Report No. 36595-BR Brazil Improving Fiscal Circumstances for Growth (In Two Volumes) Volume I: Stylized Facts and Recommendations March 2, 2007 Brazil Country Management Unit Poverty Reduction and Economic Management Unit Latin America and the Caribbean Region Document of the World Bank ABBREVIATIONS AND ACRONYMS ANP AgCncia Nacional do Petr6leo National Oil Agency BNDES Banco Nacional de Desenvolvimento National Bank for Economc and Social Econdmco e Social Development CIDE Contribuiqgo de Intervenego no Dominio Contribution for Interventioninthe Econdmco Economc Domain COFINS Contribuiqgo Para o Financiamento da Contribution for Financing Social Secunty Segundade Social CPMF Contribuiqiio Providria sobre Temporary Contribution onFinancial MovimentaqgoFinanceira Transactions CSLL Contribuiqgo Social sobre o Lucro Liquid0 Social Contribution onNet Profits DRU Desvinculaqgo de Receitas da Unigo De-earmarlungo f Federal Revenues GDP Produto Interno Bruto Gross Domestic Product FAT Fundo de Arnparo ao Trabalhador Fundfor Worker Assistance FCO Fundo do Financiamento do Centro Oeste Fundfor Financingofthe Center-West Region FNDE FundoNacional de Desenvolvimento da National Development EducationalFund Educaqiio FNE Fundo do Financiamento do Nordeste Fundfor FinancingoftheNortheast Region FNO Fundo de Financiamento do Norte Fundfor FinancingoftheNorthRegion FNDCT Fundo Nacional para o Desenvolvimento National Fundfor Scienceand Technology da CiCncia e da Tecnologia Development FUNDEF Fundo Nacional de Desenvolvimento e Fundfor Maintenance andDevelopmentof ManutenqZioda Educaqiio Fundamental Basic Educationand Traimng o f Teachers ICMS Imposto sobre Comercializaqgo de Tax on Commercialization of Goods and Mercadorias e Serviqos Services INSS Instituto Nacional de Seguridade Social National Social Security Institute IOF/IOF-OWO Imposto sobre OperaqdesFinanceiras Tax on Financial Operations IPI Imposto sobre Produtos Industnalizados Tax on IndustnalizedProducts IR Imposto de Renda Income Tax ITR ImpostoTerritorial Rural Rural Property Tax LDO L e i de Diretrizes Orqamenthrias Budgetary Directives Law LOA L e i Orqamenthna Anual Annual Budget Law LOAS L e i Organica da AssistCncia Social Organic Law of Social Assistance MMA MimstCno do Meio Ambiente Mimstryof Environment OECD Organizaqgo para Cooperaqgo Econdmca Organisationfor Economc Cooperation e Desenvolvimento and Development PIS/PASEP Programa de Integraqgo Social / Programa Program of Social Integration/ Program de Formaqgo do Patrimdnio do Servidor for the Formation of Assets of Public Publico Servants PPA Plano Plunanual Multi-Year Plan PPI Projeto Piloto para Infra-estrutura Pilot Project for Infrastructure PPP Parcerias Publico-Privadas Public-Private Partnershps RGPS Regime Geral de PrevidCncia Social General System of Social Secunty RMV Renda Mensa1Vitalicia Lifetime Monthly Income Program RPPS RegimePr6prio de PrevidCncia- Social Secunty System for Public Servants sus Servidores Publico Sistema Unico de Sa6de Unified Health System ACKOWLEDGEMENTS Ths volume (I) wntten by Ethan Weisman, Lead Economst for Brazil, with Fernando was Blanco, Economst. It was based on the detailed assessments contained involume I1of t h s report. Volume I1was prepared by Yasuhiko Matsuda, Fernando Blanco and JosC GuilhermeReis with inputs from Santiago Hererra and Ethan Weisman. Background papers were comssioned to Silvia Valadares, Sol Garson, Claudio Fnschtack, Gerhard Glomm (Department o f Economcs Indiana University) and Felix Rioja (Department of Economcs Georgia State University). h a n d Rajaram, Aymeric-AlbinMeyer and SantiagoHerrerawere the peer reviewers for the study ... 111 iv TABLE OFCONTENTS 1 EXECUTIVE SUMMARY ................................................................................................... 1 2 INTRODUCTION ................................................................................................................. 3 4.... CURRENTMACROECONOMIC CIRCUMSTANCESFACINGBRAZIL ................79 STYLIZED FACTS ............................................................................................................ 11 The public sector i s large........................................................................................................... 11 Spending trends are explosive ................................................................................................... 12 The current compositionof spending inhibits faster. more robust growth., 13 Infrastructureinvestment i s small ............................................................................................. .............................. 13 Certain extra-budgetary spending i s large ................................................................................. 14 The budget i s inflexible ............................................................................................................. 15 The budget i s not well linked to planning and execution o f public expenditure ....................... Public spending i s not achievingthe desired results ................................................................. 16 18 5 THE EXAMPLE OF ROADTRANSPORTATION ....................................................... 19 6.. RECOMMENDATIONS .................................................................................................... 21 Reduce the size o f the public sector (and control explosive spending) ..................................... Increase public (and related private) sector investment ............................................................ 21 21 More aggressively de-earmark revenue in combination with reduced mandatory spending to allow a better composition of public sector expenditure........................................................... 22 evaluation to enhance the quality o f spending........................................................................... Improve public sector management through strengthemng plannmg. budgeting. execution. and 23 ANNEX 1: BRAZIL'S FEDERALBUDGETRIGIDITIES INA NUTSHELL ..................-25 Mandatory expenditures ............................................................................................................ Federal revenue earmarhng mechanisms (2004) ...................................................................... 25 28 LIST OFFIGURES FIGURE 3.1 TAX REVENUEINPERCENTAGEGDP. 2002 ........................................... 5 LIST OFTABLES TABLE A.l FEDERALREVENUEEARMARKING ......................................................... 