 130 0.66 0.8 29.1 28.2 G. SUMMARY AND CONCLUSIONS 4.98 Based on simulations o f various policy scenarios, this Chapter estimates potential fiscal costs associated with various developments in Russia's pension system. It finds that these costs are likely to emerge as a result of the declining relative value o f old age pension and associated political pressures for budget support to the pension system. These results are based on the projections that, within the framework of the basic macroeconomic projections and with the current demographic trends, the long-term financial self-sustainability o f the existing Russian pension system may be ensured only at the cost of a substantial reduction in the average replacement rate, from 33 percent in 2002 to 24.4-27.8 percent in 2030 and to 23.5-25.5 percent in2050. 4.99 This replacement rate reduction is driven primarily by demographic factors, while its dynamics is less sensitive to variations in macroeconomic parameters, such as growth and unemployment rates. Even under the most favorable macroeconomic scenario without an increase inthe retirementage, the overall replacement rate declines below 30 percent by 2020 and by 2050 it goes further down to 25 percent. The rate o f decline would be the most pronounced in the period 2016-2025, when it is the most likely to expect that political pressures to support the pension system through budget transfers would emerge. While higher economic growth results ina much higher real value of pensions, it does not help to close the gap between growth in wages and pensions. Additional immigration and any realistic improvements in demographic and employment trends do not change much these basic results. 4.100 Table 4.13 summarizes our results for expected replacement rates and associated potential fiscal costs. Fiscal costs are measured against a target of maintaining the replacement rate at 30 percent. Inthe baseline scenarios, annual fiscal costs amount to 0.25- 0.55 percent of GDP in 2020 and to 0.55-0.90 percent o f GDP in 2030. However, potential costs increase rapidly in all scenarios with the reduced contribution rates. In the most 136 pessimistic case, with 8 p.p. cut in the USR rates and no expansion in the tax base, annual costs to the budget exceed 2 percent of GDP in 2030. In contrast, the scenarios with the increased retirementage do not requireany budgetsupport to provide a replacementrate of 30 percent. 4.101 However, the above fiscal costs would expand quite rapidly if the Government implements significant cuts in the pension contribution rates. The simulation results suggest that even under the most optimistic assumptions about potential expansion in the tax base (reaching 35 percent o f GDP), this factor cannot compensate for a cut inthe UST rate of four percentage points if it i s not supported by additional pension reforms. Political demands for maintaining simultaneously higherbenefits and low USTrates can result inquick widening o f the deficit o f the pension system, which would become a contingent liability of the government. Even with a modest rate cut of four p.p., annual fiscal costs would increase to 1.35-1.70 percent o f GDP. If the current Government proposal to cut the UST rate by eight pap.(6+2) is fully implemented,this would raise the fiscal costs above 2 percent of GDP per annum. 4.102 At the same time, all scenarios show a gradual increase inthe value o f average pension relative to the subsistence minimum. This indicates that with economic growth the real incomes o f pensioners will also grow steadily. However, there are two concerns with the patterns o f this future growth: a) in the initial period till 2015, despite a low level o f current pensions, the growth will be rather slow; and b) as reflected in the falling replacement rate, the growth inpensions will be lagging growth inreal wages, which may become a politically sensitive issue. 4.103 A gradual increase inthe retirementage by 5 years for men and 10 years for women in the long-term perspective would reduce the system dependency ratio by approximately 40 percent. This makes it possible to provide practically full compensation for the negative impact o f the both UST rate reduction of 4 pp. and demographic factors on average pensions. The replacement rate (assuming the same reduction o f the UST rate o f 4 pap.)in 2030 would grow to 33-35 percent, Le., would be the same or higher than its 2002 level. Even inthe case with no expansion inthe tax base, the replacement rate would remain above 30 percent. This drastically reduces a risk for budget support to the pension system. Table 4.13: Summary of the simulationresults for replacementrate and potentialfiscal costs Scenarios Replacement rate Fiscal costs per annum, 2020 2030 2020 2030 Scenarios of Group I 12: base case, no reforms, low growth 27.7 26.4 0.55 0.90 14: base case, no reforms, high growth 28.9 27.8 0.25 0.55 12 8 p.p.cut inUST rate + 2.45 14 + 8 p.p.cut inUSTrate 21.o 22.0 2.15 Scenarios of Group I1 22: 4 p.p.cut in USTrate, expandedtax base, low growth 23.4 22.7 1.70 2.05 24: 4 p.p.cut in USTrate, expanded tax base, highgrowth 24.7 23.8 1.35 1.70 32: Scen. 22 + increase inthe retirement age, low growth Scenarios of Group I11 33.0 34: Scen. 24 + increase inthe retirement age, highgrowth 35.5 37.1 34.8 42: Scen. 32 + reduction o f USTprivileges, low growth Scenarios of Group IV 36.4 33.8 44: Scen. 34 + reduction of USTprivileges, highgrowth 38.0 35.7 Scenarios of Group V Scenario 12 annual immigration o f 300,000 + 28.0 0.50 Scenario 12 + annual immigration of 500,000 29.1 0.20 Note: Fiscal costs measured against a target of maintaining the replacement rate at 30 percent. 137 4.104 Additional policy reforms, relatedmainly to a twofold increase of the contribution rate for the self-employed and further elimination of tax exemptions, may lead to a modest additional growth inthe affordable replacement rate by about 1percentage point. 4.105 Based on these findings the following recommendations on pension policy could be suggested: The simulations confirm the need to supplement the proposed reduction in the pension contribution rate with a set of decisions, such as a gradual increase inthe retirement age, aimed at a longer-term sustainability of the pension system and an increase in the real value o f future pension benefits. A decision on the retirement age i s critical for longer- term prospects o f the pension system. Further delaying this decision leads to high social and political costs in terms of reduction o f the relative value of future pension benefits, and, as argued in this Chapter, is likely to trigger significant longer-term budget liabilities. Moreover, the political difficulties of rising the retirement age may increase in the future as the population continues to age. The second best solution with respect to the retirement age increase, given the political sensitivity o f the issue, would be the introduction of amendments to the Pension Law, which would modify the formula that defines the amount o f retirement benefit in a way that create much stronger incentives to delay retirementvoluntarily. Trends inthe share of the taxable payroll in GDP will play a critical role in determining future results o f pension reform. This highlights the importance o f policies aimed at stabilizing income and payroll taxation (to facilitate stability o f taxpayers' expectations), as well as at the removal o f various administrative barriers inthe economy that currently hold back reduction o f shadow incomeslwages. It seems unlikely that more radical cuts in the UST rate (in excess of 4 percent points) couldbe compensated for by the combined positive impact of the tax base expansion and an increase in the retirement age. Under the circumstances, the first best strategy for reducing the UST rates would be to avoid longer-term commitments regarding future tax rate reductions in stages -- giving the system some time to stabilize after the initial additional cuts inthe contribution rate that would exceed 4 p.p. It may be rational to do round of UST rate cut of 4 p.p., and after accumulating certain experiences of operating with newparameters, planning for next steps of tax cuts andpensionreforms. At the same time, ifthe government is convinced that the USTrate cut larger than 4 pup. i s urgently requiredto accelerate the formalization of the economy and sustain growth, it should be prepared to provide a substantial compensation to the Pension Fund from regular budget revenues (including the Stabilization Fund). In particular, if the current proposal i s finalized to cut the average UST rate by 8 p.p., an annual budget transfer of 1.0-1.1 percent of GDP may be needed immediately to avoid either decline inreal value o f current pensions or accumulation of pension arrears. The compensation has to be provided until considerable improvements in the revenue performance o f the pension system materialize and/or policy decisions are made, which reduce Pension Fund's financing needs. 138 REFERENCES Andrews, Emily S. 2001. "Kazakhstan: An Ambitious Pension Reform." Social Protection Discussion Paper No. 0104. The World Bank. Washington DC. ,Anusic, Zoran and Katerina Petrina. 2003. "Reforming Ukrainian Pension System.'' Background paper for the Ukrainian CEM. Mimeo. The World Bank. Washington D C. Balcerzhak-Paradovska B., A. Chlon-Dominczak, I.Kotovska, A. Olejnichuk-Merta, I. Topinska, and I. Woycicka. 2003, "The Gender Dimension of Social Security Reform inPoland." InE.Fultz, M.Ruck, and S. Steinhibler (Eds.) The Gender Dimension of Social Security reforms in Central and Eastern Europe: Case Studies of Czech Republic, Hungary, and Poland. Budapest: ILO. Baskakov V.N., A.L. Lelchuk, and D.V. Pomazkin. 2003. "Modeling the Pension System o f the Russian Federation." In "The Pension System: A Model for Russia and International Experience." Moscow: Moscow Public Science Foundation. pp. 10-90. Balls, Ed, and Gus O'Donnell. 2001. Reforming Britain's Economic and Financial Policy: Towards Greater Economic Stability, HMTreasury, Palgrave. Beetsma, Roe1M.W.J., and Xavier Debrun.2004. "Reconciling Stability and Growth: Smart Pacts and Structural Reforms," Stajf Papers, International Monetary Fund, Vol. 51, No. 3. Bolshakova, Ekaterina. 2004. "Participation in the Pension System by Employees of Small Enterprises. Summary o f the existing regulatory framework." Mimeo. Moscow, IET (InRussian). Business Environment and Enterprise Performance Survey (BEEPS). 1999. World Bank and European Bank for Reconstruction and Development. httu://info.worldbank.org;/g;overnance/beeps/ Business Environment and Enterprise Performance Survey (BEEPS). 2002. World Bank, and European Bank for Reconstruction and Development. http://info.worldbank.org/governance/beeps2002/ Chand, Sheetal K. and Albert Jaenger. 1999. "Reform Options for Pay-As-You-Go Public Pension Systems". 