Report No. 36595-BR Brazil Improving Fiscal Circumstances for Growth (In Two Volumes) Volume II: Main Report March 2, 2007 Brazil Country Management Unit Poverty Reduction and Economic Management Unit Latin America and the Caribbean Region Document of the World Bank AbbreviationsandAcronyms AET Anuhrio Estatistico dos Transportes StatisticalAnnals of Transportation AGF Aquisiqdes do Governo Federal Federal Government Acquisitions ALADI Associaqgo Latino-Amencanade Integraqgo Latin-AmericanAssociation for Integration ANTAQ AgCncia Nacional de Transportes National Agency for Water Transport Aquaticos ANTT AgCncia Nacional de Transportes National Agency for LandTransport Terrestres BB Banco do B r a d Bank of Brazil BCB Banco Central do Brasil Central Bank of Brazil BGU Balanqo Geral da Unigo General Government Balances BNDES Banco Nacional de Desenvolvimento National Bank for Economc and Social EconGmco e Social Development CACAO Programa para estabilizaqgo de preqos do Prize-stabilization subsidy for cacao ccc cacau Conta de Consumo de Combustiveis FuelConsumptionAccount CDE Conta de Desenvolvimento EnergCtico Energy Development Account CEL Centro de Estudos emLogistica Center for Logistics Studies CGU Controladona-Geral da Unigo Comptroller-General's Office CIA AgCncia Amencana de InteligCncia Central Intelligence Agency CIDE Contribuiqgo de Intervenqgo no Dominio Contribution for Intervention inthe Econdmco Economc Domain CNT ConfederaqgoNacional dos Transportes National Confederationof Transports COFINS Contribuiqgo para o Financiamento da Contribution for Financing Social Secunty Seguridade Social CONIT Conselho Nacional de Integraqgo de National Council for Integration of Policies Politica de Transportes Terrestres on LandTransport COPPEAD Instituto de Estudos Avanqados de Institute for Advanced Studies on Admnistraqgo Admnistration CPMF Contribuiqgo Provis6ria sobre Temporary Contribution onFinancial MowmentaggoFinanceira Transactions CPSS Contribuiqgo para o Plano de Seguridade Contribution for Social Security of Public Social do Servidor Pcblico Sector Employees CSLL Contribuiqgo Social sobre o Lucro Liquid0 Social Contribution onNet Profits DER Departamentos Estaduais de Rodovias State-Level Departments for Highways DITIAS Diagn6stico del Transporte Internacionaly Diagnostic for InternationalTransport and suInfraestructwa enAmCrica del Sur Infrastructurein South America DNER Departamento Nacional de Estradase National Department for Roads and Rodovias Highways DNIT Departamento Nacional de Infra-estrutura National Department for Land Terrestre Infrastructure DRU Desvinculaqiio de Receitas da UniGo De-earmarlung o f Federal Revenues EGF Emprestimos do Governo Federal Loans of the Federal Government FAT Fundo de Amparo ao Trabalhador Fundfor Worker Assistance FCO Fundo de Financiamento do Centro Oeste Fundfor Financingofthe Center-West Region FCVS Fundo de Compensaqgo de Vanaqdes Fundfor Compensation ofVmance in Salariais Salaries FEF Fundo de Estabilizaqgo Fiscal Fundfor Fiscal Stabilization FGTS Fundo de Garantia por Tempo de Serviqo Fundfor Employees basedonDurationof Employment FIES Programa de Financiamento Estudantil Program for Student Finance FINSOCIAL Fundo de Investimento Social Fundfor Social Investment FIRJAN Federaqgo de Indhtnas do Estado de R o Federation of Industnesof the State of Rio de Janeiro de Janeiro FMM Fundo da Marinha Mercante Fundofthe MercantileMannes FND Fundo Nacional de Desenvolvimento National Fundfor Development FNDE Fundo Nacional de Desenvolvimento da National Fundfor Educational Educaqgo Development FNE Fundo de Financiamento do Nordeste Fundfor Financingofthe Northeast Region FNO Fundo de Financiamento do Norte Fundfor Financingofthe NorthRegion FPE Fundo de Participaqgo dos Estados Fundbasedon StateParticipation FPM Fundo de Participaqgo dos Municipios FundbasedonMumcipalParticipation FRD Fundo para o Desenvolvimento Regional Fundfor RegionalDevelopment using com Recursos da Desestatizaqgo Resources from Privatizationof State- Owned Enterpnses FRN Fundo RodoviarioNacional National Highway Fund FSE Fundo Social de Emergsncia Social Fundfor Emergencies FUNCAFE Fundo de Defesa da Economa Cafeeira Fundfor CoffeePrice Stabilizationand support FUNDEF Fundo de Manutenqgo e Desenvolvimento Fundfor Maintenance andDevelopment of do Ensino Fundamental e Valonzaqgo do Basic Educationand Traimng of Teachers MagistCno GDP Produto Interno Bruto Gross Domestic Product GEIPOT Grupo Executivo de Integraqgo da Politica ExecutiveGroup for Integration of Policies de Transportes onTransport GFS Estatisticas Financeiras do Govern0 Government Financial Statistics IADB Banco Interamericano de Desenvolvimento Inter-Amencan Development Bank IBGE Instituto Brasileiro de Geografia e Brazilian Institute of Geography and Estatistica Statistics ICMS Imposto sobre Comercializaqiio de Tax on Commercialization of Goods and Mercadorias e Serviqos Services IDET-FIPE Instituto de Estudos de Transporte - Institute for Study of Transport - Fundaqgo Instituto de Pesquisas Foundation of the Institute of Economc EconBmcas Research IGP-DI hdice Geral de Preqos-Disponibilidade General Pnce Index -Internal Availability Interna IMF Fundo Monetano Intemacional International Monetary Fund INPC hdice Nacional de Preqos ao Consumdor National Consumer Pnce Index INSS Instituto Nacionalde Seguridade Social National Social Security Institute IOF Impostos sobre OperaqdesFinanceiras Tax on Financial Operations IPEA Instituto de PesquisaEcondmca Aplicada Institute of Applied Economc Research PI Imposto sobre Produtos Industnalizados Tax on IndustnalizedProducts IPI-EE Imposto sobre Produtos Industnalizados - Tax on IndustnalizedProducts -Exports Exportaqao IPVA Imposto sobre a Propriedade de Veiculos Tax on Ownershp of Automotive Vehicles Automotores IR Imposto de Renda Income Tax .. 11 ISTR Imposto sobre Serviqos de Transporte Tax on Highway Transport Services Rodoviario WLCLG Imposto Unico sobre Lubrificantes e UnifiedTax on Lubricants and Liquidand Combustiveis Liquidos e Gasosos Gaseous Combustibles LDO Lei de Diretrizes Orqamentdrias Budgetary Directives Law L O A L e i Orqamentana Anual Annual Budget Law LOAS L e i Orgiinica de AssistCncia Social Organic Law of Social Assistance LRF L e i de Responsabilidade Fiscal FiscalResponsibility Law M&E Monitoramento e Avaliaqgo Momtoringand Evaluation MECON MimstCno de Economa da Argentina Mimstryofthe Economy of Argentina MPAS MimstCno de PrevidCncia e Assist&ncia Mimstryof Social Secunty and Social Social Assistance MPF MimstCno Pdblico Federal Attorney General's Office MPOG MimstCno de Planejamento, Orqamento e Mimstry of Plamng, Budget, and Gestgo Management MT MimstCno dos Transportes MimstryofTransports MTEF Arcabouqo Fiscal de Meio Prazo MediumTerm ExpenditureFramework NGO Organizaqgo n9o governamental Non-Governmental Organization OECD Organizaqgo para Cooperaqgo EconBmca Organizationfor Economc Cooperation e Desenvolvimento and Development PAB Piso de Atendimento BBsico Floor for Basic Services PDG Programa de DispCndios Globais Global Expenditures Program PISDASEP Programa de Integraqgo Social / Programa Programof Social Integration/ Program de Formaqgo do PatrimBnio do Servidor for the Formationo f Assets o f Public Pdblico Servants PPA Plano Plunanual Multi-Year Plan PPE Parcela de Preqo Especifico Share of Specified Pnce PPI Projeto Piloto para Infra-estrutura Pilot Project for Infrastructure PPP Parcerias Publico-Privadas Public-Pnvate Partnershps PROEX Programa de Financiamento das Programfor FinancingExports Exportaqdes PRONAF Programa Nacional da Agricultura NationalProgramfor Family Agriculture Familiar RECOOP Programa de Revitalizaqgo das Program for Revitalization of Cooperatives Cooperativas de Produqgo Agropecuba for Cattle and Agricultural Production RGPS RegimeGeral de Previd&nciaSocial General System of Social Secunty RGR Reserva Geral de Reversgo General Reserve o f Reversion RJU Regime Juridic0 Unico UnifiedSocial Secunty System for Judicial Employees RPPS Regime Pr6prio de PrevidCncia- Social Secunty System for Public Servants Servidores Publico SE Standard Error Standard Error SELIC Taxa referencial dejuros do Sistema Central Bank interest rate Especial de Liquidaqgo e de Cust6dia S I A F I Sistema Integrado de Admnistraqio IntegratedSystemfor Financial Financeira Management SIGA Sistema de Informaqdes Orqamentanas do System for Budget Information o fthe Senado Federal Federal Senate SIGPLAN Sistemas de Informaqdes Gerencias e de System of InformationonManagement Planejamento and Plannmg SOF Secretariade Orqamento Federal Secretariat of Federal Budget ... 111 SRF Secretaria da Receita Federal FederalTax RevenueOffice SRH ss Secretaria de Recursos Humanos Secretariat o f Human Resources Seguridade Social Social Secunty S T N sus Secretaria do Tesouro Nacional Secretariat o f the National Treasury Sistema Unico de Sa6de UnifiedHealth System TCU Tribunal de Contas da Uni5o Federal Accounting Umt TEL Niveis de tributaqgo e despesa Taxation and Expenditure Level TKU Toneladas por Kilometro Usado Tons-Kilometers Used TRU Taxa Rodoviaria Unica UnifiedHighway Tax UFRJ UniversidadeFederal do Rio de Janeiro FederalUmversity o f k o de Janeiro VAR Vetor Auto Regressivo Vector Auto Regressive iv Acknowledgements Ths report was prepared by Yasuhiko Matsuda, Fernando Blanco and JosC Guilhenne Reis with inputs from Santiago Hererra and Ethan Weisman. Background papers were comssioned to Silvia Valadares, Sol Garson, Claudio Frischtack, Gerhard Glomm (Department of Economcs Indiana University) and Felix Rioja (Department o f Economcs Georgia State University). Anand Rajaram, Aymeric-Albin Meyer and Santiago Herrera were the peer reviewers for the study. V vi TABLE OFCONTENTS EXECUTIVESUMMARY XI 1 THE LEVEL OFPUBLICSPENDINGINBRAZIL ........................................... 1 2 PUBLIC SPENDINGAND ECONOMICGROWTH ........................................ 19 3 4 PUBLICEXPENDITUREINTHE ROADTRANSPORT SECTOR ..............37 61 5..... ............................................................................................ SOURCESAND EFFECTSOFEXPENDITURERIGIDITY .......................... CONCLUSIONS 85 REFERENCES ................................................................................................................ ..................................................................................................... 97 LIST OFFIGURES Figure 1.1 Primary and Total Outlays inBrazil. 1995-2004.............................................. Evolution of Tax BurdeninBrazil. 1947-2004 (YOof GDP) ............................ 1 Figure 1.2 Figure 1.3 General Government Outlays: Brazil and Comparator Countries (2000) .....44 Figure 1.4 Evolution of Total Pension Expenditures, 1990-2004 ....................................... 7 Figure 1.5 Government........................................................................................................................... Evolution of Public Sector Pension Expenditures, 1990-2004 by Level of 7 Figure 1.6 Age Structures for the BrazilianPopulation 1980-2000-2030 ....................... 16 Figure2.1 Per capita GDPResponse to one S.E. shock inPublic Capital Stock ...........29 Figure2.2 Per capita GDPResponse to one S.E. shock inPrivate Capital Stock..........29 Figure2.3 Per capita GDPResponse to one S.E. shock inGovernment Consumption.30 Private Capital Stock Response to one S.E. shock inPublic Capital Stock .29 Figure2.4 Figure 2.5 Per capita GDP Response to one S.E. shock inSocial Security Transfers ...30 Figure 2.6 Consumption........................................................................................................................ Public Capital Stock Response to one S.E. shock inGovernment Figure2.7 Tax BurdenResponse to one S.E. shock inGovernment Consumption.......30 31 Figure2.8 Public Capital Stock Response to one S.E. shock inSocial Security Transfers Tax BurdenResponse to one S.E. shock in Social Security Transfers..........31 31 Figure2.9 Figure 2.10 Per capita GDPResponseto one S.E. shock inGovernment Subsidies ........32 Figure 2.11 Private Capital Stock Response to one S.E. shock inGovernment Subsidies 32 Figure2.12 Per capita GDPResponseto one S.E. shock inthe Tax Burden ...................32 Figure 2.13 Private Capital Stock Response to one S.E. shock inthe Tax Burden..........33 Figure3.1 Participation of earmarked revenues as a share of federal government revenues (YO) ........................................................................................................................ 38 Figure3.2 Compositionof Transfers to States and Municipalities ................................. 39 Figure 3.3 Earmarked revenues by function (2002-2003) ................................................ 41 Figure 3.4 Evolution of the Main Contributions ............................................................... 42 Figure 3.5 Mandatory Expenditures -2004 ...................................................................... Figure 3.6 Federal Government Personnel Expenditures (IGP - DIdeflator) ...............44 45 Figure3.7 Federal government employment (thousands) ............................................... -46 Figure 3.8 Personnel Expenditure Composition ............................................................... 46 Figure 3.9 INSS Revenues and Expenditures (% of GDP) ............................................... 47 Figure 3.10 Real MinimumWage (*)................................................................................... 48 Figure3.11 Health Ministry expenditures (YOof GDP)...................................................... 49 Figure3.12 Health Ministry Expenditures .......................................................................... 49 vii Figure3.13 Evolutionof FederalGovernment PrimaryExpenditure .............................. 50 Figure3.14 LOAS/ LifetimeIncome ................................................................................... 51 Figure3.15 Ministryof Transport Investmentsand CIDE collection .............................. 54 Figure3.16 Unusedearmarkedresourcesbyfunction (2002/2003) .................................. 56 Figure3.17 Share of PersonnelExpendituresandTransfers to Municipalitieson PrimaryExpenditures(2004) ............................................................................................. 57 Figure3.18 MinasGerais State: expenditurebyfunction totaland earmarked - resources(2004) ................................................................................................................... Figure4.1 Brazil-Transport Matrix inbillionof tons-km-used (TKU), 2004..............61 58 Figure4.2 RoadNetworkExtension,2004 (km) ............................................................... 63 Figure4.3 FederalpavedRoadExpansionRate, 1965-2004............................................ 63 Figure4.4 FederalPavedNetworkCondition ................................................................... 64 Figure4.5 Federal GovernmentInvestments, 1975-2004 ................................................. 66 Figure4.6 Transport Sector investmentsComposition(average2000-05) .................... -70 Figure 4.7 Compositionof RoadInvestmentExpenditures2000-05............................... 71 Figure4.8 Transport MinistryTotalExpendituresLOAS 2003-05 (inR$biof 2004) 74 Figure4.9 RoadInvestmentsLOAS2003-05 (inR$biof 2004) ...................................... 74 Figure4.10 CongressionalAmendments inPercentageof AnnualBudget Laws' Allocations by Type of Program ........................................................................................ 75 Figure4.11 Commitmentsas percentageof totalbudget(LOA+Credits) ........................ 76 Figure4.12 Compositionof budget allocationsand committedresourcebytype of investment, (averages2000-05) .......................................................................................... 76 Figure4.13 Settlementof Congressionalamendmentsinpercentageof totalsettlement by type of program(averages 2000-2005) ......................................................................... 78 Figure4.14 Payment inproportionof settlement bytype of program .............................. 79 Figure4.15 Budget andExpenditureExecutionon roadinvestments,2000-2005 ...........79 Figure5.1 TotalNominalSpendingandTax Burden, 1995-2004................................... 85 LIST OFTABLES Table 1.1: EvolutionofPrimaryExpendituresandTax Burden. 1950s-2000s................3 Table 1.2: ConsolidatedGovernment Expenditures. 1995-2004by Economic Classification(%of GDP) .................................................................................................... 5 Table 1.3: Social Security Expendituresand SelectedIndicators: Braziland Comparator Countries (2000) .............................................................................................. Table 1.4: Evolutionof the FederalPersonnelandPensionExpenditures, 1995-2004 ...68 Table 1.5: Levels ofPrimarySpendinginthe BrazilianFederation, 1995-2004 .............9 Table 1.6: Administrative ExpendituresbyLevelof Government ................................. 10 Table 1.7: FiscalPerformanceof Non-PublicFinancialCorporations: Consolidated, 1995-200111 Table 1.8: of Government, 1995-2001.................................................................................................. Expendituresof Non-financialPublic Corporationsby Sector andbyLevel Table 1.9: Expendituresof financial public corporations, 1995-2001(%of GDP) 14 Table 1.10: Demonstrativeof Financialand CreditBenefits,2004 ..........................................12 15 Table 2.1: GovernmentNon-FinancialExpendituresandper capita Growth, 1950-2000 21 Table 2.2: (% of GDP).......................................................................................................................... FunctionalDistributionof Public Expenditures:InternationalComparison 24 Table 2.3: SimulatedGrowth Effectsof ExpenditureReallocations .............................. -35 ... Vlll Table 2.4: Simulated Growth Effects of 1percent of GDP Increase inExpenditures through debt financing ....................................................................................................... Table 3.1: Rigid Portion of Total Non-Financial Central Government Expenditure ....36 37 Table 3.2: Ministry of Health: Expenditure Distribution according to Budget Source 53 Table 4.1: Participation of Modes inthe Transport Matrix and Territorial Extension Selected Countries, in percentage and million of km2,most recent years ...................... 62 Table 4.2: Road density in selected countries (most recent years) .................................. 64 Table 4.3: Condition ofthe Paved Road Network. 2004 .................................................. 65 Table 4.4: Road quality indicators, 2004........................................................................... 65 Table 4.5: Logistic Costs and its main components Brazil and USA, 2004 .................... 66 Table 4.6: CIDE-collected revenue and its allocation, 2002-2004 (inReais million) ..68 Staffinglevels inthe Federal Road Administration (DNER-DNIT) ............-68 Table 4.7: Table 4.8: Federal investments inRoads by category, 2000-2005 ................................... 71 Table 4.9: PPAs and Annual Budget Law allocations, 2000-03 (in billion of Reais) .....72 Table 4.10: Total Budget Allocations by type of investment, 2000-05 .............................. 73 Table 4.11: Commitments and settlement amounts ............................................................ Table 4.12: Pilot Project programmed investments, 2005-2007 (R%million) ...................77 81 Table 4.13: PPIBudget and Financial Execution(Figures of December 2005) ...............82 Table 5.1: Federal Government Primary Surplus Changes, 1995 2005 - ....................... 87 ix X EXECUTIVESUMMARY 1. Thrs report offers a perspective on Brazil's public expenditure policy and its effects on economc growth. Specifically, it (i) assesses the overall level o f public expenditures in Brazil and its economc consequences, (ii) analyzes the growth impact o f the expenditure composition; (iii)discusses the nature o f budget rigidity, a major obstacle to expenditure reforms inBrazil, and (iv) evaluates the quality of expenditure management inthe road transport sector. It concludes with a set o f institutional reform recommendations to improve the expenditure management cycle as ways to address the weaknesses diagnosed inthe first four chapters. THESIZE OFTHE PUBLIC SECTOR: ITS CAUSESAND ECONOMIC EFFECTS 2. Brazil's public sector has grown rapidly over the last half century and become one o f the largest among the mddle income countries. From around 1950through the 1970s, the expansion o f the public sector was driven by an explicit development strategy that relied on strong state interventions inthe economy Duringthis period, Brazil recorded one o f the hrghest growth rates in the world. However, the expansionary fiscal policy that underpinned thrs growth strategy proved unsustainable. The growing fiscal imbalance and the resultant macroeconomc instability slowed econormc growth in the 198Os, as the country found itself m r e d in fast accumulating public debt and climbing inflation. Brazil's public sector continued to grow even after economc growth stagnated. The continued growth o f the state has been dnven by the constant increase in current expenditure, many o f which were mandated in the 1988 Constitution that embodied society's strong demand for an activist state, especially insocial areas. 3. Since the late 199Os, Brazil's public spending has surpassed 40 percent o f GDP, while the tax burden has reached 36 percent o f GDP Both figures are comparable to some o f the OECD countries with sizable welfare states. Data suggest that decentralization increased the overall size o f the public sector. Spending by municipalities has increased steadily duringthe 1990s because o f the effective transfers o f a number o f social sector responsibilities, as specified in the 1988 Constitution. At the same time the revenue shanng system has forced the federal government to seek additional sources o f revenue to balance its accounts in the context o f fiscal adjustments implemented since the late 1990s. 4. Public enterpnses remain important actors. Even after large-scale pnvatization in the 1990s, gross expenditure and revenue by non-financial corporations reachedmore than 14percent o f GDP in the early 2000s, although their investment has declined from more than 2 percent o f GDP registered in the md-1990s to around 1 percent o f GDP inrecent years. Available data do not p e m t precise estimation o f the extent o f quasi-fiscal operations by these non-financial public enterpnses. However, it seems reasonable to conclude that the overall magnitude o f such operations have dirmnished in importance as a result o f privatizations and the enhanced commercial onentation o f the largest public enterprise, the Petrobrhs Group. 5. Brazil's public financial corporations engage in public policy functions through the application o f directed lending and subsidized credits. Implicit subsidies estimated by the government amount to 0.55 percent o f GDP (or roughly 14 percent o f the revenue o f the public banks) in 2002. Even though the public policy roles o f public banks in Brazil are not negligible, the picture is far from one o f ubiquitous public financial institutions directing huge shares o f the country's financial resources in response to government directions. Nevertheless, the financial sector and fiscal-financial impacts o f public banlung operations have yet to be fully assessed. x1 6. In short, the expansion o f Brazil's public sector has been driven by several structural factors. The major dnvers o f Brazil's fiscal expansion in the 1990s include (i)the 1988 Constitution that significantly expanded the social roles o f the state, especially through the relatively generous and ngidly designed social entitlement programs; (ii) the consequences o f the past fiscal laxity that translated into hgh interest payments, an the inevitable fiscal adjustment, which, given the budget ngidity and the revenue sharing arrangement, has encouraged the federal government to resort to alternative revenue increases; and (iii)fiscal decentralization that enlarged the expenditure responsibilities o f the sub-national governments without offsetting them with a reduction in the size o f the federal government. Moreover, pressures for future spending increases are mounting, due to the aging of the populationandthe corresponding need to increase social security and health expenditures, similar to the pattern observed in a number o f OECD countnes. These pressures would tend to increase the size o f the public sector. However, simple projections based on these pressures clearly show further expansion o f the public sector would be unsustainable. THECOMPOSITION OFPUBLICEXPENDITURES ITS EFFECTS AND GROWTH 7 The present level and composition o f public spending are not conducive to long-term economc growth. The analyses on the effects o f expenditure composition on economc growth indicate that the current composition o f expenditure i s inefficient from the point o f view o f promoting growth. Specifically, the econornetnc analysis o f the relationshps between selected fiscal variables and per capita GDP growth over 1950-2000 finds: 0 A positive effect o f public capital stock on private capital stock (the reverse relationship i s also positive) and a positive effect o f both on economc growth - i.e., public investments contributedpositively to growth.] 0 A positive effect o f government subsidies on pnvate capital accumulation and thereby on growth due to the short-run fiscal impulse, which dampens over time due to debt accumulation. 0 A negative effect on growth o f government consumption (short and long run) and social security and assistance transfers (long run because o f the greater financing need that translates into a higher tax burden). 0 A negative effect ofthe tax burdenon capitalaccumulation and economc growth. 8. Over time, the composition o f public spending inBrazil has shfted from a more growth- oriented nux o f hgher public investments and targeted, capital subsidies to ever larger shares o f government consumption and transfers for social security and assistance. Additional taxation needed to finance the enlarged state (with a more inefficient expenditure composition) also has dampened growth. 9 We also simulated possible effects o f expenditure reallocations to further test the hypothesis that the current expenditure composition does not stimulate growth. Using realistic parameters denved from existing empirical studies on the Brazilian economy, the simulation exercise confirms the commonly-held view that Brazil would benefit from higher public investments within a tight fiscal constraint (increasing taxation or borrowing would limt the long-run growth prospects). Especially given the current low level o f public infrastructure investments, a reallocation from social security transfers to increase spending on economc infrastructure would likely have sizeable growth effects. On the other hand, runninga public debt ' While such results are consistent with most theoretical models, empirical evidence establishing this relationship is rare. xii to finance transfers decreases growth substantially, and runninga public debt to finance education expenditure would increase growth in the short runbut debt accumulation would slow growth in the long run. Also, spending on government consumption and transfers for social security and assistance dampens growthe2 10. Both the econometnc analysis and the simulation exercise support the same conclusion: to enhance economc growth Brazil should increase its public investment by reallocating expenditure from other less productive areas, while keeping the overall tax burden and indebtedness under control. Expenditure reallocations, however, are a difficult proposition from both political and t e c h c a l points o f view not only in Brazil but in any country. Nonetheless, over the long runBrazil must gain more effective control over the growing current expenditure in order to arrest the decades o f fiscal expansion and to reallocate more resources to public investment. The instruments at the government's disposal, at least in the short run, are very limted, especially due to existing budget n g i d i t i e ~ . ~ BUDGET RIGIDITY 11. A major challenge for expenditure reform in Brazil is the excessive degree o f ngidity arising from extensive revenue earmarkmg and the hlgh share o f constitutionally or legally mandated expenditure. For example, in 2003 for the federal government 80 percent o f the revenue (20 percent o f GDP) was earmarked for specific purposes and 90 percent o f total non- financial expenditure (19.5 percent o f GDP) was non-discretionary Earmarked revenue i s assigned to line mnistnes, w h c h are not required to spend the entire amount. Effectively, the remaining earmarked funds are effectively used to reduce the net borrowing requirement; however, these funds cannot be used to reduce debt liabilities. In contrast, non-discretionary expenditure i s mandated to fully fund government programs. Thus the overlaps between the revenue and expenditure rigidities are large, but not complete, the actual degree o f budget flexibility i s less than the "free" portion o f either the expenditures (10 percent in 2003) or the revenues (20 percent in 2003). In2002 for example, the actually "free" portion o f the budget was estimated to be between 4.5 percent o f revenue and 8.1 percent o f expenditure. Whatever the precise estimate, it i s clear that the portion o f the budget that the government can allocate "freely" withn an annual budget process i s quite limted. 12. In Brazil, there are essentially three broad sources o f budget ngidity: (i) earmarkmg o f tax revenue, the most important o f whlch are for constitutionally mandated transfers to states and municipalities and for education expenditure; (ii) contributions which by defirution must be social earmarked and need not be shared with sub-national governments, (a principal instrument usedby the federal government to close its financing gap in recent years); and (iii)non-discretionary expenditures that include legal or constitutional obligations o f the government such as interest, salary payments, entitlements (such as social secunty), and social assistance benefits. Of these, the share o f social contributions intotal federal revenue and the weight o f the non-discretionary expenditure have increased from 6 percent o f GDP in 1986 to around 14 percent o f GDP in2002. These simulations are based on the functional classification o f expenditure. While the results contrast with the recent findings inWorld Bank (2005) "Povertv Reduction and Growth: Virtuous and Vicious Circles", specifically that reductions in inequality and poverty tend to have positive growth impacts, the simulation model did not explicitly take these impacts into account. In May 2006, the Minister of Planning announced cuts to budgeted amounts for 2006 totaling R$14.2 billion (R$3.7 billion on current spending and R$6.2 on investment). xiii 13. The growth o f non-discretionary expenditure i s largely due to (i)the so-called "vegetative" growth o f social secunty benefits payment, which grew from 10.5 percent o f GDP in 1998 to 13 percent o f GDP in 2003, and (ii) designation o f the federal health expenditure as the mandatory under Constitutional Amendment no. 29 in 2000. The steady increase in the social security expenditures i s related to both the expansion o f population coverage - from 11.1 million in 1988 to 21.3 million in 2002 - and the increase m benefit levels, largely influenced by the successive upward adjustments o f the mnimum wage to w h c h the mnimum social secunty benefit is tied. Although the reforms in the late 1990s slowed the rate o f increase, the benefit level still remains quite generous by international comparison. 14. A key policy question regardingbudget ngidity is whether greater flexibility is attainable through reforms on the revenue side (Le., "de-earmarlung") or o f the expenditure side (i.e., "de- constitutionalizing" certain expenditures). So far government efforts in this realm have concentrated on ad hoc "de-earmarlung" measures, called DRU. The use o f formerly earmarked revenue has been directed to non-discretionary expenditure, as much as possible, malung the independent effect o f earmarlung on budget ngidity rather innocuous. The effects o f DRU are partially offset by the mandated increases for certain non-discretionary expenditure items, most notably the INSS benefit payments (covering in part the social security deficit; R$17.2 billion in 2002) rather than to increase infrastructure investments. 15. Proponents o f specific earmarlung generally expect that earmarlung will guarantee a certain level o f budget allocation to the favored expenditure item. However, evidence in Brazil and internationally shows that revenue earmarlung i s often an ineffective means for "protecting" funding for specific purposes. At the same time, earmarlung contributes to budget rigidity that can complicate day-to-day management o f government budgets. In this context, de-earmarlung (DRU) can have beneficial effects. In2002, de-earmarked revenue was reportedly usedto cover about 40 percent o f the pnmary surplus. Tlvs does not mean, however, that increasing the extent o f de-earmarlung would bnng about correspondingly larger benefits because the additional de- earmarlung may simply be absorbed inlarge part by rapidly nsing non-discretionary expenditure. PUBLIC SECTOR MANAGEMENT OF INVESTMENT EXPENDITURE: THEEXAMPLE ROADTRANSPORTATION OF 16. Assumng existing budget rigidities can be dimnished, more funds could become available for public investment. Currently over 80 percent o f total central government investment i s dedicated to the transportation sector. Nevertheless, the report shows that greater efficiency could be obtained, even from existing levels o f investment. 17 Data for the road transport sub-sector, one o f the most critical bottlenecks to sustained economc growth, supports the inference drawn from the economc analyses reportedabove. The extent and the quality o f road infrastructure is insufficient to support the country's economc activities, especially exports, and thus increased public investments in roads and other transport infrastructure should enhance Brazil's growth prospects. 18. A closer look into the way in which budgetary resources are managed in the roads sub- sector shows, however, that a simple increase in budget allocation may not lead to immediate expansion in the formation o f public capital in the sub-sector. Besides the resource scarcity, weaknesses in planning, budgeting and expenditure execution make production o f infrastructure highly inefficient. Expenditures inthe sub-sector, inspite o f their scarcity, are not well-planned and pnoritized, and their execution suffers from a variety o f causes for delays and cost escalation. xiv 19 Prioritization and resource allocations defined in the multi-year plans (PPAs) do not guide annual budget formulation. Ths is shown by the inconsistency between the allocations to the programs included in the PPAs for 2000-2003 and 2004-2007 and those set in the government's annual budget proposals and the Annual Budget Laws (LOAs) approved by Congress. This inconsistency has several roots including: (i) technical imperfections in medium- term revenue forecasting and program costing; (ii) particular vulnerability o f the transport budget to Congressional amendments and discretionary management o f expenditure execution by the Executive (e.g., so-called "c~llti~ge~czumeYt~", or partial freeze on spending authorizations); and (iii)frequentdelaysinphysicalexecutionofpublicworksduetoirregulantyinresourceflows, adverse weather (which tends to coincide with the time when the budgeted funds are released duringthe fiscal year), and (sometimes) detailed external oversight for fear o f corruption. 