21 V Vi 1. EXECUTIVESUMMARY 1.1 Why has Brazil not been able to grow faster for an extended period of time? Long- runreal economc growth for Brazil seems stuck around 2-3 percent per year (2.4 for the penod 1980-2004), incontrast with other mddle-income countnes (e.g. China (9.8%), India (5.8%), and Chile (5.1%)). Although Brazil expenenced rapid annual average economc growth dunng the late 1960s and 1970s (8.5 percent), domestic and international events ledto much weaker growth inthe decades that followed. * 1.2 Fiscal circumstances in Brazil are a key constraint to faster and more robust economic growth. Inpart, this i s due to accumulated public sector deficits in earlier penods that are represented in Brazil's persistently highnet public sector debt (currently about 50 percent of GDP). During the 1980s and md-1990s high debt was associated with debt cnses and hyperinflation. Several attempts to control inflation failed, until the 1994Real Plan. Moreover, to inhibit the accumulation of debt, and to promote economc confidence in capital markets, two successive admnistrations have pursued pnmary fiscal surplus targets (4.25 percent of GDP in 2006). However, the quality of fiscal adjustment has beenpoor. 1.3 Legally mandated spending has increased rapidly, and without significant reform this type of spending is projected to become explosive. An important part of this mandatory spending is tied to the 1988 Constitution. Less than 20 percent o f the budget is discretionary Moreover, other policy decisions, such as mnimumwage increases, have aggravated increased current spending. Rising mnimum wages increase spending under mandatory programs, especially social secunty. 1.4 Given the fiscal target and constraints on current spending, the most important part of the fiscal adjustment in recent years has been the increase in tax revenue, reachng about 38 percent of GDP in 2005. Hightaxation provides disincentives to investment, and encourages tax evasion and informal sector growth, which inturn leads the authonties to search for additional tax revenue on a narrower set of formal activities. Thus, there i s a vicious cycle of ever increasing mandated spending with increased taxation to meet the fiscal target. However, further tax increaseswould be difficult to implement and enforce. 1.5 Budget rigidity, especially revenue earmarking, restricts policy action to resolve the fiscal conundrum. Revenue earmarlung i s linked to transfers to sub-national governments, social sector spending, and specific taxes to cover the social security deficit. The federal government i s required under the 1988 Constitution to share 47 percent of its tax revenue (personal and corporate income taxes, and taxes on industnal products) with lower levels of government. These transfers represent about 5 percent of GDP Social sector spending i s mandated as a share of current revenue. 1.6 Total public investment since 1998 has dropped by more than 2 percent of GDPa2 Private investment has also been falling duringths penod. The drop ininvestment was causedby fiscal constraints and high interest rates (caused in turn by the high level of public debt and associated risks, weak regulatory environments, and public sector distortions in the domestic financial market). * The centralgovernmentdropped ' Seealso the World Bank ReportNo 22950-BR(2002) Brazil: TheNew GrowthAgenda. - !hapercentofGDP, over !hapercentofGDP from sub-nationalgovernments, and about 1 percent of GDP from public corporations. 1 1.7 Spending levels are high relative to the quality of public services, pointing to weaknesses in expenditure management.These weaknesses adversely affect public investment and spending on public services. Thus, increasing public investment alone will not accelerate economc growth; measures are also required to address expenditure management shortcomngs. FIVESTYLIZED FACTS A. Brazil's Spendingis Large andIncreasing 1.8 Brazil has the largest public sector of any other middle income country. General government spending exceeds 40 percent o f GDP and tax revenue was about 38 percent o f GDP in2005 (an increaseo fover 10percentagepoints since the 1990s). Thus is comparable to sizeable welfare states in northern Europe, such as Finland. When combined with nonfinancial public corporations, the size o f the public sector represents about halfo f the economy. 1.9 The bulk of public sector spending is devotedto primary current spending (roughly 33 percent of GDP) and i s in~reasing.~ Social secunty schemes alone represent roughly 12% percent o f GDP Sub-national government spending on social services also has been expanding; this spending devolved to state and municipal governments under the 1988 Constitution. Pressures to increase social secunty and social spending are mounting, due in part to the aging population. 1.10 Off balance sheet transactions also have contributed to increasing the size of the public sector Contingent liabilities include pending judicial decisions. The discovery and recognition o f liabilities not previously recorded (so called "skeletons") mcreased the stock o f public debt substantially (with estimates ranging from 12-19 percentage points of GDP). The larger debt stock has increased debt servicingpayments. Tax expenditures also contribute to fiscal pressures; the federal government tax expenditure was 2.5 percent o f GDP in2004. B. Social Security SpendingRepresentsa Major Challenge 1.11 Social security spending is mandated through the 1988 Constitution, which made the systemgenerous and rigid. Pension spending covers: 1) social security for all public sector workers (RPPS); 2) social security for pnvate sector workers (RGPS, the largest segment o f the Brazilian pension system), and 3) a pnvate complementary pension system. Brazil i s an international outlier inalmost every class o f social security spending (old age, survivor, disability, and sick pay), due to extremely generous benefits. For example, survivor pension benefits represent 3.3 percent o f GDP, which i s over three times as large as OECD countnes that have older populations and higher incomes. 1.12 Pension reforms have had limited impacts. The December 2003 reform reduced somewhat the RPPS pension deficit, inpart due to increases inthe average age o f pensioners and slower growth innew pensioners. It also instituted contributions on retirement benefits, increased the mnimum age for retirement, and equalized the benefits o f public servants to the general RGPS system (but only for new entrants, so the effect has been delayed). On the other hand, admnistrative reforms to the RGPS system in 1998 had more limted impacts. They introduced changes in the formula (the fator prevzdenczarzo) to reduce incentives for early retirement, increased revenues (through recertification efforts) and reduced the number o f sickness benefits (that took effect recently). However, both the average retirement age (54 for men and 51 for women) and average length o f service remain low. RGPS expenditure and deficits continue to increase; so further parametric reforms are needed. Interest payments represented approximately 7 percent of GDP per annum dunng 1995-2004. 