1999. Social Protection Discussion Paper Series. No. 9927. The World Bank. Washington, DC. Chen, Shaohua and Martin Ravallion. 2000. "How Did the World's Poorest Fare in the 1990s?" The World Bank. Washington, DC. Collier, Paul, and Jan W. Gunning. 1999. "Trade Shocks: Theory and Evidence," in "Trade Shocks in Developing Countries,'' OxfordUniversity Press. 139 Concept o f the Pension Reform. Employers' Position. 2001. Russian Union o f Industrialists and Entrepreneurs. Mimeo. May 31.Moscow. (InRussian). De Castello-Branco, Marta. 1998. "Pension Reform in the Baltics, Russia, and Other Countries o f the Former Soviet Union". IMF Workingpaper No. 98/1, Washington, DC. Denisova, Julia. 2001. "We arepaid: where and how much? Results o f comparative analysis " o f base salaries of key employment categories in Moscow, St. Petersburg, Nizhny Novgorod, Ekaterinburg and Perm. Ekonomika I Zhizn'. Vol. 13-14. July. (In Russian). Dmitriev M., D. Pomazkin, 0. Sinyavskaya, and A. Stolyarov. 1999. Reform of Particular Social Sectors in Russia. Moscow, Institute of Economy inTransition (InRussian). Easterly, William and Luis Serven, Eds. 2003. The Limits of Stabilization. Infrastructure, Public DeJicits, and Growth in Latin America. The World Bank. Washington, DC. Eismont O., S. Chemyavsky, and J. Rotwell, 2003. Overview of a Possible Impact of Russian Energy Sector Restructuring. The Russian School o f Economics. Moscow. Gimpelson, V.E. 2002. Employment and Structure of the Russian Bureaucracy: between the Soviet Nomenklatura and Public Service to Civil Society. Higher School o f Economics, Moscow (InRussian). Gray, Cheryl, Joel Hellman, and Randi Rytennan. 2004. Anticorruption in Transition 2: Corruption in Enterprise-State Interactions in Europe and Central Asia 1999-2002. The World Bank. Washington DC. Hamilton, Ellen.2004. "Exploring Housing Subsidies to Households inRussia." Mimeo. The World Bank. Washington, DC. Higher School o f Economics (HSE). 2003. "Comparative Pay and Benefits Survey, Public and Private Sectors." March. Higher School of Economics (HSE). 2004. "Urgent Tasks o f the Reform o f Executive Authorities Apparatus." Kommersant. February 12. Holzmann, Robert. 1998. Financing the Transition to Multipillar. The World Bank. December.Washington, DC. Holzmann Robert, Indermit Gill, Richard Him, Gregorio Impavido, Alberto R. Musalem, Michal Rutkowski, and Anita Schwarz. 2004. "Old Age Income Support in the 21st Century: The World Bank's Perspective on Pension Systems and Reform." The World Bank. May. Washington, DC. IMF. 2002. Slovak Republic: Selected Issues and Statistical Appendix. IMF Country Report No. 02/210. September, Washington DC. IMF. 2003a. Russia - Medium Term Fiscal Stance and the Potential Role of an Oil Stabilization Fund. February. Washington, DC. 140 IMF. 2003b. Russian Federation. Staff Report for the 2003 Article IV Consultation. Washington, DC. IMF. 2004. Russian Federation. Staff Report for the 2004 Article IV Consultation. Washington, DC. Information For Democracy Foundation (INDEM).2002. Regional Public Fund"Informatics for Democracy." Pay and Corruption: How to Pay to Russian Public Servants. Moscow. Institute o f Urban Economy. 2003a. Practices in Reforming the Housing and Utility Sector. Moscow (inRussian). Institute o f Urban Economy. 2003b. Review of Government Expenditures for Russia. Final Report for the component "Housing and Utility Sector." Report prepared for the TACIS. Moscow. Kakwani N.and Zurab Sajaia. 2004. "New Poverty Thresholds for Russia." The World Bank. Mimeo. May. Washington DC. Kolodeznikova, Anastasia, and Raymond Struyk. 1997. Proxy Means Testing for Russia: Challenges and Perspectives. Urban Institute. Moscow. September. Kopits, George F. Ed. 2004. Rules-Bused Fiscal Policy in Emerging Markets: Background, Anulysis, and Prospects. London, Palgrave Macmillan. Kopits, George F. and Steven A. Symansky. 1998. "Fiscal Policy Rules", IMF Occasional Paper No. 162, Washington, DC. Kwon, Goohoon 2003. "Post-crisis Fiscal Revenue Developments in Russia: From an Oil Perspective," Public Finance and Management, #3(4). Misikhina, Svetlana. 2002. "Poverty, Housing Subsidies, Discount Payments for Housing and Communal Services in Six Regions of the Russian Federation in 2000." Mimeo. Moscow. OECD. 2002. OECD Economic Outlook, Chapter 4. OECDBIGMA OECD (Organization for Economic Co-operation and Development/Support for Improvement in Governance and Management). 1997. "Budgeting and Monitoring of Personnel Costs." SIGMA Paper No. 11. Perry, Guillermo. 2004. "Can Fiscal Rules Help Reduce Macroeconomic Volatility?" In Kopits (2004). Pinto, Brian. 1987. "Nigeria During and After the Oil Boom: A Policy Comparison with Indonesia," The WorldBank Economic Review, 1 (3). Plevko, Snjezana. 2004. "Romania Pension Policy Note." Draft. The World Bank. Washington, DC. 141 Poletaev, Andrey, Arganovich, Mark, and Zharova, Ljubov. 2003. "Russian Education inthe Context of International Indicators." Moscow (InRussian). Populationo f Russia. 2002. Moscow. Rosstat (InRussian). Posarac, Aleksandra and Mansoora Rashid. 2002. "The Non-income Dimension of Poverty: Russia." Mimeo. The World Bank. Washington, DC. Rofman, Rafael. 2002. The Pension System and the Crisis in Argentina: Learning the Lessons. Background Paper for Regional Study on Social Security Reform. The World Bank. Mimeo. Roik, Valentin. 2003. Financing o f Social Extra-budgetary Funds. Pensionnyue Fondy i Investitsii, issue 6 (InRussian) Rosstat 2002a. "Cadre of State Authorities in Russia. " Moscow. Rosstat 2002b. Employment and Pay to the Employees of State and Municipal Entities as of October 2002. Statistical Bulletin, Moscow. Rosstat 2003. Employment and Pay to the Employees of State Authorities and Local Self- Government. www.gks.ru. Rutkowski, Michal. 2002. Pensions in Europe: Paradigmatic and parametric reforms in EU Accession countries in the Context of EU Pension System Changes. Journal of Transforming Economies and Societies, 9: 1: 2-26. Rutkowski, Michal. 2004. "Home-made Pension Reforms in Central and EasternEurope and the Evolution o f the World Bank Approach to Modem Pension Systems." In Martin Reinand Winfrid Schmahl(Eds.) Rethinking the Welfare State: thePolitical Economy of Pension Reform. Martin Edward Elgar. Cheltenhamand Northhampton. Schwarz, Anita M. and Asli Demirguc-Kunt 1999. "Taking Stock of Pension Reforms Around the World". Social Protection Discussion Paper No. 9917. The World Bank. Washington, DC. Schiavo-Campo, et al. 1997. Government Employment and Pay in Global Perspective: A Selective Synthesis of International Facts, Policies and Experience. The World Bank. Washington, DC. Simulation and Actuarial Estimates of Corporate Pension Programs. 2003. Moscow: Prosveshcheniye. (InRussian). Sinelnikov-Murylev, S., S. Batkibekov, P. Kadotchnikov, and D. Nikepelov, 2003. Otsenka Rezul 'tatov Reformy Podokhodnogo Naloga v Rossiiskoi Federatsii. Moscow. Institute o f Economic Transition. (inRussian). Sinelnikov Sergei, Said Batkibekov, Pave1 Kadochnikov, and Dmitri Nekipelov. 2003. "Evaluation o f Income Tax Reform in the Russian Federation", Voprosy Ekonomiki, issue 6 (InRussian). 142 Smirnov S.N. and N.I.Isaev. 2003. Pension Reform: Social and Economic Aspects. Research Publications by the Institute of Economy in Transition. Issue 59R. Moscow (In Russian). Stott, Rachel. 2002. "Analysis o f Pension Modeling Undertaken by the MOEDT". TACIS. Moscow. September 24. Starodubrovskaya, Irina. 2003. "Reforms in Housing and Utility Services: InQuest for New Approaches." In Economics of Transition. Essays of Economic Policies of Post- communist Russia, 1998-2002,Moscow. Del0 Publishers (inRussian). Urban Institute. 1996. Housing Allowance Program. Russia 's Longitudinal Household Survey. Moscow. May. Vatolkin, E., Ljuboshits, E., Khmstalev, E., and Tsymbal, V. 2002. "Reform o f the Systemo f the Armed Forces Rank and File and Junior Management Staffing." Research Paper No. 39R. Instituteof Economic Transition. Moscow. World Bank. 1996a. Russian Federation: Towards Medium-Term Viability. Country Economic Memorandum. Washington, D.C. World Bank. 1996b. Fiscal Management in Russia. A World Bank Country Study. Washington, D.C. World Bank: 1998a. Russian Federation. Housing and Utility Services: Policy Priorities for the New Stage of Reforms. Washington, DC. World Bank. 1998b. Russian Federation. Affordability of Full Cost Coverage in Housing and Communal Services. Washington, DC. World Bank. 1998c. "Russian Federation: Analysis of Public Debt Sustainability." Mimeo, April, Washington, D.C, World Bank. 2000. "Benchmarking Public Expenditure Analysis inthe Russian Federation." Mimeo. Washington, D.C. World Bank. 2001. Russia: Towards Improving the Efficiency of Public Investment Expenditures. Report No.22693-RU, September 18. Washington, D.C World Bank. 2003a. Housing and Communal Services in Russia: Completing the Transition to a Market Economy.A Policy Note. Washington, DC. World Bank. 2003b. Russian Economic Report #6, www.wor1dbank.ora.m World Bank. 2003c. Pension Reform in Russia: Design and Implementation. Washington, DC. World Bank. 2004a. "Ruble Strengthening, Dutch Disease and Natural Resource Abundance: Real Threats to Russia's Economy?," The World Bank Policy Note. 143 World Bank. 2004b. Russia: Transition Meets Development. Country Economic Memorandum for the Russian Federation. Washington, DC. World Bank. 2004c. Russian Federation Poverty Assessment. Report No. 28923-RU. Washington, DC. World Bank Administrative and Civil Service Reform Web-Site. http://www 1.worldbank.org/publicsector/civilservice/index.html World Bank Public Sector Employment and Wage Database. http://www 1.worldbank.org/publicsector/civilservice/development.htm World HealthOrganization (WHO) http://www.who.org. Yasin, Yevgeny. 2003. Nerynochnyi Sektor. Structurnuye Reformy i Ekonomicheskii Rost. Moscow: Liberal'naya MissiyaFoundation (inRussian). Zolotareva, Anna. 2002. "Nefinansiruemuye Federal'nuye Mandaty." Institute of Economic Transition. Moscow. Mimeo. 