20. The government has introduced a new pilot initiative (Pilot Project for Infrastructure, or PPI) to improve the efficacy o f public investment execution. The objective i s to improve budget execution of selected infrastructureprojects by protecting them from funding freezes and closely monitoring their execution by central economc agencies. The implementation experience of the PPI itself suggests that the lack of funding i s far from the only, or perhaps not even the most important, constraint to efficient execution of the transport investment portfolio. The lack o f institutional capacity for appraisal, implementation and evaluation o f public investment projects continues to be a major obstacle for the enhancement o f the efficiency in those public infrastructure projects. MEDIUM-TERM AGENDAFOREXPENDITURE REFORMSBRAZIL IN 21. The findingsof this report lead to the followingset ofpolicyrecommendations. First, the growth of the non-discretionary expenditures needs to be controlled, and if possible, reversed. Second, gradual increases in the budget allocations to economc infrastructure seem desirable. Tlurd, increased allocation to economc infrastructure should be accompanied by systematic efforts to strengthen institutionalcapacities for plamng and execution of public mvestments. 22. The containment of the growth of non-discretionary expenditures depends on deepemng of structural fiscal reforms. These include further reforms of the social security system, rationalization of other social obligation (entitlement) programs such as those defined in the Organic Law of Social Assistance (LOAS), and admnistrative reforms to allow governments to adjust their staffing levels and composition with greater flexibility. Better targeting o f spendingin social assistance programs and in health and education, could provide some efficiencies that could augment discretionary spending. 23. Although each one of these measures would require a solid political comrmtment and consensus as well as sound analytical bases for reforms, certain other improvements in Brazil's institutional arrangement for planning and budgeting mght help the government to control the path of expenditure and their composition over the mediumterm. The most common instruments that serve t h s purpose are a medium-term expenditure framework (MTEF), combined with a politically well supported budget rule to guarantee fiscal equilibnum and to control the growth in the size of government. An MTEF would keep government expenditure and revenue levels withn targeted levels because it makes medium-term fiscal impacts of policy decisions explicit and makes this information available dunng the budget plamng process. Withn a well functiomng MTEF, explicit targets for the levels of expenditure and taxation (taxation and expenditure level, or TEL) could result from policy debates and decisions. Given that fiscal rules that are excessively rigid runthe risk of frequent non-compliance and thus a loss of credibility, it xv may be advisable for Brazil to adopt indicative TEL targets within the context of a medium-term budget strategy 24. The analysis inthis report indicates that without changing the expenditure composition in favor of investments in economc infrastructure, the budget's contribution to long-term growth will remain limted. An MTEF could also facilitate reallocations of budget resources over time, provided that the structural causes of budget ngidity are liRed. Even in the case of successful constitutional reforms to bringgreater budget flexibility, however, it i s not clear there i s sufficient information to detemne exactly where the budget should be cut and where it should be preservedor increased in the absence of an MTEF An institutionalized process of expenditure reviews for generating mcro-level information about saving options would therefore be a useful addition to Brazil's planning and budgetingprocess. 25. Finally, the review of expenditure management for the road transport sub-sector reveals considerable needs for institutional strengthemng. While the report did not examne other types of infrastructure (in part because transportation represents the bulk of current investment spending), it is likely that the situation could be similar for other public investments. The government clearly sees infrastructure investment as a prionty (as suggested by the PPI), however, a clear strategy for improving infrastructure mnistries' capabilities to manage these limted resources efficiently and effectively i s laclung. Such a strategy could involve the following elements: (i) a medium-term infrastructure investment strategy (linked to the MTEF); (ii)setofperformanceindicatorstomonitorefficiencyandeffectivenessoftheinfrastructure a portfolio as a whole, as well as by sector and sub-sector; and (iii)an organizational reform, including adoption of a human resource policy to enhance the professional quality of the staff worlung in the infrastructure sector and to streamline the mnistries' business processes and organizational structures, ifnecessary xv1 1. THE LEVEL OFPUBLIC SPENDINGINBRAZIL HISTORICAL EXPANSIONOFTHE STATE SECTOR 1.1 The Brazilian economy has been growing fast in the last decades, but the size o f the public sector has been increasing even faster. Evenwith the deceleration that took place since the early 1980s, the average annual growth rate o f the Brazilian economy from 1947 to 2000 was 5.5 percent. Inthe same penod, however, the share of the Brazilian government inthe economy has grown at a faster rate o f 7.5 percent. As a result, the share of the public sector in the economy more than doubled from 15 percent o f GDP in 1947to close to 35 percent in2000. 1.2 The actual evolution of the size of the public sector in Brazil during this period is hardto estimate. Onone hand, the tax burdenis probably a poor indicator of t h s evolution, and tends to underestimate its rate o f increase. Fiscal deficits and their corollary, debt accumulation, have been a mark o f the period post WWII, although the latter has been rmtigated by inflationary surges up to the stabilization process in m d 1990s. Moreover, a substantial volume o f unrecorded liabilities were generated along t h s period, through different mechanisms - controlled prices of state owned utilities, subsidies in credit operations, etc. In this sense, the current generation i s paying the taxes that previous generations did not pay. On the other hand, adopting a broader concept of public sector makes it clear that there was an enlargement since the 1970s, with the creation of several state enterpnses, funds and other mechanisms of state intervention. Thus process was reversed in the 1990s, with privatization and incorporation o f hidden liabilities to public debt statistics. Figure 1.1 Evolutionof Tax BurdeninBrazil,1947-2004(%of GDP) I 401---- 10 I , , , ,, , , / / , , , / I , , I , , , , , , , I, , , , , , , , , , , , , , , , , , , , / , , , , 1 , , , , , , 1 Source: IBGEand Ministryof Finance: Federal Revenue Secretanat (SRF). 1 1.3 The activist state underpinned economic growth and state expansion in the mid 20th century. In fact, public expenditures played a major role inthe development process dunng the inward-onented expansion o f the 1960s and 1970s inBrazil. The active role o f the government, accentuated during the military penod, i s considered one o f the explanatory factors for the spectacular growth performance inthe second half o f the 1960s and the 1970s. The enhancement o f tax collection capacity tnggered by the Tax Reform o f 1967 allowed the government to restore its fiscal equilibnum and to boost expenditures. Primary expenditures (or non-financial expenditures, which exclude amortizations and interest payments on debt) rose from 20 percent o f GDP in the early 1960s to more than 25 percent at the end o f this decade. Inthe 1970s, primary expenditures accompamed the strong growth o f the economy, fluctuating between 20 and 25 percent o f GDP Government savings also experienced a strong increase from -1 percent o f GDP in the early 1960s to 5 percent of GDP inthe first half o f the 1970s. Backed by the increased government savings, investment expenditures reached a peak o f 5 percent o f GDP and subsidies climbed to 2 percent o f GDP duringthe 1970s. 1.4 The second round of state expansion, in the 1990s, was driven more by political and social concerns. Low growth, highinflation, and increasing fiscal disequilibria charactenzed the macroeconomc context in which the country initiated its re-democratization process in 1985. The increasing social demand that resulted from political openness created additional pressure for expenditure e~pansion.~ The 1988 Constitution reinforced the ngidity o f government expenditures through the increase in mandatory federal expenditures, the expansion o f intergovernmental revenue sharing, and the earmarking o f an increasing part o f tax revenues (World Bank 2002,5 Velloso 2004). The 1988 Constitution also expanded the social responsibilities o f the state, guaranteeing free access to social services, establishing higher social security benefits, creating entitlements for rural workers, elderly, and disabled people, and expanding the generous employment regime for all public sector employees, including employment tenure, hgher compensations and an extraordinarily generous social secunty system (Bevilaqua and Werneck, 2000). The effect o f all these measures was reflected immediately in the marked increase in primary expenditures, w h c h jumped from 25 percent o f GDP in 1988 to more than 30 percent o f GDP in 1990. A main cause was the increase inpersonnel expenditures and social security benefits for public employees, which increased from 8.5 percent o f GDP in 1988 to 12.5 percent in 1990. For retired pnvate sector employees, social secunty transfers and social assistance programs contributed to an increase o f expenditures by more than 3 percent o f GDP between 1988 and 1991. Meanwhile, investment expenditures and subsidies suffered strong reductions. 1.5 The third round of government expansion coincided with the stabilization of the economy and the fiscal adjustment effort of the late nineties. The detnmental effects o f the 1988 Constitution on fiscal sustainability would have been immediately evident, but inflation masked the fiscal disequilibna duringthe early nineties. With the end o f the inflationary process, the deterioration o f the fiscal situation became more visible. Previously, inflation served not only as a source o f revenue but also as a useful mechamsm to control government expenditures inreal terms. This loss o f maneuvering space to adjust expenditures, coupled with the increasing path o f mandatory expenditures and the lack o f fiscal reform, resulted in nsing government deficits and the strong m e o f public debt from 29 percent o f GDP in 1994 to almost 42 percent o f GDP in 1998 (Blanc0 and Herrera, 2004). 4 Increasing fiscal disequilibria were a main reason for the failure o f the two stabilization attempts implementedby the democratic government in 1986 and 1987 Brazil: Issues inFiscal Federalism. Report No. 22523-BR. May 2002. 2 1.6 Since 1999, the Brazilian government has carried out rigorous fiscal adjustments, and achieved a remarkable turnaroundin its fiscal performance from a primary deficit of 0.2 percent of GDP in 1998 to a surplus of 4.6 percent in 2004. Giventhe strong expenditure rigidity, this fiscal turnaround was accomplished by a strong increase inrevenues-the tax burden grew from 29 percent of GDP in 1998 to more than 36 percent of GDP in 2004 - as well as by curtailing investment expenditures - public investments by the Federal Government fell from 1.1 percent o f GDP in 1998 to 0.5 percent o f GDP in2004. Table 1.1: Evolutionof PrimaryExpendituresandTax Burden, 1950s-2000s 1950s 1960s 1970s 1980s 1990s 2000s Primary expenditures (as % o f GDP) 20.1 22.7 22.9 25.1 33.2 35.5 Tax burden 15.8 194 24.4 24.3 28.1 35.2 Decade averages _ _ _ _ _ _ ~ ~ ~ _ _ _ _ _ _ _ ~ ~ ~ Source: IBGE; National Accounts and Ministryo f Finance FederalRevenue Secretariat (SRF) 1.7 Given its current high level, it is unlikelythat Brazil can withstand a new round of increase of public spending without considerable efficiency loss in the economy. But there are some built-inpressures to increase spending that must be dealt with, inorder to avoid either a worsenmg of the quality o f public expenditures or a new increase in the tax burden, both with possibly deletenous impacts on economc growth in Brazil. As pointed out elsewhere in this report, the way public spending can help economc development in Brazil today i s through an improvement inits quality. THEGENERALGOVERNMENT SECTOR 1.8 After continuous state expansion, the tax burdeninBrazil has reached 36 percent of GDP in 2004 and aggregate spending has been over 40 percent of GDP since 199EL6 While the increase inthe interest payments, from 4 to 6 percent of GDP in 1995-97 to around 7 to 8 percent of GDP in 1998-2002, i s partially responsible for ths increase in the aggregate spending level, the data also shows a steady upward trend inthe level o f primary spending from around 32 to 33 percent to 35 to 36 percent of GDP for the 1995-2002 pen~ds.~ O f t h s total, we estimate that roughly the equivalent of 2 percent of GDP that i s classified as primary expenditures by the GFS methodology is actually best charactenzed as financial expenditurese8Even after discounting ths portion, the government expenditures since 2001 have beenabove 40 percent of GDP. As noted, using the tax burden as an indicator ofthe size o f the state canbe misleading. Measured by the tax burden, the public sector size has increased at an impressive pace after the stabilization o f the economy (see Figure 1.1). But aggregate expenditures show a flatter profile. This is so because the bulk o f the increase in the tax burden, especially after 1998, was directed to generate primary surpluses required to guarantee the long runsolvency of the public sector, by attempting to stabilize the debt to GDP ratio. We used the series that disregards the effects o f exchange rate volatility ininterest payments. These large fluctuations - in 2002 the exchange rate depreciated by more than 50 percent (between January and October), going from R$ 2.40 to R$ 3.90, then rolling back to a level R$3.00 in early 2003 - have strong impacts on interest payments calculated on an accrual basis, but are best characterized as asset changes. Recognizing this fact, the Brazilian Central Bank calculates a series o f interest payments disregarding the exchange rate effects. * This is probably the case of "financial and fiscal affairs", which reached 4 percent of the GDP and may include expenditures that are not currently recorded as prmary expenditures. After a detailed examination 3 1.9 This level of aggregate public spendingis unusually high givenBrazil's incomelevel, and i s actually close to the European average (45 percent), and above the OECD average (36.5 percent o f GDP in 20009), where the welfare state is far more developed than Brazil. If we restnct the comparison to countries o f similar levels o f GDP per capita, the differences are strilung. For example, Mexico, Chile and South Korea, all economes with higher growth rates than Brazil in recent years, all spend less than 25 percent o f GDP overall, although at least inthe case o fMexico, this level i s considered a little low Figure 1.2 Primary andTotalOutlaysinBrazil, 1995-2004 m TOMPrimary Spend. (% GDP) Net Deb4 IntmealPayments(% GDP) -Total PrimarySpending and Net Dsbl I n l e ~ l&Primary Surplus 50 45 40 35 30 25 20 15 10 5 0 1665 1888 1697 1698 1869 20W 2001 2002 2003 2W4 1695 1888 1887 1698 1698 2000 2W1 2W2 2003 2W4 Source:Basedon Garson (2004) andCentral Bank. Figure 1.3 General Government Outlays:Braziland Comparator Countries(2000) 80 - Fmnw Denmark E 5 0 - * bustna -sg n . 0 Bra21 Greaoa M y tEelgiu~ 'Finland Garmany.) a-- 4 0 - J p " Ponugal I 301 t + U K Noway Spain Canada 30 - Chile Ch:e Australia USA wEB f5E E Korea Ireland 20 i 0 ' 5000 10000 15000 20000 25000 30000 35000 40000 45000 Source OECD(2002) GDP per Capita (Us$) ECONOMICCOMPOSITIONOFPUBLICSPENDING 1-10 Interms of the compositionof the public spending,the observedincrease inthe total spending from figures hovering near 39 percent of GDP in 1995-97 to figures around 42 of the items included, we estimate at 2 percent the maximum value that could be considered as financial, rather than primary expenditures. OECD 2003. 4 percent in 2000-2002 is due to increases in both non-financial public expenditures and interest payments. "Compensations of employees" remained stable throughout the penod. The upward trend shown in the payroll of municipal governments was compensated by the slight reduction observed at the federal level. There i s an upward trend in expenditures on "use of goods and services." However, the sharp increases observed in2001 and 2002 can be associated with methodological differences introduced with the accounting change in 2000 (e.g., likely incorporation of "subsidies" into "use of goods and services")." 1.11 "Social benefits" - ths does not include public sector pensions, which are included as part of "compensation of employees" - show a clear upward trend up to 2000, driven mainly by federal spending, that accounts for 90 percent of t h s type o f spending, due to the INSS comrmtments. When we add 2001 and 2002, even after incorporating FGTS and the S System, there i s a reduction, leading to a flat behavior dung the period. Methodological differences between the two data sets may explaint h s change of pattern. 1.12 Investments ("acquisitions of non-financial assets") show a slight decreasing trend, from 2.5 percent of GDP in the beginning o f the penod to 2.2 percent at the end. The decrease was more intense at the federal level (see Garson, 2004, p. 8), partially offset by the increase observed at the state level. Table 1.2: ConsolidatedGovernment Expenditures, 1995-2004 by Economic Classification (YOof GDP) 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Compensation o f employees [GFS] 15.2 14.5 14.7 14.7 14.2 14.9 14.9 15.0 14.8 14.1 Use o f goods and services 5.9 6.2 6.1 6.4 6.4 6.6 8.5 9.2 n.a n.a Subsidies 0.6 0.4 0.7 0.4 0.3 0.3 n.a n.a n.a n.a Grants 1.1 0.9 0.6 0.6 0.8 1.5 1.0 1.1 n.a n.a Social benefits [GFS] 8.2 8.4 7.6 9.3 9 7 9.0 8.1 8.3 n.a n.a Other expenses 0.2 0.2 0.1 0.2 0.2 0.1 0.5 0.0 n.a n.a Current primary expenditures 31.2 30.5 29.8 31.5 31.5 32.4 32.9 33.6 33.6 33.2 Acquisitions ofnon-financial assets 2.5 2.5 2.3 2.9 1.9 2.1 2.3 2.2 1.9 2.2 Total primaryexpenditures 33.7 33.0 32.1 34.4 33.4 34.5 35.2 35.8 35.5 35.4 Interest payments 6.3 5.1 4.7 7.3 8.5 6.8 6.8 7 7 9.3 7.3 Total 40.0 38.1 36.8 41.7 41.9 41.3 42.0 43.5 44.8 42.7 DRIVERSOFFISCAL EXPANSION 1.13 The observed increase in aggregate public pending is due to the increase in current expenditures. Data compiled by Velloso (2004) based on National Accounts for current expenditures of the general government show a clear upward trend from 1970 to 1999. Up to 1988 there was a moderate increase (from 20 percent to 22.4 percent of GDP) in current expenditures, while public investments in infrastructure were kept at a level slightly below 5 loIt should be noted that according to the 2001 Government Finance Statistics method for expenditure classification, "Subsidies" should not be considered as part o f "Use o f Goods and Services," but the Brazilian fiscal data do not allow the distinction between the two since 2001. 5 percent o f GDP. From 1988 on, the upward trend incurrent expenditures has been much steeper, with these expenditures eventually reachng 30 percent o f GDP at the end o f the 1990s while investment expenditures were drastically reduced at levels around 2 percent o f GDP 1.14 Effects o f these changes, which are attributable to a large extent to the fiscal effects o f the 1988 Constitution, as discussed above, could have led to a more rapid imbalance in fiscal accounts, but part o f the impact was kept hdden by inflation. Actually, inflation in Brazil postponed the collapse o f this fiscal regime and masked the fiscal disequilibria during the early nineties. Dunng this hgh-inflation penod, indexation of taxes and sub indexation of expenditures helped the government to obtain positive fiscal outcomes, despite the msmatch between limted fiscal resources and increasing obligations (Cardoso 2002 and Giambiagi 2001). 1.15 Social obligations: Of the new social obligations enshrined inthe 1988 Constitution, the most outstanding example in terms o f its fiscal impact i s the pensionbenefits. The imbalance in the social secunty system i s well known, but nonetheless, it i s worth talung a closer look at the figures. Public spending for the pension system reaches more than 11 percent of GDP, in a country in w h c h the population i s still very young - less than 10 percent are above 60 years o f age. Ths level o f spending compares to those of mature countnes, as shown in Table 1.3 below.'] In this table, countnes are ranked by the level o f expenditures with public sector employees' pensions to hghlight the role played by these pensions in explaimng the extremely highlevel of spending inBrazil. Table 1.3: Social Security Expenditures and Selected Indicators: Brazil and Comparator Countries (2000) Social Security Social Security Expenditures Population above Expenditures O/O Public Employees 60 years old of GDP Yo of GDP YOof Total Brazil 11.5 4.7 8.4 Austria 13.4 3.4 20.7 France 12.3 3.0 20.5 Belgium 10.8 2.7 22.1 Portugal 8.7 2.6 20.8 Greece 12.1 1.9 23.4 Germany 11.2 1.7 23.2 Netherlands 9.3 1.6 18.3 USA 6.3 1.6 16.1 Australia 6.2 1.3 16.3 Japan 7.0 1.1 23.2 Denmark 8.4 1.o 20.0 Spain 9 4 0.9 21.8 Canada 6.0 0.7 16.7 Ireland 3.8 0.6 15.2 Source: OECD and Ministry of Social Security, MPAS (2001) III t can be argued that in Brazil, some pensions are similar to pure cash transfers, since many pension earners did not or do not contribute to the system. This i s particularly true for rural pensions, in spite of the fact that even in this case there is indeedsome contribution. Inany case, this distortion is not large enough to invalidate the comparison, nor does it affect the ranlung presented inthe table. 6 1.16 Not only are the public expenditures for pensionbenefits hgh, but they have also grown steadily in recent years because o f the built-in mechanisms for automatic increases. Pension payments for both retired public servants and retired private workers have increased steadily between 1995 and 2003 (Figures 1.4 and 1S), Expenditures with pensions increased at all levels o f the public sector in the last 13 years. Total expenditures with the public sector system are hovenng around 5 percent o f GDP since 1999, and the federal government and the states spend more than 2 percent o f GDP each. Although a series o f reforms in the public pension system since the late 1990s seems to have succeeded in stabilizing total spending, both spending and the deficit remainquite hgh. Figure 1.4 Evolutionof TotalPensionExpenditures, 1990-2004 Total PensionErpendlturer as %of GDP, 19!MkZW 10?6 - . . .. 0% - . - ..________.-I. , I990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 --cRGPS(privateworkers) -a- RPPS(public workas) +Total 1 Source: Ministry of Social Security, MPAS (2003) and World Bank (2005) Figure1.5 EvolutionofPublic Sector PensionExpenditures, 1990-2004 byLevelof Government RPPSExpendituresas */ir of GDP, 1990-2004 6% 1 I Source: Ministry of Social Security, MPAS (2003) and World Bank (2005) 1.17 Remarkably, there was actually a decline in the wage bill at the federal level during this period, but this decline was fully offset by the correspondingincrease in the spending for retirement benefits. Public expenditures for these benefits grow steadily because they are indexed to the mnimum wage which i s itself set by the legislative act o f Congress, and because 7 both the pension benefits and the mnimum wage have been adjusted at least by the rate o f inflation on an annual basis (Velloso 2002). Besides, the pension benefits for retired public servants are automatically adjusted at the same rates as the salary increases for the active-duty public servants in the same employment categories. Since these pension benefits are constitutionally guaranteed entitlements, they have also contributed to a significant increase inthe structural ngidity o f the Brazilian public budgets which has lirmted the scope for fiscal adjustments by way o f expenditure reductions, and instead contributed to the steady increases in tax revenues (see Chapter 3 for greater detail). Table 1.4: Evolution of the Federal Personnel and Pension Expenditures, 1995-2004 (Yoof GDP) 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Public SectorPersonnel 5.6 5.3 4.8 5.0 5.1 5.0 5.4 5.5 5.0 4.8 Active 3.3 2.9 2.6 2.6 2.6 2.7 2.8 2.7 2.7 2.6 Retired 2.3 2.3 2.2 2.4 2.5 2.4 2.7 2.7 2.3 2.2 INSS Pension 5.0 5.3 5.4 5.8 6.0 6.0 6.3 6.5 6.9 7.1 Source: Velloso 2004 and IPEA. 1.18 Interest payments: Another expenditure item that has grown in recent years and pushed up the overall level o f public spending is interest payments. Today's relatively hgh interest payments reflect both the continued fiscal vulnerability o f the country and consequences o f lax fiscal policies in the past which have resulted in accumulation o f a large stock o f public debt. A recent World Bank report (2003) shows that long-term interest rates are related to the market's perception of the public sector solvency. The report argues that the most prudent measure to reduce interest rates would indeed be to sustain credibility for fiscal responsibility (e.g., by maintaimng sufficiently high primary surpluses as a signaling device, precisely as the two successive governments have done), while stimulatmg growth with a fiscally neutral policy (World Bank, 2003"). 1.19 An important share of the current stock of public debt originated at the state level through the series of bailout operations since the 1980s. States' lax fiscal behaviors were already evident by the m d 1980s when state governments would freely issue bonds or borrow from their own state banks to finance their expenditures, including wage bills, and rollover o f existing debts. The situation worsened with the promulgation o f the 1988 Constitution that gave states (and municipalities) larger expenditure responsibilities than revenue autonomy. Combined with the relatively highdegree o f political autonomy o f sub-national governments, the new fiscal federalism arrangement created built-in incentives whereby states and municipalities would increase their spending without corresponding efforts to augment their own revenues but instead with reliance on federal transfers or borrowing to finance the expanded expenditure responsibilities. 1.20 The most evident manifestations o f the problematic nature o f this fiscal regime were the three episodes o f bailouts o f state governments between 1989 and 1997. Inthese bailouts, state debts equivalent to significant shares o f GDP (1.38 percent in 1989, 7.24 percent in 1993, and 11-6.5 percent in 1997) were renegotiated in terms favorable to states. In addition, state debts equivalent o f 1.03 percent o f GDP were forgiven entirely in 1997 Only the 1997 bailout was conditioned to states' implementing rigorous fiscal adjustment programs, including liquidation of state-owned banks that had frequently been sources o f soft loans to state governments. Sub- '`World Bank (2003), Country Economic Memorandum: Brazil Stability for Growth and Poverty Reduction. Report No. 25278-BR. 8 national fiscal discipline has improved since then thanks to a set of measures taken - e.g., liquidation o f most state-owned banks which usedto be major sources of supply of credit to state governments, the introduction of the Fiscal Responsibility Law in 2000, and the relatively tight- fisted management of sub-national borrowing by the federal government under bilateral debt refinancingagreements, including an application o f sanctions to the State o f Minas Gerais when it defaulted in 1999 (Bevilaqua 1999). Nonetheless, the burdens o f past fiscal indiscipline continue to compromse the quality of expenditure composition today. 1.21 Fiscal federalism: Since the md-l99Os, expenditures have evolved differently among the three levels of government. The federal government and the municipal governments have shown a clear upward trend inexpenditures, while state governments have kept their expenditures roughly stable since the m d 1990s. Expenditures by the state governments have been restrained partly because (i) their revenues were kept stable (as a proportion of GDP) - states did not profit from the decentralization that took place after the 1988 Constitution nor from the re-centralization that occurred inthe md-nineties; (ii) most o f them currently have no fiscal room for borrowing, a situation that has been tightened up fitrther with the debt agreement in 1997 and subsequently the Law o f Fiscal Responsibility in 2000; and (iii) the capping of the interest payments as part o f the bilateral debt re-negotiation agreements most states have signed with the National Treasury But a number of the states struggle to keep their current spending, especially personnel expenditures, under control, and to make more resourcesavailable for public investments. Table 1.5: Levels of Primary Spending in the Brazilian Federation, 1995-2004 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Consolidated general 33.7 33.0 32.1 34.4 33.4 34.5 35.2 35.8 35.5 35.4 government Federal government 20.9 20.2 20.7 22.0 22.1 22.4 21.3 21.8 20.9 21.3 State governments 12.3 12.4 11.7 12.8 11.8 12.5 12.6 12.8 12.2 12.2 Municipal governments 6.4 6.5 5.8 6.5 6.6 7.3 7.2 7.2 8.0 7.6 Source: Garson (2004) based on IBGE figures until 2000. From 2001-04 Ministry of Finance: National Treasury Secretariat (STN). Note: figures do not add up because o f intergovernmental transfers. 1.22 The increase inmunicipal spendingcan be seen as a positiveoutcome, inprinciple, to the extent it reflects the movement towards greater social sector decentralization. There i s indeed evidence that greater municipalization in certain areas of health and education has improved delivery of services, at least in quantitative term^.'^ However, the data on the aggregate spending level is not sufficient to establish whether each municipal admnistration i s actually taking advantage of the presumed informational advantage and incentives for greater accountability to local citizen needs so as to improve efficiency inexpenditure allocations and execution. There i s recent evidence from ongoing World Bank research that the central mmstry's capacity for traclung and controlling transferred resources and for demanding performance in exchange i s extremely limted in the health sector.I4 The recent controversy related to the allegedly ineffective monitoring of the Bolsa Familia conditionalities i s another example of the weak central capacity for controlling federal expenditures at the sub-national level. A more reliable assessment of the possible efficiency gains from greater municipalization o f fiscal resources would requiredetailed exammations of expenditure allocation and execution at eachmunicipality, which i s beyondthe scopeof ths report giventhe data limtation. I3See, for example, World Bank (2002), Brazil Municipal education resources, incentives, and results. - - Report No 24413-BR,shows positive impact o f municipalization on educational outcomes. l4 "Pesquisa sobre Rastreamento go Gasto Pdblico em Sadde." Interhealth Ltda. 2004 (unpublished draft). 9 1.23 There appears to be a trade-off between presumed gain in expenditure efficiency from greater municipalization on the one hand and the loss of scale of economy and increased administrative overhead that results from the proliferation of small municipalities on the other. The 1988 Constitution introduced flexible conditions for the creation o f new municipalities, and unleashed an immediate trend o f municipal proliferation (i.e., a 33 percent increase between 1988 and 2001). The guarantee o f receiving revenue transfers from States and the Federal Government created a perverse incentive for municipalities to split into ever-smaller, autonomous units - for municipalities with less than 10,000 inhabitants, these fiscal transfers account for 95 percent o f total revenues. The net effect o f the proliferation i s a redistribution o f resources away from larger municipalities, where the expenditure needs are arguably greater, to smaller municipalities, where the capacity to spend effectively i s often limted. 1.24 As a consequence of this process, there has been a clear increase in administrative expenses with executive and legislative organs, which contributes to a worsening in the quality of public expenditure. More than 18 percent o f total expenses o f municipalities were devoted to ths lundo f admnistrative expenses with the head o f the executive and with legislative organs. For municipalities up to 5,000 inhabitants, this number was equal to 22.7 percent. These figures are much higher than the comparable numbers for states and the federal government, indicating an association between the proliferation o f small municipalities and the diversion o f resources to less focused activities.15 Table 1.6: Administrative Expenditures by Level of Government Expenditure with Executive and Legislative Organs, 2002 Federal Government States 12.6 I Municipalities 18.3 Upio 5000 inhabitants 22.7 Source: Garson (2004) THEBROADER PUBLICSECTOR 1.25 The size of the Brazilian public sector is even larger once we take into account the activities of the entities outside the general government sector. These include financial and non-financial state enterprises whose operations can result inliabilities and assets that will affect the public sector's overall fiscal health inthe long run. Moreover, governments frequently pursue fiscal policy objectives through extra budgetary funds, as well as through tax expenditures. 1.26 Off-budget or quasi-fiscal activities have been extensively used by past governments inBrazil. One way to measure the incurrence or realizationo f contingent liabilities, off-balance- sheet comtments, or other fiscal risks, i s to see what part o f the growth in debt as a share o f GDP i s unexplained by the accumulationo f fiscal deficits, valuation changes inthe stock o f debt, privatization receipts and GDP growth. Goldfajn and Guardia (2004) show that almost 45 percent o f the expansion o f the consolidated net debt between 1994 and 2002, equivalent to 12 l5Although the Constitutional Amendment number 15 in 1997 introduced more restrictive rules to limit the process o f creation of new municipalities, the distortion in terms of revenues distribution among the existing municipalities still exists. 10 percent o f GDP, was due to debt recognition, the so-called hdden liabilities. Giambiagi and Ronci (2004) estimated that balance sheet adjustments were responsible for a net addition in the public sector debt of no less than 19 percentage points o f GDP between 1994 and 2002.16 While the accumulation of hdden liabilities and their explicit recognition at a later point does not by itself lead to larger expenditures, the resultant increase in the debt stock can negatively affect the level of the current fiscal activities through higher explicit debt payments. 