2 C. Effectsof PublicInvestmentand Consumptionon Growth 1.13 The present level and composition of public spending are not conducive to long- term economicgrowth. Econometnc analysis ofper capita GDP growthover 1950-2000 shows: Public (and private) investment increases growth; Government subsidies increase pnvate capital accumulation (and thus growth), but t h s effect dampens over time as debt accumulates. The tax burdenreduces capital accumulation (and thus growth). Government consumption, social secunty, and social assistance transfers all reduce growth (the latter inthe long runbecause more financing translates into hgher taxes). 1.14 Over time, the composition of public spending has shifted to larger shares of government consumption, social security, and social assistance and away from public investment and targeted capital subsidies. Additional taxation needed to finance the enlarged public sector (along with inefficient expenditure composition) has dampened growth. 1.15 Simulations show that Brazilwould benefit from higher public investment within a tight fiscal constraint (however, increasing taxation or borrowing to pay for such investmentwould limit long-run growth prospects). Given the low level o f public investment, a reallocation from social security to spending on infrastructure would likely have sizeable growth effects. On the other hand, increasing public debt to finance transfers would decrease growth. Moreover, increasing public debt to finance education expenditure would increase growth inthe short run, but the debt accumulation would slow growth inthe long run. D.The Budgetis Inflexible 1.16 A major challenge is the excessive rigidity arising from legally mandated expenditure and extensive revenue earmarking.There are essentially three broadsources of budget rigidity. (i)earmarking of tux revenue, (the most important are Constitutionally mandated spending on social protection, health and education, and transfers to states and municipalities); (ii)social contributions, which are used to cover part of the social secunty deficit; and (iii)non-discretionary expenditure that includes legal or Constitutional obligations for interest payments, wages and salanes, entitlements (such as social secunty), and social assistance benefits. 1.17 So far government efforts in this realm have concentrated on an ad hoc measure to "de-earmark"federal revenue, called DRU (Desvznculuqa`ode Receztas da Uniiio). However, given simultaneous increases in certain non-discretionary expenditure items (most notably social security benefits), the DRUhas had innocuous impacts on budget rigidity and overall spending. E. The Budget is Not WellLinkedto PlanningandExecutionof Public Expenditure 1.18 An expenditure management review highlighted institutional weaknesses in planning, budgeting, expenditure execution, and evaluation. In particular, there is a lack o f congruence between expenditure allocations set inmulti-year plans, annual budgets, and amounts of investment effectively executed. Since 1996, the government's primary instrument for expenditure p l a m n g has been its Multi-year Plan (PPA, Plano Plunannual), which does not contain a medium-term fiscal framework. The Budgetary Guidelines Law (LDO) defines revenue, expenditure, and balances in a three-year rolling framework, but the LDO targets are not well linked to the PPA, nor do they integrate the impacts o f policy decisions, such as increases inthe 3 mnimumwage. Moreover, Congress makes changes when the budget proposal gets finalized 111 the Annual Budget Law (LOA). In addition, a reversal o f pnonties results from reductions to the budget and restnctions onthe release of funds ("contingenczamento 'I)dmng execution. 1.19 As a result, the quality of spending is poor relative to the amount of spending. For example, health indicators are well below the average for mddle-high income countries, even though per capita health spending i s above average. Moreover, the execution of spending i s not well monitoredand evaluated for outputs or results. FOUR RECOMMENDATIONS A. Reducethe Size of the Public Sector (and ControlExplosiveSpending) 1.20 To address the fiscal conundrum nondiscretionary expenditures need to be controlled and the risingtrendreversed.Changes to legal mandatesare requiredto restrain the growth of the public sector and to reduce its size. This will require legal changes to break the rigidity insocial security andto add flexibility insocial spending. Measures should focus on: 0 reducingspendingon the pensionsystem; 0 reducingcertainsubsidies, (e.g. credit throughpublic financial entities); 0 increasing revenue by selectively applying fees for services and elinmating tax expenditures; 0 revitalizingthe privatizationprocessto add efficiency and lirmt quasi-fiscal deficits; 0 taking into account the full fiscal impacts of further increases in the minimum wage, including impacts on social protection spending; (ideally, spending offsets should be contemplated for such increases); and 0 improvingthe quality of spendingthrough better targeting insocial assistanceprograms, health, and education. 1.21 To make these adjustments more persuasive, a certain share of the fiscal saving should be directed to increasepublic investment. B.More Aggressively De-earmarkRevenueinCombination with ReducedMandatorySpending 1.22 Brazil should more aggressively pursue revenue de-earmarking, but not through DRU. As a first step, all earmarked funding should be set interms of reazs inreal terms (that is indexed to inflation) rather than as a share of revenue. Unused hnds in a given year should be used for domestic debt reduction, or saved to fund additional infrastructure. De-earmarhng should accompany reductions inmandatory spending. C. Improve Public SectorManagementthrough Strengthening, Planning,Budgeting,Execution,andEvaluation 1.23 Reforms to improvethe coherenceof planning, budgeting,execution, and evaluation should be implemented across all levels of the public sector. Efforts could focus on improving the budget cycle and developing a consistent medium-termmacroeconomc framework consistent with the budget. 1.24 The LDO is the most appropriate vehicle to develop a medium-term framework. Eveninthe case of successful constitutional reforms to bnnggreaterbudget flexibility, there may not be sufficient information to detemne where the budget shouldbe cut, preserved, or increased 4 inthe absence of a medium-term framework. An institutionalizedprocess of expenditure reviews for generating mcro-level information about saving options would be a useful addition to Brazil's plamng and budget process. 1.25 To strengthen executing agencies, measures should improve financial management, increase flexibility regarding human resource management, and implement a solid set of monitoring and evaluation systems. Eventually, a medium-term framework and the monitoring and evaluation system could be developed in a harmonized manner to support results-based budgetingfor the entire general government. I).IncreasePublic(andRelatedPrivate)SectorInvestment 1.26 Part of the fiscal saving should be directed'to increase public investment, while keeping the overall tax burden and indebtedness under control. Systematic efforts also will be required to strengthen institutional capacities for plamng, budgeting, and executing public investments. Improving infrastructure mnistnes' capabilities to manage these limted resources efficiently and effectively could involve: (i)a medium-term infrastructure investment strategy; (ii)setofperformanceindicatorstomonitorefficiencyandeffectivenessoftheinfrastructure a portfolio as a whole, as well as by 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 mnistnes' business processes and organizational structures. 