144 ANNEXES ANNEX2.1. DEFINITIONS CIVILSERVICE REFORM ON The following definitions on Civil Service are used inChapter 2. Federal Civil Service includes all employees o f the federal executive, legislative, judiciary and other federal bodies (these bodies include the Accounting Chamber, Central Election Commission, and the Apparatus o f the Plenipotentiary on Human Rights) that enjoy the status o f federal civil servants and are financed from the federal budget. Employment o f federal civil servants i s regulated by the framework L a w on State Service System enacted inMay 2003. SubnationalCivil Service includes all employees o f the regional and municipal executive, legislative, judiciary and other regional and municipal bodies that have the status o f civil servants o f subjects o f Federation or municipal servants and are financed from the regional and municipal budgetsg2.Employment in subnational civil service i s regulatedby civil service and municipal service legislation (including federal legislation on municipal service and regional legislation on civil service at the regional level). Categoriesof Civil Servantsrefer to the groups o f civil service positions used, inter alia, for determination o f cash compensation, in-kindbenefits, and the like. A detailed table on the existing categories o f federal civil servants i s presented inAnnex 2.2. Core Government Administration Employment includes federal and subnational civil service, as well as other employees o f the executive, legislative, judiciary and other authorities that do not have the status o f federal/subnational civil servants (employment o f such employees i s usually regulated by Labor Code)93 that are performing functions not related to maintaining public order and internal security. Core government administration employment includes employees o f the federal authorities that are either headquarter- basedor deconcentrated(but inany case financed directly from the federal budget) and employees o f regional and municipal authorities (financed from subnational budgets). Civilian Public Sector Employmentincludes employees o f health, education, research and development, and other social sector branches financed from the federal, regional, and municipal budgets. Staff o f the Ministry o f Education, Ministry o f Health, etc., i s not 92The term includesboth "civil service of a subject of the Federation" - a category used in accordancewith the new Law "On the System of State Service in the Russian Federation" No.58FZ dated May 27, 2003 - and municipal service. Civil service of a subject of the Federationis to be financed from the regional budgets. Civil servants ofthe subjects of the Federationmay also be financed from the federal budget, ifthere is an appropriate provision to this effect included ina federal law. Municipal servants are financed from sub-nationalbudgets. 93Insome cases, government employment may be regulatedby specific legislation, such as a Law on Customs Service. For example, all uniformed and managerial staff (division heads and above) in the State Customs Committee are not counted as civil servants, but as customs servants. At the same time, some technical and support staff, as well as temporary employees working in the government agencies are not considered civil servants, andtheir employment is regulatedby the Labor Code. It is expectedthat during the implementation of civil servicereform, the legal framework would become simpler andmore systematic. 145 included in this category as they are part o f the above mentioned category o f core government administration employment. Police Service includes all employees o f the authorities responsible for maintaining public order and internal security. Police service includes, inter alia, Federal Security Service, MinistryofInterior, etc. Armed Forces include all personnel employed by military authorities (i.e., Ministry o f Defense, military officers and solders). General Government Employment includes core government administration employment, civilian public sector employment, police service and armed forces94. Total Public Employment includes general government employment and SOE employees (employees o f enterprises that are majority-owned by the government). SOE employment and wages are beyond the scope o f this Chapter. The definition i s listed here primarily for the purposes o f cross-country comparisons. 94This definition corresponds to the one adopted inthe International Standard of Industrial Classification of All Economic Activities (ISIC). According to the ISIC, General Government refers to employment in "all government departments, offices, organizations and other bodies which are agencies or instruments of the central or local authorities whether accounted for or financed in ordinary or extraordinary budgets or extra- budgetary funds. They are not solely engaged in public administration, but also in defense and public order, in the promotion o f economic growth and in the provision of education, health, cultural and social services." (International Standard o f Industrial Classification o f All Economic Activities (ISIC), Series M No.4, Rev 3 - 1990). 146 The main components of the government employmentare illustratedinFigureA2.1, : FigureA2.1: TotalPublicEmploymentinthe RussianFederation I I. a I TOTAL PUBLIC EMPLOYMENT General Government SOE Employees Employment t/ Civilian Public Core Government I Sector Administration 1 Police Service I Empolyment Employment II IIIIIIIIIIIII Health, Physical Culture and Sports, Federal Social Services Authorities Armed Forces - -based HI2 -federal IIIIII I civil servantd I -other employees II I I I I Research and Development Municipal subnational civil service i I \ '-_______-_-_-______ other employees ; General Government Sectors Analysed in this Note Notes: Sectors in Civilian Public Sectors are divided in accordancewith Rosstat (2002b). Sectors of the general government analyzed Chapter 2 are depicted inside the dot line area. Cash Compensation refers to an average monthly amount of the base salary and all other payments incash receivedby an employee for services rendered, includingpayroll taxes. Wage Bill refers to the total annual amount of cash payments to all public employees in return for services rendered before deduction of payroll taxes. It i s calculated by multiplyingcash compensation by the number o f employees and by twelve (the number o f months). Non-Monetary Benefits refer to all in-kindbenefits, such as transportation, housing, meals, travel, but do not include intangible rewards, including job security, prestige, social privileges, training, trips abroad (see Annex 2.3 for classification o f public servants' both contractually provided and intangible rewards). Non-Wage Expenditures refer to overhead costs o f providing public services and include costs of office maintenance, equipment, staff training, etc. For purposes o f this Chapter, 147 these expenditures are calculated for core government administration only and represent a difference between total expenditures on core government administration and the sum of the wage bill, payroll taxes, and identifiable non-monetarybenefits. Compression Ratio for Federal Executive HQ-Based Civil Service i s estimated as a ratio between the average pay to federal executive HQ-based servants holding positions classified as Category C Top Group and average pay to federal executive HQ-based servants holding positions classified as Category C Junior Group. In this case, pay includes all monetary payments to civil servants. Compression Ratio for Federal Executive Civil Service i s estimated as a ratio between the average pay to federal executive HQ-based servants holding positions classified as Category C Top Group and average pay to federal executive servants employed in deconcentrated unitsand holdingpositions classified as Category C Junior Group. Inthis case, pay includes all monetary payments to civil servants. Internationally, there i s no methodological consensus for estimation o f compression ratios, The World Bank website and database estimate the compression ratio as a ratio of the highestsalary to the lowest on the central government's main salary scale. The OECD measures wage compression in OECD countries as the mean o f ninth decile salaries dividedby the mean of first decile salaries. The OECD's approach ensures that a handful of salaries will not dramatically skew the compression ratio. Given that the base salary paid to the civil servants in Russia represents only a part o f total monetary payments made to civil servants and the provisions o f the draft Federal Law "On State Civil Sewice of the Russian Federation suggest that this situation would continue to be the case inthe " future (see HSE (2003) for more specific analysis), for the purpose of this Chapter the compression ratios were estimated based on the average monetary payments to the officials holding positions in Category C Top Group (a proxy for high level officials) as compared to the average payment in Category C Junior Group (a proxy for the lowest grades in the service). Because o f this methodological difference, the cross-country comparisons in this Chapter may be used primarily for reference rather than for policy deliberations Civil Service Reform includes all reform efforts affecting the functions, structure, and employment o f core government administration employees. In Russia, unlike in most other countries, the government makes a clear distinction betweencivil service and public administration reforms, which derives from the division o f responsibilities for preparation of these two reform packages among different commissions within the Govemment. However, the two are interrelated (infact, the pay reform, which forms an integral part of civil service reform, i s an important condition of implementation o f the administrative reform, as it would supposedly reduce the internal opposition within the civil service to restructuring efforts). For the purpose of this Note, administration reform i s considered as a component o f the overall civil service reform agenda. Attrition denotes the amount o f actual staff reduction and i s estimated as a difference between the staffing levels at the beginning o f the reform and staffing levels at the end of thereform. 148 I I 4 : 4t! Y I I 3 t ANNEX2.3. CLASSIFICATION OF CIVIL SERVANTSREWARDS The basic classification o f civil servants rewards used for cross-country comparisons i s presented inthe figure below. FigureA2.2: Civil Servants Rewards Classification Source: World Bank Public Administration Website. Most o f the rewards listed in the figure are also applicable to federal and subnational civil servants inRussia. However, the structure o f monetary rewards included inthe wage bill is more complicated than shown in this figure. In addition to base wageisalary it usually includes allowances for grade, years in employment, as well as performance bonuses and other cash payments. It should be noted that there is a tendency to substituting in-kind allowances for transportation, housing, meals and travel (especially inthe police service and armed forces) with cash allowances. 