1.27 Non-financial public corporations: Historically, Brazil's state-owned enterprises have played important roles in the country's development model. Through direct investments in key infrastructure and rnanufactunng sectors, non-financial public corporations used to contribute a significant share of the total public investments (above 70 percent), dunngthe 1980s. Their roles in the Brazilian economy remain important today For the three levels of government, total revenue of the non-financial public corporations represented 14.7 percent of GDP in 2001 and total expenditure 14.2 percent of GDP.I7 Federal public corporations account for the largest shares of both revenues (11.3 percent of GDP) and expenditures (11.4 percent of GDP), while the shares of the state and municipal public corporations declined from 1995 to 2001. The participation of the government in commercial activities in 2001 concentrated on rnanufactunng (with total expenditures representing 6 percent of GDP) followed by energy (4.2 percent of GDP), trade (1.8 percent of GDP) and public utilities (1.1 percent of GDP). At the same time, gross investment by non-financial sector public corporations as a share of GDP decreased for all levels o f government, reaching 1.3 percent of GDP in2001 (Table 1.7). Table 1.7: Fiscal Performance of Non-Public Financial Corporations: Consolidated, 1995-2001 1995 1996 1997 1998 1999 2000 2001 Consolidated Total revenue 14.06 14.31 15.89 10.63 12.06 12.94 14.71 Total expenditures 14.16 14.68 15.80 10.86 12.72 13.58 14.88 Gross investment 2.18 2.20 2.40 1.51 1.20 1.02 1.26 Overall fiscal balance (0.29) (0.55) 0.00 (0.58) (1.19) (0.80) (0.38) Federal Total revenue 9 49 9.62 10.40 7.17 8.43 9.86 11.35 Total expenditures 9.18 9.50 10.12 7.08 9.01 10.40 11.42 Grossinvestment 1.56 1.46 1.so 0.70 0.58 0.59 0.70 Overall fiscal balance 0.14 (0.03) 0.23 (0.22) (1-09) (0.68) (0.24) State andMunicipal Total revenue 4.57 4.69 5.50 3.47 3.58 3.02 3.23 Total expenditures 4.99 5.19 5.68 3.77 3.70 3.15 3.41 Gross investment 0.62 0.73 0.90 0.81 0.62 0.43 0.56 Overall fiscal balance (0.43) (0.52) (0.22) (0.35) (0.14) (0.15) (0.23) Source: Valadares (2004), based on IBGE information. 1.28 It is thus clear that the importance of non-financial public sector corporations in Brazil did not decrease when measured by total revenue or total expenditure as percentage of GDP, but investments by these corporations have declined. This result is somewhat l6 The corresponding gross figure is 23 percentage points, privatization being responsible for a debt reduction equivalent to 4 percentage points o f GDP mthe period. I7These statistics consider the commercial and non-commercial activities operated by these companies in the same way. Ideally, we should consider as public expenditures only the quasi-fiscal activities, or non- commercial activities. Inpractice, however, it is very difficult to put apart all quasi-fiscal operations made by government owned enterprises. 11 surprising, since Brazil has undertaken a massive pnvatization program duringthe 1990s. From 1991 to 2002, revenues from this program, including the transferred debt, reached US$105.6 billion, 67 percent of which accrued at the federal level covering telecommunications (32 percent o f the total revenues), electricity (30 percent), mning(8 percent), oil and gas (7 percent), banlung (6 percent) and others. The decrease in the participation of non-financial public sector corporation in 1998 as shown in Table 1.8 was a direct result of the pnvatization of the telecommunication sector (a decrease of about 3 percent of GDP) and state energy companies (1.7 percent of GDP). 1.29 The continued importance of the non-financial public corporations in these fiscal statistics does not necessarily mean that quasi-fiscal operations through them continue at the same level as in the past. A good share o f these activities i s accounted for by the largely commercial activity in the manufacturing industry, where the group Petrobras i s the most important actor. Total revenues inthis segment achieved 6.3 percent of GDP in2001, up from 3 percent on 1995, and total expenditures from 2.9 percent of GDP in 1995 to 6 percent in2001.I8 1.30 Once we exclude the manufacturing sector on the assumption that most of its activities are commercial, rather than quasi-fiscal, in nature, the total expenditure of the non-financial public corporations has actually declined from 11.3 percent of GDP in 1995 to 8.8 percent of GDP in 2001 at the consolidated government level. At the federal level, the energy sector, where Eletrobras maintains important presence and perhaps trade are the only sectors that retain significant roles for the public enterpnse sector. For the state and mumcipal governments, public utilities (e.g., water supply and sanitation) and energy are the sectors that continue to rely on public enterpnse, though in the latter sector, the pnvatizations in 1997-98 reduced the role of public enterpnses. It therefore seems reasonable to conclude that quasi-fiscal activities by state-owned enterprises have probably decreasedinrecent years. Table 1.8: Expendituresof Non-financial Public Corporationsby Sector and by Level of Government, 1995-2001 1995 1996 1997 1998 1999 2000 2001 Mining. 0.37 0.33 0.04 0.03 0.01 0.01 0.02 Federal 0.36 0.32 0.02 0.03 State & Municipal 0.01 0.01 0.01 0.00 0.01 0.01 0.01 Manufactunng 2.91 3.15 3.35 3.03 4.03 5.14 6.03 Federal 2.91 3.15 3.34 3.02 4.03 5.14 6.03 State & Municipal 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Publicutilities 1.02 1.12 1.30 1.25 1.10 1.08 1.13 Federal 0.02 0.01 0.02 0.02 State & Municipal 1.00 1.11 1.28 1.23 1.09 1.06 1.10 Energy 4.97 4.80 5.04 3.46 4.24 3.43 4.20 Federal 1.77 1.57 1.56 1.48 2.26 1.84 2.51 State & Municipal 3.20 3.23 3.48 1.98 1.99 1.59 1.70 Transportation 0.93 0.90 0.88 0.87 0.90 1.36 0.80 Federal 0.52 0.45 0.42 0.55 0.54 1.09 0.38 State & Municipal 0.41 0.45 0.46 0.33 0.36 0.27 0.42 Communication 2.29 2.78 3.29 0.38 0.38 0.39 0.43 Federal 2.18 2.62 3.09 0.38 0.36 0.37 0.41 State & Municipal 0.12 0.16 0.20 0.00 0.02 0.02 0.02 Trade 1.29 1.21 1.38 1.27 1.47 1.66 1.78 l8 These changes are due to the transformation that took place in the oil sector during the period, with the abolition o f the monopoly o f Petrobras in the oil sector, the definition of a new policy for Petrobris' oil derivatives prices, the increase in the international oil price, the devaluation o f the Real and also the increase in the production o f oil. 12 Federal 1.26 1.17 1.33 1.24 1.43 1.62 1.75 State & Municipal 0.03 0.04 0.04 0.03 0.04 0.03 0.02 Others 0.39 0.40 0.54 0.56 0.60 0.52 0.49 Federal 0.16 0.21 0.33 0.37 0.39 0.35 0.34 State & Municipal 0.23 0.20 0.21 0.20 0.20 0.17 0.15 Source: Valadares (2004), based on IBGE information. Box 1.1 Quasi-fiscal Operations inthe Energy Sector Nowadays there are three funds managed by Eletrobrhs: Global Reserve o f Reversion (Reserva Geral de Revers50 -RGR), Fuel Consumption Account (Conta de Consumo de Combustiveis - CCC) and Energy Development Account (Conta de Desenvolvimento Energktico -DE). Thelr resources come from electricity consumption. None of them have their expenditures annually approved by the National Congress. CCC and CDE are not even considered in the Eletrobrhs budget. The resources from RGR are considered in the Global Expenditures Program (PDG - Programa de DispCndios Globais), a decree issued by the Federal Government, but only in aggregatedterms. The RGR fund i s used to finance specific programs in the electncity sector, according to the legislation. When Eletrobrhs withdraws finds from the RGR account to finance any project, Eletrobrhs assumes a debt with the Treasury for the same amount. This debt i s remuneratedby an interest rate o f 5 percent per year, without any monetary correction. Projects are normally financed with an mterest rate of 5 percent nommal plus a management fee charged for Eletrobrhs. According to the Eletrobrhs Management Report for 2001," the RGR Fund collected revenue for that year was o f R$ 1,030 millions, or 0.09 percent of GDP. Inthe same year the total outlays of RGR resources was of R$ 608.6 millions. The amount of subsidy for 2001 can be calculated by talung the difference between the estimated debt at the end of 2001 and the actual level at the same date. The estimated debt i s calculated as the stock at the end of 2000 adjusted by the Selic rate dunng the year plus the outlays of the Fund.*' The subsidy calculated for 2001 i s of R$ 241 millions (0.02 percent of GDP). Since 2002, part of the RGR resources has been used to subsidize the electncity tariff of low- income population. Inthis case, the resources are not lent to concessionaires but transferred as outlays, and all the expenditure with this program should be considered as subsidy. The Fuel Consumption Account (CCC) resources are collected from the interconnected system's electncity sector concessionaires to finance expenses with diesel for thermoelectnc plants from the isolated system, which i s concentrated in the North region o f Brazil. It is used for cross subsidypolicy. Inthe last few years, the amount ofrevenue collectedvaned from 0.20 percent of GDP in2001 to 0.12 percent o f GDP in2003. The CDE was created in2002, with the purpose of stmulating the development o f some specific sources o f energy, like wind, small power plants, biomass, natural gas and national rmneral coal; promoting a program with the purpose of making electncity supply mversal and subsidizing electncity for low income population. The value to be collected in 2004 i s R$ 1,782.1 millions, or 0.11percent of GDP? There is no Report available after this year. *'Based The data for the debt are available at the annual financial reports o f Eletrobras. on market consensus GDP growth and inflation forecasts. 13 The revenue collectedby the three contributions o f the electricity secto?2 represents0.32 percent of GDP. The amount of subsidy i s not the same, as it depends on how much was spent on financing projects andat whch rates andhow much i s spent as outlays. 1.31 Financial public corporations: In Brazil, a large part of the financial sector i s still controlled by the government. There are large federal banks - BB, BNDES, etc - and some residual financial institutions at the state level. As of December 2004, government-owned banks were responsible for 39 percent of the total lending of the financial system. The total revenue of financial public sector corporation represented 15.5 percent of GDP in 2001 (14.7 percent were revenues on financial intermediation) for all levels of government. O f the total, the federal government financial institutions' total revenue was 14.6 percent o f GDP and the states and municipalities represented 0.9 percent of GDP. At the same time, the total expenditure represented 15.4 percent of GDP (12.6 percent of GDP for financial intermediation revenue) for all levels of government, with the federal level representing 14.5 percent of GDP and the states and municipalities, 0.9 percent. The net lending for all three levels of government was 0.09 percent of GDP (0.13 percent for the federal government financial corporations and -0.03 percent for the state and municipalities financial corporations). Observed volatility of federal-level expenditures i s largely due to the exchange rate-related fluctuations in the Central Bank operations. 1995 1996 1997 1998 1999 2000 2001 Consolidated 19.16 11.62 14.89 17 49 28.11 11.56 15.44 Federal 15.94 8.14 10.23 13.86 25.75 8.59 14.50 State & municipal 3.22 0.68 4.66 3.62 2.36 2.96 0.94 1.32 The extensive presence of public banks d o a not by itself mean that these banks actually engage in significant quasi-fiscal activities, such as providing loans to specific sectors at below-market rates of interest. As these activities have impact on the aggregated demand, it i s a form of public policy and should be reported inthe fiscal results. According to the Demonstrativeof the Financial and Credit Benefits, publishedinthe Annex of the Budget Law of 2004, total implicit credits and financial subsidies by public banks (excluding the Central Bank) were estimated at 0.55 percent of GDP for 2004. This amount, roughly equal to 14percent of the total revenues of these public banks, should be considered de facto public expenditures in estimating total public spending in 1.33 Tax expenditures: Another means for governments to engage in extra-budgetary fiscal activities is through the use of tax expenditures. Since tax expenditures are often identical intheir effects to explicit expenditure programs, or represent a benefit to a specific group of the society, it i s important to have estimates about their volume and distribution, in order to get a more complete view of public policy and public expenditures. Tax expenditures normally include 22 We are summing data o f different years. 23 The methodology used to estimate the implicit subsidies was the same used inthe paper published by the Economic Policy Secretariat (2000). A large part o f these subsidies were calculated as the difference between the net worth estimated if the resources were applied at the interest rate charged by the government inits bonds (the SELIC rate) and the net worth inthe end o f the period, considering the interest rate actually charged in the operations and the expenses o f the fund. This methodology represents the opportunity cost o f the government fund, but does not represent the total subsidy which would be the difference between the interest rates he could have inthe market and the one charged in the operation. 14 exemptions from the tax base, allowances deducted from gross income, tax credits deducted from tax liability, tax rate reductions, and tax deferrals (such as accelerateddepreciation, for instance). 1.34 Usinga broad concept of tax e~penditure:~including social security subsidies,total tax expenditures for the federal government can be estimated at 2.5 percent of GDP. O f this, 1.4 percent of GDP are Table 1.10: Demonstrativeof Financialand Credit related to economc affairs, about Benefits,2004 a third (31 percent or about 0.45 percent of GDP) of which is R$ %of allocated to regional development. millions GDP The other two large groups inthe Direct or explicit subsidies 16,327 0.92 AGF and strategic inventories 153 0.01 Economc Affairs division are F a m n g costs 115 0.01 agriculture (0.37 percent of GDP) Federal Government Lending 3 0.00 and mcro and small companies (EGF) (0.33 percent of GDP). Also 0.33 PRONAF 818 0.05 percent of GDP i s allocated to Price guarantee policy 118 0.01 social protection and 0.24 percent Investment 354 0.02 to health. For state governments, PROEX (Equalization) 1,180 0.07 the available data indicates tax PESA (Rate equalization 2%) 46 0.00 expenditures at around 0.5 percent Housing subsidy 325 0.02 Social interest projects 139 0.01 of GDP Thus, for the public CACAO (Equalization) 8 0.00 sector as a whole, a conservative FAT (Equalization) 28 0.00 estimate of the total amount o f tax Alcohol inventory 39 0.00 expenditures is 2.9 percent of FCVS 13,000 0.73 GDP, Indirect or implicit subsidies 9,686 0.55 FRD 31 0.00 PRESSURESAFFECTINGFUTURE FMM 248 0.01 PUBLICSPENDING FND (18) (0.00) Banco da Terra (Land Bank) 119 0.01 1.35 In spite of spending FAT 2,921 0.17 more than 40 percent of the Regional Funds 2,862 0.16 GDP, the Brazilian public FIES 350 0.02 PROEX (Financing) 648 0.04 sector faces pressures stemming CACAO 72 0.00 from different causes to RECOOP 64 0.00 increase spending (as a FUNCAFE 208 0.01 proportion of the GDP). The Agricultural securities 2,181 0.12 major force causing upward Source: Federal Government'sAnnual Budget Law Annex pressure on spending i s the aging o f population and consequent demands this process implies on pensions and health care. As one can see inthe graphs below, the demographic composition of the Brazilian population is changing fast, and ths i s putting, and will continue to put, strong pressures on the social security system and ultimately change the composition of social spending, towards more healthcare and associatedpersonal services. The populationaged 60 years or more i s expected to grow at an annual average rate of 3 percent in the next 20 years, while total populationi s growing at less than 1.4 percent annually. 24 This concept encompasses the items included in the Tax Benefit Report and in the Tax Expenditures Report (SRF), both publishedby the Ministry o f Finance. The only exception will be the exclusion of the non-taxation of dividends and profit participation as tax expenditure (tax benefits), since such taxation should mean double taxation and could be considered inthe normal tax structure. 15 and the low economc Their simulations show that even with the economy growing at a rate of 3 percent annually, the aging o f population will put pressures on spending: inthe absence of reforms: total spending with pensions and LOAS would rise from the current 7.8 percent o f GDP to level of 9.2 percent of GDP in 2030. This ratio would decline to 7 4 percent if the economy grows at a more vigorous rate of 4 percent per year. 1.37 These simulations also show the strongimpact of real rises inthe minimumwage in the pension system. Due to the links with the mnimumpensions and mnimumbenefits of a variety of social programs, an increase in real terms in the mnimum wage has a strong fiscal impact. Even in the scenario of a more buoyant economy, total spending with pensions would rise to a level over 9 percent of GDP if the mnimumwage were readjusted systematically inreal terms (by 4.5 percent, the averagereal growth observed inthe last 10 years). 1.38 Pressures ansing from aging population are unavoidable and will maintain the level of spending and the deficit of the pensions systemat very highlevels, as well as producingimpact in the health care area. Fromthe point of view of public policies, it therefore seems essentialto take measures to offset these pressures. This would involve a new set of reforms inthe social secunty for pnvate sector workers. It seems clear that a reduction of public spending (as a proportion of GDP) i s only feasible in Brazil in a long run. Ths conclusion reinforces the importance of actions to improve the quality of public spending, by operung space to some recovery in public investments, increasing the effectiveness of social policies, and reducing the level of expenditure rigidity 25 Perhaps it would be more appropriate to say, instead o f low economic growth, poor performance o f the labor market. Contributions for the social security depend on the generation o f formal jobs, and there are factors related to the microeconomics o f the labor market that have been leading to a reduction in the creation o f formal jobs. 17 18 2. PUBLIC SPENDINGAND ECONOMIC GROWTH 2.1 The size and composition o f government expenditures have a sigmficant impact on the economc performance o f a country. It is generally understood that public investments in infrastructure, education and health expenditures that result in physical and human capital accumulation should have more positive effects on growth than current transfers or interest payments because the former enter in the economy's production function while the later do not (Barro and Sala-I-Martin 1995,Fischer and Tumovsky 1995, Canning 1999).26 2.2 While theoretical models predict clear conclusions, empirical studies exammng the relationshp between fiscal variables and growth have not produced robust results. Ina review o f the literature, Bleaney et a1 (1999) have found wide vanations in the results o f empirical studies that examne effects o f fiscal vanables (tax revenue, government consumption, transfers, government investment, etc). This non-robustness reflects differences in terms of data set, econometnc techques and specification used in the estimation o f the impact o f fiscal vanables on 2.3 An alternative to these cross-national comparisons is to conduct an in-depth analysis o f growth effects o f fiscal variables in a single country using time series information, which i s the approach taken in t h s chapter. To explore growth effects o f government fiscal vanables in Brazil, this chapter presents (i) results o f an empincal analysis o f the impacts o f the government expenditures and taxation on economc growth; and (ii) results o f a simulation on expected effects o f shfting the expenditure composition from the current pattern. EVOLUTION THE COMPOSITION,GROWTH, OF AND PERFORMANCEOFGOVERNMENT EXPENDITURESBRAZIL,1950-20002* IN 2.4 The evolution of government expenditures and taxation have had strong influence on Brazil's economic growth, although the effects have been differentiated between two sub- periods since 1950: (i)1950 to 1980, when Brazil expenenced a high-growth period; and (ii) from 1980 up to now, when the economy suffered a slowdown. Throughout the two sub-periods, the government expenditures and taxation have shown an overall increasing trend. Yet in the earlier period, Brazil's growth performance was much stronger than inthe latter period. 2.5 To understand why the evolution o f government expenditures and taxation have affected the growth performance differently, it i s necessary to analyze the context in which the increasing 26 Government expenditures can be differentiated according to whether they are included or not as an input inthe economy's production function. Ifthey are part of the production function they can be classified as productive as they will positively affect the level of output and the growth rate. Public expenditures that are not included in the production function are classified as unproductive. On the revenue side, a similar distinction between distortionary and non-distortionary taxation is used. Taxes that distort the consumption-saving-investmentand labor-leisure decisions have negative effects on efficiency and long- rungrowth while non-distortionary (neutral) taxes that do not affect these decisions have neutral effects on *'For growth. example, in an empirical analysis o f a large sample o f countries, Devarajan et al. (1993) found no statistically significant effect o f those expenditures often considered "productive" (e.g., transport, communication, health, education, or capital spending more broadly), possibly because cross-national data on expenditure allocations do not adequately capture the extent to which these actually lead to accumulation o f human and physical capitals, the variables empirically shown to have positive effects on economic growth. 28This section draws from the backgroundpaper preparedbyBlanco and Herrera (2004). 19 trend unfolded. Care must be taken to interpret the presumed relationshps between public expenditure policies and economc growth in a given penod because other vanables are also at work to influence the country's growth performance. Nonetheless, it i s logical to conclude that, to the extent public expenditures have had any effect on growth, these effects were positive inthe earlier, high-growth period, while in the latter low-growth penod, either the public expenditures have been less productive or other non-expenditure factors' negative effects have outweighed whatever beneficial effects o f the public expenditures. 2.6 There are, in fact, important differences in the expenditure and tax compositions between the two periods. Inthe highgrowth penod, the share o f investment expenditures and subsidies increased in order to foster private accumulation and growth. On the taxation side, during this penod, the quality o f the tax system was improved with the substitution of distortionary cascading indirect taxation by modern value-added tax instruments and the strengthemng o f income taxation that m n i m z e d the negative impact o f taxation on economc efficiency. It i s commonly accepted that the active role o f the government in the economy was responsible for the strong growth process expenenced by the country dunng the penod 1950-80. The increase in investment expenditures (especially by state-owned enterpnses whose investments averaged more than 6 percent o f GDP through the 1970s), the massive utilization o f direct and indirect subsidies to stimulate private sector activities, the rationalization o f the tax system, and the fiscal equilibrium preserved dunng the penod were all conducive to growth (at least in the short run). However, in many cases, the instruments used to foster economc activities such as subsidized credits resulted in the accumulation o f contingent liabilities that worsened the financial situation with negative effects on the macroeconomc situation in the following decades2' 2.7 The expansionist fiscal policy pursued in the high-growth period turned out to be unsustainable. Throughout the 1970s and the 1980s, the persistent fiscal imbalance contributed to macroeconomc detenoration, marked by high and increasing inflation. In this low-growth period, current expenditures, especially government consumptionexpenditures and social secunty and welfare transfers, increased (from 70 percent o f the total non-financial expenditures in 1950 to 90 percent in 2000), while investment expenditures were compressed (from 9 percent in 1950 to below 6 percent in 2000). The increase o f tax revenue was based on the reintroduction and expansion o f distortionary indirect cumulative taxes.30 Dunngt h s penod, the government began implementing a series o f measures to liberalize the domestic economy and privatize important shares o f the public assets to stimulate a more market-oriented economc policy Besides, the political democratization since the rmd 1980s has clearly left an impnnt on the country's revealed preference for greater state roles in social areas and poverty reduction, and less explicit emphasis (than inthe high-growth period) on economc growth strategies as 29 Besides the expenditure and taxation channels, the Brazilian government induced the acceleration o f economic growth through its regulatory policy. Pnce policies, banking credit policies, banlung regulation, tax deductions, fiscal incentives for specific sectors, regional development incentives were tools used intensively by the government to foster growth and capital accumulation 30 On the tax revenue side, it is difficult to delineate a clear pattern. Before 1967, the Brazilian tax system was concentrated inindirect cascading instrumentsthat were drastically reduced with their substitution with less distortionary value-added tax instruments. However, since the early 1980s, the cascading indirect taxation has recovered importance, combined with the expansion of the social security system and progressive improvements inincome taxation on the direct tax side. 3' The growth performance has been erratic and disappointing, partly because the country has yet to overcome the perception o f country nsk that has its roots in its fiscal vulnerability. The market perception i s apparently influenced not only by the high debt-to-GDP ratio and the resultant fear of default, but also by 20 Table 2.1: GovernmentNon-FinancialExpendituresandper capita Growth, 1950-2000 1950s 1960s 1970s 1980s 1990s Governmentnonfinancial expenditure (YOof GDP) 20.1 22.7 22.9 25.1 33.2 Government consumption (% o f GDP) 11.7 11.2 10.0 12.4 18.3 Social Security and Welfare Transfers (% o f GDP) 4.2 6.2 7.3 8.4 12.8 Government investment (% o f GDP) 3.5 4.3 3.2 2.9 1.8 Subsidies (% o f GDP) 0.8 1.o 2.4 1.4 0.4 Per capita GDP growth (YOAnnual Average) 4.0 3.1 5.9 1.0 0.4 Source: Reis et al, 2003. In IBGE: Estatisticas do seculo XX. 2.8 The quality of public expenditures measured in terms o f both composition of expenditures and the efficiency with which these are executed i s logically expected to relate to the country's economc performance. Yet researchershave struggled to obtain robust ernpincal results that link expenditure composition to economc outcomes because of the difficulty of obtaining expenditure data o f good quality and coverage as well as the complexity of specifying the causal relations between particular expenditure items and growth, or poverty and inequality reduction. 2.9 There is no magic formula for a technically optimal composition of public spending. Besides, budgeting i s fundamentally a political process through which countries arrive at their desired allocational patterns, although not all budgetary processes are efficient in generating welfare-enhancing policy decisions. Still from an economc point of view, a key consideration would include whether the composition of public spending conforms to the basic pnnciples of public economcs whch guide decisions about public-private roles in the economy. The public economcs literature suggest that government interventions should address market failures (e.g., provision of public goods, compensation for externalities that result in under-investment in valuable goods) or specific policy objectives such as redistribution or creation of an enabling environment for the private sector to operate.32 2.10 Inthe case of Brazil, the analysespresentedinths report support the common contention that Brazilunder-invests ineconomc infrastructure and that the growth effects of public spending would likely be enhancedby increasingthe amount of public resources allocated to infrastructure investments, a typical expenditure category for creation of an enabling environment for private sector activities. They also show that government consumptions and transfer payments such as social secunty benefits do not leadto economc growth. Although these analyses are not detailed enough interms of their treatment of the expenditure categones, they clearly suggest large room for improvements inBrazil's public spendingcomposition. 2.11 While the growth-enhancing role of public expenditures has effectively taken a back seat, the successive Brazilian admnistrations smce the re-democratization in 1985 have repeatedly prioritized the reduction of inequality and the fight against poverty through public interventions. With the rhetonc o f repayingthe "social debt" the country "owed" to the hstoncally less favored, the high and growing share o f current expenditures, especially the deficitary social security systems, that are believed to limit the government's ability to maintain tight fiscal policy over the medium to long run. See, for example, "Brazil on the Edge" Deutsch Bank Global Market Research, M a y 24, 2002. 32 See, for example, the World Bank's guideline for preparing Public Expenditure Reviews, (http://www1.worldbank,org/publicsector/pe/peanalysis.htm#1) 21 the 1988 Constitution established guaranteed free and universal access to public services as constitutional nghts and createda set of mechanisms to finance social programs. 2.12 Against t h s backdrop of explicit priontization of social expenditures, peremal questions are whether Brazil spends enough inthe social areas to address its needs and whether it does so with high enough efficiency and effectiveness to obtain deslred outcomes. Advocates of larger stateroles inthe social sectors point to the still large m e t social needs as a mainjustification for calling for even higher social spending.33As the chapter 1 showed, however, public expenditures inBrazil surpass 40 percent ofGDP, with social expenditures alone accounting for morethanhalf of the total spending (Le., more than 20 percent of GDP, more than half of which are devoted to social security payments). These figures place Brazil among the highest public spenders among the mddle-income countnes, especially in the social sectors (Comnetti and Ruiz 1996). This may constitute a prima facie ground for questioning the need, and the room, for even larger amounts to be channeled into social spending. 2.13 In terms of efficiency and effectiveness, evidence abounds that a sizable portion of Brazil's social spending is ineffectual in reducing poverty. A World Bank estimate found that only 14 percent of the resourcesdedicated to social programs inthe m d 1990s accrued to the first quintile o f the income distribution (World Bank 2001). A study by PEA estimated that in 2002 the elimnation ofpoverty would theoretically have cost R$ 36.5 billion, or 2.7 percent of GDP, if perfect targeting were possible.34 There is also evidence, both anecdotal and more systematic, that the available resources in the social sectors are not always utilized in the most prudent and efficient manner.3s 2.14 Both the international comparison presented and estimations such as these indicate that the resources dedicated to social areas in Brazil are quite generous relative to the size of the economy and that their overall efficiency and effectiveness tends to be low Thus, the main challenge for Brazil's social policies and expenditures i s not the insufficiency of resources allocated to these areas but their relative inefficiency and inadequate targeting towards the poor. 2.15 In Table 2.2, we divide the expenditures into the following groups to facilitate international comparison: public goods (essential "pure" public goods and services that are non- rival and non-excl~dable~~,such as national defense, environmental protection); merit goods and services (goods that in principle could be made available through markets, but that would be provided in less than the efficient amount by the markets because of market imperfections, such as education, healthcare); social protection (transfers that provide support for income and living standards); and economc affairs (provision of co-funding o f private goods or services by the government) (Atlunson and Noord, 2001). 2.16 Ths comparison with selected OECD and Latin American countries highlight the following features of Brazil's public expenditure pattern: 33 See, for example, de Castro and Sadeck (2003), de Castro et al. (2003) for advocacy for higher social spending inBrazil. 34 For the elimination o f extreme poverty, the volume o f resources necessary would be R$18.3 billion or 1.35 percent o f GDP Paes de Barros et a1(2003). 35 An ongoing World Bank study on public expenditure management in the health sector has documented widespread weaknesses inpublic resource management at the state, municipal, and facility levels. 36Non-rival means that consumption o f the good by one individual does not reduce the amount o f the good available for consumption by others. Non-excludable refers that mdividuals can not be excluded from the consumption o f these goods. 22 . Although the total outlay does not reach the level of European countries where welfare states are far more developed, the data do confirm the widely-held perception that the level of public spending (as a share o f GDP) inBrazil i s quite hgh compared not only to . other Latin American countnes but even to some of the OECD low spenders (Le., Australia, US). O f the total outlays (42.5 percent of GDP) recorded in2002, about a thrd i s allocated to public goods (17.9 percent of GDP), where the item general public service i s the most . important at 11.9 percent of GDP O f this, 7 7 percent of GDP corresponds to interest payments. The allocation for merit good is of 8.7 percent of GDP, or 20 percent of total outlays, which is relatively low compared to the OECD countnes. This i s due mainly to the . generally higher level of health spending inthe OECD countnes due to the hgh costs of complex healthcare systems and the aging population. Social protection accounts for 12.6 percent of GDP, or 29 percent o f total outlays. The OECD countnes spend far more on social protection as a whole, but this i s mainly because they have more advanced welfare programs (i.e., larger shares of social . assistance expenditures) as well as hgher total spending. Many of them also spend less on general public services, reflectingperhaps their lower interest payments than Brazil. In economc affairs (3.3 percent of GDP), Brazil ranks toward the low end of the comparison, though clearly infrastructure gaps are larger in Brazil than in a typical OECD country 37 2.17 Ths pattern of expenditure composition hardly indicates low pnontization o f social expenditures. If anything, the observed distributional pattern of Brazil's public expenditures reinforces the common view that Brazil has lately been under-investing in economc infrastructure. . Giventhat Brazil's social expenditures have reached a level that appears to be near the limt of the country's current capacity to finance them - and the sum total spent on the social sectors is a reasonable amount by international comparison - a key consideration i s the quality of these relativelyhighsocial expenditures. 