1.27 Given the fiscal constraints, the private sector will need to play a more active role. Public-private-partnershlps; solid and reliable contracts (public and pnvate); and public sector reforms that strengthen the regulatory environment are needed. Moreover, getting the size of the public sector under control and continuously worlung to reduce indebtedness would set enabling conditions for continued reductions ininterest rates, leadingto more private sector investment. 5 6 2. INTRODUCTION 2.1 The size and compositionof public sector spendinghave servedas a brakeon robust economic growth inBrazil. Moreover, the present fiscal adjustment inBrazil is not sustainable in the long run, despite the improving macroeconomc conditions in recent years. Based on current trends and comtments embedded in the 1988 Constitution and other laws, Brazil's fiscal spending i s on an explosive path. Without corrective spending measures, further tax increases to maintain the primary fiscal target would choke off economc growth. No immediate crisis i s envisaged; however, medium-term economc growth will continue at anemc rates as long as policies do not address the conundrum of hgh public sector debt, expansive and expanding public sector tax and spending policies, low investment, budget ngidities, and weak public sector management and institutional arrangements. 2.2 This document is part of a series of reports by the WorldBankon Brazil's potential to foment more robust economicgrowth and reducepoverty andinequality.Ths report does not attempt to directly address these outcomes. The main focus of ths report i s on the interrelated fiscal circumstances facing Brazil, with an emphasis onpublic sector spending. 2.3 The report is divided into two volumes. This first volume distills the essential stylized facts (that are described in detail in volume two) and focuses on key recommendations (discussed in detail in Chapter 5 of the second volume). These recommendations hghlight not only areas where improvements could be achieved, but also suggest options on how to go about implementing such modifications, based in part on international expenence. The second volume of the report contains: more detail on these stylized facts; a first attempt to compile data for the entire non-financial public sector; quantitative analysis of the fiscal impacts on growth using a dynamc general equilibrium model and a simulation based on that model; specific examples of key budget rigidities; and, as an example, a review of road transport spending with a view to strengthening the impacts of this infrastructure spending. Other related World Bank reports concentrate on social secunty, health, financial markets, infrastructure, poverty, and the issues surrounding the monitonng and evaluation of public sector services in generaL4 Volume one of this report reflects some of the prelimnary findings inthese other reports. 2.4 The innovative contributions of this report are to: 1) put the information on Brazil's fiscal circumstances (some of whch i s common knowledge to specialists) into a single report; 2) compile the profile of the entire non-financial public sector and the entire public sector for the first time (chapter 1 of volume 2); 3) provide empirical evidence on the overall economc growth impacts o f the composition of public spending and simulate the growth impacts of possible expenditure reallocations (chapter 2 of volume 2); and 4) draw practical policy recommendations from t h s analysis. The report pulls together analysis that has been inprogress over several years. While some of the data are not fully up-to-date, ths reflects in part the difficulties to compile For example, see the World Bank reports: "Brazil: How to Revitalize Infrastructure Investments inBrazil Public Policies for Better Private Participation" Report No 36624-BR; "Brazil: Interest Rates and Intermediation Spreads" Report No 36628-BR; "Brazil: More Efficiency for Better Quality- Resource Management in Brazil's Unified Health System (SUS)" Report No 36601-BR; "Brazil: Enhancing Performance in Brazil's Health Sector: Lessons from Innovations in the State o f Si30 Paulo and the City o f Curitiba" Report No 35691-BR; and "Brazil Measuring Poverty Using Household Consumption" Report NO36358-BR. 7 such coherent data and points to the need to strengthen publicly dissemnated fiscal datan5This document aims to serve as a useful focal point inforthcomng debates on Brazil's macroeconomc (and especially fiscal) policies and related structural reforms to promote enhanced economc growth and reductions inpoverty and inequality ' The International Monetary Fund's Government Finance Statistics Manual 2001 (http://www.imf.org/external/pubs/ft/gfs/manual/index.htm)and related reports on standards and codes on both fiscal and statistical practices provides useful guidance for such efforts. 8 3. CURRENT MACROECONOMIC CIRCUMSTANCESFACINGBRAZIL 3.1 Even though Brazil expenenced relatively rapid economc growth dunng the late 1960s and 1970s (at an annual average rate of 8.5 percent), domestic and international events led to much weaker growth in the decades that followed (for example, 3 percent per year dunng the 1980s). Moreover, since then, economc growth has been steady, but modest. Long-run real economc growth for Brazil seems to remain stuck around 2-3 percent real economic growth per year (2.4 for the penod 1980-2004), while in contrast, many mddle-income countnes expenenced more rapid and robust growth (for example, China (9.8%), India (5.8%), Turkey (4.1%), Chile (5.1%), Indonesia (5.4%), and Thailand (6.1%) duringthe period 1980-2004). 