152 ANNEX2.4. INTERNATIONAL PUBLIC SECTOR PAY AND EMPLOYMENT DATA, 1996-2000" Countries Public Sector Employment (%of Total Central Govt. Average population) Wage bill (Civilian central andArmedForces) govt. wage to per Civilian Central Govemment and Health Subnational Education % of GDP %of govt. capitaGDP Govemment Employees expend. ratio High-IncomeOECD Countries Australia 0.8 2.1 3.8 2.1 2.6 1.3 Canada 1.1 2.1 5.0 1.6 10.4 1.5 France 3.6 2.4 3.2 4.5 15.9 1.o Germany* * 1.o 5.1 1.o 7.9 0.9 UnitedKingdom 3.1 3.4 3.0 1.5** 5.9 1.4 UnitedStates 1.o 5.9 3.8 1.4 8.4 1.4 TransitionEconomies Bulgaria 0.4 0.4 2.2 5.2 7.9 1.o Czech Republic 4.8 2.3 1.o 2.7 8.1 0.8 Hungary 1.5 1.6 4.5 7.3 8.6 0.7 Kazakhstan 0.0 0.5 5.1 4.9 9.7 0.9 Poland 0.4 0.3 2.1 7.6 1.4 Russia*** 0.4 0.5 6.0 1.2 4.5 1.0 Ukraine 1.1 3.9 8.7 1.2 DevelopingCountries with FederalSystem Brazil 0.3** 1.3** 1.8** 12.2 16.5 India 0.3 0.6 0.4 1.8 11.0 4.8 Mexico 0.7 0.7 0.3 2.6** 24.9 1.1 RegionalAverages**** Africa 0.9 0.3 0.8 6.7 5.7 Asia 0.9 0.7 1.o 4.7 3.0 EastemEurope and 1.o 0.8 5.1 3.7 1.3 Former USSR Latin America and 1.2 0.7 1.1 4.9 2.5 Caribbean Middle East & 1.4 0.9 1.6 9.8 3.4 North Africa OECD 1.8 2.5 3.4 4.5 1.6 Overall 1.2 1.1 2.4 5.4 3.0 *Data are for the latest year available. ** Latest data available is for 1991-95. ***For Russia data is for 2002, Central Govemment includes employees ofthe federal authorities (both HQ- basedand deconcentrated, but does not includeMinistry of Interior. ****Regional data is for early 1990sbasedon sample cross-countrysurvey. Source: World Bank Public Sector Employment and WageDatabase, Rosstat, Schiavo-Campo, et al., "Government Employment and Pay in Global Perspective: A Selective Synthesisof International Facts, Policies and Experience," WorldBank (1997). 153 ANNEX2.5. ASSUMPTIONONCORE GOVERNMENTADMINISTRATION ATTRITION RATES Table A2.3: Significant AdministrativeReform (Attrition Rates) Employment Reduction Assumptions,Executive, Category C Group FederalHQ FederalDeconcentrated Subnational TOP 25% 25% 35% Chief 20% 20% 30% Lead 20% 20% 30% Senior 15% 15% 25% Junior 15% 15% 25% Other 20% 20% 30% Table A2.4: Fair AdministrativeReform (Attrition Rates) Employment Reduction Assumptions,Executive, Category C Group Federal HQ FederalDeconcentrated Subnational TOP 10% 10% 20% Chief 10% 10% 20% Lead 10% 10% 20% Senior 7% 7% 17% Junior 7% 7% 17% Other 7% 7% 17% 155 E N OI e z a c c N e I- c N =e z w zc 0 c p1 Cr c c p1 p1 c 5 Ec p1 ANNEX2.6. INTERPRETATION AND ANALYSISOF RESULTS: IMPACT OF INDIVIDUAL FACTORS As shown in Table 2.12 of the main text, the estimated annual fiscal costs of the reform scenarios vary significantly both in the mid-term perspective (between 1.3 and 3.3 GDP p.p. for 2006 as compared to 2003) and inthe long term (between 1.2 and 4.2 GDP p.p. for 2010 as comparedto 2003). Inthis annex we analyze the sensitivity o f these results to the key parameters used in the simulations (macroeconomic parameters, extent of pay reform, pace o f reform and extent of administrative reform). Our cost estimates are generated as additional government expenditures (in pep.of GDP) needed to finance operations of both government administrationand civilian public sector as a result of the proposedreforms. Sensitivitv to macroeconomicparameters. The results of the simulations proved that potential fiscal costs o f civil service reform are quite sensitive to relative growth rate o f private sector wages, and respectively to the changes inthe share o f the total payroll inGDP. To illustrate this sensitivity, three groups of reform scenarios were selected: (i) radical pay reform combined with significant administrative reform implemented at a highpace, Le., within 2004-2006 (scenarios 4, 16, and 28); (ii) moderate pay reform combined with fair administrative reforms and implemented at a mediumpace, Le., inthe period from 2004 to 2008 (scenarios 8,20, and 32); and (iii) moderate pay reform implementedduring 2004 - 2010 with no administrative reform (scenarios 5, 17 and 29). Within each group, there i s a considerable variation o f the growth (relative to GDP growth) in - - private wages, ashescribed inTable 2.5 inSection D. FigureA2.3: Sensitivity of CivilService ReformCosts to RealWage Growth" 4 5 bc?4 0 c f 3 5 Scenario16 3 0 0Scenario28 -f2 1 2 5 0Scenario6 w Scenario20 2 0 Scenario32 -55 1 1 5 w Scenario 5 0 0 Scenario17 Y c 0 5 0 0 I 2004 2005 2006 2008 2010 Source: Staff estimates. 95 Scenarios 4, 16, and 28 all assume a radical pay reform undertaken at high pace and accompanied by significant administrative reforms, but they differ by the growth rate in private sector wages. Respectively, scenarios 8, 20, and 32 all assume moderate pay adjustment at medium pace, and fair administrative reform effort, but the same difference in private wage growth. Scenarios 5, 17, and 19 assume moderate pay reform at a low pace and no administrative reform at all, while the same variation by wage growth remains intact. 157 As may be seen from Figure A2.3, the same reform scenario implemented in a situation when the real wages growth rate exceeds the GDP growth rate by 2 p.p. (Scenarios 4, 8 and 5) will be significantly more expensive for the budget that the one implemented ina situation when the share of real wages in GDP doesn't change over time (Scenarios 28, 32, and 29). As a result, variation intotal budget costs within each group o f scenarios amounts to 1.1 to 1.4 pap,of GDP. A decrease inrelative real wages growth by 1p.p. leads to about 0.4 -0.7percentofGDPinannualbudgetsavingfortheendofthereformperiod. It is notable that the degree of sensitivity to changes in relative wage growth varies among different groups of scenarios: it i s higher for scenarios without any administrative adjustments (Scenarios 5, 17, and 29), and lower for scenarios with administrative reform (Scenarios 4, 16 and 28 that assume radical pay adjustment and Scenarios 8, 20, and 32 that assume moderate pay changes). Such cost variation i s explained by the fact that Scenarios 5, 17 and 29, while leaving a significant residualpay gap, assume no any attrition. Thus, a much larger number o f core government administration and particularly civilian public sector employees benefit from the pay increase, and the stronger effort to close the compensation gap leads to more serious fiscal implications relative to the scenarios assuming significant attrition in civilian public sector and at least some attrition in core government administration. Overall, the more ambitious the pay reform, the more volatile are its fiscal costs. The above findings with respect to sensitivity of total reform costs to changes in the real wage growth remain valid when we look separately at the variation of costs associated with a narrower task -reformofthe core govemment administration only (Figure A2.4). FigureA2.4: Sensitivityof Civil Service ReformCosts to RealWage Growth(for Core Government Administration Only)96 2.0 1 Scenario 4 0 4 1.5 ( 3 - Scenario 16 0Scenario28 0Scenario8 W Scenario 2C D l Scenario 3; Scenario 5 0Scenario 17 2004 2005 2006 2008 2010 Source: StafJ'estimates. 96 See the previous footnote for a description of the scenarios. 158 Sensitivitv to the depth ofpav reform. It i s clear that the scenarios assuming radical pay reform tend to be associated with larger fiscal costs than those assuming a more moderate approach to closing the public-private compensation gap. To illustrate the difference incosts, two pairs o f reform scenarios were selected: (i) radical pay reform to be implemented a within 2004 - 2008 with moderate administrative changes (Scenario 13) and the same administrative reform effort but combined with a moderate pay reform (Scenario 20); and (ii) a radical pay reform to be implemented within 2004-2006 with significant administrative reforms (Scenario 16) and the same administrative reform effort but combined with a moderate pay reform option (Scenario 24). All these scenarios are cost out under the same wage growth assumptions. FigureA2.5: TotalFiscalCosts of Civil Service Reform:Radicalversus. ModeratePay Reform9' 2 2 3.00 2.50 -c t$ 5 Scenario 13 n 2.00 1.50 6Scenario20 gE =I+ HScenario 16 O 1.00 Scenario24 r" 0.50 0.00 + c) 2004 2005 2006 2008 2010 I Source: Stafestimates. The results of the comparison are presented in Figure A2.5. that confirms that the radical pay reform aiming at a 50 percent residual public-private compensation gap i s on average about 1.5 times as expensive as the moderate pay reform that allows for 100 percent public-private compensation gap at the end o f the reform process. This difference becomes less extreme when we consider separately the costs of reform for core government administration (Figure A2.6). Although the proposedpay increase inthe civilian public sector i s lower than the one in core government administration, the much larger number of civilian public sector employees has a major influence onthe total cost difference. 97 Scenario 13 assumes a radical pay reform to be implemented within 2004-2008 with moderate administrative changes. Scenario 20 assumes a moderate pay reform under the same administrative reform effort. Scenario 16: a radical pay reform to be implemented within 2004-2006 combined with significant administrative reforms. Scenario 24: a moderate pay reform but the same administrative reform effort. 159 FigureA2.6: FiscalCosts of Civil Service Reform:Radicalversus ModeratePay Reform(Only Core GovernmentAdministration Covered)'' E 2.00 , Scenario 13 Scenario 20 Scenario 16 0Scenario24 - 2004 2005 2006 2008 2010 Source: Staffestimates. The extent of the pay reform influences several components o f the overall fiscal costs inour model, including both direct implications (suchas increase incash compensation, staff attrition costs, and increased PIT collection), and indirect implications (increased non-wage expenditures), which in our model are a function o f the wage bill and the administrative reform pattern. Table A2.6 presents information on relative importance o f different components o f incremental costs. See the previous footnote for a description o f the scenarios. 160 Table A2.6: Fiscal Costs of Civil Service Reform for the Selected Scenarios (as p.p. of GDP) Scenario 2004 2005 2006 2008 2010 Medium-PacedRadicalPayReformwith Fair Administrative Adjustment (Scenario 13) Core GovernmentAdministration 0.58 0.82 1.05 1.47 1.58 Cash Compensation 0.50 0.64 0.78 1.05 1.10 Non- WageExpenditures 0.13 0.24 0.34 0.52 0.59 StaffAttrition Costs 0.01 0.01 0.01 0.01 0.00 Increased PIT Collection -0.05 -0.07 -0.08 -0.11 -0.12 Civilian Public Sector Employment 1.09 1.18 1.23 1.16 1.09 Cash Compensation 1.17 1.27 1.31 1.22 1.20 StaffAttrition Costs 0.03 0.04 0.04 0.05 0.01 Increased PIT Collection -0.11 -0.12 -0.13 -0.12 -0.11 Total Costs 1.67 2.00 2.27 2.63 2.67 Medium-PacedModeratePay Reform with Fair Administrative Adjustment (Scenario 20) Core GovernmentAdministration 0.49 0.63 0.78 1.03 1.13 Cash Compensation 0.43 0.51 0.58 0.73 0.77 Non-WageExpenditures 0.09 0.17 0.25 0.37 0.44 Staff Attrition Costs 0.01 0.01 0.01 0.01 0.00 Increased PIT Collection -0.04 -0.05 -0.06 -0.08 -0.08 Civilian Public SectorEmployment 0.96 0.94 0.88 0.64 0.58 Cash Compensation 1.03 1.oo 0.93 0.65 0.63 StaffAttrition Costs 0.03 0.03 0.04 0.05 0.01 Increased PIT Collection -0.10 -0.10 -0.09 -0.06 -0.06 Total Costs 1.45 1.58 1.66 1.67 1.71 High-PacedRadicalPay Reformwith SignificantAdministrative Adjustment (Scenario 16) Core GovernmentAdministration 0.90 1.25 1.53 1.65 1.77 Cash Compensation 0.58 0.79 0.96 1.01 1.06 Non- WageExpenditures 0.37 0.53 0.67 0.75 0.82 StaffAttrition Costs 0.01 0.01 0.01 0.00 0.00 Increased PIT Collection -0.06 -0.08 -0.10 -0.11 -0.11 Civilian Public SectorEmployment 1.16 1.24 1.19 1.11 1.09 Cash Compensation 1.22 1.29 1.24 1.22 1.20 Staff Attrition Costs 0.06 0.07 0.07 0.01 0.01 Increased PIT Collection -0.12 -0.12 -0.12 -0.12 -0.11 Total Costs 2.06 2.49 2.73 2.11 2.86 High-PacedModeratePay Reformwith SignificantAdministrativeAdjustment (Scenario 24) Core GovernmentAdministration 0.69 0.84 0.95 1.06 1.16 Cash Compensation 0.43 0.50 0.56 0.60 0.64 Non- WageExpenditures 0.29 0.37 0.44 0.52 0.59 Staff Attrition Costs 0.01 0.02 0.01 0.00 0.00 Increased PIT Collection -0.04 -0.05 -0.06 -0.06 -0.07 CivilianPublic Sector Employment 0.95 0.84 0.67 0.60 0.58 Cash Compensation 0.99 0.86 0.67 0.65 0.63 StaffAttrition Costs 0.05 0.06 0.06 0.01 0.01 Increased PIT Collection -0.09 -0.08 -0.06 -0.06 -0.06 Total Costs 1.64 1.68 1.62 1.65 1.74 Source: Staffestimates. 