37 However, agricultural subsidies may be accounting for an important share o f OECD spending on economic affairs. 23 +. 5e, 0 E E E 2.18 Following a similar classification scheme, social expenditures can be examned separately among social protection (social secuntyhsurance and assistance, mcluding labor and employment programs), merit good provision (health and education), and other social public good provision (e.g., housing and urban development, water and samtation). Each group has its own distinct charactenstics, both positive and negative from the point o f view o f impacts on poverty reduction. Social protection (12.6 percent of GDP): Ths is by far the largest category o f social spending in Brazil, domnated by the social secunty benefits that tend to benefit upper income groups. Although some o f the social assistance programs have become increasingly innovative and thus effective in addressing poverty and inequality, these remain small in size, and thus limted intheir overall impact. Merit goods provision (8.2 percent of GDP): Similar to social protection, t h s category o f social spending i s also divided between those expenditures that tend to benefit the well-off in society (e.g., higher education, complex hospital care) and those that are de facto targeted to the poor (e.g., fundamental education, preventive health care). The latter have increasingly been decentralized over the 1990s with the states and the municipalities currently responsible for 80 percent o f education expenditures and for 60 percent o f health expenditures. Since the m d 199Os, some o f these resources have been channeled through innovative formula-based funding (e.g., FUNDEF, PAB), which on average seems to have improved the effectiveness o f these expenditures. 2.19 Public goods provision and other social spending (2.7 percent of GDP): Unlike the expenditures in the first two categories, most o f whose funding are either constitutionally or legally guaranteed (e.g., through revenue earmarlung), this thrd category i s charactenzed by expenditures (e.g., housing, water & sanitation) that are relatively small inthe total quantity, and that tend to suffer from resource unpredictability (partly because many o f these are capital expenditures). Unlike the other two categories o f social spending, this thrd category tends to suffer from the absence of a clear policy framework and explicit prioritization by successive governments. 2.20 Inspite of the indications that the quality of social spending has improved somewhat since the m d 1990s, the overall picture remains that a large portion o f the total social spending i s consumed by relatively regressive programs, most notably the bulk o f the pension benefit programs (11 percent o f GDP), tertiary education (0.5 percent o f GDP at the Federal level only), and certain kmds o f complex hospital care (hospital care absorbs more than half o f the Mimstry o f Health budget, or about 1 percent o f GDP, though these resources benefit the poor as well). Lack o f detailed data on the consolidated government spending prevents us from presenting detailed figures for Brazil's tertiary education and complex health care spending, much o f which i s executed at the sub-national level. But there i s no denying that these programs, along with the social security benefits, account for the largest share o f the country's social expenditures. 2.21 Health expenditure in Brazil has grown consistently, putting severe pressure on already scarce public resources. Health indicators in Brazil are well below the average for mddle-high income countnes, even though per capita health spending i s above the average (US$631 against US$585 o f the group). International expenence shows that the cntena and ways in w h c h the resources are allocated, transferred from central authorities to provider m t s , and how they are used thereafter, significantly affect the impact that financial resources can have on health services and outcomes. 25 2.22 Given the country's federal structure and the decentralized nature of the health management system (SUS), the financial flows are extremely complex. Health plannmg, budgeting, and execution are disjoint. The plamng and budgeting processes in the Unified Health System (SUS) are fairly well structured and formal; however, complexity and bureaucratic formalism limt the effectiveness of planning and budgeting as a management tool. The execution of spending is complicated by numerous laws and norms covering purchasing and contracting processes, the management of human resources, and accounting requirements. In addition, poor managerial capacity in local umts, scarce information, and a msaligned incentive structure are other factors affecting the quality of health service provision under the SUS system. 2.23 Most of the social expenditures are hghly rigid either becausethey are explicitly defined as constitutionalAega1 entitlements (e.g., social secunty, health budget overall), defined by cost drivers beyond the government's immediate control (e.g., hospital care), and are backed by powerful constituencies that have become vested interests to preserve the status quo and the acquired rights, irrespective of the overall welfare effects of these policies for society as a whole.38 Not only do these policies and programs tend to be regressive intheir own distributive impacts, but they constrain the government' s ability to allocate more resourcesto providing other types of expenditures withn the social sector and beyond (e.g., economc infrastructure) that can have positive effects on promoting growth and thus reducing poverty inthe long run. The nature and the effects of the structural ngidity of Brazil's public finance are discussed inChapter 3. Table 2.3: Social Expendituresby Level of Government, 2002 (YOof GDP) Federal State Local TOTAL Social Protection 10.5 1.5 0.6 12.6 Social Security Benefits 9.2 1.3 0.3 10.9 Social Security Benefits -RGPS (private sector workers) 6.6 6.6 Social Security Benefits - RJU (public sector workers) 2.6 1.3 0.3 4.3 Social Assistance 0.7 0.2 0.2 1.o Labor and Employment 0.7 0.04 0.03 0.8 Merit GoodsProvision 2.6 2.6 3.0 8.2 Education 0.8 1.7 1.7 4.2 Health 1.8 0.9 1.3 4.0 Other Social Programs 2.I 0.3 0.3 2.7 Sanitation 0.1 0.2 0.20 0.48 Land Reform and LandPolicies 0.3 0.01 0.00 0.34 Housing 0.02 0.03 0.1 0.12 Tax Deductions, SS exemptions, housing subsidies an1 S system 1.7 1.70 TOTAL 15.2 4.4 3.9 23.5 - Treasury Secretanat for lower goiernment levels' information. EFFECTS GOVERNMENTEXPENDITURES ECONOMICGROWTHINBRAZIL OF ON 2.24 What are the empincal relationshps between the observed patterns o f expenditure composition and Brazil's growth performance? Would rigorous econometnc analysis confirm the causal conclusions one would be tempted to draw from the discussion above (i) that the ever 38 For political economy analyses o f social sector reforms, see Weyland (1996), Melo (2000). 26 increasing size of the Brazilian public sector has dampened long-term growth; and (ii) the that increasing share of government consumptions and transfers and the crowding-out of public investments has exacerbated the negative growth impacts of the growing public sector? The findingsof our analysis support these conclusions - during the period 1950-2000, the growth of the government and the increasing share of current expenditures (consumption and transfers) vis-his the reduction in the share of capital expenditures in the government non-financial expenditures, reduced the growth rate of the Brazilian economy. Box 2.1 Modeling Growth Impacts of FiscalVariables: A Methodological Consideration Numerous empincal studies exarmrung the relationships between government fiscal variables and growth have yielded inconclusive results. Results estlmating the effect of government expenditures and taxes on growth are non-robust and varying widely due to the utilization o f different data sets (periods and countnes), specifications and different econometric techniques. As diverse results from different data sets are natural, the specification of the equations to be estimated andeconometric techniques issues deserve some attention. Part of the non-robustness of the empirical findings cited above reflects ad-hoc ways in which fiscal vanables are added inthe right handside ofthe equation. Inparticular, many specifications presented inseveral studies do not take into consideration the government budget constraint that imposes restnctions on the specifications. The non-inclusion o f either expenditures or revenues in the equation creates a bias inthe coefficients that measure the impact of fiscal variables on growth. Thus, specifications that result inaccurate estimates o fthe impact of expenditures should include the tax side inorder to estimate the net effect of them on growth. The other problem associated with the econometric techniques i s the need to consider the endogeneity in growth equations between the fiscal vanables (nght hand side) and growth (left hand side). If fiscal vanables respond to growth (via Wagner's Law for example), the results obtained from simple static regression methods are non-robust and can lead to misleading conclusions. 2.25 To exarmne the relationship between public and private capital stocks, government consumption social security and assistance transfers, government direct subsidies, tax revenue and the GDP per capita, thls report has adopted a multiple equation co-integratingVAR approach. The model also includes tax revenues to take into consideration the government budget constraint. All the variables are expressedinlevels.39 2.26 The results of the estimation confirm the classic positive effect of the public capital stock on economc growth. An interesting finding i s that the long-run elasticity of per capita GDP with respect to the public capital stock i s larger than of the private capital stock. As the public capital stock results from the accumulation of public investment flows, it i s possible to infer that public investment expenditures have a positive effect on growth. On the other hand, the results show a significantly negative effect of taxation on growth. In the same direction, government consumption expenditures and social security and assistance transfers have a negative effect on growth, while government subsidies presenteda positive but non sigmficant effect on growth. 2.27 It i s interesting to analyze the relationship between all the variables included in the model. For this, the response of each variable to changes in the other variables was estimated through the Generalized ImpulseResponse Function. Beginning with the public capital stock, a 39The stocks o f private and public capital were obtained from the Reis and Morandi (2004) who used the perpetual inventory method. Government expenditures and GDP figures come from the National Accounts System - IBGE. 27 shock that produces an increase of 7 percent of public capital stock in ten years (which can be associatedwith an increase inpublic investment) leadto a 5 percent hgher per capita GDP, while an equivalent shock on pnvate capital leads to an increase of 4 percent in GDP 40 However, Figures 2.1 and 2.2 show that inthe short run, pnvate capital stock has a larger effect than public capital. 40 The elasticity o f GDP to public capital i s about 0.7 which is considered very high. This fact IS also observed inseveral o f the classic studies for the US and OECD economies, such as Aschauer (1989), Ram and Ramsey (1989), Eisner (1994), Sturn and de Haan (1995), and Vijverberg et a1(1997). Hurlin(2001a, 2001b) shows that, in general, results based on time series analysis o f variables in levels, like the present one, tend to find large output elasticities o f public capital. 28 Figure 2.1 Per capita GDP Response to one S.E. shock inPublic Capital Stock - - - - - 0.06 - - 0 . 0 4 / Percapita GDP 0 . 0 2 - - - - - - Figure 2.2 Per capita GDP Response to one S.E. shock inPrivate Capital Stock 0.08 0 . 0 6 0 . 0 4 / PercapitaGDP 0 . 0 2 0.00 f -0.02 0 1 2 3 4 5 6 7 8 9 1 0 -I + T i m e H o r i z o n 2.28 The strong effect of public capital can be explained by the fact that it also affects positively the private capital stock. As the approach used allows for the exammation o f the relationshps between all the variables involved, it shows that increases inthe public capital stock are also associated with increases in pnvate capital and vice-versa, which confirms for Brazil the complementary relation between both types o f capital and reinforces the larger direct effect o f public capital on per capita GDP Figure 2.3 Private Capital Stock Response to one S.E. shock in Public Capital Stock - - 0 . 0 6 0 . 0 4 / P r i v a t e C a p i t a l S t o c k 0 . 0 2 - - - - - 2.29 On the contrary, government consumption as well as transfers for social security and assistance is negatively correlated with per capita GDP. Increases in both government expenditure categories are associated with falls in per capita GDP Government consumption expenditures seem to have been more harmful for growth than transfers for social security and assistance as increases inconsumption expenditures have an immediate and strong negative effect on GDP Social security transfers have negligible effects on GDP in the short run, but become increasingly negative inthe medium term. 29 Figure2.4 Per capita GDPResponseto one S.E. shockinGovernment Consumption - -0 .o 1 - _ _ / LGDPPC -0 .o 2 - - - _ _ _ Figure 2.5 Per capita GDPResponseto one S.E. shockin Social Security Transfers 0 . 0 0 - , - 0 . o 1 l - i i P e r c a p i l l G D P 2.30 The negative impacts o f consumption and transfers for social security and assistance on GDP growth take place through two channels: through the s h f t they impose on public investment, which leads to slower public capital accumulation; and through the need to increase taxation to finance them, which dampens private capital accumulation. An increase in government consumption expenditures o f 5 percent leads to a reduction o f 2 percent in public capital stock in ten years and an increase of tax burden (tax revenue to GDP ratio) o f 0.5 percent inthe same period. The same increase insocial security transfers promotes a fall o f 3 percent in the public capital stock and to an increase o f 0.25 percent o f the tax burdenV4' Figure 2.6 Public CapitalStockResponseto one S.E. shockinGovernment Consumption - - -0 0 1 - - / P u b l i o C a p i l a i S l o s k -0 . o 2 -0 . o 3 - - - - - _ _ 4' I t is worth to note that the effect on taxburdenis small because the tax burden is a ratio that varies from 0 to 1. 30 Figure2.7 Tax BurdenResponseto one S.E. shockinGovernment Consumption 0.008 T 0.006 - - . . . - - 0.004 - - - - - - / T a x B u r d s n 0 .oo - - - - 7 -0.02 / PublicCapitalStock -0.04- - - - - & Figure 2.9 Tax BurdenResponseto one S.E. shock in Social Security Transfers 0.015 0.010 -0.005 0 1 2 3 4 5 6 7 8 9 10 Time Horizon 2.31 The short runeffect of government subsidiesto the privatesector is negligible but in the long run, subsidies contributed to an increase in per capita GDP by 2 percent. The channel for the positive impact of subsidies i s through a positive long-run effect on private capital accumulation. On the other hand, increases in subsidies also lead to increases in the tax burden which have a negative impact on pnvate capital accumulation and economc growth. However, the net effect of subsidies is positive. 31 Figure 2.10 Per capita GDPResponseto one S.E. shockinGovernment Subsidies 0.04 - - - - - 0.02 - - / Percapita GDP Figure 2.11 PrivateCapitalStockResponseto one S.E. shockinGovernment Subsidies / ] / / I ITl l 0 . 0 4 0 . 0 3 0 . 0 2 T 0 . 0 1 0 P w a l e C a p i l i l S l o c k 0 . 0 0 - 0 ,o 1 - 0 . 0 20 1 2 3 4 5 6 7 8 Q 1 0I T i m e H o r i z o n 2.32 A noteworthy finding is the negative impact of increasedtax burden on per capita GDP. An increase in the tax ratio by 1.5 percent of GDP is associated with a lower GDP per capita o f more than 1 percent in the short run and o f 0.8 percent in the long run. This negative effect is explained by the impact o f hgher tax burden on pnvate capital accumulation. An increase 111the tax burden by 1.5 percent o f GDP reduces pnvate capital by 0.3 percent in the short runand 0.1 percent inthe long run. Ths result has important policy implications. The first i s that taxation in Brazil i s costly in terms o f per capita GDP growth. Thus, in assessing growth impacts o f public spending, it i s also imperative to analyze how financing o f such spending can have indirect negative effects on growth. For example, the negative impacts o f government consumption and social security transfers on GDP worsen over time as a result o f the s h f t away from public investments as well as the increase in the tax burden that dampens pnvate capital accumulation. Similarly, an analysis o f the effects o f government subsidies and public capital stock on growth should take into account the negative impact o f taxation (as a financing source of subsidies and government investments) on growth. Figure2.12 Per capita GDPResponseto one S.E. shock inthe Tax Burden -0.006 -0.012 7 -0.014 2 3 4 5 6 7 8 9 10 Time Horizon 32 Figure 2.13 Private Capital Stock Response to one S.E. shock inthe Tax Burden 2.33 In summary, the results of the VAR model estimation suggest that for Brazil, the following long-run effects are observed: 0 A positive effect o f public capital stock on pnvate capital stock (the reverse relationship i s also positive). 0 A positive effect o fpublic andpnvate capital stock on economc growth. 0 A negative effect on growth o f government consumption (short and long run) and social security and assistance transfers (long run). 0 A positive effect onpnvate capital accumulationand growth o fgovernment subsidies. A negative effect o f the tax burdenon capital accumulationand economc growth. 0 A positive effect o f government expenditure variables, which are investment (or public capital accumulation), consumption, social secunty transfers and subsidies, on the tax burden. 2.34 As a net effect of these empirical relationships, we can conclude that the evolution of the size and the composition of the government expenditures during 1950-2000 affected Brazil's economic growth rate negatively. The increasing share o f government consumption and social secunty and assistance transfers and the corresponding decrease in government investments and subsidies constrained the contribution o f government expenditures to growth, while the increase in the tax burden needed to finance higher expenditures, increasingly less productive, had a negative effect on pnvate capital accumulation and growth. These ernpincal findings are consistent with, and thus reinforcedby, the results o fthe simulation exercisereported inthe next section. 33 SIMULATING GROWTH EFFECTS PUBLICEXPENDITURE OF REALLOCATIONS4' 2.35 Thls section analyzes the simulated effects on growth of reallocations of government expenditures. Using a simple overlapping generations model calibrated to Brazilian data and parameters,two rounds of simulation exercises were performedto investigate the growth effect of changes in the composition of government expenditures. The model also captures the effect of government expenditure increases financed by debt that affects interest rates and private capital accumulation. IBox 2.2 Simulating Growth Effects of Fiscal Variables: an Overlapping Generations Model I The model replicates an open economy that consists o f a production side, a government and consumers. The production side i s described by a technology function where public and private capital stock and humancapital are the factors of production and the technology exhibits constant returns to scale in the private factors. The government collects taxes on labor and capital earmngs, and spends them m public infrastructure, education and transfers. The former two government expenditures are considered as productive as their stocks are mputs inthe production function while the latter i s considered unproductive as it i s not part o f the production function. Consumers live for three penods: inthe fxst they accumulate humancapital allocating their time to learning or leisure activities, inthe second they produce consumption goods and allocate their earnings to consumption and savings, in the third they do not work and receive government transfers whlch are integrally consumed. Thus, individuals' preferences define the quantities of consumption, savings andhumancapital accumulation. Finally, the real interest rate i s a function of the internationalinterest rate andthe domestic public debt. The parameters of the model were extracted from different studies that estimated production functions for Brazil. The public finance variables such as tax burden, the ratios o f education and infrastructureexpenditures and transfers to GDP correspond to averages o f the period 1980-2000. Other parameters such as interest rates and discount rates were defined according to the specialized literature. In order to test the robustness of the results, sensitivity analysis was performed usingdifferent values for the parameters. 2.36 As the baseline scenario, the simulation used the current level o f Brazilian public expenditures (education expenditures 4 percent of GDP, 1percent of GDP for infrastructure and 9 percent for transfers) and the average growth rate of the Brazilian per capita GDP for the period 1980-2000 (2.3 percent per year). The first exercise simulates the effect on growth of simple reallocations of public resources from social secunty transfers to education and to public infrastructure. As the exercise assumes simple reallocations, the government budget i s balanced ineachperiod. The secondexercise simulates the effect ongrowth of increasesinsocial security transfers, education expenditures and public infrastructure investments that generate government budget deficit financed by debt. 2.37 Reallocating expenditureswithin a balanced-budgetconstraint: The results show that given Brazil's current expenditure composition, reallocating public expenditures from transfers to education or infrastructure expenditures has positive growth effects. Table 2.3 shows that reallocating expenditure from social security transfers to education by 1 and 2 percent of GDP would increase the growth rate from the observed 2.3 percent to 2.4 and 2.5 percent, respectively Reallocation from social secunty transfers to public infrastructure has a sizeable 42This section draws from the backgroundpaper preparedby GlommandRloja (2004). 34 growth effect, with the rate going from the current 2.3 percent to 2.9 percent and 3.3 percent.43 The greater effect o f the reallocation to infrastructure investment compared with the reallocation to education depends on the imtial expenditure composition. As education expenditures represent 4 percent o f GDP while infrastructure investment i s only 1 percent o f GDP, a reallocation to infrastructure i s found to have a greater marginal impact. Table 2.3: SimulatedGrowthEffectsof ExpenditureReallocations I GDP growthrate with reallocations from Social Security I 2.38 On the contrary, allocating more public funds to transfers and reducing the amounts directed to public education or public infrastructureimposes substantial costs in terms of lower growth. Inthe first case, from education to transfers, the growth rate would decrease from 2.3 percent to 2.2 percent while the reallocation from infrastructure investments to transfers reduces GDP growth to 1.7percent44 2.39 Increasing expendituresthrough debtfinancing: The second simulation exercise shows the effect o f increasing debt financing o f expenditures. Allowing the interest rate to be an increasing function o f the government debt to GDP ratio, the negative effects o f increases in transfers on growth are greater while the positive effects o f education and infrastructure expenditures on growth are smaller. 2.40 Table 2.7 shows the growth effects o f raising debt by 1 percent o f GDP and using the proceeds to finance one o f the three expenditure items: transfers, education or infrastructure investment. The results show that financing additional transfers through public debt is harmfulto growth. This is especially true inthe long runwhen the growth rate would achieve only 1.8 percent and 0.7 percent. The channel by which the increase in transfers financed with indebtedness affects growth i s that increasing the debt to GDP ratio raises the interest rate, which causes the reduction in physical capital accumulation. Ths decline in physical capital has also negative effects on human capital accumulation, w h c h in turn deepens the negative impact on growth. 2.41 Financing extra infrastructure investment or extra public education has modest positive growth effects. These growth effects decline over time as higher interest charges accrue. Initially, increasing public education spending increases human capital accumulation, which in turn increases the marginal product o f physical capital and hence the demand for ~~~ ~ ~ 43 This result implies that also a reallocation of government expenditures from education to infrastructure investments would increase the growth rate from 2.3 percent to 2.6 percent. The opposite reallocation would reduce growth rate to 1.7 percent. 44 Note that the model does not consider a possible relationship between transfers (which are directed only to retired workers) and human capital accumulation. An interesting extension o f the model would be the inclusion o f conditional transfers to stimulate human capital accumulation. Inthis case, these transfers may have positive effects on growth. 35 physical capital. The result i s a positive growth effect, which i s superseded by a negative interest rate effect after a while (the thrd generation in the simulation). In the case of public infrastructure investments, the growth effects are more positive because the effects o f more public infrastructure on the productivity o f physical capital are stronger than the effect o f more education on the productivity o f physical capital, while the negative effect o f debt accumulation i s identical inboth cases. Table 2.4: Simulated Growth Effects of 1percent of GDP IncreaseinExpenditures through debt financing Generations Current Scenario Social Security Education Public (Average per capita GDP Transfers Infrastructure growth rate 80-99) 1 2.3 2.3 2.4 3.3 2 2.3 2.3 2.4 3.0 3 2.3 1.8 1.9 2.7 4 2.3 0.7 1.1 2.5 2.42 In summary, the results indicate that reallocating public expenditures from transfers to education and infrastructure expenditures would increase the growth rate. Especially given the current low level o f public infrastructure investments, a reallocation from social secunty transfers to increase spending on economc infrastructure would likely have sizeable growth effects. On the other hand, rumng a public debt to finance transfers decreases growth substantially, and runninga public debt to finance education expenditures would increase growth inthe short runbut debt accumulation would dampen growth inthe long run. SUMMARY 2.43 The analyses presented in this chapter support the contention that Brazil may benefit from increased public spending in the area of economic infrastructure. Both the econometm analysis using historical data from Brazil and the simulation exercises with data calibrated for Brazil indicate positive and strong growth effects o f public physical capital stock and public investments. Even a cursory review o f the available data shows that Brazil i s likely to be under-spending on key infrastructure items such as road maintenance, in spite o f its expected hgh economc returns, and that its budget management system inordinately burdens program managers in the infrastructure sectors by failing to provide predictable flow o f funding. It i s therefore likely that growth contributions o f expenditure policies would include both gradual increase in public investments and improvements in the efficiency in budget execution. The Federal Government has imtiated a pilot effort to increase efficiency with which pnority public investment projects are executed. Ths i s a welcome development that should be pursued as a step toward developing a more effective system for managing the government's entire public investment portfolio. 2.44 Both the empirical analysis and the simulations point out clearly the negative effects of increasing taxation on economic growth. Thus, it is not advisable for Brazil to pursue its need for mcreasmg public mvestments ma expansionary fiscal policy that results m a hgher tax burden than the current level. Instead, a long-run solution to recovemg an adequate level o f public mvestments mustbe sought mreallocationofpublic spending w i t hthe fixedoverall fiscal envelope. l h s means the need to re-examme the composition o f the current expenditures, mcluding those allocated to the social sectors that today consume a lion's share o f Brazil's public expenditures. 36 3. SOURCES AND EFFECTS OF EXPENDITURERIGIDITY 3.1 The findings of the previous two chapters - that the level and the composition of public expenditures inBrazil are not conduciveto sustained growth - suggest a need for both controlling the level of total public spending and adjusting its composition. Ths i s necessarily a medium- term reform agenda that, even under the best of circumstances, would take several years to accomplish. Inthe case of Brazil, however, the highdegree of budget ngidity pose considerable obstacles to impede governments, no matter how well-intentioned they may be, trying to gain better control over their expenditure level and composition. 3.2 The considerable level of budget rigidity in Brazil is due to the extensive use of revenue earmarking as well as existence of large non-discretionary expenditures. In 2003, non-discretionary expenditures represented 89 percent of total non-financial federal expenditures while earmarked revenues represented 80 percent of total revenues. Rajkumar (2004) calculated that the "hlly free" portion of the executed budget in 2002 was R$ 25.5 billion, or just 8.1 percent of non-financial expenditures, In a different calculation, using the approved budget (budget law), rather than the executedbudget, Silva (2003) found that in2002, only 4.5 percent of the non-financial revenues were considered of free allocation in the sense that they were not earmarked and were not allocated to mandatory expenditures. 3.3 Comparing the degree of budget rigidity across countnes i s not a straightforward exercise. There is no commonly accepted methodology, and the sources of rigidity, or what constitutes rigidity, are illdefined in the literatures. In spite of the methodological pitfalls, however, available information suggests (i)that budget ngidity i s high among mddle-income Latin American countnes with a high degree of fiscal decentralization; and (ii)that even in comparison to these countries the level of budget ngidity inBrazil i s on the hgh end. Table 3.1: Rigid Portion of Total Non-Financial Central Government Expenditure4' ~~Country and Year Rigid Portion Data Source Argentina (2003) 85% World Bank (2003) Colombia (2000) 84% Lozano (2000) Chile (2003) 76% MinistryofFinance, Chile (2003) Philippines (1994) 57% World Bank (1995) United States (2000) 64% Office o f Management and Budget (2000) Brazil 89% Rajkumar (2004) 3.4 The level of budget rigidity increased considerably in the last twenty years. For instance, earmarked revenues represented 30 percent of revenues in 1984 compared to 80 percent in2003.46 The same patternholds for mandatory expenditures. There was an increase insocial 45Inall cases except for the Philippines, the figures are based on Budget Law numbers. Inmost cases, the original source document included interest and debt payments in the calculation; the latter were then excluded from the calculation (Le,, from the numerator as well as denommator) to arrive at the figures in the table. The types o f items included m the definition o f "rigid portion" vary depending on the country and data source. Generally included are non-discretionary spending items such as wages and pensions, as well as certain easily definable categories o f earmarked revenues, such as earmarked transfers to sub- national governments. However, not all rigid items or items funded from earmarked revenues are necessarily included, and thus the figures are likely to underestimate the true portion o f the budget that i s rigid, inmost cases. 46Total revenues without privatization revenues and credit operations. 37 security e x ~ e n ~ ~ofi 5.8~ptrccr11sof GDP from 1'388 to ~ 0 0 3(4.4 perceiit relative to private ~ ~ r ~ sector Rorkers {lNSS) and t 4 percent for public sector ~ ~ ~ p ~ Also, eother~~ ~a n d a ~ o ~ o ~ e . expend~t~res were created lazes, such as the ~ ~ 1floor1of ~~ ~~s ~~ae ~~~l ~d t~~~~ 1 .~~ ~ ~ Figurc3.1 icipation of e as a share of federal 80 t 38 Municbios). Until then, these transfers constituted 14 percent and 16 percent, respectively (Constitutional Amendment n. 23, 1983)48 Also, 3 percent of the two taxes above were earmarked to regional funds (FCO, FNE and FNO) and an additional revenue sharing of 10 percent of PIand 50 percent of rural property tax (ITR -Imposto TerritorialRural), 100 percent of the financial operations tax with gold transactions (I@- were established. ouro) 3.8 Other mechamsms of revenue sharing regulated or created later include: 0 The distribution of two thuds of salirzo-educaqa'o, a contribution charged over payroll (2.5 percent) to finance education, established inthe Constitution and regulated by Law No. 9424, 1996. 0 CIDE (Contribuiqa'o de Intervenqa'o no Domi'nzo EconoImico, an economic contribution charged over commercialization and importation of petroleum and its derivatives, natural gas and its derivatives and alcohol created by the Constitutional Amendment No. 33, 2001, 29 percent of whch must be transferred to state governments according to the Constitutional Amendment No. 44 (2004); and 0 Financial compensations, or transfers of a substantial amount of revenues collected from water and mneral resource explorationby states to mumcipalities (between 80 percent and 90 percent of the revenues collected, representing 0.5 percent of GDP in2003), which increased inrecent years, mainly withthe riseofthe oilprices andproduction. 3.9 As Figure 3.2 shows most of the transfers to states and municipalities that are constitutionally/legally sanctioned. Voluntary transfers were almost nil. Also, transfers to lower levels of government represented22 percent of the total earmarked revenues in2002. Figure 3.2 Composition of Transfers to States and Municipalities 4.5 4 3.5 2B0 3 2.5 $ 2 1.5 1 0.5 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 0 FPE/FPM/IPI-EE 0 Regionalfunds W Education-Salary W Financial compensations W Others 0 CIDE 1 Source: Ministryof Planning and Budget. 48For a deeper discussionof revenue sharing history, see Varsano (1996). 