3.2 Why has Brazil not been able to grow faster for an extended period of time? The answer to ths question may be long and complex (see also the World Bank Report (2002) - Brazil; The New Growth Agenda); however, there seems to be broad consensus that the fiscal position remains at the crux of this issue. Inpart, t h s is due to accumulated public sector deficits inprevious periods that are represented inBrazil's persistently highpublic sector debt (roughly 50 percent of GDP). The large debt stock means that the government must spend an important share of revenue collection eachyear on servicingthis debt. The highdebt burden initially caused severe economc turbulence through a senes of debt crises andhyperinflation. Several attempts to control inflation failed, until the introduction of the Real Plan in 1994. Moreover, to inhibit the accumulation of even more debt, and to promote economc confidence in capital markets, recent governments spanning ahnistrations with different political leadership have endorsed the pursuit of announced primary fiscal surplustargets (4.25 percent of GDP for 2006).6 3.3 Legally mandated and policy driven spending has increased rapidly and, without significant reform, this type of spendingis projected to become explosive. An important part of ths mandatory spending i s tied to the 1988 Constitution (including specifically the social security scheme), as well as mandatory mnimum levels o f spending in social areas of education and health. As a result, less than 20 percent o f the budget remains at the discretion of fiscal policy makers. Moreover, other policy decisions have aggravated increased current spending, especially real increasesinthe mnimumwage. The risingmnimumwage has impacts on levels of spending under the mandatory programs, especially social security and social assistanceprograms. 3.4 Giventhe fiscal target and constraintson current spending, the mostimportant part of the fiscal adjustment in recent years has been the increase in tax revenue, reachng about 38 percent of GDP in 2005, an increase of around 10 percentage points since the 1990s. The high level of average taxation (which i s unevenly distributed across tax payers) provides disincentives to investment, and encourages tax evasion and informal sector growth, whch in turn leads the authorities to search for additional tax revenue (including through new or hgher taxes) on a decreasing set o f formal activities. This vicious cycle of taxation could be avoided if ample flexibility was available on spending, but as described indetail int h s report, ths i s not the case. 3.5 Budget rigidity,especially revenue earmarking, restrictspolicy action to resolve the fiscal puzzle (see Annex 1).Revenue earmarlung i s linked to social sector spending, designed to protect this type of expenditure. Inaddition, transfers to sub-national governments are also based on earmarked revenue shanng mechanisms. The federal government i s required under the 1988 Constitution to share a set percentage of its tax revenue (specifically, 47 percent of personal and corporate income taxes, and taxes on industnal products) with these lower levels of government. These and other transfers taken together represent about 5 percent of GDP Thus, the federal The fiscal year is identical to the calendar year. 9 government has been inducedto develop new tax mechanisms (on top of previous ones). Ths has contributed to a vicious cycle of spend and tax policies. Insum, about 80 percent of total federal tax revenue i s earmarked, mahngthe budget extremely rigid. 3.6 A small amount of fiscal adjustment has taken place through cuts to public investment.Central government investment has dropped to around !4 percent of GDP in recent years, from only 1percent of GDP in 1998. Similarly, sub-national governments' investment has declined to 1.5 percent of GDP from 2.3 percent of GDP during ths penod, and public corporations' investment dropped to 1 percent of GDP from 2 percent of GDP dunng the same period. Thus, total public investment has dropped by more than 2 percent of GDP. Moreover, private investment has not picked up to compensate for this reduced public investment, and in fact has also been falling during this period. As a result, analysts note the sharp decline in total (public and pnvate) investment since the late 1980s. Ths was driveninpart by the drop inpublic sector investment and by the hgh interest rate environment (caused in turn by the hgh level of public debt and associated risks, weak regulatory environments, as well as public sector distortions inthe domestic financial market). 3.7 Spending levels are disproportionate to the quality of results obtained, pointing to weaknesses in expenditure management (planning, budgeting, implementation, and evaluation). These institutional weaknesses adversely affect public investment, as well as spending on education and health. Therefore, increasing public investment spending alone will not accelerate economc growth; measures are also required to address expenditure management shortcomngs. The report illustrates this point usingthe example o f roadtransportation. 3.8 The policyresponseto these fiscal circumstanceshasresultedina fiscal conundrum. Consistently acheving the headline fiscal targets, withn a context of ngid and expanding current expenditure constraints, represents a fiscal conundrum. Recently the authonties have increased tax collection significantly and reduced public investment spending slightly to meet the fiscal target. However, this fiscal adjustment i s of poor quality. Hightaxation has reduced the growthof formal sector activity, ceterzsparibus, further narrowing the tax base. Spendingngidities channel resources to less than optimal allocations, including a shift away from infrastructure investment, hrther reducing economc growth. Moreover, ths fiscal adjustment i s not sustainable. With average taxation equivalent to about 38 percent of GDP, further tax increases would be difficult to implement and enforce. 3.9 Thus far the solution to the fiscal conundrum is one of the root causes for low economic growth, and therefore, it will not be possiblefor Brazilto grow its way out of this fiscal challenge. Higher growth cannot be attained without first addressing the spend-and-tax cycle. Given this context, focusing on public sector spending (its size, pnorities, and qualities, as well as the budgetary and other institutional circumstances under whch this spending takes place), is central to resolving the fiscal conundrum that may be the lead cause for slower than deslred economc growth. 10 4. STYLIZED FACTS THEPUBLICSECTORISLARGE 4.1 Brazil has the largest public sector of any other middle income country (Figure 1). General government spending exceeds 40 percent o f GDP and tax revenue was about 38 percent o f GDP in 2005. This i s comparable to sizeable welfare states in northern Europe, such as Finland. When combined with non-financial public corporations, the size o f the public sector represents about half o f the economy Moreover, the high tax burden provides strong disincentives to private investment, formal sector employment, and economc growth. 4.2 Part of this spending is needed to cover debt service (that is, payments for deficits accumulated in previous periods). Interest payments on the debt represented approximately 7 percent o f GDP per annum during 1995-2004. Figure 3.1: Tax Revenueinpercentageof GDP, 2002 .. . ....-....-.... - . . __ .._ . 