161 The affordability issue of radical pay reforms becomes even more critical when one considers the distribution of the additionalfiscal burden between thefederal and sub-federal levels of the budget system presented inFigure A2.7. To illustrate the difference inthe fiscal implications we considered radical and moderate pay reform scenarios implemented at a medium pace, Le., within 2004-2008, and accompanied by fair administrative reform efforts (scenarios 13 and 20). Figure A2.7: FiscalImplicationsfor Federaland Sub-FederalBudgets for Radical and Moderate Pay Reforms" zdE 1.8 7 1 ' 1.6 1.4 1.2 p 1.0 - Pc) 0.8 0.6 ;00.2 5 4 -d+0 0.0 2004 2005 2006 2008 2010 Scenario 13 (Federal Budget) 0Scenario13(Sub-FederalBudgets) Scenario 20 (Federal Budget) Scenario20 (Sub-FederalBudgets) Source: Staff estimates. Figure A2.7 shows that the sub-federal budgets would be much more affected by higher fiscal pressures than the federal budget: costs for subnational budgets are almost 1.5 times higher than for the federal budget. Even increased collection of PIT by subnational governments doesn't compensate for the fact that most o f public employees in Russia are paid from subnational budgets, which as a result become more affected by the proposed pay adjustment. Sustainability of the sub-federal finances would call for a substantial increase in transfers from the federal budget complemented by a significant increase inthe revenue base for sub-federal budgets. As discussed in Section D, the pay reform has two major tasks: improving external competitiveness o f the core government administration and the civilian public sector and instilling internal performance incentives through internal decompression o f pay in the core government administration. The results of our simulations confirm that the magnitude o f pay adjustment affects the compression ratios in the executive branch (Figures A2.8 and A2.9). To illustrate this effect, we selected 6 reform scenarios: (i) Scenarios with radical pay adjustments implemented invarious macroeconomic conditions (scenarios 4, 16, and 28); and (ii) Scenarios with moderate pay adjustments implemented invarious macroeconomic conditions (scenarios 8,20, and 32). Figure A7.6 shows that both radical and moderate pay adjustments result in significant decompression in HQ-based civil service, from the ratio between average wages inTop and Junior Groups of 2.5 in 2002 to 5.2-8.3 in2006 and to 6.8-8.3 in 2010. At the 99Both scenarios 13 and 20 assume implementation of reforms at a medium pace and accompanied by fair administrative reform efforts and by medium growth rates of private wages. However, scenario 13 provides for aradical pay reform, while scenario 20 for a moderate one. 162 same time, the decompression effect i s stronger for the scenarios with radical pay adjustment. This can be explained by a combined effect o f the two parameters used for simulating pay adjustment: residual public-private sector gap and minimumpay increase. In the scenarios that assume a radical pay reform (Scenarios 4, 16, and 28), the residual public- private pay gap is smaller, and hence, most o f the wages in core government administration are increased by a factor exceeding the minimumpay increase o f 30 percent. As a result, the salary structure is brought closer to the private sector comparators, and the decompression ratios are higher. Reversely, in the scenarios that assume moderate pay adjustment (Scenarios 8, 20 and 32), the rule o f minimum pay increase (20 %) plays a greater role in determining expected wage growth for a lower level staff. Hence, the residual pay gap i s smaller than the average for lower grade staff, which moderates adjustment in the pay structure and results in smaller decompression. FigureA2.8: Internal Decompressionin FederalExecutiveHQ-BasedCivil Service: Radicalversus ModeratePayAdjustment"O I a 9.0 1 i [3Scenario4 HScenario16 0Scenario28 0Scenario 8 HScenario20 u E 2004 2005 2006 2008 2010 Source: Staff estimates, This differentiation in decompression effort could also be seen from the comparison o f the proposed pay increases for moderate and radical pay reform scenarios"' (Table A2.7). As one may notice, the ratio betweenthe pay adjustment proposed for Top and Junior Groups i s lower in the case o f moderate pay reform than in the case o f radical pay adjustments. Considering that for non-HQ-based staff (Le., for a great majority o f affected employees) we propose pay increases that are only 50 percent o f those presented in Table A2.7, application o f the minimumpay increase rule i s more frequent when we simulate the dynamics o f wages for employees in both deconcentrated units and regional and municipal core government administration. loo Scenarios 4, 16, and 28 all assume a radical pay reform, but different growth rates of private wages. Scenarios 8, 20, and 32 assume a moderatepay reform, but the same variation inprivate wage growth. lo' Note that in actual simulations these increases are further adjusted to reflect annual real wage growth in the private sector. 163 Table A2.7: gxisting Pay Gap and Proposed Pay Increasesin Federal ExecutiveHQ-based Civil Service, times Category C Actual Net EstimatedNet Proposed Pay Increase Group Public-Private Public-Private Pay Pay Gap, 2002 for actualpay Gap,2003 (adjusted Radical Moderate increases in 2003) Scenario Scenario TOP 7.8 6.5 4.3 3.3 Chief 5.7 4.7 3.1 2.4 Lead 5.1 4.2 2.8 2.1 Senior 3.6 3.0 2.0 1.5 *Minimum Junior 2.3 1.9 1.3 1.2* pay increase is applied. Source: Staffestimates; see also Table 4.2. This trend towards decompression is even stronger in case the compression index captures both H Q and deconcentrated units of federal executive civil service (Figure A2.9), because, as was mentioned above, the proposed pay adjustment for deconcentrated service i s smaller and based on local labor market prices. Adopting separate pay scales for HQ-based and deconcentrated civil servants leads to a drastic increase in internal compression ratios from 3.7 in2002 to about 10-12 in2010. Figure A2.9: Internal Decompressionin FederalExecutiveCivil Service: Radicalversus Moderate Pay Adjustment102 'E3 130 120 v) ?= 11.0 Scenario 4 8 0 Scenario 16 a 8 7 0 0Scenario 28 6 0 0Scenario 8 ;$ 5 0 Scenario 20 Scenario 32 T m 3 0 55 2 0 'f 1 0 0.0 2004 2005 2006 2008 2010 V ~~~~ ~ Source: Staffestimates. Pay reform ,Dace and overall affordability of reforms. A brief review of summary results for all 36 scenarios presented in Table 5.4 points to several reform combinations that do not seem at the moment affordable to the Russian government budget. If we assume that the increase in total fiscal costs should be lower than 1.7 GDP p.p.lo3in2004 as compared to '02See the previous footnote for a description o f the scenarios. lo'Significant pay increase in core government administration and civilian public sector implemented in October 2003 already suggests substantial increase o f financing in 2004 as compared to 2003. Therefore, the assumption for fiscal costs growth for 2004 is set at quite a highlevel. 164 2003, and/or in2006 - 2.3 p.p. o f GDP respectively, andor in 2010 - 2.7 p.p. of GDP, then the following two groups of scenariosbecome hardly affordable for the budget: (9 scenarios assumin1 implementationo f radicalpay reforms (Scenarios 1-4, 13-1 6 and26-28 %4 ) especially those assuming implementation o fradical pay adjustments at a highspeed, Le., within 2004-2006 (Scenarios 3,4, 15, 16, 27, and 28); (ii) implementationo f pay reforms without any administrative reform actions (Scenarios 5,7, 10, 17, 19,22,29, 31,34) regardless o f options for private wage growth dynamics. This preliminary review leads us to two major conclusions. Firstly, signzjkant attrition in civilian public sector and at least some attrition in core government administration is a precondition for successful implementation of civil service reform. Secondly, radical pay increasefor public sewants is not affordable to the budget, especially if considered in the 3-year implementationframework. Taking the above into account, we focus on a moderate pay reform with fair administrative reform efforts to illustrate the impact of the pay reform pace on fiscal costs in 2004-2010 (FiguresA2.10 andA2.11). FigureA2.10: Distributionof TotalFiscalBurdenof Civil Service ReformScenarios, Dependingupon Pay ReformPace105 8 CL I 2.0 4 1.8 1.6 1.4 -e Scenario 20 $ 0.8 0Scenario 23 0.6 0.4 w 0.2 I- t 0.0 2004 2005 2006 2008 201o Source: Staff estimates. Figures A2.10 and A2.11 illustrate that although there i s some variation in annual fiscal costs depending on the pace of reform implementation, and it should be taken into account in the context o f other forthcoming fiscal challenges (such as peaks in public debt reduction), yet the variation i s not very large. This can be explained by the fact that the fiscal implications o f the civil service reform consist mostly of the increased current expenditures: indeed, the only type o f one-time expenditures captured by our simulations are the costs associated with attrition, and those are not very high (see Table 2.12)'06. Moreover, to sustain the reform achievements, the consolidatedbudget would have to accommodate annual indexation o f pay in the core government administration and civilian public sector in line I O 4However, the simulations show that implementation o f radical pay adjustments at moderate pace is a viable option incase the share of the real wages in GDP remains constant (see Scenario 25). 