39 3.10 Intergovernmental transfers are important sources of revenues for states and municipalities. In 2000, 21.6 percent of state primary revenues and 65.8 percent o f municipal revenues came from intergovernmental transfer^.^' There is a discussion being held now to increase the transfers for municipalities, increasing the percentage of earmarlung to municipal government (FPM) from 22.5 percent to 23.5 percent of the income tax (IR) and the industnalized product tax (IPI). These discussions are inevitably influenced as much by political consideration, reflecting the balance of power and tug of war for resources between the three levels of government inthe Brazilian federation. 3.11 Education: Anther important earmarked mechanism is the allocation of 18 percent of total tax after transfers to education expenditures. In addition, one thrd of the wage education contribution, charged over the payroll i s earmarked to national development educational fund - FNDE (Fundo Nacional de Desenvolvimento du EducuqEo). Earmarked expenditures on education represented3 percent of primary earmarked revenues. 3.12 Social and economic contributions: The 1988 Constitution created the Social Secunty Budget to finance expenditures with social security, social assistance and health. Ths budget would be financed by the following contributions: social secunty contributions paid by employees and employers; social contributions over the revenues (COFINS) and profits (CSLL) and revenues from lotteries. While some of these (e.g., the social security contribution and COFINS under the name FINSOCIAL) have existed for some time, there has recently been a substantial increase in the revenues collected through these contributions. Today, COFINS and social secunty contributions are the most important sources of federal government revenues, and represent 3.8 percent and 5.2 percent of GDP in2003, respectively 50 3.13 The recent trend for a nsing share of social contributions is not only due to increases in amounts collected through existing contributions. To close the financing gap of the ever increasing social secunty expenditures, the government created a new contribution, the CPMF - Provisory Contribution on Financial Transactions - first as a tax in 1993 and then as a contribution with its revenues earmarked first to health (1996), later including social secunty (1999), and then to the Poverty Eradication Fund created in 2000. Today, of the rate of 0.38 percent, 0.2 percent is earmarked to health, 0.1 percent to social contribution and 0.08 percent to the Poverty Eradication Fund. Contributing around 1.5 percent of GDP in additional revenues, CPMF has become a very important source of revenue for the federal government. 3.14 Earmarlungto social protection and health represented 61 percent of total earmarlung on average in2002 and 2003. From this, COFINS, CSLL, CPMF, plus social secunty contributions represented a lion's share (92 percent of total). Eighty-three percent of the total earmarked revenues go to social protection, health, and transfers to states and municipalities. The other main destinations o f earmarked revenues are education (3 percent), economc affairs (6 percent), where the most important item is the contribution CIDE discussed below, and others (7 percent). Ths last category includes the PISRASEP, earmarked to FAT (Fundo de Amparo ao Trabalhador) (60 percent), which finances unemployment insurance program and programs that finance the economc development at BNDES (Banco Nacional de Desenvolvimento Econ6mico e Social) (40 percent). 49For municipal governments, state transfers are an important part o f total intergovernmental transfers. 50 The only tax responsible for more revenues alone is Income Tax. But from this tax, 47 percent is transferred to other levels o f government and to regional development funds. 40 3.15 The most important contribution earmarked to econormc affairs i s CIDE, charged over petroleum and its denvatives, natural gas and its derivatives and alcohol. It was created in2OOI5' to substitute the PPE (Purcela de P r e p Especzpco) charged over oil denvatives. But differently from the PPE, CIDE has its revenues earmarked to the transportation infra-structure expenditures, subsidies for alcohol, natural gas and its denvatives and oil denvatives. Onginally, CIDE, likeall the other contributions, was not subject to revenue shanng with sub-national governments. However, the Constitutional Amendment No. 44, 2004, passedas part o f the tax reformpackage, stipulated that 29 percent of its revenues shouldbetransferred to state governments. Figure3.3 Earmarkedrevenuesbyfunction (2002-2003) r: Economic Protection and Health 61% Source: Federal Government General Balance 2002 and 2003 3.16 Since 1986, the contributionrevenues as a whole have increased by 7.8 percent of GDP with the greatest increase accounted for by COFINS (3.1 percent of GDP), followed by CPMF (1.5 percent of GDP), PISPASEP and CSLL (1.1 percent of GDP each), CIDE (0.5 percent of GDP) and social security contributions (0.2 percent o f GDP). In most of the items (e.g., CSLL, CIDE), the increase is mainly due to the creation of new contributions. But also there has beenrate increases (e.g., CPMF, COFINS). 3.17 Two important events explained the increase of COFINS/FINSOCIAL since around 1993. First, even though COFINS substituted FINSOCIAL with a hgher rate in 1991, its actual implementationhad to wait for ajudicial decision to confirm COFINS' constitutionality in 1993. As a result, in 1995, COFINS collection was 1.3 percent of GDP higher than in 1992. Later, in 1999, the government increased the rate from 2 percent to 3 percent and expanded its base by levying it on financial institutions. COFINS collection increased by 1.6 percent of GDP from 1998 to 2000. Constitutional Amendment No. 33,2001 andLaw No. 10,336,2001. 41 3.18 Overall, two factors explain the increase of contribution revenues: the need for more resources to finance the social security budget and the fact contributions are not shared with sub-national governments. The increase in contribution collection has been crucial for the fiscal adjustments since 1999.52 Figure 3.4 Evolution of the M a i n Contributions 14.00% 12.00% 10.00% 8.00% 6.00% 4.00% 2.00% 0.00% 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 Cide Cofins CSLL CPMF PIS/PASEP o CPSS Pnwte workers SS Contributions Source: Ministry of Planningand Budget. 3.19 Apart from the increased revenue collection that has aided the government's fiscal adjustment efforts, economic consequences o f the increased share o f contribution revenues have not been entirely neutral. First a larger share o f contribution revenues with cumulative incidence on economc activities means worsemng o f the quality o f tax policy COFINS and PISPASEP are charged on invoicing and CPMF i s charged on financial transaction^.^^ These three contributions represented 6.6 percent o f GDP The other effect i s a more rigidbudget, as they are earmarked to specific expenditure^.^^ 3.20 Others: In addition to the examples above, there is a substantial number o f other earmarlung in the federal budget. One example i s the revenues collected directly by different agencies and entities which are earmarked to expenditures o f the collector agency or entity. On average, in2002 and 2003, own revenues represented 2.8 percent o f total pnmary expenditures. 3.21 FSE/FEF/DRU: With the aim o f decreasing the level o f earmarlung o f the federal budget, the government created a mechanism, which "de-earmarked" 20 percent o f the earmarked 52 The increase on contribution as a way o f enhance revenue collection avoiding sharing resources with lower levels o f government may have today more political opposition. The new transfer of CIDE resources to states may be an example. 53Recent legislation reduced the cumulative effect o f contributions COFINS and PIS. j4As a substantial part of their revenues is allocated to finance mandatory expenditures, like social security and the health floor, the net effect on budget rigidity may be lower than at first sight, at least in the short run. 42 taxes and contribution^.^^ Inthe first years of implementation, the fund had two implications: the increase in the federal government resources, with the decrease ~fl transfers to states and municipalities, and the mcreased flexibility o f the federal budget. When the FEFwas replacedby the current DRU, the new regulation excluded intergovernmental transfers (or general revenue shanng) from its appli~ation.~~ 3.22 In2002 and in2003, the total amount allocated to DRUwas R$24.7 billions and R$ 27.6 billions, respectively (SOF, 2003). It should be noted, however, that not everything represents an increase in budget flexibility, because part of the revenues de-earmarked was used to finance ngid expenditures. In some cases, the de-earmarked revenue was reallocated to finance the same expenditure to which it was onginally linked, as the case o f the social secunty budget, which has sources of financing earmarked, but has a funding deficit. In this case, de- earmarlung has a smaller net impact on rigidity than its nomnal amount implies, since most of the de-earmarked resourceswill be necessary to finance the onginal mandatory outlays. 3.23 After an initial increase in the share of the de-earmarked revenue in the total revenue with the introduction of the FSE/FEF, there was again a decline in its share in the total revenues. In 1995, after the introduction of FSE, The share o f the total de-earmarked revenues, including FSE/FEF, in the total revenues increased from 22.9 percent in 1993 to 30.5 percent in 1995, but declined again to 20 percent in 2003. This i s because o f the subsequent exclusion of intergovernmental transfers, as mentioned above, and the increase in the shares of other earmarked revenues, particularly social and economc contributions. NON-DISCRETIONARYEXPENDITURES 3-24 Another source of budget rigidity is the non-discretionary expenditures." The definition of the set of non-discretionary expenditures i s not obvious since some expenditures may be non-discretionary in the short run but closer to discretionary in the medium or long run. Here, we use the same classification used by the Ministry of Plamng, Budget and Management (MPOG) in its statistics, with the addition of the Mimstry of Health expenditures. The MPOG definition is a budgetary one that i s intended to measure the expenditures that the government cannot avoid in the short run, but i s slightly different from those annually established by the Budgetary Directives Law (Lei de Diretrzzes Orqamentdrias-LDO).'* "FirstcreatedunderthenameofFSE(FundodeSocialdeErnerg&cra), in1994,thismechanismhasgone through several name changes to FEF (Fundo de Estabilizaqdo Fiscal) in 1996 and DRU (Desvmculaqdo de Receitas da Unido) in2000. In2003, DRUwas extended to 2007 56 For municipalities, intergovernmental transfers represented 65.8 percent of total pnmary revenues in 2000. For state governments this number was equal to 21.6 percent. (Regionalizaqdo das Transaqces do Setor Pliblico - 2000-IBGE) 57 Intergovernmental transfers and minimum application o f resources on education are not considered in this section. Although being obligatory, as they are caused by earmarked revenues mechanism, they were discussed above. This difference can be very subtle, but this division was made to make possible an attempt of evaluating both components separately. The Budgetary Directives L a w (LDO) each year identifies a list o f "obligatory" expenditures that are protected from the "contingenclamento" decrees, as determined by the Fiscal Responsibility Law, with the aim o f guaranteeing achievements o f the primary balance target established annually in the LDO. This list includes expenditures that are clearly mandatory, like pension benefits, but others that are not clearly so, like some programs o f the federal government. The concept used in the LDO used to be consistent with those used by MPOG in the past, but the LDO concept does not follow a consistent definition and changes every year and is less strict than the MPOG concept used here. 43 3.25 Figure 3.5 below shows the main non-discretionary expenditures in 2003, excluding transfers to states and municipalities. The most important items are social security outlays, which are responsible for 53 percent o f total pnmary expenditures in 2003, divided between the social security benefits for pnvate sector employees (40 percent) and for federal government employees (13 percent). At the same time, personnel outlays represented 16 percent o f the total. Health expenditures, w h c h have to hlfilla floor according to the Constitution, represented 9 percent. The others are unemployment insurance (3 percent), LOAS / lifetime income program (2 percent) and others (5 percent).59 Other non-discretionary expenditures represented only 6 percent o f total non-discretionary expendituresV6' These are the items whose classifications as mandatory are more open to question. Without them, total non-discretionary expenditure would represent 84 percent o f total pnmary outlays, excluding transfers to other levels o f governments, a still very highnumber. Figure3.5 MandatoryExpenditures 2004 - Others3% mandatory mandatory 14% Health 9% Social Securityfor Private Sector Unemployment Employees40% Insurance 3% LOASl Lifetime Income Program 2% Personnel 15% Social Security for public Sector Employees13% Source: Ministry of Planningand Budget 3.26 Not only is the level of non-discretionary expenditures very high today, but also increased considerablyinrecent years. The social secunty expenditures for pnvate and federal government employees increased by 5.8 percent o f GDP from 1988 to 2003 due to the generous rules for benefit concession and increase inthe minimumwage. Another important cause was the definition o f the mandatory mnimumexpenditure levels for health, w h c h represented 1.9 percent 59Inthe case of LOAS, the guarantee of a minimumwage to the elderly and the handicapped is also inthe constitution. However, its regulation is done by infra-constitutional legislation: The defimtion of the minimum age, for instance, is done by ordinary law. Finally, the unemployment insurance is also guaranteed by the Constitution, but all of its rules are regulated by ordinary law, which implies more flexibility inthe short run. 6o 96 percent o f them are represented by subsidies (40 percent), Complementary Law No. 87/96 (29 percent), payment to settled juridical cases (7 percent), FUNDEF complementation (5 percent), FGTS (15 percent). 44 o f GDP in 2003 (excluding personnel expenditures in the health sector). For the future, some o f these items still require special attention. The process o f population aging have a direct impact on social secunty and assistance as well as health spending, responsible today for 51percent o f total expenditures. 3.27 Public sector personnel expenditures: Since the late 1980s, there has been an increase inthe share o f social security pensioners inthe total personnel expenditures from 26.8 percent on 1987 to 46 percent on 2003. From 1987 to 2003, expenditures with employees increased 74 percent at constant pnces, while expenditures with retirees andpensioners rose 300 percent. Figure 3.6 Federal Government Personnel Expenditures (IGP DI deflator) - 1 -Employees -Retirees and pensioners 1 Source:SRHI Ministryof Planning and Budget 3.28 Both expenditures also increased as percentage o f GDP from 1988 to 2003. Wages and salaries increased by 0.2 percent o f GDP (from 2.5 percent to 2.7 percent) and outlays for retirement and survivor benefits rose by 1.4 percent o f GDP (from 0.9 percent to 2.3 percent). Ths shows us that the social secunty expenditures were the main cause for the nse inpersonnel outlays. At the same time, the number o f federal government employees decreased dunng the period analyzed. 45 700 J - - . - I 600 iii500 "a 400 g8 801 300 200 I I I 2004 55 28% Retirees and pensm El Employees 2987 73 25% 46 7.0% 1.5% S.0% 7 " 0% 5.0% 0.5% 0.0% 4.0% -0.5% 3.0% -1.O% 2.05i0 -1'5% IQ%5 -2,Qofo 0.0% -2,5% Figure3.10 Kc81XIinimum\1C'age (*) Source Central Bank C) INPC deflator and the lowest 1.8 percentI1119 2,513 1995 1998 1997 1998 1999 2000 2001 2002 2003 Figure3.13 350,000 998 7999 2000 2001 x x x x 2002 2003 f tl Heaitl-1~ W I ~ personneff ~ ~ # t 13 Psrsonnei ce Sofp 1995 1996 1997 199% 1999 21300 2001 2002 2003 2004 2005 Source.5OF t Mnrstry of Ranntngand Budget i s that even if mandatory expenditures were somehow reduced, revenue earmarlung would not allow the government to re-allocate revenues to other types of expenditures.68 3.42 The most important among the earmarked revenues that finance non-discretionary expenditures are those that comprise the so-called social security budget that finances social security (pensions), social assistance and health expenditures with earmarked revenues (e.g., social secunty contributions, COFINS, et^.).^' When the social secunty budget is in deficit, the revenue earmarlung to these expenditures does not worsen the budget rigidity According to the budget law of 2002,70the deficit of the social secunty budget was of R$17.2 billions, or 11.5 percent of its revenues. As a result,,the effect of the DRU on the reduction of earmarlung has been partially offset by this deficit. Of the R$ 20.3 billions de-earmarked by DRU from social contributions, the portion that was onginally earmarked to the social security budget (amounting to R$ 17.2) was simply allocated back to cover the financing gap inthe social secunty budgets7' Inthe end, the net effect of de-earmarlung was only R$3.1 billion, or 7.5 percent of the total discretionary budget inthat year (SOF, 2003), and 11percent of the pnmary surplus predictedfor that year (1 percent of total revenues and expenditures). 3.43 A good example of new earmarked revenue simply substituting other revenues was the creation, in 1996, of the CPMF, a contribution charged over financial transactions, earmarked to health.72 The CPMF revenue directed to the Health Mimstry was equal to 31 percent of the average of the mnistry expenditures in 1995-1996 (the two years just before its implementation). But inthe three subsequentyears (1997 - 1999), average total expenditures of the health mmstry were only 6 percent higherthan the 1995-1996 average (Table 3.2). Inthis sense, only 20 percent o f the new source of financing actually meant an increase in expenditure (and thus increased rigidity). The rest simply substituted the existing sources of revenues (e.g., COFINS and CSLL), dueto the fungibility effect (Reis et al., 2001).73 3.44 A flip side of the phenomenon discussed above is that when the earmarked revenues are insufficient to cover the non-discretionary expenditures that they are supposed to finance, earmarlung itself can become an ineffective means to "protect" funding of at least a portion of these non-discretionary expenditures. Ultimately, however, if the government i s comrmtted to ensunng a certain level of spending in a particular area, it could do so without either revenue earmarlung or constitutional or legal designation of mandatory expenditures. A case in point appears to be, once again, the pattern observed inthe Health Ministry budget in 1998-99, before the Constitutional Amendment No. 29 established the spending "floor." In these years, there were significant shortfalls inthe collection of some of the social contributions. The government In2002, 60.8 percentofthe earmarked revenues, composed of social security contributions over payroll, CPMF earmarked to social security and health, COFINS and CSLL financing social expenditures and PISPASEP contributions financing formal-sector worker benefits, financed only non-discretionary expenditures (Rajkumar 2004). These are not the complete intersection, as there are earmarked revenues that finances discretionary as well non discretionary expenditures as education expenditures. 69 In Brazil, the social security budget is considered separate from the "fiscal" budget based on the ordinary tax revenues. 70 Law n. 10.407, o f 2002. 71 According to Federal Constitution, social security budget includes social security, assistance and health ''expenditures.revenues Later, its where earmarked to social security and to the hungry eradication fund also. 73 Of course, we could not infer that the increase in expenditures only happened due to the new source of financing, but this new source did not result in higher expenditures by the same amount o f resources collected. 52 compensated for the reduced revenues from these sources by increasing the allocation from the ordinary tax revenues, and thus maintainedthe spendinglevel (Table 3.2). Table 3.2: Ministryof Health:ExpenditureDistributionaccordingto BudgetSource Financing sources 1993 1994 1995 1996 1997 1998 1999 Ordinary taxes 1376.3 156.1 703.2 38.3 237.3 2285.3 3201.4 NationalTreasury 2319.8 251.9 587.3 648.2 648.2 98.3 165.9 Secretariat (STN) Domestic credit operations. 401.5 - 1704.0 1598.1 -Currency External credit operations- 71.2 187.3 237.6 165.4 115.9 223.6 315.7 Currency Resources directly 155.0 333.4 542.6 484.3 560.7 548.3 753.3 collected Social Contributions 10377.3 8511.1 15465.6 12268.2 16984.9 15043.3 13621.3 Social Contrib. on Firms 1644.2 2013.7 4503.3 4022.2 4491.9 1690.3 2903.2 Profits Social Contrib. for the SS 3868.5 5316.7 10962.3 8246.1 6025.7 5501.3 5797.9 System Employees Contrib. for SS 4864.7 Financial Transactions Tax - 6467.3 7851.8 4920.1 CPMF Social Contrib. for 1180.7 financing SS and Assistance Systems Fiscal Stabilization Fund 5745.8 2665.2 3422.9 4530.9 2814.0 3085.3 Others 684.2 413.2 393.3 725.4 155.3 156.7 648.2 Total 15,385.4 15,598.7 22,298.9 19,350.8 23,233.1 21,169.5 21,791,l Source: Castro et a1 (2003) 3.45 Does earmarking increase or protect spending? An important policy question is, even when earmarlung revenues are financing non-discretionary expenditures, if the existence of sources of resources guaranteed somehow contributes to the high elevation of these expenditures. Ths is a difficult question to answer because it is not straightforward to establish either an adequate (theoretically optimal) amount of spending or an amount that would have been maintained in the absence of earmarlung. Anecdotally, there appear to be instances, in which relatively well-endowed municipalities find it unnecessary to spend 18 percent of their revenues on education as requiredby the Constitution, and "scramble" to find ways to spend the resources inthe sector simply to fulfill the formal requirement. We do not know, however, whether these municipalities are actually tending toward under-investing in education (Le., the theoretically optimal amount is at least as much as 18 percent o f the revenues), nor do we know how widespread such a situation is across the country 3.46 For the group of earmarked revenues that do not finance mandatory expenditures, an important question i s whether all the resources are being applied as stipulated inthe law, whch should have the effect of keeping the specific expenditure at a certain level corresponding to the amounts of the earmarked revenuescollected. CIDE, earmarked to non-personnel expenditures in the transport sector, offers an example of the lirmted effect o f earmarlung on expenditure "protection," a phenomenon widely recognized in the 1iterat~u-e.'~Contrary to the proponents' expectation, the introduction of CIDE did not result in an increase in budget allocation for 14 53 transport investments, not because the level of investment was already above the CIDE collection, as inthe case of CPMF to finance health expenditures (Figure 3.15). Infact, the level of transport investment decreased in 2002 (by 37 percent compared to 2001) and 2003 (by 57 percent compared to 2001). This indicates that, at least in the short run, whether funding i s protected depends more on types of expenditures (e.g., current vs. capital) and the government's explicit or implicit priorities, than on whether the sector has "guaranteed" funding through earmarked revenues. But in the long run the existence of revenues earmarked for specific purposes is likely to keep, or possibly increase, pressure for hgher spending. In the case of CIDE, the low levels of investment by the federal government tnggered a reaction from the sub- national governments, whch demanded, and obtained, a new arrangement to make CIDE sharable between the federal and state government^.^^ In short, earmarking does not seem sufficient in the short run for protecting expenditures when the government priorities are elsewhere, but can add to political pressure for higher spending over time. Figure 3.15 Ministry of Transport Investments and CIDE collection `I7.000 - 8.000 -. W 6.000 / /1 N 5.000 1-Ministry of Transport lnvestments -CIDE CollectionI Source: SOFI Ministyof Planning and Budget I Box 4.1: Dealing withRevenue Earmarking I The federal government has managed to get around constraints on budget management imposed by the earmarlung legslation in a number of ways. In the last few years, the Fiscal Responsibility Law has made these procedures more difficult since it forbade the use o f any resource legally earmarked to specific expenditures for other types of expenditures, even inother years. But even then, in 2002, the Executive promulgated a decree (medidu provisonu), later converted into Law No. 10.595, whch allowed all National Treasury cash stock originated from non-constitutionally earmarlung to be used for debt amortization. This allowed resources not used in any year before 2002 in the specific area to be used in debt amortization the following years. Resourcesnot applied to the specific expenditure can also be "used" to generate apnmary surplus andbecome part of the Treasure cash stock. According to the Federal Budget Secretariat (SOF), 40 percent o f the pnmary surplus in2002 originated from earmarked sources. Thls implies that, ifall the earmarked resources were used inthe predetermined allocation, the primary surplus in 75The Constitutional AmendmentNo. 4412004 was passed as part of broader political negotiations on the tax reform proposed by the federal government in 2003, and earmarked 29 percent o f CIDE collection to states. 54 2002 would have been 60 percent o f the actual level. Inother words, if Brazil were to adopt a mandatory budget system whereby the executive i s legally obligated to execute annual budgets as approved by Congress - as inthe United States -the current level o f revenue marking might not allow sustenance of the responsible fiscal policy that the government has maintamed for the past several years. 3.47 The discussionabove shows that earmarked revenue contributesto budget rigidity, but in the short run, at least, its net effects seem to be swallowed by the larger effects on rigiditycaused by the high level of mandatory expendituresin Brazil's budget. Most of the earmarked revenues are tied to these mandatory expenditures, but since the total of the latter i s larger, a significant portion of the non-discretionary expenditures is also financed by non- earmarked revenues. From 1998 to 2003, from 73 percent to 80 percent of the non-earmarlung revenues were allocated to finance mandatory expenditures (Silva, 2003).76 3.48 How are earmarked revenues allocated? It i s interestingto highlight here how the non- use (retention) of earmarked resources i s distributed by functions of government. Here we compare the earmarked sources collected in 2002 and 2003 with expenditures that had the same budget sources in those years. We worked with pnmary revenues and exclude free resources. The government functions are: i)economc affairs; ii)health, social security and social assistance; iii)education; andiv) others (composedofsecurity andgeneralpublic servicesandrevenues earmarked to more than one function, including the items (i), and (iii)). (ii) 3.49 Ifall the earmarkedrevenueswere fully executed as specified, or ifthe non-utilizationof earmarked revenues were equally shared across different categones, then 61 percent would be spent on social protection and health, 22 percent on revenue sharing with the sub-national governments, 6 percent on economc affairs, 3 percent on education, and 7 percent to others ( Figure3.3). Inreality, however, nearly half(48 percent) of the total nonutilization of earmarked revenues was related to economc affairs (Figure 3.16). This can be particularly important if we consider that economc affairs representedonly 6 percent o f the earmarked revenues collected in 2002 and 2003. If all the earmarlung resources had been used to finance expenditures, the level of expenditures on economc affairs would have nearly doubled the actual amount spent (2.75 percent of primary expenditure^).^^ 3.50 The remaininghalf(52 percent) of the non-utilizationfell on"other" brokendown into: i) 9 percent from direct collection by different entities of government; and ii)45 percent from others. Social security and health functions are responsible for -2 percent7*o f the non-utilization of resources and transfers to states and municipalities for -3 percent, meaning that these categones absorbed all their earmarked revenues in the years considered and also stocks o f non- usedresources from previous years. Three percent of the earmarked primary resourcesthat were not used were related to the education sector, roughly equivalent to the sector's share inthe total earmarkedrevenues inthe first place (i-e., the "cut" was proportional"). 76From 1998 to 2001, data are from the Annual Budget Law. For 2002 and 2003, data were from Budget Law Proposal. 77 This analysis was done not considering the DRU resources. The resources de-earmarked by DRU are classified as source 100, or free resources, malung more difficult to infer its effect by function o f government. 78Negative numbers are possible. We compared the resources collected on the years to the used on the same years. As expenditures could be financed by resources collected inprevious years and non-used then, negative figure are possible. We worked with the average o f two years to reduce this effect. 55 Figure 3.16 Unused earmarked resources by function (2002/2003) 60%7--- I z 50% 40% -g - 30% 20% % 10% ;;e 0% 3.5 1 Budget rigidity at the state level: Budget rigidity i s not a problem unique to the federal government. For state governments, for instance, the most important source o f rigidity is personnel expenditures. In 2003, an average state spent 51 percent o f primary expenditures on personnel, with 38 percent to employees and 14 percent to retirees and survivors o f public servant^.^' The actual share o f personnel expenditures vary from 36 percent in Tocantins, a relatively new state, created in 1988, to 62 percent in Paraiba. Another important source o f rigidity in state budgets is transfers to municipalities. On average these represent 14 percent o f primary expenditures. These two items together represented 65 percent o f pnmary expenditures. Other sources are education and health earmarlung. According to the 1988 Constitution, state and municipalities must apply at least 25 percent o f its tax and transfer revenues ineducation, and 12 percent to healthago However, as these two areas (especially education) have an important component o f personnel expenditures, it i s difficult to establish their net effects on budget rigidity l9S T N data. Constitutional Amendment No. 29,2000. 56 Figure3.17 Share of PersonnelExpenditures andTransfers to Municipalitieson PrimaryExpenditures(2004) 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Source: Ministryof Finance: National Treasury Secretanat. 3.52 To take the case o f Minas Gerais as an example,81 personnel expenditures in the Educational Secretariats2reached 86.7 percent o f the government's total personnel expenditures, without counting retirees and pensioner^.'^ This could imply that the net effect o f revenue earmarlung to education in the short run i s lower than imtially thought, due to the intersection between earmarlung and mandatory (personnel) expenditures. But also in this case, we should remember that inthe medium to long run, the rmmmumresources guaranteedmay have an impact on personnel hmng, with more effect on budget rigidthan at first. 3.53 In Minas Gerais, 52 percent of primary expenditures are financed with earmarked resources, including transfers to municipalities.84 Ths level is lower than in the federal budget. But here, the rmnimum levels o f application in health and education are not considered as part o f earmarlung, except FUNDEF The mnimumhealth spending represented 6 percent o f the total primary spending in 2003. In education, net o f FUNDEF resources, t h s number was 7 percent. 3.54 The distribution o f earmarked resources by function was: 42 percent to intergovernmental transfers, 26 percent to education, 10 percent to health and sanitation, and 7 percent to social security and social assistance (Figure 3.18). We can infer from Figure 3.18 that for some areas, an important part o f its financing i s from non-earmarked resources, where its share in the total primary expenditures i s much bigger than the share of the earmarked resources in this area. The 'I For Minas Gerais State, total personnel expenditures and transfers to municipalities represented 70 percent o f primary expenditures. '*The Educational Secretariat is responsible for 92.2 percent o f expenditures to comply with the constitutional floor. 83Data for Minas Gerais State are from 2003 balance. 84The new transfer CIDE, would have represented and increase o f 1.15 percent of earmarlung resources. 57 opposite occurred with transfers to municipalities, where its share inthe pnmary expenditures i s substantially hgher than the share o f earmarlung revenues, w h c h probably implies that most o f the resources transferred to municipalities came from earmarked sources. The use o f free resources on health and education may be a consequence o f the constitutional minimumlevel o f spending inthese sectors. Figure3.18 MinasGerais State: expenditure byfunction total and earmarked - resources(2004) 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Ill!klEarmarked expenditures =Total primary expendituresI 3.55 Most of the non-discretionary expenditures in Minas are determined by federal legislation. Besides the items above, the only mandatory expenditures not due to federal legislation i s the mnimum application o f 1 percent o f current budgetary revenues to the state research entity. Ths was a rather small amount compared to the spending due to federal earmarlung in some other areas. In 2003, the mnimum amount earmarked to research represented 8.9 percent and 3.6 percent o f the constitutional mnimum amount earmarked to health and education, respectively. CONCLUSIONS 3.56 Federalgovernment budged is distinguished by a high level of rigidity,consequence o f obligatory expenditures, as well as earmarlung mechanisms, and both increased substantially over the last twenty years. The 1988 Constitution was an important component to t h s budget rigidity increase, on both sides. Social nghts were enlarged which implied an increase on expenditures on social security and assistance. Actually, a separate budget was created to social security, social assistance and health, showing the new importance o f these areas inthe spending. Since then, social spending had increased substantially. The financial resources to finance t h s new level o f spending were created later, by increases o f the contribution rates already predicted to finance social security budget or by creation o f new ones, like the CPMF 3.57 As a result, social contributions increased considerably during the same period. These revenues were earmarked to social expenditures, representing a rise in earmarlung mechanisms. Inaddition to financing social expenditures, social contributions became "revenue instruments o f choice" for the federal government when Brazil began rigorous fiscal adjustments 58 inthe late 1990s because o fthe legal characteristic o fthe social contributionswhich didnot need to be shared with sub-national governments. 