1 30.0 I 1 I 35'0 15.0 c 5.0 0.0 Brazil China India Korea Russia 4.3 The bulk of public sector spendingis devotedto primary current spending (roughly 33 percent of GDP). The largest share o f this spending i s devoted to social secunty schemes (roughly 12.5 percent o f GDP), which are generous and rigidly designed. Another important part o f the increase in current spending i s attributable to sub-national governments, because social sector spending has been devolved to state and municipal governments as mandated by the 1988 Constitution. In addition, admnistrative costs due to decentralization have been increasing rapidly At the same time that federal funds were devolved to sub-national governments, the federal government has not been able to make equivalent cuts in its spending, resulting in an increase inthe overall size o f government. 4.4 Public enterprises remainimportant actors, despite privatizations inthe 1990s.Gross revenue and expenditure by non-financial corporations reached more than 14 percent o f GDP in the early 2000s. While quasi-fiscal operations may have been reduced due to these pnvatizations, rough estimates suggest that they may still be significant. Ths i s particularly plausible when the effects o f public financial corporations are taken into account. The public policy functions through these financial agents, especially the use of directed and subsidized credit operations are not negligible, but have yet to be studied thoroughly. Directed credit represents approximately one-third o f Brazil's total credit by the banlung system. 11 4.5 Off balance sheet transactions also have contributed to increasing the size of the public sector In addition to contingent liabilities, some of which are due to pending judicial decisions, the discovery and recognition o f liabilities not previously recorded (so called skeletons) has increased the stock o f public debt substantially (with estimates ranging from 12-19 percentage points o f GDP). Only some o f the skeletons were discovered dmng the process o f privatizations. The larger debt stock has increased debt servicing payments, leading to a larger public sector. SPENDING TRENDS ARE EXPLOSIVE 4.6 Public sector spendingis increasingrapidly. General government spending, the largest component o f the public sector increased spending to 43 percent o f GDP in2004 (from about 40 percent o f GDP in 1995). Most o f the increase in spending i s attributed to social secunty and social assistance benefits, interest payments, and expense on goods and services. 4.7 Moreover, pressuresfor future spending increases are mounting even faster, due in part to the aging o f the population and the corresponding need to increase social security and health expenditures, similar to the pattern observed in OECD countries. These pressures would tend to increase the size o f the public sector. Uncontrolled, these spending trends could increase the Size o fthe public sector to over 50 percent o f GDP ina decade. 4.8 Social security, as mandated through the 1988 Constitution, reflects a generous and rigid system. The entire pension system covers: 1) the RPPS (Social Secunty System for Public Servants) for civil servants at all levels o f government, 2) the RGPS (General System o f Social Secunty) for private sector workers (the largest segment o f the Brazilian pension system), and 3) the relatively new complementary pension system (a defined-contribution scheme, the third pillar). Fiscal cash flows are restricted to the RPPS and RGPS. Prelimnary analysis shows that Brazil i s an outlier in every class o f expenditure (old age, survivor, disability, and sick pay), due to extremely generous benefits. The most obvious outlier concerns survivor pension benefits, which represent 3.3 percent o f GDP (out o f 12.5 percent o f GDP for total Brazilian pension expenditure). This i s over three times as large as OECD countnes that have older populations and higher incomes, and even more disproportionate comparedwith non-OECD countnes. 4.9 The December 2003 reform of the RPPS has had some impacts. It reduced pension expenditure (and the RPPS pension deficit), in part due to increases in the average age o f pensioners and slower growth in new pensioners. The RPPS reform also instituted contributions on retirement benefits, increased the mnimum age for retirement, and equalized the benefits o f public servants to the general RGPS system (but only for new entrants, so the effect will be delayed). 4.10 On the other hand, reformsto the RGPS system have hadlimitedimpacts, and alone they will not be able to containthe growthin expenditure and deficits. They created thefator prevzdenczarzo to reduce incentives for early retirement, increased revenues (through recertification efforts) and reduced the number o f sickness benefits (that only took effect recently). However RGPS expenditure and deficits have continued to increase. The average retirement age (54 for men and 51 for women) and the average length o f service remain low. Further parametnc reforms are needed. 12 THECURRENTCOMPOSITION OFSPENDINGINHIBITSFASTER,MOREROBUSTGROWTH 4.11 The present level and composition of public spending are not conducive to long- term economic growth. The present composition of expenditure, with low investment and hgh government consumption i s inefficient for enhancing economc growth. Specifically, econometnc analysis of per capita GDP growth over 1950-2000 finds: A positive effect on economc growth of (public and pnvate) investment7(and a positive correlationbetweenthe public and private capital stocks); A positive effect of government subsidies on private capital accumulation (and thereby on growth) due to the short-run fiscal impulse; however, this effect dampens over time due to debt accumulation; A negative effect on growth o f government consumption (inthe short and longrun), A negative effect on growth of social secunty, and social assistancetransfers (inthe long runbecauseofthe greater financing needthat translatesinto ahgher tax burden).* 4.12 Over time, the composition of public spending in Brazil has shifted to ever larger shares of government consumption and transfers for social security and social assistance, from a more growth-oriented m x of hgher public investments and targeted capital subsidies. Additional taxation needed to finance the enlarged state (with a more inefficient expenditure composition) has dampenedgrowth. 4.13 Simulated possibleeffects of expenditure reallocations further tested the hypothesis that the present 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 investment withn a tight fiscal constraint (however, increasing taxation or borrowing to pay for such investment would l i m t long-run growth prospects). Especially given the current low level of public infrastructure investment, a reallocation from social secunty transfers to increase spending on economc infrastructure would likely have sizeable growth effects. On the other hand, increasing the public debt to finance transfers decreases growth substantially Moreover runninga public debt to finance education expenditure would increase growth inthe short run, but the debt accumulation would slow growth in the long run. Also, spending on government consumption and transfers for social security and assistancedampen growthsg INFRASTRUCTURE INVESTMENT I S SMALL 4.14 A smaller percentageof GDPis spent on public sector investment,(about 3 percent of GDP), broken down into roughly 0.5 percent of GDP by the federal government, 1.5 percent of GDP by sub-national governments, and 1percent of GDP by public corporations. 