'Os Scenarios 18, 20, and 23 assume different pace o f pay reforms, but all provide for a moderate pay adjustment, fair administrative reform effort, and medium growth inprivate sector wages. I O 6As discussedabove, we do not try to evaluate investment costs associated with public service modemization inthe scope ofthis Note. 165 with real wage growth inthe private sector. Becauseinour example the scenarios 18,20, and 23 are implemented under the conditions of accelerated real wage growth as compared to GDP growth, the fiscal costs of reforms inscenario 18 and 20 (whichassume the completion of reforms in 2006 and 2008, respectively) continue growing even after the reforms have been completed. FigureA2.11: Distributionof FiscalBurdenof CivilService ReformScenarios,DependinguponPay Reform Pace(Only Core GovernmentAdministrationCovered)"' B 1.2, 0 2004 2005 2006 2008 2010 Source: Staff estimates. Administrative reform magnitude. Three additional scenarios illustrate potential cost differences associated with different scopes of administrative reform: (9 scenarios with "no administrative reforms" assumed some increase in the employment levels in core government administration and civilian public sector with a reduction of the share o f non-wage expenditures in total expenditures on core government administrationto 25 percent; (ii) "fair effort" inadministrative reformassumedthat executivecore government employment will be slightly reduced, and employment inthe civilian public sector will be reduced (based on the defined ratios between staffing levels inthe sectors that provide these services and numbers of their clients) with the share of non- wage expenditures in total costs on core government administration moderately reducedto 32 percent from 37 percent in2002; (iii) "significantadministrativechanges"thatassume,inadditiontotheadjustmentfor civilian public sector employment described above, higher attrition rates for core government administration and sustainability o f the share of non-wage expenditures in total costs o f core government administration at the level o f 36 percent. To illustrate the cost variation under different scenarios o f administrative reform implementation, we compared three options Figure A2.12) for a moderate pay reform scenario implemented at a medium pace (i.e., between 2004 and 2008) and accompanied with: (i) administrative reform (scenario 19); (ii) administrative reform (scenario 20); no fair and (iii)significant administrative reform (scenario 21). lo'See the previousfootnote for a description of the scenarios. 166 FigureA2.12: Total FiscalCosts of Civil Service Reformfor DifferentScenarios of Administrative ReformImplementation(p.p. of GDP as comparedto 2003)10s p 3.6 0 3.4 3.2 20 3.0 2.8 #g,g - 2.6 2.4 2.2 2.0 1.8 5 1.6 0.8 10.6 2 0.4 0.2 0.0 2004 2005 2006 2008 2010 Source: Staff estimates, Figure A2.12 shows that pay reform implemented without attrition clearly drives the total fiscal costs o f the reform too high (with additional costs reaching about 2 percent o f GDP by 2010). The results of the estimates for fair and significant reform scenarios (20 and 21) do not vary significantly becausethe key administrative reform component (attrition rates for the civilian public sector employment) is the same for both scenarios. Civil service reform with significant administrative changes requires more financing for non-wage expenditures, which makes scenario 21 slightly more expensive than scenario 20. 'OsScenarios 19, 20 and 21 assume different scope of administrative reform, but all provide for a moderatepay adjustment,mediumpace of pay reforms, and medium growth inprivate sector wages. 167 FigureA2.13: FiscalCostsof Civil Service Reformin Core GovernmentAdministrationfor Different Scenariosof Administrative ReformImplementation,pap.of GDP'" r I 6.-e$ c 1.4 1 I 1.2 : E n 1.o 0.8 3 5 0.2 '0 0.0 2004 2005 2006 2008 2010 Source: Staff estimates. Similar estimates for the same reform scenarios but applied only to core government administration surprisingly yield a different cost pattern: in this case scenario 19 i s less expensive than scenario 20 (see Figure A2.13). To explain the reason for such a difference between the results in Figures A2.12 and A2.13, we looked at the structure o f this expenditure increase. Figure A2.14 illustrates the dynamics o f the two key components o f total fiscal expenditures on core government administration: increased spending for cash compensation o f employees (depicted by columns) and increased financing o f non-wage expenditures (depicted by lines). FigureA2.14: FiscalCostsof Civil Service Reformin Core GovernmentAdministration:Dynamicsof CashCompensationand Non-WageExpenditureIncreasefor DifferentAdministrativeReformScenarios - (p.p. of GDP as comparedto 2003)"0 1.2 , 1 0 P 2 1.0 0 d Sc. 19, conpensation Q 0.8 0Sc.20,corrpensation d 0.6 Sc. 21, conpensation .-uf +FSc. 19, non-wage costs 0.4 -n-Sc.20,non-wagecosts z5 CI +Sc. 21, non-wage costs Ir" 0.2 0.0 2004 2005 2006 2008 2010 Source: Staffestimates. Figure A2.14 illustrates two trends. Firstly, the more signiJicant the administrative changes and respectively employment cuts are, the lower is theJiscal burden caused by pay increase. Secondly, the more signijkant the administrative changes are, the higher is the log See the previousfootnote for a description of the scenarios. I t 0See footnote 10 for a description of the scenarios. 168 pressure to increase non-wage expenditures on core government administration andfinance public service modernization, HR reforms, etc. Therefore, the total fiscal implications of the civil service reform incore government administration will highly depend on the dynamics of the share of non-wage expenditures in total structure of core government administration financing - this dependence i s in fact so high that it reversed the costs pattern and brings a different trend for the total increase infiscal expenditures on core government administration as illustrated by the difference betweenFiguresA2.12and A2.13. Given that the scenario of pay reform without any administrative changes i s unlikely to be selected because of the higher overall fiscal costs (i.e., when increased financing o f the civilian public sector i s taken into consideration), we did additional analysis o f the fair and significant administrative reform scenarios in order to find a "breaking point" - the share o f non-wage expenditures in total expenditures on core government administration that would still make the "significant administrative reforms" scenario more affordable than the scenario with the "fair administrative reforms". To do this, we modified the assumptions used for estimating the costs of scenario 21, The estimates showed that when the share o f non-wage expenditures i s equal or lower than 35 percent in the total structure of core government administration expenditures, the scenario with significant administrative reforms becomes less costly than the one implyinglower attrition rates (Figure A2.15). FigureA2.15: Fiscal Costsof Civil Service Reformin Core GovernmentAdministrationfor Different Scenarios of AdministrativeReformImplementation (measured in pap.of GDP as comparedto 2003)"' Scenario 20 2004 2005 2006 2008 2010 Source: Staff estimates. This observation underlines the fact that, while non-wage expenditures are often neglected inpolicy debates, they do play a key role in determining overall trends in costs of core government administration, and thus they should be fully taken into account at the design stage of civil service reforms. This finding also suggests that the total fiscal costs of reforms (that account for the entire civilian public sector costs) may be underestimatedinour `I1See footnote 10 for a description of the scenarios. Scenario 21 (adjusted) is similar t o Scenario 21, but has a share o f non-wage spending reduced to 35%. 169 model because the simulations do not take into account the dynamics of non-wage expenditures for the civilianpublic sector. 170 ANNEX3.1 MACROECONOMIC FRAMEWORK FORTHE ANALYSISOF FISCALCOSTSOF USED STRUCTURALREFORMS This work i s based on the following approach to the development of macroeconomic scenarios for the analysis o f the fiscal costs of structural reforms. We took the government's baseline macroeconomic projections for the period 2004-06, and used them as a basis for building a set of macroeconomic scenarios, each of which reflects a specific combination of factors that are primary determinants o f Russia's macroeconomic performance. We identified two such determinants o f Russia's performance in the medium to longer term: external, which i s outside the government control, and internal, which depends on government policies. Giventhe high dependency of both the Russian economy and the government budget on world oil prices, their level was used as the external determinant of macroeconomic trends. Following the government's macroeconomic projections, we selected the same two benchmark values for the Urals oil price: US$l8.5 per barrel and US$22.5 per barrel. With ongoing global economic recovery, it seems unlikely that oil prices might decline below US$18.5 per barrel in the next two to three years. Beyond that, temporary price drops are possible, but it i s still unlikely that they would reduce the 10-year moving average o f the Urals oil price to below the benchmark US$18.5 per barrel, which was the long-term average price before 1999. Thereby, our assumptions about future oil prices appear to be fairly conservative and to allow a rather accurate reflection o f the possible "low case," (Le., a macroeconomic slowdown causedby low oil prices). The speed of implementing reforms was the internal determinant o f macroeconomic dynamics. For this determinant we distinguish two cases: `no reforms' and "advanced reforms" scenarios. The scenario without reforms reflects an inertial strategy o f dealing with major structural challenges, which i s similar to the reform strategy pursued by the government in 2001-03, when progress in most key structural reform areas was slow and incomplete. The advanced reforms scenario was built on the assumption that the implementation of reforms would be accelerated, primarily in sectors such as energy, housing and utilities, public administration, pensions, and the investment climate. The characteristic features o f the advanced reforms scenario include, inter alia higher growth in domestic energy prices in the initial period (owing to aggressive reforms in energy and utilities) and highinvestmentrates (owing to improvements inthe investment climate and, more generally, to improved investment confidence in the environment of stronger government reform commitment). At the same time, in the advanced reforms scenario we assume that the acceleration of reforms may cause some shocks for the real sector, which would result in lower growth rates for the initial period. However, growth i s expected to pick up considerably inthe medium term, when structural reforms wouldbringa significant pay-off. Itis assumed that by 2010 the annual rates of GDP growth under the advanced reforms scenario would exceed the rates under the no reforms scenario, and by 2015 cumulative GDP growth under the advanced reforms scenario, would exceed that under the no reforms scenario. The detailedpresentationo f specific quantitative macroeconomic