3.58 The current patternof budget rigiditydisproportionately favors social expenditures at the cost of spending for non-social items (e.g., economic infrastructure). Not only is earmarlung concentrated on social spending, but also the earmarlung sources not spent are normally non-social. This could indicate that earmarlung mechanism sometimes does not protect expenditures at the level expected when implemented. But even when specific earmarked revenues are not actually spent in a given fiscal year, they still constitute an obstacle to efficient budget management - e.g., the Treasury could not use non-used earmarked revenues to retire public debts. 3.59 A naturaldecrease in the level of budget rigidity in the foreseeablefuture is highly unlikely. Some o f the sources o f rigidity follow real economc growth, like health mnimum spending level and earmarlung mechanisms. Non-discretionary expenditure items such as social security benefits will m e steadily, especially in conjunction with the inevitable aging o f the Brazilian population, and will further worsen the overall rigidity o f the Brazilian budget and threaten fiscal sustainability Yet the general perception that Brazil does not spend enough in various sectors persists and translates into demands for spending increase and revenue sharing between different levels o f government. The new transfer to states governments o f CIDE collection and new pressures for increases on the level o f mumcipalities' transfers may be examples that the strategy o f concentration on the mcrease o f revenues on non-shared sources may have an end. 59 4. PUBLIC EXPENDITUREINTHE ROADTRANSPORT SECTOR 4.1 The compression o f public investments as a result o f the fiscal adjustment effort in a context o f high expenditure rigidity has exacerbated the infrastructure bottlenecks which are one of the oft-mentioned constraints to growth in Brazil. These bottlenecks are most acute inthe transport sector, inmaintaimngand upgrading roads, rail, and waterway networks, and are especially obvious in the lack o f infrastructure necessary to support Brazil's increasing integration into international trade, such as ports and modern cross-border links. These bottlenecks have become even more acute with the recent resumption o f stronger economc activity, and they threaten the sustainability o f the current Brazilian economc growth cycle. 4.2 An increase ingovernment investment expenditures does not imply,however, an immediate expansion o f the public capital infrastructure with positive growth effects. Besides the resourcescarcity, weaknesses inplanning,budgetingand expenditureexecutionin infrastructuresectors are highlyinefficient. Thus, inadditionto the resource scarcity, expenditures inthe sub-sector are not well-plannedand pnoritized, and their execution suffers from a variety o f causes for delays and cost escalation. 4.3 Ths chapter analyzes the quality o f expenditure management inthe road transport sub- sector given its role as the main transport mode in Brazil, its importance in the federal government investments and the potential positive growth impacts o f the expenditure management enhancement o f the sector. THEROADTRANSPORT SECTORE5 4.4 The transport sector accounted for almost US$ 25 billion or 3 percent o f GDP in 2004. The sector moves 825 billion tons-km o f freight per year and 890 billion passengers-km. Road transportation i s the domnant mode in the Brazilian transport matrix (responsible for almost 60 percent o f freights) and accounts for about 2.5 percent o f GDP Despite the competitive nature o f the transport service industry (e.g., inexistence o f significant entry bamers), the road transport sector i s inefficient with unitary cost two or three times higher than the railroad and waterborne systems. Railroads and the waterborne modes, which are more cost effective, transport only 38 percent o f total freight. Inpassenger transportation, the domnance o f road transportation is even stronger with more than 90 percent o f passengers transported by roads. Figure 4.1 Brazil Transport Matrix inbillionof tons-km-used (TKU), 2004 - 0 3 3 . 3 4.0% n 4% Total: 825 bi TKU w Roads .Railroads 0 Waterborne 0 Other .Air Source: IDET-FIPEKNT and MT 85This section draws from the background paper preparedby Frischtack (2005). 61 4.5 Given the geographlc characteristics o f the country, the predominance of road transportation in the Brazilian transport matrix leads to inefficient logistics. High freight volume and long distances would favor the use of railroads, and the navigability conditions of Brazilian rivers would suggest a more intensive use of waterways. The comparison with other continental countries with similar geographical patterns shows Brazil's high dependence on road transportation, similar to the charactenstics of small industrialized countnes. Table 4.1: Participation of Modes inthe Transport Matrix and Territorial Extension Selected Countries, in percentage and million of km2,most recent years Roads Railroads Waterborne I Terr. extension(km2million) Continental Countries I RussianFederation 8 81 11 17.0 China 50 37 13 9.6 USA 32 43 25 9.2 Canada 43 46 11 9.2 Australia 53 43 4 7.6 Brazil 60 26 14 8.5 Countries of small territorial extension Japan (2) 55 4 41 0.4 Poland('I 55 44 1 0.3 Switzerland(2) 58 28 14 0.4 Germany (I) 63 20 17 0.3 France(2) 74 22 4 0.5 4.6 Most of the Brazilian roads, spamng 1.6 million km in total, are non-paved (88 percent o f total road network) and managed by municipalities (82 percent of total road network). The pavedroads (182,000 kmintotal) are mostly kept by state governments (55 percent of total paved roads). The federal road network is mainly paved (80 percent of federal roads are paved) and has an extension of 58,000 km. About 10,000 km of the federal and state networks are managed by pnvate concessionaires. 4.7 Inspite of its relatively limted extension (36 percent of the total paved road network), the federal network i s crucial as it domnates medium and long distance transportation that interconnects regions and states. More than 70 percent of the transported goods flow via the federal network. 4.8 Traffic figures also show the relevance of the federal road network. Traffic on federal paved roads vanes from less than 1,500 vehicles up to 50,000 vehicles per day in the most heavily trafficked sections of the Southeast region. Ingeneral, the traffic intensity of state roads i s lower than the federal roads (World Bank 1997). 62 Figure 4.2 Road Network Extension, 2004 (km) Source, CNT 4.9 The logistic inefficiency arising from the inadequate composition of the Brazilian transport matrix is magnified by the low density of the road network and the deterioration of its conditions, reflections of inadequate investments in the sub-sector over the years. Given the hgh dependence on road transportation and the limted participation o f the pnvate sector in the road infrastructure provision, the fall in government transport investment expenditures has strongly impacted road transportation costs. Because of the size of the transport industry and its impact on other productive sectors, the fall of government investment in the most important transport infrastructure should have negative impact on the efficiency of the economy. 4.10 The public investments in the 1970s-80s that led to expansion of the road network to the intenor o f the country However, there remain large dispanties in road network density among regions, with the South and Southeast having a dense and congested system o f highways while the North and Center West regions having a density three times lower than the national average, given their territonal extension and lesser economc importance. Since the 1980s, the compression of public investment has led to slower pace of the expansion of the road network (Figure 4.3). Figure 4.3 Federal pavedRoad ExpansionRate, 1965-2004 12 10 8 6 4 2 0 1965-75 1975-85 1985-95 1995-04 Source: IBGE.Annual Statistics Report. 63 RoadDensity RoadDensity Continental Countries Brazil and regions USA 0.3795 Brazil 0.0194 Canadd 0.0906 North 0.0025 Mexico 0.0487 North-west 0.0254 Argentina 0.0205 Southeast 0.0517 Brazil 0.0194 South 0.0798 Countries of small territorial extension Center-West 0.0140 Germany 0.6472 France 1.7896 Brazil w/o North 0.0334 Spain 0.3225 Source: GEIPOT / Annual Transport Statistics Report. DITIAS/ ALADI -2000. IBGE 4.11 Even more important than the low road density are the poor conditions of the road network. Even duringthe 1950s to 1970s, the penod o f substantial investments inthe sector, the government priority in the road sub-sector was to expand the network, at the expense o f maintenance and rehabilitation. Although priorities shfted from new road construction to maintenance and rehabilitation in the 1990s, the aggravation o f the fiscal problems led to the postponement o f maintenance and rehabilitation. Furthermore, the sector agencies have been unable to adjust the road sector policy to the decrease in available resources and to create a management structure onented to maintenance and rehabilitation rather than new road construction. The inadequate response by the public sector was not compensated by pnvate participation.86 Figure 4.4 Federal Paved Network Condition 100% P 90% gE 80% -E 70% 0 g6 60% 50% z 1 40% 30% U j 20% 10% 0% 1979 1984 1986 1989 1992 1996 2000 2001 2002 2003 2004 Year IEdGood(IRIc3) mFair (3cIRM) OPoor (IRb5)I Source:World Bank 2006 4.12 According to a survey conducted by the National Transport Chamber (CNT), only 25 percent of the paved roads were considered in good conditions as of 2004. Only 17 percent of the paved roads under government management are in good condition while 78 percent of the network administrated by private concessionaires is in good condition (Table 4.3). 86In 1996, the government conceded to private sector the administration of some pavedroads. However, only 10,000 km of the 160,000 km of federal and state paved roads are currently administrated by private concessionaires. 64 Underprivate All Government concessionaires Km YO Km % Km % Excellent 8,692 11.6 3,474 5.4 5,218 51.5 Good 10,227 13.7 7,506 11.6 2,721 26.9 Deficient 27,148 36.4 25,371 39.3 1,777 17.5 Bad 17,686 23.7 17,319 26.8 367 3.6 Worst 10,928 14.6 10,878 16.9 50 0.5 Total 74,68 1 100.0 64,548 100.0 10,133 100.0 4.13 Besides the fact that 90 percent of the roads are single lanes with two-direction traffic, which implies low speed and high accident risks,the bad conditions of the pavement and the precarious signage are identified as the most critical problems. The quality of the pavement is considered the most cntical problem as almost 42.000 km are considered deficient, bad and worst with severedamages in 8,000 km. Table 4.4: Road quality indicators, 2004 I I Km % Pavement incntical condition 41.911 56.1 Lack of signalization 48,788 65.4 Without secunty lane 30,072 39.8 Signals covered by vegetation 18,355 24.6 Segments with severe damage 8,280 11.1 Simple lane with two directiontraffic 67,239 I 90.1 4.14 The direct and indirect costs of the bad conditions of the road network are difficult to estimate. As part of the direct costs, the CNT estimates that the deficiencies of the road transport infrastructure increases operational costs (piece substitutions, truck repair and maintenance and fuel) by more than 30 percent. In terms of indirect costs, a recent study showed that logistic costs associated with transport (including stock, storage and administrative costs) reached about 12 percent of the GDPaX7 Incontrast, intheUnitedStates, these costs amount to 8 percent of GDP 87Research results from the Centro de Estudos de Logistica, at CoppeadllTFRJ (Federal University o f Rio de Janeiro), presented at the XI F6rumNacional de Logistica, Rio de Janeiro, August o f 2004. According to this study, expenses with fuel, tolls and vehicle maintenance amounted to R$ 104.3 billion for roads cargo. Adding the railroads revenue (R$ 7.5 billion), the price o f waterborne freight (R$ 6.9 billion), o f duct (R$ 2.1 billion) and air freight (R$ 1.9 billion), transport costs was almost R$ 122.5 billion, or 7 percent o f GDP The expenses with storage, stocks and administration were respectively R$ 11.2 billion or 0.6 percent, R$ 71 billion or 4 percent and 8.2 billion or 0.5 percent o f GDP 65 Brasil (A) EUA (B) AB Administrative 0.5 0.3 1.67 Storage 0.6 0.7 0.86 Stock 4.0 2.1 1.91 Transport 7.0 5.0 1.40 FEDERAL INVESTMENTEXPENDITURES INSTITUTIONS INTHE TRANSPORT AND INFRASTRUCTURE SECTOR: HISTORICAL TRENDS 4.15 Since the 1960s, the transport sector's mstitutional development and the level o f resources available to the sector have evolved hand-in-hand with each other, as a result o f shfts over time in the sector's financing mechanisms. The trend since the rmd 1970s i s clear: both the total public investments and transport investments have steadily declined (Figure 4.5). With t h s declinmg trend, the sector has also lost some o f the institutional capacities it had developed inthe earlier period. Figure4.5 Federal Government Investments, 1975-2004 (inpercentof GDP) Sources: R. Velloso, opxit., Ministerio dos Transportes;SOF 4.16 The declimng trend i s marked with two distinct moments. First, the end o f the high growth penod in the early 1980s (partly caused by the second oil shock) ledto a strong reduction in federal investments. Second, the resumptionof investment expenditures inthe second halfo f the 1980s proved to be short-lived as the fiscal effects of the 1988 Constitution that prioritized social expenditures and the fiscal crisis in the 1990s aborted this trend. The transport sector was particularly hard-hit by the fiscal contraction. 4.17 On the financing side, the sector's investments were funded by a senes o f revenue earmarlung arrangements that provided predictable flow o f funding.88Yet inthe 1970s, the strict In 1945 the National Road Fund (Fundo Rodoviario Naczonal -FRN) was created with the funds earmarked from the Tax on Oil and Gas and Liquid Combustibles (Impost0 Unico sobre Lubrijcantes and Combustiveis Liquidos e Gasosos-IULCLG). In 1967, the National Road Fund gained another earmarked 66 revenue earmarlung system that guaranteed resources for the sector began to weaken progressively. In 1974, part of the earmarked resources was redirected to the National Fund of Development managed by the Planning Mimstry which had broader objectives than the development of the transport sector. Combined with the reduction of the rates of the IULCLG tax, used to alleviate the impact of the oil shocks, the gradual weakening o f earmarlung resulted in the reduction of resources and increasing uncertainty of the financial flows to the sector. In 1988, the new Constitution fully elimmated the earmarlung mechanism that financed the transport formed the basesof earmarking for transport expenditure^.^' sector. The tax reformincludedinthe Constitution of 1988 abolished some of the taxes that 4.18 On the institutional side, the 1960s-1970s was a period in whch the sector saw development of a structure for planning and management o f its investment portfolio. The National Road Fund encouraged formulation o f multi-annual plans for transport investments, strengthemng o f the Transport Mimstry, as well as the creationof the National Roads Department (Departamento Nacional de Estradas e Rodovzas-DNER), the 27 state roads departments (Departamentos Estaduazs de Rodovzas DERs), and the municipal transport ~ecretanes.'~The - sector planning was further strengthened with the creation o f the Executive Group of Integration of the Transport Policy (Grupo Executivo de Integraqlio da Politica de Transportes - GEIPOT) in 1965 as a multi-mmisterial instance responsible for formulation and implementation of the national transport policy '' 4.19 The sector's institutional development weakened in parallel with its declining funding. The agencies ability to recruit competent staff was kept high when it could count on secure sources of funding from earmarked revenues. But over time, the erosion of the sector's fundingcapability has resulted inits institutional weakening as well, as reflected inthe declimng size of the DNER/DNIT92'sstaffing, for example (Table 4.6).93 revenue source with the Tax on Road Transport Services (Imposto sobre Sewicos de Transporte Rodoviririo-ISTR). Finally in 1970, the revenue from the Roads Charge (Taxa Rodovirirza Unica-TRU) became a third financing source of the National Road Fund. The revenue from these three federal taxes represented about 1.5 percent of GDP and the revenue earmarlung mechanisms governing them strictly guaranteedthat they were to be usedexclusively for roadtransportation expenditures. Additional resources were financedby government borrowing. 89 The IULCLG was outright eliminated, the ISTR became part of the base of state value-added tax, (Imposto sobre Comercializap?ode Mercadorias e Sewicos - ICMS),and the TRU was converted to the state vehicleregistrytax (IPVA). 'O The DNER was responsible for execution of the National Transportation Plan and approval of the transportation plans of the state governments which were receiving 36 percent of the earmarked revenues ''directedGEIPOT to the transport sector. The was composed of the ministnes of Transport and Public Infrastructure Works, Finance, Planning, and the Presidency following the TechnicalAssistance Loan agreement signed by the Brazilian government with the World Bank. In 1969, the GEIPOT was subordinated to the Transport Ministry and changed its orientation from implementationactivities towards an analytical and assessment role for the policy makers supporting the coordinationamong transport modals, long term planning and evaluationof the transport sector policies. The word Executive was changed to Study (Estudo). In 1973 GEIPOT changedits name to Brazilian EnterpriseofTransport Planning. 92 NationalDepartment of Transport Infrastructure-DNIT is the successor of DNER, see below. 93 The prevalence of political interference has also been alleged to be another cause of the weakened institutional capacity of the transport sector. The political visibility of transport infrastructure made the DNER and other transport government agencies attractive for political parties that composed the support coalition of the federal administrations. Political appointees with little relevant technical and managerial 67 Year 1987 1992 1997 2002 2005 Staff 16,400 8,000 4,000 3,700 2,400 4.20 In order to reestablish a substantial and regular flow of resources to the sector, a new revenue was created in 2001 (Contribuipio de Intervenp7o no Dominio Econ6mico - CIDE).~~ But CIDE has had limitedeffect inimprovingthe level andthe predictabilityof fundingfor the sector. Although the CIDE collection has beenhigher than expected since 2002, only a small portion has actually been allocated to transport investment expenditures. As Table 4.7 shows, an important part of CIDE collection revenue is incorporated to the non-financial balance of the federal government. Source: Sistema Integradode Administra@oFinanceira(Siafi) * Law 10.866of2004 establishedthetransfer of percentof CIDE revenueto state and municipalities. 29 4.21 While the government attempted to revamp the funding base for the transport sector with the creation of CIDE, it also attempted an institutional overhaul of the sector. The main aspects of this institutionalreformwere: . Decentralizationthrough the transfer of federal roads to state governments, . which assumed the responsibility for the maintenance and rehabilitation expenditures of the roads transferred. Increasing participation of the private sector in the transportation sector through the concession of several highways or sections to pnvate operators, who . will became responsible for operation, maintenance, and upgrading directly from tolls. Replacement of the DNER with the National Department of Transport Infrastructure(DNIT), amulti-modal agency for executing transport projects ; experience, rapid turnover and political interference inoperational decisions characterized the institutional reality o f the DNERand the Transport Ministry. 94The CIDE refers to an economic contribution charged over commercialization and importation o f oil and its derivatives, natural gas and its derivatives and alcohol, and was earmarked to the transport sector. Unlike the National Roads Fund that was earmarked to road transport expenditures, the earmarlung o f CIDE is broader as its revenues are directed not only to the road transportation infra-structure but also to subsidies on alcohol, natural gas and its derivatives and oil derivatives, urban transport projects and environmental programs related to exploration and production o f oil and its derivatives. 95 Law 10.866 o f 2004 established the transfer o f 29 percent o f CIDE revenue to state (25 percent) and municipalities (4 percent). 68 Internal reorganizationof the Transport Ministry; Nacional de IntegraqBode Politica de Transportes Terrestres - CONIT); Creation of the National Council of Integration of Road Transport (Conselho Creation of new regulatory agencies for land transportation(ANTT) and water transport (ANTAQ); and Liquidation of GEIPOT and transfer of its activities to the DNIT, ANTT, ANTAQ CONIT. The launch of the federal program for maintenance and rehabilitation of the federal network, which consists in contracting out rehabilitation works and maintenance services combined on entire routes of the remaimng network through (five-year) output-based agreements with the contractors responsible for achieving specified levels of service. 4.22 While these government initiatives can be considered adequate, their progress has been slow. In terms of the decentralization of roads, the federal government has transferred 14 thousand km of roads to 14 state governmentsg6.However, the state governments, arguing the lack of resources, did not carry out the maintenance and rehabilitation of the transferred roads, thus, the detenorationtrend of the roads was not stoppedwith the decentralization. 4.23 Given the lack of commitment of the state governments with the transferred roads, the federal government i s currently considenng the devolution of these roads, whch would imply a backward step inthe decentralization process. 4.24 Better results were obtained with the concession of highways to the private sector. Including roads transferred to states, about 5 thousand kilometers have been conceded to the private initiative with positive results interms of enhancing quality of roads and user satisfaction. However, contract renegotiations have transferred the nsks to users through the increases intolls, which have generated public outcry. As a result, the Federal Accounting Tribunal (TCU) has objected to the concession contracts and has paralyzed the advance of pnvate participationinroad management. 4.25 The reorganizationof the sector's institutional arrangement has not so far produced visible improvements in its institutional capacity The DNIT inhented most of the fragilities o f its predecessor, the DNER. The problem of inadequate staffing continuedto worsen (see table 3.6 above)." DNIT, like DNER before it, has beenunable to re-adapt its structure to the changes in the main focus of the transport investments from road building to maintenance and rehabilitation activities. With a reduced staff, DNIT, too, has been unable to manage a large number of increasingly complex contracts effectively The lack o f an information system to integrate budgeting, financial and physical execution, and accounting continues to be a binding constraint for efficient project management. The closure of GEIPOT apparently had limted 96 I t is important to remember mention that the Constitution o f 1988 transferred to the state governments the tax revenues that previously had financed the sector, thus, the decentralization o f roads was nothing more than the completion o f the decentralization process launched by the Constitution o f 1988 w h c h had decentralized revenues but not expenditure responsibilities inthe road sub-sector. 97 Observers and ministry staff attribute t h s to the lack o f recruitment processes for new staff and the unattractive salary policy. 69 70 4.29 The federal transport programs can be classified in three main categories: construction, improvement, and maintenance. Among the construction programs,. the most important are construction of new roads, duplication, and elimnation o f cntical points.99 O f these, construction was the most affected investment category with a tend-fold decrease in the amount executed (from R$880million in 2000 to R$80million in2005). Allocations to maintenance expenditures were preserved relative to other categories, although these, too, dropped somewhat (from R$890 million in 2000 to R$ 710 million in 2005), while improvement investments passed from R$ 57 million to R$ 30 million. The overall decline inthe level of federal investments inroads was 55 percent between 2000 and 2005. I Table 4.8: FederalinvestmentsinRoadsby category, 2000-2005 I 2000 2001 2002 2003 2004 I 2005 Maintenance 0.89 1.18 0.66 0.9 0.85 0.71 Improvement 0.57 0.58 0.34 0.15 0.47 0.3 Construction 0.88 1.13 0.85 0.19 0.39 0.08 Others 0.14 0.08 0.05 0.02 0.05 0 Total 2.480 2.970 1.900 1.260 I 1.760 I 1.090 4.30 The larger relative decline of the programs for constructing new roads compared with the other programs acceleratedthe s h f t of the pnonties that hadbeenoccumng inthe previous years. As a result, the share of maintenance grew from 38 percent of total roads investment in 2000 to 65 percent in 2005, while investments inconstruction were reduced from 35 percent in2000 to 7 percent in 2005. The share of improvement investments sharply fluctuated averaging about 20 percent of total roads investments. Figure4.7 Compositionof RoadInvestmentExpenditures2000-05 100% 2000 2001 2002 2003 2004 2005 Maintenance Improvement Construction Others Source: Brazilian Senate Site. Budgetinformation. SIGA System. Calculationsby World Bank Staff 4.31 The data on expenditure execution clearly show a shift in the composition of the transport expenditures over the 2000s (Figure 4.7). While on the surface, the gradual shift 99 Improvement programs include improvement and emergency works, while maintenance encompasses programs o f road recovery, restoration, conservation, direct maintenance and maintenance trough contracting out. 71 budget allocations in favor o f maintenance activities makes sense given ther relatively high returns, especially in the context of resource constramts, it i s not clear to what extent thls re- allocation has been guidedby conscious forward planning. 4.32 The government's primary instrument for forward expenditure plamng is its Multi-year Plan (PPA). The PPA 2000-03, for example, was partially based on a comprehensive analysis of the country's investment needs and potentials, with heavy emphasis on economc infrastructure investments along geographically defined "axis" of development.'00 However, there i s evidence that the PPAs have not beenthe techmcal basis for actual resource allocations. The allocations to the programs included in the PPAs 2000-20003 and 2004-2007 have not been reflected in the allocations set in the annual budgets, whether in the government's budget proposals or in the Annual Budget Laws (LOAs) approved by Congress. 4.33 At the aggregate level, total transport investments inthe PPA 2000-2003 were set at R$ 38 billion while the sum of the government budget proposals reached only R$ 22 billion (58 percent of the PPA's indicative allocations). Transport investments inthe LOAs duringthe four- year penod were R$ 25 billion (67 percent of the PPA allocations). Figures are similar in the period 2004-2007 as the PPA allocations are much higher than the established inthe government budget proposals and inthe approved Annual Budget Laws."' 4.34 At the level of individual programs, the inconsistency i s even more pronounced. Some programs, such as the construction of federal roads and transport services, received multiples of the PPA's indicative allocations, while others, notably the strategic programs o f the axis, were consistently under budgeted in both plamng penods with cumulative L O A allocations representingbetween 7 percent and 95 percent of the PPA allocations (Tables 4.9 and 4.10). Table 4.9: PPAs and Annual BudgetLaw allocations, 2000-03 (inbillionof Reais) Program PPA Sum o f Gov Sum LOAs I 2000- Gov Budget LOAs PPA (%) 2003 Budget proposals 2000103 Proposals I PPA(%) 2000l03 Improvement o f federal roads 0.018 0.018 100 0.069 383 Construction o f federal roads 0.083 0.83 1000 0.91 1096 Decentralization o f federal roads (S) 0.066 0.079 120 0.079 120 Maintenance o f federal network (S) 3.84 3.52 92 3.94 102 Security in federal roads 0.62 0.44 71 0.55 89 SBo Francisco Axis (S) 1.34 0.24 18 0.36 27 Leste A x i s (S) 2.09 1.10 53 1.53 73 Transmetropolitano Axis (S) 1.75 0.63 36 0.68 39 Sudeste A x i s (S) 4.49 0.16 4 0.33 7 Mercosul Axis (S) 3.66 1.18 32 1.77 48 Nordeste A x i s (S) 3.88 0.83 21 1.37 35 Oeste Norte Axis (S) 1.23 0.57 46 0.88 72 Araguaia-Tocantins Axis 3.91 0.92 24 1.56 40 Fronteira Nortes Axis 0.73 0.30 41 0.69 95 Waterways Transport Services 0.98 1.87 191 1.17 119 Passenger railways - urban transport (9 3.89 2.06 53 2.40 62 loo"National Axes o f Integration and Development Study." Io'Inthis case, 2007 was not included inthe sum o f LOAs, thus the difference would be less pronounced than the observed intable below. 72 Maintenance of waterways 0.076 0.12 158 0.11 145 Managementof transport policy (S) 0.11 0.31 282 0.30 273 Cargo railway services 0.057 0.035 61 0.057 100 Roads transport services 0.007 0.058 829 0.059 843 Interior navigation 0.005 0.011 220 0.01 200 Portsupgradingand expansion 0.015 0.015 100 0.039 260 Environmental quality 0.002 0. 97 4850 0.002 100 Others (Administrative) 4.84 6.32 131 6.23 129 Total 37.72 21.9 58 25.10 67 ( S ) Strategicprogram Source:PlanningMinistry and Brazilian Senate Site. Budget information. SIGA System.Calculationsby World Bank Staff 4.35 As mentioned above, the strong fiscal adjustment and the increasing budget rigidity resulted in the reduction of budget allocations to the transport sector and thus broadening of the gaps betweenplannedand budgetedinvestment expenditures. From 2000 to 2005, the budgeted resources for the Transport Mimstry fell by 20 percent (with most of the fall concentrated in 2004). Inthe case of road investments, after reaching a peak of R$ 6 billion in 2002, the budgeted amounts fell sharply to R$ 3.2 billion in 2003 and to R$2.1 billion in 2004. In2005, however, the introductionofthePilotProjectfor Investments(see below) broughtroads investments back to R$ 5 billion (the same amount as in 2000). In terms of categones, duplication, critical point elimnation, restoration and maintenance were the road investment categones that suffered the greatest budgetary reductions while improvement allocations were doubled inthe same penod (Table 4.12). Total Roads Investments 5.11 4.52 6.18 3.24 2.09 5.16 26.30 Total Ministryof Transport 12 11.1 13.5 13.5 8.58 9.63 68.31 4.36 The inconsistency between planning and budgeting appears already when the Executive submts its budget proposal to Congress. The differences between PPA allocations and government budget proposals can be explained partly becausethe firsts become out of date as the 73 Figure 4.10 Congressional Amendments in Percentage of Annual Budget Laws' Allocations by Type of Program (Averages 2000-05) Reventive Conservation bmtenance through Contracting Out Duplication h r g e n c y Works 0% 25% 50% 75% 1007 Source: Brazilian Senate Site. Budget Information. SIGA System. Calculations by World Bank Staff 4.38 Once budgeted, expenditure allocations for the transport sector investments tend to be characterized by low levels of execution. At first, the decree of budget and financial execution issuedjointly by the Mimstnes o f Planning and Finance freezes a significant part of the budget authorized in the LOA. In practical terms, this decree limts the amounts of new expenditure comtments the rmnistry can make during the year, until a new decree revises the ceiling. The financial programng establishesmonthly lirmts for cashpayments withn the level compatible with the projected financial flows.'02 Budget execution is delayed further by the complexity involved in managing a large number of small and complex contracts, the rigid rules of the national Procurement L a w 8,666, and the weak institutional capacity of the transport agencies to design and manage contracts. The delay in the contracting process frequently necessitates ex post changes to the project design and re-negotiations o f the contracts, which further delay project execution and also increase per unit cost o f the public works contracts. 4.39 The low limits for spending authorization and monthly payments at the beginning of the year keep the ministry and the sector agencies from making large expenditure commitments. Besides, the small amounts liberated for financial execution at the beginning of the year are typically earmarked for settling bills pending from the previous year (so-called restos apagar). As a result, comtments are significantly lower than total budget allocations. Figure 4.11 shows that from 2000 to 2005, the total commitments for road investments, on average, amounted only to 60 percent of total budget allocations to the sub-sector. I O 2Given that at the beginning o f the year the revenue collection i s low and, the Finance and Planning Ministries adopt a conservative stance expressed inthe reduction of about 30 percent the budget allocations and the establishment of low limits for monthly expenditure payments to guarantee the achievement o f fiscal target surpluses. A recent study by the Credit Suisse shows that about 70 percent o f the primary surplus is obtained inthe first semester o f the year due to the freeze on budget and financial execution. 75 Figure 4.11 Commitments as percentage of total budget (LOA+Credits) (2000-05 averages) Reventive Conservatlon Recovery tinergency Works Others Restoration Total Roads hvestmnts hprovemnt Construction h pication Critica Fbnts Himnation Maintenance Maintenancethrough ContractngOut 0 20 40 60 80 100 Source: Brazilian Senate Site. Budget information. SIGA System. Calculations by World Bank Staff 4.40 Other effects o f the budget reductions and the restnctive financial releases are the so- called reversal o f pnonties by the sector agencies. The budgetary and financial constraints faced by the sector agencies open the opportumty for justifying changes inthe resource allocations set inthe LOAs. The mnistry's budget proposal is altered several times duringthe budget process - first in the intra-executive negotiations with the Ministry o f Plamng, then by Congress in the LOA approval process, and finally through the discretionary control of spending limts dunng budget execution. As a result, there are significant differences between the composition of investment allocations that was originally planned, authorized in the annual budget, and actually executed. Figure 4.12 Composition of budget allocations and committed resource by type of investment, (averages 2000-05) Others Cintenancethrough Contracting Out Restoration Recovery EmergencyWorks Wintenance Critical hints Himnation Dualicatian PreventiveConservation hmrovemnt 0% 5% 10% 15% 20% 25% 300, 0 Budget m Comdtmnt Source: Brazilian Senate Site. Budget Information. SIGA System. Calculations by World Bank Staff 76 4.41 Physical execution expenences frequent delays mainly due to the irregulanty of resource flows for the payments and seasonal problems that affect physical execution o f works. In a typical year, the first months are almost exclusively dedicated to the payments for the contracts settled inthe previous year. New comtments begin after these settlements are cleared and once the budget ceiling i s somewhat relaxed m d year. This leaves a rather short period of time for actual execution, whch i s also affected by adverse climate conditions (i.e., rains) in the last months of the year. As a consequence, road contractors have a penod of about 6 months to execute their contracts. The irregular flows of payments for the settled contracts lead to frequent interruptions of the physical execution, contract extensions, pnce adjustments, and other additional costs. 