7While such results are consistent with most theoretical models, emprrical evidence establishing this relationship is rare. The econometric analysis also shows a negative effect o f the tax burdenon capital accumulation and economic growth. These simulations are based on a functional classification o f expenditure. While the results contrast with the recent findings inWorld Bank (2005) "Poverty Reduction and Growth: Virtuous and Vicious Circles", specifically that reductions ininequality and poverty tend to have positive growth impacts, the simulation model did not explicitly take these impacts into account. 13 4.15 Increasing infrastructure investment i s not the sole responsibility of the public sector; the private sector will need to play a larger role. Given fiscal constraints, the public sector will remain limted inthe amount of infrastructure investment that it can finance directly 4.16 Public sector policies to establish partnerships with the private sector have been initiated. In addition to the pnvatization program in previous years, examples include the Pilot Project for Infrastructure (PPI), public-private-partnerships (PPPs), and concessions and similar contracts. In particular, the PPI was designed to protect public infrastructure investment from budget cuts and to improve the expenditure management cycle. However, prelimnary results from this exercise show that merely increasing public investment (for example inroad transport) i s not sufficient to overcome the institutional weaknesses embedded in the public sector expenditure management framework. Moreover, all of these incipient government programs will not be enough to meet the large infrastructure demands of the economy Other public sector policies to address bottlenecks to pnvate sector investment will be needed. At the same time, the highinterest rate environment, due inpart to the high level of public sector debt and other risk factors, impedes substantial increases in investment" Again, addressing the fiscal conundrum would contribute to reducing these nsk factors and lower the interest rate environment for investment. CERTAINEXTRA-BUDGETARYSPENDINGIS LARGE 4.17 The amount of public sector spending is affected by the operations of quasi-fiscal public corporations and tax expenditures. The activities of the entities outside the general government sector include non-financial and financial public enterprises (at all levels of government) whose operations result inthe accumulation of public liabilities and assets that will affect the public sector's overall fiscal health in the long run. In addition, several policy objectives are being pursued through the utilization of tax expenditures by the public sector that representsabout 3 percent of GDP. 4.18 Off-budget or quasi-fiscal activities have been extensively used by past governments inBrazil and generatedlarge contingent liabilities whch led to debt accumulation. Estimates of the contribution of contingent liabilities to debt accumulation are inthe range of 12-19 percent of GDP 4.19 Non-financial public corporations have played important roles in the country's development model. Through direct investment inkey infrastructureand manufactunng sectors, non-financial public corporations made most of the public sector's total investment (above 70 percent) duringthe 1980s. Thelr role in the Brazilian economy remains important. For the three levels of government in 2001, non-financial public corporations' total revenue represented 14.7 percent of GDP and total expenditure represented 14.2 percent of GDP However, a good share of this was associatedwith largely commercial activities. 4.20 Public financial corporations control a large part of the financial sector. As of December 2004, government-owned banks were responsible for 39 percent of the total lending of the financial system. The total revenue of public financial corporations represented 15.5 percent of GDP in2001 for all levels of government. lo Also see the separate Bank report ("Brazil: InterestRates and Intermediation Spreads". Report No 36628-BR). 14 4.21 The extensive presence of public banks does not necessarily mean that these banks engage in significant quasi-fiscal activities. Quasi-fiscal activities have impact on aggregate demand; it i s a form of public policy and should be reported inthe fiscal accounts. According to government statistics, one estimate o f total implicit credits and financial subsidies by public banks (excluding the Central Bank) was 0.55 percent of GDP for 2004." Thls amount, roughly equal to 14 percent of the total revenues of public banks, should be considered de facto public expenditure inestimating total public spending inBrazil. 4.22 Tax expenditures are another way for governments to engage in extra-budgetary fiscal activities. Tax expenditures normally include exemptions from taxes, allowances deducted from gross income, tax credits deducted from tax liabilities, tax rate reductions, and tax deferrals. Total tax expenditures for the federal government reached 2.5 percent of GDP in 2004. Almost 60 percent of tax expenditures are related to economc affairs, with companies investing in the North and Northeast, agricultural firms, and mcro-enterprises being the most important beneficiaries.'2 THEBUDGETIS INFLEXIBLE 4.23 A major challenge for expenditure reform in Brazil is the excessive and increasing degree of rigidity arising from the high share of constitutionally or legally mandated expenditure and extensive revenue earmarking (Annex 1). For example, in 2003 for the federal government, 80 percent of the revenue (20 percent of GDP) was earmarked for specific purposes and 90 percent of total primary expenditure (19.5 percent of GDP) was non- discretionary. Earmarked revenue i s assigned to line mnistnes, whch are not required to spend the entire amount. The remaining earmarked funds are effectively used to reduce the net borrowing requirement; however, these funds cannot be used to reduce debt liabilities. Non- discretionary expenditure is mandated to fully fund government programs. Thus the overlaps between revenue and expenditure ngidities are large, but do not match exactly. The actual degree of budget flexibility i s less than the "free" portion of either the expenditure (10 percent in 2003) or the revenue (20 percent in 2003). For example, in 2002, the "free" portion of the budget was somewhere above 4.5 percent of revenue, but below 8.1 percent of expenditure. In 2004, combining revenue earmarlung and mandatory expenditures, the "fully free" portion of the executed federal budget was just 2.2% of GDP (14 % of primary expenditures). Whatever the precise estimate for the actual "free" portion of the budget, it is clear that the government i s quite limted inits ability to "freely" allocate funds within the budget process. 