parameters used in simulations could be found inTables 2.5, 3.13, 3.15, and 4.4 o f the main text, 171 It is worth noting that GDP growth rates assumed in our scenarios for the period 2004-06 are lower than those assumed inthe corresponding government projections. Overall, we believe that without advancing the reforms growth rates *ill be declining: better utilization o f existing reserves in the economy, which was a critical growth factor in 1999- 2003, cannot support future growth in the same way as before because the reserves are to a large extent exhausted. At the same time, the advance reforms scenario implies that reforms are likely to temporarily slow down GDP growth compared to the no reform scenario, other thingsbeing equal. Therefore, inthis case growth rates will also be lower than those assumed by the government. The consensus medium-term estimates for Russia's economy made by investment companies, IFIs and NGOs for 2004-05 are rather favorable; everyone expects that in the most likely scenario the economy would grow at about 5.2 percent a year under the assumption o f a modest reform effort (no breakthrough) and favorable oil prices (23-25 $/bbl). At the same time, almost all of them predict the slowdown o f economic growth along with declining oil prices. Inthis respect, our growth scenarios would not differ muchfrom the alternative projections ifthe latter are recalibratedat our lower oil prices. At the same time, it is worth noting that the two out of three models, used for costing structural reforms in this report, designed in a way that the main results (i.e. incremental fiscal costs) are not sensitive to the assumptions on economic growth. The growth parameter i s critical only for modeling the pension reform in Chapter 4. To reflect this, in the latter case we undertook an alternative set o f simulations based on the assumption o f 6 percent average growth for the period 2004-09. But even in this case we found quite a modest sensitivity o f the results to the variation inthe growth rate. A combination of the oil price scenarios with the reforms speed scenarios provides for the following set o f four possible macroeconomic scenarios. While each of these four scenarios is internally consistent, their combination allows the consideration o f rather a broad variation in possible macroeconomic trends. For instance, across-scenario variation in the projected average growth rate for 2004-06 would be between 2.3 and 5.0 percent, annual average inflation would vary between 9.0 and 12.3 percent, and real investment growth would be between- 5.7 and 7.7 percent. No reforms Advancedreforms Moderateoil prices (18.5), low growth Scenario 11 Scenario 12 High oil prices (22.5 andhigher),highgrowth Scenario 13 Scenario 14 We used the above framework for the analysis o f the fiscal implications o f specific structural reforms -pension reform, public administration reform, and housinghtility reform. Ineach of these cases, we used the four macroeconomic scenarios as a basis on which we design andelaborate further more detailedsub-scenarios o f specific structuralreforms. 172 ANNEX3.2 MAINDATA SOURCESAND DATA ASSUMPTIONS The simulation o f future costs o f the delivery of housing and utility services, as well as estimates o f the associated demands for budgetary support were undertaken based on the following data sources and assumptions (Table A3.1). Table A3.1: Basic DataUsed in Simulations and RelatedAssumptions Indicator Comments 1.Average per capitahousehold We used 2002 household income data as reported by the Rosstat (2003) income by region. Household Regiony Rossii. Social 'no-economicheskiye Pokazateli. income distribution, by region Related assumptions:(a) For the period of forecast, the average real per capita income assumedto grow at an identical rate acrossregions. (b) The income distribution by octile in each region would remain unchanged during the forecast period, Le., the shares of households in each octile remain the same (but region specific). Average income for each octile changesis basedon the assumptions about real income growth. The forecast was based on the actual household income distribution by octile in2002. 2. Share of population groups that Two specialpopulation groups were identified: havemuch lower liabilities with respect to housing and utility (a) Families eligible for benefits on payments (Igoty) for housing and payments,by region utility services, established by the federal legislation. These data were based onthe 2003 NOBUS survey. (b) Families that occupy housing units with permanent access to running water. These data were based on Rosstat reports for 9 months of 2003 (statistical report 22-ZhKH). The importance of separating this group relates to the fact that families without access to running water are, as a rule, occupants of sub-standardhousing who do not have access to other most expensive utilities, such as district heat, sewage, and hot water. As a result, the overall costs of housing and utility services for the residents of this type of housing are considerably lower than the average. In practical terms, it means that those households in the units without running water never apply for housingallowances. 3. Monthly costs of delivering The model provides estimates for unit costs of delivering housing and housing and utility services per utility services using information on the federal standards for maximum 1sq. m. ofthe occupiedhousing housing and utility costs per 1 sq, m. of housing in 2002, approved by the stock, by region federal government for each region (Government Decree No. 804 of November 19, 2001). The model (a) estimates the future national average costs of housing under chosen assumptions, and then (b) differentiates these costs by region, based on the historical cross-regional cos1 differentiation in 2002. Relatedassumption: The regional proportions betweenhousing costs will remain unchangedduring the forecastperiod for all simulated scenarios. 4. Cross-subsidizationintariffs Estimatesare based on the dataon actualenergy andutility tariffs by types of consumers in early 2003 as reportedby the Rosstat in Tseny i Tarijjfi I Zhilishno-kommunal 'nomKhozyaistve 2003. Issue 1(15). January-March. Related assumDtion: Cross-subsidization in tariffs is phased out in t w c years (2005-06) intwo equal steps. 5. Information on housing Rosstat Bulletin.Tseny i Tarffi v Zhilishno-kommunal 'nom Khozyaistve. allowancesprograms(total costs and number ofrecipients), by region 173 6. Estimatesfor budget Expendituresonutility services were estimatedas the share ofthe reported expenditures on utility services budget expenditures on particular government functions and were based providedto budget organizations upon: data (in economic classification) on the execution of the consolidatedregionalbudgets in2001 (source:www.budietrf.ru) data on the federal budget expenditures in 2001-03 and reports on the consolidated regional budget execution in 2002 (in functional classification)(source: www.minfin.ru) the findings of the IUE survey for the Vologodskaya and Rostovskayaoblasts in 1999-2001 data on housing and utility prices and the norms of per capita consumption of these services in 2002 (source: Rosstat Bulletin. Tseny i Tarijjj v Zhilishno-kommunal 'nom Khozyaistve: 1(15), 2003) data ofthe Center for Facilitation of PenalReform on the number of inmates and personnel in the penitentiary institutions as of 1 July 2003 data on the number of military and civil personnel in the Armed Forces, the Federal Border Guard, the Interior Troops of the Ministry of Interior and on the number of staff in law enforcement agencies, the Federal Security Service and tax police (source: www.budietrfm). Informationonthe incidence of housingprivileges (lgoty) The information on a number o f beneficiaries of lgoty by region i s inprinciple available from three different sources: 0 Rosstat: the federal government's statistical survey o f housing and utilities (Form 26- ZKH). The latest available information is for the nine months of 2003. 0 Ministry of Finance: A specially prepared data set used by the Ministry for the estimation of regional fiscal needs. The latest available information i s for the first six months o f 2002. 0 NOBUS: National survey o f household budgets and participation in social programs (Russian acronym NOBUS), undertaken for the first time inspring 2003. Informationabout the number o f people enjoying discounts on payments for housing and utility services has been traditionally reported by the Rosstat inForm 26-ZhKH. A major deficiency o f this source i s that housing and utility providers that file Form 26-ZhKH have incentives to report larger numbers of lgoty recipients. Moreover, lack o f effective control over these reports results in quite distorted statistics. In the past, despite its reliability problem, most research on the topic was based on this source just because no nationwide alternative was available. Data from the Finance Ministry do not represent an independent source of information. They are based on Rosstat data, but a considerable effort was made to clean up the original information and eliminate most inconsistencies. However, this data set is the most outdated. Inparticular, it does not reflect some reduction ina number o f lgoty recipients that took place in2002-03 owing to the monetization o f their benefits. 174 The NOBUS survey provides the best available data to date on lgoty recipients. The survey was undertaken for the first time in April-May 2003. Forty-five thousand households inall Russianregions participated inthe survey. It is believed to be much more reliable than the 26-ZhKH data, in part because the parties responsible for collecting and processing the survey returns were not interested in misreporting the results. This i s the reason why this report uses the NOBUS data for the simulation of fiscal effects related to lgoty. 