4.42 The delays in the physical execution explain gaps between c o m t m e n t and settlement figures. Table 4.11 shows that in December of 2005 only 41 percent of the committed resources directed to road investments were effectively settled, although the level of c o m t t e d resources was unusually high in 2005 because the Pilot Project for Investments protected specific road investment programs from the budgetary freeze. 4.43 The Congressional influence does not end at the budgetingprocess but also extends into the expenditure execution stage. Figure 4.13 shows a similar patternof influence inthe execution phase as in the budget approval stage. Congress guarantees the execution o f its amendments in construction, cntical point elimnation, and restorationprograms as expressedby the highweight of the settled amendments111 relationtotal settlement. Source: Brazilian Senate Site. Budget information. SIGA System. Calculations by World Bank Staff 77 Figure 4.13 Settlement of Congressional amendments inpercentage of total settlement by type of program (averages 2000-2005) Construction Critical Points Elimination Restoration Improvement Total Road Investments Maintenance Other Preventive Conservation Maintenancethrough ContractingOut Duplication EmergencyWorks 0% 25% 50% 75% 100% Source: Brazilian SenateSite.Budget information. SIGA System. Calculationsby World Bank Staff 4.44 Finally, differences can also be observed between the settlement and payment stages. Only 60 percent of the settled road investments are effectively paid during the fiscal year (Figure 4.14). The critical points elimnation, emergency works, maintenance through contracting out, and improvement programs pay less than half of the settlement of their investments. 4.45 The differences between settlement and payment reflect weakness in the financial programng of the sector agencies. Absence of reliable revenue flow estimations withm the year and the uncertainty about the completion o f physical execution generates substantial amount of payments in arrears that in turn disturb the financial execution inthe following year. Infact, there i s a vicious circle. As payments are delayed, providers slow down the physical execution of contracts whch justifies the low proportion of settlement relative to comrrutment. In turn, the flow of revenues dunngthe first months of the year is directed to pay contracts already settled in the previous year, thus contributingto the low comtment levels new cycle. 78 Figure 4.14 Payment in proportionof settlement by type of program (averages 2000-2005) Total Transport Ministry Total roads Others Wiritenanc:e through contractingout Restoration h r g e n c y Works Wintenance Critical hints Himnation Duplication Construction Reventive Conservation hprovemnt 0% 20% 40% 60% 80% 100% Source: Brazilian Senate Site. Budget information. SIGA System. Calculationsby World Bank Staff 4.46 The problems analyzed in each o f the stages generate large differences between budget, the physical execution (settlement), and the financial execution (payment). Infact, in the last five years, less than 50 ercent of budgeted road investments were settled and only 30 percent effectively paid.1O P Figure 4.15 Budget and Expenditure Execution on road investments, 2000-2005 (inbillionof R$of 2004) 6 5 4 3 2 1 0 2000 2001 2002 2003 2004 2005 Source: Brazilian Senate Site. Budgetinformation. SIGA System. Calculationsby World Bank Staff RECENT GOVERNMENT INITIATIVES 4.47 Recognizing that infrastructure bottlenecks, especially in the transport sector, are h a m n g the country's economc growth prospects, the government launched a couple o f irutiatives to I O 3The lack o f availability of annual figures for the PPA avoids the comparison between the multi-annual planned and expenditure execution figures. However, showed above, the large differences between the planning and annual budgeting figures should reveal the existence of broader gaps between planning and expenditure execution than the observed between planning and expenditure execution. 79 increase the level and the efficiency of investments in the transport sector (as well as in a few other infrastructure sectors). One i s an effort to improve execution of selected infrastructure investment projects, most of them in the transport sector, through special treatments of these projects within the budget cycle.Io4 The Public Investment Pilot Project (PPI) 4.48 An important current government initiative to enhanceefficiency of public investments i s the Pilot Project for Infrastructure (PPI) with the programmed annual investments of R$ 10 billion (about 0.15 percent of GDP) over the penod 2005-2007 `05 M a i n innovations of the PPI are (i)protection of the selected investment projects from funding freezes;lo6 (ii)project selection based, in theory at least, on the potential returns of the projects in terms of their effects on growth and fiscal revenue^;'^' and (iii)the introduction of intensive monitoring of the actions under the PPI portfolio. 4.49 The selection of imtial projects under the PPI was based on the following critena: i) increased fiscal revenues or reduction o f future expenditures; ii)maintenance and recovery of strategic infrastructure assets, iii)reduction of logistic bottlenecks; iv) enhancement of the country competitiveness; and, v) progress inthe project execution schedule to see if its inclusion inthe PPIwouldresult inmoreagile execution. 4.50 The sector mnistnes prepared a potential portfolio of projects, which was then screened by the Planning and Finance Mimstries usingthe above-mentioned cntena. htially, 97 strategic projects were selected, 90 of which were in the transport sector - divided among 49 for road recovery, 10 for road improvement and construction, one for waterways and 30 for ports (Table 4.12).Io* IO4 Another notable initiative IS promotion o f private-sector participation in provision o f infrastructure services through public-private partnerships (PPP). This is discussed extensively in a separate report on infrastructure in Brazil "How to revitalize infrastructure investments in Brazil: Public policies for better rivate participation," a forthcoming World Bank report. The PPI originated from a discussion betweenthe Brazilian Government and the International Monetary Fund(IMF) in2004, inwhich the parties agreed to a special arrangement to select, implement and monitor execution o f a set o f priority public investment projects. In particular the Brazilian government and the IMF agreed to exclude these strategic public investment projects from the government's fiscal adjustment targets. I O 6Specifically, inclusion in the PPI guarantees that budget allocations for the selected projects will be excluded from the ministry-by-ministry commitment ceilings (Le., the total budget allocations can y$. Brazil immediately be committed and that settled contracts corresponding to these projects will be automatically has not developed a centrally coordinated process whereby each new proposed project i s systematically scrutinized for its fiscal and economic impacts. As such, enhancement o f government capacities to appraise, execute, monitor and evaluate public infrastructure projects has became one o f the most important objectives o f the PPI. lo* The other projects included were in the areas o f urban transport, irrigation infrastructure, research and development, and improvement o f tax and social security revenuecollection efficiency. 80 GP stand survey 29 0 29 29 0 58 Bio science Lab 11 3 14 14 14 41 Efficiency Payroll tax rate reduction 0 100 100 0 0 100 improv.in Tax integrationwith states 0 60 60 0 0 60 revenue collection TOTAL 1,545 2,820 4,365 3,633 3,609 11,607 4.51 For the purpose of intensive monitonng, the government created a task force across the Mimstnes of Finance and Planning and the Chief Cabinet (Cam Civil, Presidency) for PPI implementation, and signaled the PPI as a high pnority A special module was designed in the PPA Management Information System (SIGPlan''') to closely monitor project execution at the level of each contract and to identify constraints and actions needed to guarantee timely implementationof the projects."' 4.52 The centralization of the PPI project management constitutes an important institutional reform fostered by the PPI. The recognition of the weak institutional capacity of the sector mnistnes and the fiscal content of the PPI has invited the Finance Ministry to oversee the PPI implementation. As a result, a unit was created withn the National Treasury Secretariat (STN) to manage the PPI project implementation inthe short run. Inthe mediumrun, the responsibility of this umt will be to appraise and select public infrastructure projects, to monitor their implementation, and to evaluate them interms of their fiscal and economc impacts."' 4.53 Performanceof the PPI during 2005 was modest. The government's imtial difficulties in identifying projects with high economc and fiscal returns caused a delay in launchng the program. Because the PPI project selection was centralized in the Plamng and Finance Mimstnes, economc cntena (Le., returns) prevailed vu-2-vu the pragmatic cntena of progress in their execution (Le., to pnoritize those projects closer to finalization) as proposed by the sector logSIGPLANis a remote projectmonitoringtool usedby the PlannlngMinistry. 'loThe Ministry of Transport is also building its own system for much closer monitoring of project implementationby the DNIT 'I1CoordenaqgoGeral de Analise EconBmicoFiscalde Projetosde InvestimentoPublico (COAPI). 81 mnistnes. Inthe case of the transport sector, some o f the selected projects had not completed c o m t m e n t phase (Le., projects without contracts) or were having techcal-implementation difficulties. As a result, the first months of implementation saw reallocation of resources among projects and changes to the portfolio composition according to the projects' readiness for irnplementation/completion."* 4.54 Given the imtial delays during the project identification phase and the institutional reorganization to manage the PPI, the implementation was imtiated only in the second part of 2005. Naturally, this was reflected in the low budget execution. While the PPI's initial premse was that low execution of public investments could be overcome by guaranteed fund flows, the gaps betweenthe c o m t m e n t and the settlement figures in 2005 show that resource availability i s only part of the problem (i.e., settlements are delayed even though cash resources are available for payments). Source: PlanningMinistry. PPIAnnual Report 4.55 The recovery projects have shown better performance than the improvement and construction projects. The more extensive use o f results-based contracts for road recovery partially explains the better performance. At the level of individual projects, information until September 2005 reveals that 15 out of the 49 road recovery projects and 11 out of 15 improvement-constructionprojects had a settlement to total budget ratio of zero. That is, the problems on expenditure execution have remained despite the intensive monitonng, although the data for late 2005 indicate acceleration of the execution. SUMMARY 4.56 The steady decline of budget allocations for infrastructure investments has been particularly pronounced in the road transport sector, historically the most important sub- sector for federal public investments. Data reveal a number of dimensions of the sector's poor performance, which constrains the country's growth prospects. 4.57 The decline of the sector's performance has been accompanied by the reduced funding level. This has been associated with the loss of earmarked revenues to finance the expenditures, and the parallel weakening of the sector's institutional foundations. However, the inability of the CIDE earmarked revenue to enhance resource allocations for the transport sector shows that the underlying problem is not necessarily the existence of earmarlung arrangements, but the relative political prioritization. Since the late 1980s' the pnorities have been to finance the social expenditures as detemned in the 1988 Constitution and maintenance 'I2 Two road improvement and construction projects were dropped and 5 others were included, thus increasing the number o f PPI projects 97 to 104. Also one railway project was added. At the program- level, these reallocations meant that 21 out o f 46 road recovery projects suffered budgetary reallocations and 6 out o f 10 improvement and constructionprojects in the road improvement and construction program sufferedbudgetary reallocations. 82 of fiscal responsibility, whch has necessitated control of discretionary spending, most notably capital investments. 4.58 The analysis of the sector's expenditure management reveals a serious disconnect betweenplanning, budgeting and execution of investment projects. Although the government has begun to address t h s problem, especially with the PPI, the implementation expenence of the PPI itself suggests that lack of finding is far from the only, or perhaps not even the most important, constraint to efficient execution of the transport investment portfolio. More detailed diagnosis of investment execution i s neededto identify causes of ths performance systematically However, some of the plausible causes include (i)the sector's weak capacities for project plamng and contract management; (ii)cumbersome and somewhat unpredictable legal environment regarding procurement and contracting and licensing (e.g., environmental); and (iii) intensified external oversight in the face of suspicions and allegations o f corruption surrounding the sector. 83 84 5. CONCLUSIONS 5.1 The previous four chapters discussed major public expenditure issues Brazil faces today, namely, the level o f public expenditures and the size o f the public sector in the economy, the composition o f public expenditures and its growth effects, the sources and the nature o f budget rigidity and the efficiency and effectiveness o f expenditure management in the roads transport sector as an example o f a critical infrastructure sector. These analyses will need to be complemented with detailed reviews o f the level and the quality o f expenditures inother specific sectors as well as assessments o f the public sector's institutional arrangements and capacities for managing public expenditures. Leaving those tasks for other reports the Bank or others may prepare, we offer in t h s concluding chapter a summary o f our findings and suggest possible measures for Brazil to regain better control over the level and the composition o f its public expenditures. MAINFINDINGS OFTHEPREVIOUSCHAPTERS 5.2 Since the 1970s, the size of the Brazilian state has increased steadily Although it i s difficult to establish an optimal size o f the state, it i s clear that the fiscal size o f the Brazilian state i s larger than any comparable country inthe world. The analysis indicates that t h s i s largely due to the on-budget activities o f the general government sector. Since 1998, the total public spending in Brazil, including interest payments, has reached above 40 percent of GDP Ths i s well-above the levels o f public spending in rmddle-income countries, and even higher than some o f the low-spending OECD countnes. Figure 5.1 Total Nominal Spending and Tax Burden, 1995-2004 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 1 UTax burden mTotal nominal srxndina I Source: Ministry of Finance: National Treasury Secretanat. 5.3 The role o f the state in the economy i s not measured only in terms o f its budgetary expenditures and tax burden. We therefore measured the extent o f the state's presence in the Brazilian economy through extra-budgetary activities. The extent of off-budget activities via tax expenditures (around 2.9 percent of GDP) and quasi-fiscal activities by public sector banks (around 0.6 percent of GDP) are relatively limited. Although we don't measure quasi- fiscal activities directly, it i s remarkable that the size of activities by non-financial public 85 enterprises remain quite extensive despite the privatizations in the 1990s. Even after discounting the petroleum sector (Le., the Petrobras group) on the assumption that their operations are largely commercial in nature, the expenditures by the remaining non-financial public enterprises reach around 9 percent of GDP. 5.4 As observed by a number o f Brazilian analysts,*I3 the main driver of the fiscal expansion is the constantly increasing current expenditures, especially social obligations enshrined in the 1988 Constitution and interest payments. Both are non-discretionary expenditure items that are difficult to reduce inthe short run. Federalismhas also encouraged the fiscal expansion. On the one hand, decentralization o f expenditure responsibilities (mainly inthe social sector) led to transfer o f resources to states and especially to municipalities over the 1990s. Instead o f reducing federal expenditures and revenues to offset the increased sub-national spending, however, this trend has been compensated by the federal government's fiscal adjustment efforts based on revenue enhancement, which have led to increases in those federal revenues that are not sharable with states and municipalities. 5.5 Growth effects o f expenditure composition were e x m n e d with an econometnc analysis and a simulation exercise usingrealistic parameters for Brazil. Both exercises reach consistent findings that growth effects of Brazil's public spending is expected to increase with greater allocations to capital investments in economic infrastructure and that the marginal growth impact of additional tax revenues or borrowing to finance higher spending is negative (Le., increasing the size o f the state reduces the country's growth potential). 5.6 Policy implications from these two sets of analyses are clear. First, Brazil would benefit from controlling, or if possible, reducing its level of public spending so as to reduce the tax burden. This means that even if the additional expenditures were themselves of good quality (e.g., it contributes to growth or income distribution), the positive effect would be attenuated by the negative effect of additional taxation needed to finance the growing public expenditures. Similarly, the simulation exercise shows that if larger public investments are financed with borrowing, their positive growth effects will dirmnish over time as the higher debt burdentranslates into other constraints such as higher interest rates. 5.7 A second implication is that the expenditure composition should be shifted to prioritize capital investments. However, since raising taxes or borrowing to finance increased capital investments constrains growth, additional funding for capital investments must come from internal re-allocation of the available resources. This i s a difficult task in any country, but the challenges are arguably more daunting in Brazil because o f the high degree o f budget rigidity 5.8 Brazil's recent track record in fiscal adjustment shows the practical difficulty o f prioritizing public investments. While at the aggregate level (i-e,, achevement o f a primary surplus), Brazil's fiscal adjustment so far has been a success. But the fiscal turnaround was achieved trough increasing revenues and at the expense of cutting capital expenditures. Total revenues increased by 25 percent, which accounted for 222 percent o f the adjustment, while total expenditures growth o f 14 percent offset some o f t h s effect, and dampened the adjustment effects by 122 percent. Decomposition o f the change in expenditures shows that the main "culprits" are current expenditures and intergovernmental transfers, which grew 32 percent and 30 percent, respectively, having contributed negatively to the adjustment with 81 percent and 43 percent, respectively In contrast, investment expenditures fell 13 percent and had a positive 'I3See Velloso (2004) or GiambiagiandRonci (2004). 86 contribution to the fiscal adjustment o f 5 percent.Il4 For Brazil to improve the growth impact o f its expenditure policy, it i s therefore necessary to reverse t h s pattern o f fiscal adjustment, which i s obviously not an easy task, not least because t h s would imply addressing the structural problem o f budget ngidity Table 5.1: Federal iovernmentPrimary Surplus Changes, 1995 2005 - Annual Averages (YO of GDP) Variation 95/98 99/05 - 1995- Percentage to the ITotalRevenue 18.6 23.2 4.6 24.6 222 Treasury Revenue 13.6 18.0 4.3 31.8 210 Tax Revenue 12.0 16.6 4.5 37.6 219 Other Treasury Revenues 1.6 1.4 -0.2 -12.0 -9 Social Security Revenue 5.1 5.3 0.2 3.3 8 I1Total Expenditure 17.6 20.1 2.5 14.3 -121 Personneland Social Contributions Social Security Benefits Other Current Expenditures Subsidies FAT Goods and Services and Assistance Benefits Investments Intergovernmental Transfers PrimaryBalance (I - 11) Source: Ministry of Finance: NationalTreasury Secretanat. Calculationsby World Bank staff 5.9 Brazil's budget rigidity has two inter-related sources: the revenue side through earmarking and the expenditure side through legal designation of certain expenditures as obligatory The government has tned to reduce ngidity on both sides, with limted effects so far. On the revenue side, portions o f the earmarked revenues have been "de-earmarked" in an arrangement now called DRU(Desvinculapk de Receitas da Unia`o). While t h s measure allows the government to apply part o f the de-earmarked revenues for rumng the primary surplus, it has limted impact on the reduction o f total spending or on the reallocation o f budgetary resources from current to capital expenditures as the de-earmarked resources needs to finance mandatory expenditures. Infact, the government has sort o f a "deficit" between the earmarked revenues and the non-discretionary expenditures (Le,, the sum total o f the non-discretionary expenditures are larger than the sum total o f earmarked revenues). So long as a significant portion o f the expenditures are legally and constitutionally mandated, the government will continue obligated to `I4 terms of functions, the sectors related to economic affairs that include mostly infrastructure sectors In carried the great part o f the adjustment. On the contrary, social security and health functions were not only protectedbut also had a negative contribution to the fiscal effort. 87 fund them no matter whether the revenues are earmarked for them. There is thus no alternative but to try to reduce the ngidity on the expenditure side. Ths would require a series o f legal and constitutionalreforms, even though these are techmcally complex and politically controversial. 5.10 To address the expenditure-side rigidity, there have been repeated attempts to reform the social security systems for retired workers from both public and private sectors. The succession o f reforms adopted since the late 1990s have improved fiscal sustainability o f the social secunty systems, mainly by tightening the eligibility cntena. However, effects o f these measures have not been sufficient to elimnate the social secunty deficit. They have also been offset by parallel increases in the mnimum wage to which the mnimum level o f social security benefit (as well as a number o f other social obligation expenditures) i s linked. The adrmnistrative reform to increase flexibility o f the personnel expenditures, an important source o f rigidity and a threat to fiscal sustainability at the sub-national levels in particular, has stalled since the late 1990s. On the other hand, there i s constant pressure to increase expenditure rigidity through proposed creations o f new non-discretionary programs. Each o f these may represent an important and worthwhile public policy intervention, but their overall fiscal impacts could be negative to the extent they add to the budget rigidity 5.11 The economic importance of increased capital expenditures (typically public investment in economic infrastructure) does not automatically mean Brazil will immediately benefit from a larger allocation of budgetary resources to infrastructure sectors. Unless the public sector i s capable o f designing economcally viable projects and executing them efficiently, capital spending does not necessarily translate into formation o f high-return economc assets and thus does not necessarily contribute to hgher and sustained growth. The review o f the expenditure management in the road transport sub-sector indicates that the institutional weaknesses in p l a m n g and executing public investments are rather senous in Brazil. It i s not clear whether, in a hypothetical situation o f windfall revenues, for example, the infrastructure sectors could efficiently absorb a sudden and dramatic increase in the budget allocation to bnng about greater growth impacts. Rather, it seems advisable that Brazil invest in developing institutional capacities for investment planning and execution, while gradually increasing allocation to the sector, should additional resources be made available. POLICYIMPLICATIONS 5.12 The analyses presented inthis report lead to the following set o f policy recommendations. First, the growth of the non-discretionary expenditures needs to be controlled, and if possible, reversed. Second, gradual increases in the budget allocation to economic infrastructure seem desirable. Third, increased allocation to economic infrastructure should be accompanied by systematic efforts to strengthen institutional capacities for planning and execution of public investments. It i s not clear whether there i s a solid consensus among Brazilian experts about these policy implications. A number o f them do advocate these and other measures to reconfigure the country's public finance structure. But many also seem to disagree, for example, by calling for hgher social spending without offsetting it with reduction elsewhere (except perhaps interest payments, which are not within a given government's discretionary control). 'l5 For example, the paper by the Economic Policy Secretariat at the Finance Ministry (2003) on the measurement and adequacy o f federal government social spending provoked a strong reaction by social sectors and debates within the own government. 88 5.13 Given the apparent absence of a solid consensus, a first step must be construction of a political agreement among key actors as to the direction of the country's fiscal policy In- depth debates about the need for and the desrable content o f specific reform measures such as social secunty reform are absolutely necessary. But, it may also be beneficial if the country had an institutional arrangement that fostered such policy debates with bothmacro-fiscal and sectoral implications of a given policy proposal taken into account. Once in place, such a framework would allow policy-makers to manage inevitable fiscal and policy trade-offs in as rational a manner as can be expected. The following paragraphs offer some concrete suggestions about how such an institutional framework may bebuilt inBrazil over the mediumterm. Institutional arrangement for facilitating budget rationalization 5.14 Containment of the growth of non-discretionary expenditures no doubt means continuing to deepen those structural fiscal reforms that have long been on the government's reform agenda in Brazil. These include further reforms of the social security system, especially for private employees (INSS), rationalization of other social obligation (entitlement) programs such as those defined in the Organic Law of Social Assistance (LOAS), and admnistrative reforms to allow governments to adjust their staffing levels and compositions with greater flexibility As has often been suggested, de-linlung specific social benefits (e.g., social secunty payments) from the mnimum wage would bnng about far-reaching effects on long-term viability o f the government budget by limtmg unsustainable fiscal impacts of what i s essentially a labor-market policy decision.Il6 5.15 Addressing specific technical details of each of these reforms would be beyond the scope of this report. It suffices to note that these are necessarily controversial reforms, many of which are seen as "conquests" achieved by groups in society seelung more progressive social policies since the 1988 Constitution. Some of them are inefficient yet effective in achieving certain sector-specific policy goals. Many of them by now have vested interests that current benefit from their continuation and would therefore resist any effort to reform them. Therefore, a necessary and inevitable first step is to have broad political discussions about the merits and demerits of these policies, including re-examnations of the relevance of their stated goals, their efficacy in achieving them, and their medium-to long-term fiscal implications. 5.16 Perhaps because of the politically controversial nature of these reforms, governments tend to address them one by one instead of in a bundle. This incremental approach is perfectly reasonable from the point of view of handling the politics of reform. It would also help the government ensure that unforeseen effects of the reforms remain manageable (e.g., a reduction in tax collection after a tax reform that broadens the tax base and lowers the tax rate); A downside, however, is that cumulative effects (especially in terms of fiscal impacts) o f a whole set of reforms are rarely taken into account when one reform is considered in isolation from the others (e.g., the eligibility criteria for social secunty benefits may be tightened while the mnimumwage increase offsets fiscal savings from the first measure, and possibly worsens the equity impacts of the policy package - fewer beneficianes receiving more - at the same time). Thus a question is `I6 The Bank's own report on labor issues Brazil - Jobs Report (Report No. 24408-BR, 2002) recommended not to de-link minimum social security benefit from the minimumwage. But t h s argument was made from the point o f view o f the pension program's effectiveness in lowering old age poverty and not o f the overall fiscal sustainability o f expenditure programs. The position o f the current report is that a decision must be guided by the direct intended effect of the policy inits sector (e.g., poverty reduction inaspecific segment ofthe population) andthe program's overall fiscal impacts. A mainproblem with the automatic linlung o f the minimum wage to minimumprogram benefits i s that it encourages policy-makers to decide on benefit increases without considering the fiscal effect. 89 whether an institutional arrangement can be devised so that whenever policy decisions with significant and lasting fiscal consequences are considered, these can be debated with their impacts on the overall fiscal balance and sustainability taken into account. 5.17 With the above-cited concern in mnd, the focus o f t h s section is to explore how the institutions o f planning and budgeting can be improved to guide and restrainpolicy decisions that can lead to net increases inthe level o f non-discretionary spending, or to find a way o f offsetting the effects o f such decisions if they cadshould not be avoided for non-budgetary reasons (i.e., increasing a particular social benefits i s deemed to be a politically desirable end). What i s important is to ensure that the government can make policy decisions (e.g., increasing social benefits) with adequate considerations about their fiscal impacts so that neither expected benefits of a specific policy decision or the overall fiscal sustainability are not undermined by negative budgetary impactsthat may result from it. 5.18 Therefore the question i s whether a separate budgetary or legal instrument can be devised to discipline such decisions. Such an instrument should possess at least the following qualities: (i) establish credible targets for levels o f expenditures, less as a bindingceiling than as a means to tngger a well-informed policy debate among policy-makers when any policy decision i s likely to threaten achievement o f these pre-established expenditure targets; (ii) serve as a framework with which budgetary impacts o f policy decisions can be simulated; and (iii)provide sufficiently detailed information regardinghow expenditure level and composition should be adjusted. 5.19 Expenences o f OECD countnes that have gone through expenditure control and re- allocations under fiscal stress show that such reforms were aided by an institutional arrangement that helped the government to know whether corrections were needed inthe overall balance and the composition o f the budget over the medium term. The most common instruments that serve this purpose are a medium-term expenditure framework, combined with a budget rule that requires an adjustment. Once the need for budget adjustment i s recogmzed, governments can conduct program reviews, typically a centrally-led process o f expenditure scrutiny with a view to identifying savings options, in order to identify which specific spending items couldshould be rationalized.] 5 -20 Medium-term expenditureframework: A medium-term expenditure framework (MTEF) specifies medium-term targets for total revenues, total expenditures, the deficit, and sometimes disaggregated components o f the expenditure budget. The framework typically covers 3 to 5 years, including the upcomng budget year, and may be adjusted on an annual basis by adding one more (outer) year (Le,, "rolling" MTEF). It may also be done penodically without "rolling" (e.g., revised every other year). Whether rolling or periodic, an MTEF could be either binding (i-e., expenditure levels for future years must be consistent with those projected in the framework) or indicative (i.e., the framework simply indicates what the levels o f expenditure aggregates would be without any other policy change).'" An MTEF aids the government effort to keep its expenditure level withn a targeted level because it makes medium-term fiscal unpacts o f policy decisions explicit and makes t h s information available dunng the budget p l a m n g process. However, this presupposes that both the medium-term macroeconomc forecast and the projection `I7"Reallocation: The Role o f Budget Institutions," OECD2005. Intheory, a strictly binding framework may not be fully consistent with a fxed primary balance target as used in Brazil since the government will be kept from adjusting its expenditure level when the revenue falls due to economic downturn. In practice, however, the highly rigid nature o f the Brazilian budget means that, at least in the short run, a strictly binding MTEF would not significantly worsen the government's ability to adjust its fiscal vanables to achieve its primary surplus target. 90 o f the key fiscal variables as well as costmg o f the expenditure elements are technically robust and politically credible. 5.21 Inthe case of Brazil, the PPA is a vanant of an MTEF, especially now that the Mimstry o f P l a m n g prepares an indicative rolling plan along with its submssion o f the annual evaluation report. However, the PPA in its current form does not quite fulfill the signaling role a well- developed MTEF plays for medium-term expenditure adjustments. On the one hand, the PPA contains a hgh level o f program-level details w h c h are not strictly necessary for strategic budget decision-malung o f the lunddiscussed here. The PPA treats all programs as ifthey were equal in terms o f their economc significance. It certainly does not distinguishbetween discretionary and non-discretionary, or even between current and capital expenditures, and offer medium-term projection o f each type. From the point o f view o f medium-term fiscal control, a critical task would be to project expected evolution o f the total revenue and the non-discretionary expenditures. Inthe current format, the PPA does not provide such information. 5.22 Besides the PPA, the Fiscal Responsibility Law (LW) requires that the government include an annex on annual fiscal targets to its LDO bill. The main fiscal targets are expressed in terms o f pnmary surpluses for 3 years including the budget year. The annex discusses implications o f a given level o f primary surplus target for other key fiscal variables, namely, the ratio o f net consolidated debt to GDP, based on certain macroeconomc assumptions (e.g., GDP growth rates, interest rate, exchange rate). LDO also includes a long-term projection o f the fiscal impacts o f the social assistance programs specified in the Organic Law o f Social Assistance (LOAS), but the annex does not situate the projected increases in LOAS spending within an overall budgetary context (Le., vis-a-vis projected revenue envelopes and projected increases in other non-discretionary expenditures), which i s necessary to assess affordability o f the increases. A separate annex estimates the "margin" for expansion of "continuous" non-discretionary expenditures. This estimate is intended to assure that "permanent" current expenditures - i.e., those created by law, decree or other admnistrative norms to be executed for more than two budget years - are backed up by corresponding increases in "permanent" revenues - i.e., those revenue increases based on increased rates, broademng o f the revenue base, or creation o f new revenue. However, the annex offers estimates only for the upcomng budget year, and thus lacks a medium-term framework. 5.23 It is therefore recommendable to review the roles of the PPA and the LDO as instruments of medium-termbudget management. There are at least two distinct paths for Brazil to develop a robust MTEFbased on the existing instruments. One option would be to turn the PPA into a penodic MTEF The estimates o f the revenues and the key expenditure aggregates (e.g., personnel, non-discretionary social benefits, constitutional transfers) can be projected on an annual basis for each o f the four years that the Plan covers, and the Plan, or at least these key revenue and expenditure aggregates, can be updated every other year, takmg into account revised macroeconomc estimates and effects o f any policy change that may have been registered in the intenm. Since the "updated" PPA would cover the 2ndand the 3" year o f the next admnistration, the projection should be seen as indicative rather than legally bindingfor the obvious reason that the current admnistration cannot legally c o m t the next one to a particular set o f policies. The objective i s simply to make explicit fiscal implications o f the existmg set o f policies and expected macroeconomc framework, without creating any legal obligation for the government to follow exactly the same set o f policies "left behnd" by the previous admnistration. 5.24 An alternative would be to develop the LDO into a rollingMTEF,by further polishing its annexes. The government could set as an explicit fiscal target not only achevement o f a certain level of primary surplus, but also levels o f expenditures, both the total level and certain 91 disaggregates. Since LDOs are annually passed, each year a new LDO would update the information, and add one more outer year in the framework. The level of expenditure disaggregates to be reported in the LDO will depend on what Brazil chooses to monitor for the purpose of medium-term budget management. It certainly would not be necessary to monitor each one of the specific budget category in the annual budget laws and the current PPA. A mnimum approach would be to emulate Chile's practice of projecting non-discretionary expenditures and the total current revenues in order to estimate the fiscal space for additional discretionary spending (for either new programs or expansion of existing program^)."^ Estimating the fiscal space in ths fashlon would be valuable, but would not fully serve the purpose of controlling the level of non-discretionary expenditures. For this, it would be necessary to monitor the expected evolution of key non-discretionary spending items such as personnel, social security benefits, and so on, as well as projected comtments for ongoing capital investment projects. 5.25 Between the two alternative paths, our recommendation would be to use the LDOs as a basis for developing an MTEF As detailed later, it would seem best to treat the PPA as a strategic planning document that guides expenditure pnontization, especially of the discretionary portion of the budget, by each admnistration. 5.26 Budgetrules: Once an MTEFmakes the projected levels and composition of spending explicit, the government, together with Congress, could voluntarily choose to control the observed pattern of expenditure evolution with its own discretionary judgment. However, there always remains a fear that a future government may be less c o m t t e d to fiscal discipline and improved quality of public expenditures. A common way to try to l i m t the discretion of future governments i s a fiscal or budget rule. 5.27 Inthe case of Brazil, some of the requirements of the LFW and the content of the LDO already provide fiscal rules. For its part, the LFW and its associated Senate resolutions establish ceilings for personnel expenditures and the level of net consolidated debt, both as a share of net current revenues. These LRF provisions are important in promoting sustainability of the government accounts, and represent major advancement in the Brazilian fiscal legislation. The main objective is clearly to avoid uncontrolled accumulation o f public debts through deficit financing of public expenditures. As significant as they are, these specific provisions are still insufficient if the objective were to restrain the growth of expenditures since they specify the fiscal targets as a share of revenues. As has been demonstrated with Brazil's fiscal management in recent years, a positive fiscal balance can be obtained by increasing revenues rather than contaiung the growth of expenditures. 5.28 A number of states in the United States, for example, have adopted expenditure rules, with apparently positive effects in controlling the growth of taxation and expenditures, although findings from available academc studies are not completely These rules l i m t the level of total taxes and/or total spending to a percentage of some measure of income (e.g., GDP), or restnct the growth of spending to that of either the population or inflation. The articles inthe LDO bill 2006 that were intended to limt the levels of federal taxation and expenditures to 16 `I9Chile makes these additional revenues available to ministries on a competitive basis through a "bidding fund" (Fondo Concursable). Ministries can present proposals for either new programs or expansion of existing ones following a standard format set by the Ministry o f Finance. The bidding fund is made available only when the fiscal projection shows such fiscal space. See Robert Krol (1997) for support for the thesis that budget rules matter (although this i s based on a survey of existing literature), andPenner and Weisner (2001) for a report of inconclusive evidence. 92 percent and 17 percent, respectively, o f GDP are a vanant o f t h s tax and expenditure limtations (TEL).12' 5.29 Whether Brazil will really benefit from an explicit TEL i s a difficult question. On the one hand, those LRFrules - the level o f personnel expenditures and the level o f net consolidated debt - seem to be effective in restraining government behavior. However, the apparent effectiveness o f these rules may depend on a host o f factors, some o f which lie outside the LFW itself. For example, the fact that these expenditure and debt ceilings are expressed interms o f a share o f the net current revenues means that the governments can meet the required levels by either reducing the numerator (personnel expenditures or net consolidated debt) or by increasing the denomnator (net current revenues). This gives the governments somewhat greater flexibility The rules themselves are not excessively rigid. Those governments that do not meet the ceilings are given a reasonable timeframe for meeting them eventually. This helps in maintaimng the credibility o f the rules. Most importantly, the enforcement o f the rules is backed up by the bilateral fiscal adjustment programs the National Treasury has with each o f the states. These programs actually predate the LRF Finally, the rules were defined in such a way that basically exempted the federal government, for which there is no external enforcer that can play the role o f the National Treasury v i s - h i s the states and municipalities. If, however, Brazil were to adopt a TEL, the main target will have to be the federal govemment whose growing tax and spending levels are the main sources o f concern. Enforceability of a TEL in such a context would be somewhat suspect.'22 5.30 These considerations raise questions about practical viability o f a rigid TEL for Brazil. It i s conceivable that, once explicit TEL targets are set as government c o m t m e n t , the international financial market will begin to pay more attention to the levels o f taxation and spending, and thus serve as a de facto enforcer. But, defirung the TEL as a percentage o f GDP, as was done in the 2006 LDO bill, could be nsky in that in case o f a sharp economc decline, the rules mght have to be violated and thus lose credibility (because the budget rigidity will not allow the govemment to reduce expenditure sharply to meet the target). 5.31 Of course, the rule could be expressed in a more flexible manner. For example, it could require that the govemment present a credible medium-tern budget strategy (Le., MTEF) and c o m t to limting the growth o f taxation and expenditures within a targeted level over time. If the government fails to meet the targeted levels or to take appropnate measures in the right direction, it may then be requiredto revise the strategy and submt it for Congressional approval. If these budget strategies become a credible instrument for public debates, discussions of these could generate necessary debates on fiscal reforms, including those controversial Constitutional In recognition of the risk of continuously nsing tax burden, which in turn has been pushed up by the ever-increasing current expenditures, the government introduced an innovative measure in the 2006 LDO bill (submitted in 2005). In addition to the by-now standard fiscal target of primary surplus, the government included new articles that set ceilings for the federal government's revenue and expenditure estimates in the annual budget law as a share o f GDP (16 percent for tax revenue and 17 percent for current primary expenditures). The ceiling is applicable only to those revenues collected by the Federal Revenue Secretariat, meaning social security contributions collected by the Ministry o f Social Security are not included. Unfortunately the proposal did not survive the congressional scrutiny without an amendment that basically undermined the original intent o f the proposal, and the Executive decided not to include the same or a smilar proposal in the LDO proposal for 2007 The 29" Constitutional Amendment that set the rules about the minimum level o f health spending is based on the same logic, except that m this case the rule s ecifies a floor rather than a ceiling. The fate o f the TEL introduced in the LDO 2006 bill is a telling example of the limitation of such an approach intoday's political context inBrazil. 93 reforms, to make the budget more flexible and sustainable. Therefore, with or without a binding rule, it would be advisable for Brazil to adopt a medium-term budget strategy that contains an explicit TEL target. As mentionedabove, an LDO annex would be a possible mstrument for the government to present such abudget strategy on a regular basis. 5.32 A relatedmeasurethat couldtightenthe budget process is a formal adoption of a two-step budgetingprocess whereby the Congress approves the revenue envelope, and perhaps ceilings for certain expenditure aggregates (e.g., non-discretionary spending) first (e.g., with the approval of the LDO), and detailed appropnations later (Le., with the approval of the LOA). Ths would be a measure to control the oft-mentioned problems o f budget "inflation" whereby the Congress increases estimated revenues in order to accommodate a higher level of discretionary spending than would be possible under the Executive's typically more conservative estimates. There has been a controversy between the two branches regardingthe accuracy of their respective revenue estimates. One way of resolving ths dispute and making the estimates more credible would be to form a techcal c o m s s i o n with representation from both branches, and possibly from outside the government, to establish a common macroeconomc and revenue framework for LDOs. This comssion should have no role in mcro-level expenditure allocations, whch should be reserved for the secondphase o f the budget process. The medium-term budget strategy as an LDO annex as proposed above would then be discussed as part of ths "first phase" of the two-step budget approval process that focuses on agreeing on the key budget aggregates. Re-allocatingresources 5.33 Suppose that Brazil somehow managed to control the current trend of constant revenue and expenditure increases. Our analysis indicates that without changing the expenditure composition more in favor of investments in economic infrastructure, the budget's contributionto long-term growth will still be limited. Changingthe expenditure composition inthe current Brazilian context inevitably means re-allocatingresources away from parts of the social sector that presently receives a lion's share of the total revenues. Once again, ths could only be possible with constitutional reforms that lift the finding protection for these expenditure items or reduce their levels. 5.34 Even if Brazil suddenly had full budget flexibility, however, it is not clear there is sufficient information to detemne exactly where the budget should be cut and where it shouldbe preserved or increased. An institutionalized process for generating mcro-level information about savings options would therefore be a useful addition to Brazil's plamng and budgeting process. The growing and widespread interests in program monitonng and evaluations (M&E) and the incipient imtiatives to develop an M&E system are welcome developments inthis regard. Well- conducted M&E would no doubt contribute to improving the quality o f public expenditures through improved program design and implementation arrangements. By its nature, however, program M&E would not be sufficient to dnve a system-wide resource re-allocation exercise. Typically program evaluations cover only selected programs and review their quality in depth - although the depth can vary according to a vanety of considerations. While in some cases, evaluations can lead to savings by recommending cancellation of program components, or even elimnation of the program as a whole, they are not the best instruments for resource re-allocation decisions simply because they tend not to assess the merit of a given program relative to the others. Ths is why countries such as Canada, whch does have a system of ongoing program evaluations, has had to resort to ad hoc, centrally-driven program reviews with the explicit aim of identifying savings options when it was faced with the need to conduct major expenditure rationalizations (e.g., 1994). 94 5.35 Expenditure reviews: Followingthe OECD's defimtion o f program reviews (ZOOS), we define expenditure reviews, different from program evaluations, as "a specific form o f policy evaluation characterized by the fact that it i s imtiated and supervised from the centre o f the government, namely the budget office or the office o f the prime mmster/president" with the main purpose o f guiding expenditure rationalization decisions. A few countries, such as the Netherlands and the UK, have regularized such reviews as part o f a medium-term expenditure p l a m n g cycle. 5.36 Expenditure reviews as defined here would scrutimze expenditure programs' rationale and allocative efficiency by aslung questions such as: (i)should the government be involved in this particular business/activity (Le., should the task be privatized, contracted out, or decentralized)? (ii)does the program use an appropriate m x o f inputs to achieve its stated objectives? (iii) what would happen to program delivery and impact if the funding were reduced by a certain amount? On an annual or bi-annual basis, each mnistry should be directed to conduct a comprehensive review o f its program portfolio. The Mimstry o f P l a m n g would coordinate the process, and depending on the targets established in the medium-term budget strategy, mght assign specific savings targets by mnistry. 5.37 For Brazil, one possibility would be to turn the current PPA annual evaluations, which basically are intended to provide public information about certain aspects of program performance, to an exercise (annual or bi-annual) with an explicit aim of expenditure rationalization. The process is already centrally-dnven and does feed into the annual PPA re-programmmg and budget formulation. But the format o f the PPA annual evaluations i s not focused on expenditure rationalization (e.g., it does not ask whether a given program's objective would be better pursued by other means), and they are not linked to any medium-term budget strategy. These mnistenal program reviews could form the basis o f the mnistnes' budget proposals, which would then be consistent with the government's overall budget strategy. Periodic reviews o f the mnistnes' program portfolio may need to be supplemented by cross-cutting reviews o f expenditure items, such as personnel expenditures, grants given to NGOs, etc., to be conducted by the Mimstryo f Plamng. Strengthening expendituremanagementinthe infrastructuresector 5.38 Our review o f expenditure management inthe road transport sector reveals considerable needs for institutional strengthening. While we did not examne other infrastructure sectors, it i s likely that the situation i s not muchbetter. The government clearly sees infrastructure investment as a pnority (e.g., as evident in the prionty accorded to the PPI), but it lacks a clear strategy for improving the infrastructure mnistnes' institutional capabilitles to manage the limited resources efficiently and effectively. Such a strategy may involve the following elements: (i) a medium- term infrastructure investment strategy; (ii) a set o f performance indicators to momtor efficiency and effectiveness o f the infrastructure portfolio as a whole as well as by sector and sub-sector; and (iii) an organizational reform, including adoption o f a strategic human resource policy to enhance the professional quality o f the staff workmg ~fl the infrastructure sector and to streamline the mnistnes' business processes and orgamzational structures. 5.39 Medium-term infrastructure strategy: Strategies for infrastructure investments could be done on a sector-by-sector basis or for the government as a whole. Ideally, each infrastructure mnistry should develop its own sectoral strategy with some central coordination to foster inter- sectoral synergy and avoid inconsistency In the short run, however, it may not be realistic to expect infrastructure mnistnes to develop robust investment plans given the uneven and generally weak capacities o f the infrastructure mnistnes. Inaddition, given the importance o f 95 ensunng consistency between the infrastructure strategy and the medium-termbudget strategy, it may be best to assign the responsibility of preparing this strategy to the Mimstry of Plannmg. The government should simultaneously initiate a medium-term plan to strengthen infrastructure mnistnes' institutional capacities with a view to gradually transfemng the sector planning function from the Ministryof Plamng to the corresponding sector mnistnes. 5.40 One option i s to hghlight an infrastructure strategy as a special chapter o f the PPA, in a way similar to the PPA 1996-99 Brazil in Action that focused on a relatively small number o f priority investment projects. The Mimstry of Plamng also has the pnor expenence with the Eixos study, from which a long-term infrastructure investment planwas derived and which served as a basis for the PPA 2000-03. Representatives o f the relevant sectors as well as the new unit created with the National Treasury to help manage priority public investment projects under the PPI could also be part of preparingt h s strategy under coordination by the Mimstry of Plamng. Ths strategy could be anoutgrowthof the current PPI. 5.41 Infrastructure performance indicators: Each mnistry should continue to develop and refine indicators to monitor performance of each o f its programs. But from the point of view of strategic management of the government-wide infrastructure portfolio, it would be desirable for the Mimstry of Planning, in collaboration with the sector mnistries, to develop and monitor certain high-level indicators. For example, it shouldbe of interest to the government to track the rate at whch the existing portfolio of infrastructureprojects i s being executed, its overall as well as program-by-programcosts, and sector-by-sector infrastructure needs as anupper boundary or a long-term target to guide the sector-by-sector investment strategies. 5.42 A basic objective is to provide policy-makers and society with objective measureof how government spending i s (or is not) contributing to achieving policy goals in the infrastructure sector. The PPA currently includes indicators that measure physical progress of specific infrastructure programs (e.g., X km of roads constructed), but these indicators do not show whether and how construction of specific segments of road (or whatever) actually farther the sector's policy objectives. Without such information, it would be difficult for both the sector mnistnes themselves and the central economc mnistnes to devise a meaningful medium-term budget strategy (on what basis should the government fund the infrastructure sector and by how much?) and monitor the sector's performance (how well i s the sector doing with the amount o f resources made available to it?). 5.43 Organizational reforms: The government should conduct a review of the human resources in the infrastructure mnistnes in order to assess whether the current pool of skills matches the requirements of better performance. Particularly cntical are areas of plamng and project management. The a l m n g decline in the DNEWDNIT's staffing level over time needs to be reversed ifthe prioritization of the infrastructuresector is to become operational reality For this, the government may need to devise ways to make a career in infrastructure management attractive for aspiring professionals. This may imply creation of a new specialized career, or simply application of the existing career of public policy management specialists preferentially in the infrastructure sector through carefully planned strategic recruitments over several years. 5.44 Finally, certain legislative reforms may be in order to allow more agile execution of public works. Possible areas for review include the procurement law and environmental regulations. Obviously, care must be taken inreviewing these pieces of legislation so as to ensure adequate balance i s maintainedbetween admnistrative agility and efficiency on the one hand and fiduciary control onthe other. 96 REFERENCES Afonso, J. and R.Varsano (2004) "Reforma Tributana: Sonhos e Frustraqdes". In:Giambiagi, F., J.G. Reis and A.Urani (eds.) "Reformas no Brasil: Balanqo e Agenda" Editora Nova Fronteira, Rio de Janeiro. Aschauer, D. (1989) "Is public expenditure productive?" Journal of Monetary Economcs 23(2):177-200. Atlunson, A.B and J.S. Stiglitz (1980) Lectures inpublic Economcs. McGraw-Hill. London. Atlunson, P and P van der Noord (2001) "Managzng public expenditure: some emerging policy issues and aframework for analysis 'I,OECD Economc Department Worlung papers No 285. Banco Central do Brasil (BCB): www.bcb.nov.br Barro, R. and X. Sala-i-Martin(1995) Economc Growth. McGraw-Hill. London. Bevilaqua, A. and R. Werneck (2000) "Delaying public sector reforms: post stabilization fiscal strains in Brazil" In: Talvi, E. and C. Vegh "Como armar o rompecabezas fiscal? Nuevos indicadores de sustentabilidad". IADB.Washington, DC. Bevilaqua, A. (2000) "State-government bailouts in Brazil" Department of Economcs PUC-Rio de Janeiro. Discussionpaper No 421. Blanco, F and S. Herrera "The Quality of Fiscal Adjustment inBrazil 1999-2003" World Bank, mmeo. Bleaney, M,N.Gemmell and R.Kneller (1999) "Fiscal Policy and growth: evidence from OECD countries" Journal of Public Economcs 74: 171-90. Bleaney, M.,N.Gemmell and R. Kneller (2001) "Testing the endogenous growth model: public expenditure, taxation, and growth over the longrun" Canadian Journal of Economics, 34: 36-57. Brazil Annual BudgetLaw (2004). Canning, D. (1999) "Infrastructure's contribution to aggregate output". World Bank. Policy ResearchWorlung PaperNo 2246. Paper for World Bank Country Economc Memorandum, Brazil - Stability for growth and poverty Cardoso, E. (2002) "Real Interest Rates, Bank Spreads and Financial Sector Issues. Background" reduction. Report No 25278-BR. Washington DC. Castro, J. and F Sadeck (2003) "Financiamento do gasto em educaqgo das trCs esferas de govern0 em 2000". IPEA.Texto para discussiio No. 955. Castro, J., M.B.Moraes, B.C. Duarte, F Sadeck and H. Simdes (2003) "Analise da evoluqiio e dinamca do gasto social federal: 1995-2001" IPEA. Texto para discussiio No. 955. Comnetti, R. and G. Ruiz (1996) "Evolucih del gasto p6blico social en Amenca Latina: 1980- 1995" World Bank LCSHD Paper Senes No 12. 97 DeutschBank (2002) "Brazil on the edge". In:Global Market Research.May Devarajan, S., S. Vinaya and H. Zou (1993) "What Do Governments Buy' The Composition o f Public Spending and Economc Performance" World Bank Policy Research Worlung Paper No 1082. Eisner, R. (1994) "Real government spending and the future". Journal of Econormc Behavior and Organization, 23: 111-126 Eletrobras (2001). Management Annual Report. Fischer, W.H. and S.J. Turnovsky (1995) "The composition of government expenditure and its consequences for macroeconomc performance" Journal of Econormc Dynamcs and Control, 19 (4): ~ ~ 7 4 7 - 8 6 . Fnshtack (2005). "A Situapiio Recente da Infraestrutura de Transportes no Brasil" Background paper prepared for the World Bank AAA Public ExpenditureReview inBrazil. Garson, S. (2004) "Ovewiew of public expenditures in Brazil" Background paper prepared for the World Bank AAA Public Expenditure Review inBrazil. Giambiagi, F. and A.C Altm (2001) "Financas Pcblicas" Ed. Campus Ltda. Giambiagi, F and M. Ronci (2004) "Fiscal Policy and Debt Sustainability- Cardoso's Brazil, 1995-2002" International Monetary Fund. Policy Development and Review Department Worlung paper No 156. Giambiagi, F., J.L. Mendonqa, K.I.BeltrBo, and V.L. Aredeo (2004) "Diagn6stico da Previd6ncia Social no Brasil: o que foi feito e o que falta reformar?" PEA. Texto para discusssoNo. 1050. Glomm, G. and F. Rioja (2004) "Fiscal Policy and long rungrowth inBrazil" Backgroundpaper prepared for the World Bank AAA Public ExpenditureReview inBrazil. Goldfjan, I.and E. Guardia (2003) "Fiscal Rules and Debt Sustainability in Brazil". Banco Central do Brasil, mmeo. Hurlin, C. (2001a) "Estimating the contribution of public capital with time series production functions: a case of unreliable inference" Applied Economcs Letters 8: 99-103 Hurlin,C. (2001b) "How to estimatethe productivity ofpublic capital?" WorldBank, mmeo. IBGE (2000) "Regionalizapiio das Transaqdes do Setor Pcblico - Atividade de Admnistrapgo Pcblica" RIo de Janeiro. Brasil. IBGE(2003) "Estatisticas do stculo XX" RIo de Janeiro. Brasil. IMF (2002) "Debt Cnses. What's different about LatinAmerica?" In:World Economc Outlook 2001: Chapter 2: 61-74. Krol, Robert. (1997). "A Survey of the Impact o fbudget Rules on StateTaxation, Spending, and Debt." Cat0 Journal, Vol. 16, No. 3. 98 Lindert,P (2004) "Growing Public: Social Spendingand Economc Growthsince the Eighteenth Century" Cambridge University Press. Lozano, I. (2000). "Colombia's Public Finance inthe 1990s: A Decade o f Reforms, Fiscal Imbalanceand Debt." Paperpresented at the POE Structure Meeting on FiscalPolicy Issues, Osaka, Japan. Melo, M.A.(2000). "When institutions matters: a comparison of the politics of adrmnistrative, social security and tax reforms". InBenRoss Schneider and Blanca Heredia (eds.), Reinventing Leviathan: the politics of admnistrative reform indevelopingcountries, North South Center Press of the Umversity of Miam, 211-250. Mendes, M.J. (2002) "A Eficacia da Vinculaqgo de Recursos no Federalism0 Brasileiro: o cas0 do FUNDEF" Secretana Nacional do Tesouro (STN). PrCmo Tesouro Nacional MimstCno do Desenvolvimento Social e Combate B Fome (MDS) 2004. Anulise Comparativa deprogramas de proteqcio social - 1995 a 2003. MimstCno da Fazenda: Secretanade Politica EconBmca. (2000) "Orqamento de Renfincias Fiscais e Subsidios da Uni2o" mmeo. Mimsteno da Fazenda: Secretana de Politica EconBmca. (2003). "Gasto Social do Governo Central 2001-2002". mmeo. Mimstry of Finance of Chile. (2003) "Informe de Finanzas Pdblicas sobre el Proyecto de Ley de Presupuestos del Sector Publico del aiio 2004" Santiago de Chile. MimstCno do Planejamento Orqamento e Gest2o: Secretaria de Orqamento Federal. (2003). "Vinculaqdes de Receitas dos Orqamentos Fiscal e da Seguridade Social e o poder discncionhrio de alocap2o dos recursos do Governo Federal", mmeo. MimstCno da PrevidCncia e AssistCncia Social. (2003), Anuano Estatistico da PrevidCncia Social. OECD (2005). "Reallocation: the Role of Budget Institutions" Office of Management and Budget. (2000). "US FinancialYear 2000 Budget." U S Government Printing Office, Washngton, DC. Paes, R.B., M. Carvalho, S. Franco and P.S Carvalho. (2003). "FocalizapBo como instrumento para uma politica social mais efetiva" PEA. SCrie Semnanos No. 148 Penner, R.G and M. Weisner. (2001). "DO State Budget Rules Affect Welfare Spending?" Assessing the New Federalism, Occasional Paper No. 43 Perry, G. (2002) "Can Fiscal Rules Help Reduce Macroeconomc Volatility in LAC?" Paper prepared for the IMF/World Bank Conference on Fiscal Rules and Institutions, Oaxaca, Mexico. Rajkumar, S. (2004) "Brazil: Land of camval, soccer and earmarlung" Background paper prepared for the World Bank AAA Public ExpenditureReview inBrazil. Ram, R and D. Ramsey (1989) "Government Capital and Pnvate Output in the United States" Economc Letters, 13: 223-226. 1989. Reis, C. 0. O., J.A.C. Ribeiro and S.F. Piola (2001) "Financiamento das Politicas Sociais nos anos 1990: 0 Caso do MinistCrio da SaGde". PEA.Texto para discuss20N o 802. 99 Reis, E., F Blanco, L. Morandi, M. Medina and M. de Paiva Abreu (2003) "0 seculo XX nas Contas Nacionais" In:Estatisticas do seculo XX. IBGE. Reis, E. and L. Morandi (2004) "Estoque de Capital Fixo no Brasil, 1950-2002" In:ANPEC. Anais do XXXII Encontro Nacional de Economa. Jog0 Pessoa, PB. Rubinfield, D. (1991) "The Economcs of the Local Public Sector". In: Auerbach, F and M. Feldstein (eds). Handbook of Public Economcs. Amsterdam: North-Holland, 1991, pp. 571 - 645. Silva, M.S. (2003) "Vinculaqdes de receitas n5o financeiras da Unigo" VI11 Congresso Intemacional DelCLAD sobre la Reforma del Estadoy de la Admnistraqgo P~blica,Panamh. Stokey, N. and S. Rebelo (1995) "Growth effects of flate-rate taxes". Journal of Political Economy, 103, pp 510-50. Sturm, J.E. and J. de Haan (1995) "Is Public Expenditure Really Productive?" Economc Modelling, 12: 60-72. Valadares, S. (2004) "Analysis of sources and effects of expenditure rigidity in Brazil". Backgroundpaper prepared for the World Bank AAA Public ExpenditureReview inBrazil. Varsano, R. (1996) "A evoluqgo do sistema Tributirio Brasileiro ao Longo do SCculo: Anotaqdes e Reflexdes para Futuras Reformas" Texto para discussgo 11.405, IPEA. Velloso, R. (2001) "Ajuste Incompleto Retarda Recuperaqgo da Economa Brasileira" Study prepared for the Instituto Futuro Brasil. Velloso, R. (2002) "Public Finance in Brazil: Policy Issues and Projections" Background Paper for World Bank Country Economc Memorandum, Brazil - Stability for growth and poverty reduction. Report No 25278-BR. Washington DC. Velloso, R. (2004) "Ajuste do gasto Pcblico e Retomada do Crescimento EconBmco". FIRJAN, mmeo. Vijverberg, W.P., C.P Vijverberg, and J.L Gamble (1997) "Public capital and pnvate productivity" Review of Economcs and Statistics, 79. 267-278. Weyland (1996) Democracy without Equity- Failures of Reform in Brazil. University of PittsburghPress. World Bank. (1995). "Philippines: Pubic Expenditure Management for Sustained and Equitable Growth." Report No. 14680-PH. World Bank (1997). "Staff Appraisal Report. Brazil - Federal highway rehabilitation and decentralization project" Report No 16425-BR. World Bank. (2001). "Brazil - Attaclung Brazil's poverty' a poverty report with a focus on urban poverty reduction policies" Report No 20475-BR. World Bank. (2002) "Issues infiscal federalism". Report No 22523-BR. 100 World Bank. (2002) "Brazil - Municipal education - resources, incentives, and results". Report NO24413-BR. World Bank. (2003). "Inequality inLatin America and Caribbean: brealungwith history?". World Bank LatinAmencan and Caribbean studies. World Bank (2003). "Brazil - Stability for Growth and Poverty Reduction" Country Economc Memorandum(CEM), Brazil. Report No 25278-BR. World Bank. (2003) "Argentina: Refomng Policies and Institutions for Efficiency and Equity of Public Expenditures." Report No. 25991-AR. World Bank. (2005a) ``Poverty Reduction and Growth: Virtuous and Vicious Circles". World Bank Latin American and Caribbean studies. World Bank. (2005b) "Project Appraisal Document on a Proposed Loan for a Programmatic Fiscal Reform Loan- Social Secunty Reform" Report No 32226-BR. World Bank. (2006) "Project Appraisal Document on a Proposed Loan for a Road Transport Project". Report No 32005-BR. World Bank. (2006) "Brazi1:Enhancing Performance in Brazil's Health Sector: Lessons from Innovations inthe State of SBo Paulo and the City of Curitiba" Report No 35691-BR World Bank. (2007) "Brazil: How to revitalize Infrastructure Investments in Brazil Public Policies for Better PrivateParticipation". ReportN o 36624-BR World Bank. (2007) "Brazil: Interest Rates and Intermediation Spreads" Report No 36628-BR World Bank. (2007) "Brazil: More Efficiency for Better Quality- Resource Management in Brazil's UnifiedHealth System (SUS)". Report No 36601-BR World Bank. (2007) "Brazil Measuring Poverty Using Household Consumption". Report No 36358-BR 101