175 A"EX4.1. SENSITIVITY ANALYSISFORPENSION SIMULATIONS Given the fact that two parameters - UST tax rate and share of payroll in GDP (i.e. the UST tax base) -have the greatest impact on the performance of the pension system, additional sensitivity analysis o f the results was undertaken to explore details o f their influence on the average pension, as well as to assesspotential linksbetween such individual influences. Figure A4.1 presents the indifference curves for the average pension in 2030, measured as its ratio to the pensioner's subsistence minimum. These estimates correspond to the scenario 44 (advanced institutional reforms under the high oil prices). Each curve corresponds to the same ratio of average pension to subsistence minimum, Le. it reflects the same purchasing power of the average pension. These results reflect the outcomes of about 100 simulations that correspond to specific values o f these two parameters (UST rate and payroll share). Inthe base scenario without UST rate cut, the share o f payroll i s expected to increase from the current 25% to 29% o f GDP, which bring the pensiodsubsistence ratio from 2.3 to about 2.5. The diagram also shows how muchthe payroll share should increase to keep the ratio roughly at the same level o f 2.5 under different assumptions regarding the magnitude o f the tax cut. If the rate i s cut by 4 p.p., the share has to reach 36% to keep the pensiodsubsistenceratio intact. Ingeneral, 1p.p.decline inthe UST rate could be compensated by an increase inthe payroll share by about 1.5 pup.At every specific tax rate the decline inthe payroll share by 2 p.p.bringsthe pensiodsubsistenceratio downby about 12points, from 2.5 to 2.38. The latter result suggests that our estimates are rather robust: substantial fluctuations in the payroll share cause modest changes inthe purchasingpower o f the average pension. Figure A4.2 presents a more accurate non-linear approximation for the relationship between these two parameters that corresponds to the indifference curve o f 2.3 at Figure A4.1. FigureA4.3 presents a similar set of indifference curves for the replacementrate. The presented approach helps expand understanding of potential dynamics of the analyzed variables. Instead o f generating specific point estimates o f particular parameters, it focuses on larger intervals o f policy variables, within which the performance characteristics of the pension system remain sensible. It also helpsto concentrate the analysis on the issue of internal consistency o f the assumptions: how realistic i s that the future joint dynamic o f main parameters would go ina way that would ensure a stable performance o f the system? 177 Figure A4.1: Indifferencecurves for the pensionsystem: the ratio betweenthe average pension and pensioner'ssubsistence, 2030 Share of payroll in GDP, Yo Additionalgrowthin payroll,triggeredby the cuts inUSTrates Growth inthe payroll wio cuts in USTrates FigureA4.2: Indifference curve for the pension system: compensatory relationshipbetween an increase in the payrollshare and cuts in the UST rate. Payrollshare in GDP, YO 33 32 31 30 29 28 27 25 26 3 -4 -3 -2 -1 0 Change in the UST rate(base pension) 178 Figure A4.3: Indifferencecurve for the pensionsystem: replacementratiointhe PAYG system, 2030 ReplacementRate tr 30 29 28 27 2Bc 5 10 11 UST Rate 179 ANNEX4.2 SIMULATION METHODOLOGY Simulation methodology used in this report i s close to the approach employed in the PROST program by the World Bank. Accordingly, the type o f model used in this study had been employed frequently by the World Bank for undertaking similar analyses in other countries. It i s also worth notingthat the model followed some implementationprinciples that are similar to those used in the ILO Social Budget Program. The latter related e.g., to the model implementation inExcelwith partial use of the Visual Basic. On the one hand, such an approach allows for more transparency in calculations because intermediate results for each program block are stored in separate spreadsheets o f Excel books making it easier to trace links between the spreadsheetsand check the formulas. On the other hand, rather complicated links between the spreadsheets present certain difficulties for the analysis of calculation sequences, and the formulas inserted directly in EXCEL cells make the model more vulnerable because it i s easier to make a mistake in copying formulas, while it i s rather time consumingto change separate formulas. A brief description of the model is provided inthe Annex to the World Bank (2003~) report, as well as in Simulation and Actuarial Estimates (2003). The earlier version o f the model was reviewed by the TACIS consultants, and their main comments (Stott, 2002), such as more accurate reflection o f pension contributions by working pensioners, were reflected in the current version o f the model. The simulation program used for analysis in Chapter 4 was fully implemented in Visual Basic. The same simulation technique was previously used for the preparation o f the World Bank Report (2003). A specific structural feature o f this version o f the computer program as compared to the earlier version described in Baskakov et al. (2003) i s its analytical focus, absence o f interface forms and reliance on bar chart reports, which substantially facilitate the analysis and quality control of results obtained. As compared with the World Bank (2003~)report, the simulations in Chapter 4 are based on a radically revised set of assumptions regarding Russia's macroeconomic projections that reflect the actual changes in the country's economic performance that occurred in 2001-03. Besides, we consider a broader range o f potential development scenarios. Inaddition, itis worthmentioning a certaindifference inthe logic betweenthismodel and the model in the World Bank (2003~)report. The latter modeled the accelerated growth o f the payroll based on the explicit assumption about difference in growth rates between average wage and productivity. Inthis paper, the trend inthe payroll share in GDP i s instead considered as a key exogenous variable. This trend i s described in the model by a conventional exponential transition process. Such an approach adds flexibility to the model. Inparticular, it allows undertaking additional analysis of the pension system's sensitivity to changes inthe basic trends intaxable wages. The following set o f basic input variables in the model determines most o f the variability within and across the scenario groups: 0 GDP growthrate Share of payroll inGDP 180 Level o f economic activity of population Birthrate Net migration Inflationrate Unemployment rate Interest rate Contribution rate (tax rate for UST) Pension system parameters (such as e.g., collection ratio) Retirement age The model employs a module principle based on the following sequence of calculations applied to each group o f contributors and pensioners.' l2 0 Estimating the number o f contributors andpensioners; 0 Estimating the average size of salaries andpensions; 0 Estimating the flows of pensioncontributions andbenefits; 0 Balancing revenues and expenditures of the pension system; 0 Determining the key performance characteristics o f the pension system, such as a replacementrate, etc. To estimate the number of contributors, all employees inthe economy are divided into three groups: (i) regular employees; (ii) who benefit from the reduced UST rates, those including self-employed; and (iii) delinquent taxpayers. , The number of pensioners inthe model i s determined usingthe Stock method, where share matrixes are definedfor each type o f pensioner, as well as for each age and sex group. Then, to determine the number of pensioners in a particular group one has to multiply the population number in each age and sex group (defined by the existing demographic projections) by a respective value inthe share matrix. The size o f pensionbenefit for each group o fpensioners is determined baseduponthe overall amount o f UST collection, the number of pensioners in each pension group, types o f pension benefits established for these groups, as well as upon the adopted pension indexation rules. The level o f administrative costs inthe system i s considered to be constant at 2 percent o f the total annual PAYG benefits. The NDC benefit i s determined as accumulated notional individual fund divided by the annuitization factor that reflects the remaining life expectancy o f the cohort and the notional interest rate. The annuitization factor i s gradually increasingbetween 2002 and 2013 from 12 to 19. The initial notional individual fund for those who are enrolled inthe system in 2002 i s equivalent o f the average monthly pension o f 830 Rbl. The overall inflow o f revenues to the pension system i s determined by the number and structure of contributors, average USTrates, as well as by the following factors: 0 Collection rate (currently amounts to 0.96); 'I2 The model considers two groups of contributors (regular employees and those who are eligible for reduced UST rates) and four groups of pensioners: old-age pensioners, disability pensioners, survival pensioners and early retirementpensioners. 181 0 Nature o f the wage distribution relative to the regression scale o f the existing UST rates; this factor i s reflected in the model by the introduction o f a separate coefficient of regressivity (currently amounts to 0.94); 0 Exemption rate for UST that reflects the average reduction ineffective UST rate for contributors due to the existing legal benefits for particular groups o f taxpayers, such as those who pay a single agricultural tax (currently amounts to 0.94) Average working life period is estimated as 30-35 years. The indexation of benefits in the model is based on current legislation: the base pension i s indexed with average inflation, while the NDC portion of the benefit i s indexed based on the growth rate of the pension fund collections per beneficiary (which on average exceeds the inflationrate due to the growth inreal wages). The model calculations were undertaken without feedback, Le., even if the pension system balance starts to accumulate considerable surpluses, the model doesnot provide for an automatic change inthe rules o f pension indexation. The logic behindsuch an approach i s in focusing the analysis on what outcomes various indexation rules may generate (in terms o f both the pension system balance and average replacement rate) if kept intact for a considerable period o f time. In addition to this, however, the model generates an alternative estimate for the replacement rate, which i s the rate that could be attained at a zero balance of the pension system. To get this estimate in the model all current surpluses in the system are distributed among the groups o f beneficiaries proportionally to their NDC pensions. This alternative estimate reflects a potential maximum pension benefit that could be funded by utilizing all collected funds (Le., without runningsurpluses), but at the same time without creating debts in the pension system. InChapter 4 we call this indicator of the maximum potential pension benefitas an "affordable replacement rate", and this i s an indicator that represents the core to our analysis. Inother words, we are especially interested inunderstanding for each particular scenario how much the pension system could afford to pay. It i s assumed that ifinthe longer term the selected indexation rules proved to be too restrictive (as they seem to be now), they couldbe adjusted to allow for a full pay-out o f available pension funds. While statistical data required for building a full quantitative model of the link between the pension contribution rate and the taxable base for UST were unavailable, a numeric experiment was undertaken, which allowed estimating sensitivity o f the basic pension systemparameters to the cuts incontributionrate. 182 c 1")9 7 0 7 (1 r) h s