4.24 Social protection, including social security, is at the center of budget inflexibility There are essentially three broad sources of budget ngidity in Brazil: (i)earmarlung of tax revenue, (the most important of which are for constitutionally mandated spending on social protection and health, as well as transfers to states and municipalities); (ii) social contributions, which are completely used to cover part of the social security deficit; and (iii) non-discretionary expenditures that include legal or constitutional obligations o f the government, specifically interest payments, wages and salaries, entitlements (such as social secunty), and social assistance benefits. l1 methodologyusedfor The this updated estimate is based on the government's approach publishedby the Economic Policy Secretariat in 2000. While that methodology represents an opportumty cost of the government funds, it does not represent the total subsidy that would be the difference between the interest rates the bank could get in the market and the one charged in the operation. The latter estimate would be significantly larger. '* For state governments, the available data indicates tax expenditures are around 0.5 percent of GDP 15 4.25 More than half of total non-discretionary expenditure was associated with social security in 2004. The growth o f non-discretionary expenditure is largely due to the growth o f social secunty benefits payments, which grew from 10.5 percent o f GDP in 1998 to 13 percent o f GDP in 2003.13 The steady increase in the social security expenditures is related to both the expansion o f population coverage - from 11.1 million in 1988 to 21.3 million in 2002 - and the increase in benefit levels, largely influenced by the successive upward adjustments o f the mnimumwage to which the mnimumsocial security benefit is indexed. Although the reforms in the late 1990s slowed the rate of increase, the benefit level still remains quite generous by international comparison. 4.26 Healthand educationspending havedevolvedto sub-nationalgovernmentsfromthe federal government, creating significant budget rigidities for these sub-national governments. Total public sector spending on health and education has increased by only 0.6 percent o f GDP between 1995 and 2004 (education remained at 4.2 percent o f GDP and health rose from 4.4 percent o f GDP to 5 percent o f GDP during the penod). There are legally mandated floors for spending on health. States and municipalities are required to spend on education 25 percent o f the sum o f constitutionally mandated transfers received from high levels o f government (federal for states, and federal and state for municipalities) plus their own tax revenue. 4.27 A key policy question regarding budget rigidity is whether greater flexibility is attainable through reforms on the revenue side (Le., "de-earmarking") or of the expenditure side (removing expenditure mandatesthrough Constitutional changes). So far government efforts int h s realmhave concentrated on an ad hoc measure to "de-earmark" federal revenue, called DRU (Desvznculaqa`ode Receztas da Unia`o). Formerly earmarked revenue has been directed to non-discretionary expenditure as much as possible. The effects o f the DRU are partially offset by the mandated increases for certain non-discretionary expenditure items, most notably social secunty benefit payments (covenng in part the social secunty deficit; R$17.2 billion in 2002) rather than to increase infrastructure investments. Thus, the DRUhas had m n o r impacts on budget ngidity and an innocuous effect on overall spending. THEBUDGETIS NOT WELL LINKEDTO PLANNINGAND EXECUTIONOFPUBLICEXPENDITURE 4.28 The review of expenditure management indicates institutional weaknesses in planning, budgeting, expenditure execution, and evaluation. Inparticular there is a lack o f correspondence among investment expenditure allocations set in the multi-year plans and in annual budgets, when compared with the amount o f investment expenditure effectively executed. These weaknesses are also evident inareas o f current spending. 4.29 The government's primary instrument for forward expenditure planning is its Multi-year Plan. Since 1996, the federal government has prepared multi-year plans, PPAs (Plano Plunannual) covering a four year penod. The PPAs are instruments to strategically direct the allocation o f federal resources to spending programs. The main innovation o f the PPA was the intention to introduce a results-oriented management model, which would include detailed programs with defined objectives, performance indicators, and a clear intention to link planning and budgetingto ensure their implementation. l3To a lesser extent, the increase was also due to the designation of federal health expenditure as mandatory under Constitutional Amendment no. 29 in2000, which established a spending floor. 16 4.30 The budgetary framework for the medium-term i s not well linked to planning. Despite considerable progress, the PPA does not contain a medium-term fiscal scenano to support the spending programs, especially for investment. Infact, programmed investment expenditure i s not based on estimates o f macroeconomc and fiscal variables that define resource availability As a result, the indicative budget allocations contained in the multi-year investment plans are frequently unrealistic. The overall fiscal framework i s contained inthe Budgetary Guidelines Law (LDO), which defines fiscal revenue, expenditure, and balances in a rolling three-year framework. It contains annexes on social secunty, contingent liabilities, and revenue projections that are used to define the fiscal envelope. However, the LDO targets are not well linked to the PPA. Moreover, the LDO does not integrate the fiscal (and macroeconomc) impacts o f policy decisions, such as increases inthe mnimumwage. Thus, the LDO does not provide a framework for strategic decision-malung. 4.31 The inconsistency between planning and budgeting appears when the Executive submits its budget proposal to Congress. There is evidence that PPAshave not beenused as the t e c h c a l basis for actual budget allocations, whether in the government's budget proposals or in the Annual Budget Laws (LOAs) approved by Congress. Infact, the strong adjustment effort that the federal government has been implementing since the late 1990s and the increasing expenditure ngidity that squeezed the fiscal space for investment reveals the disconnection between PPA and LOA allocations. 4.32 There is a lack of connection between annual budget allocations and expenditure execution. The tendency o f the Congress to overestimate revenue has consequences for the budget allocation o f expenditure. However, at the same time, the need to acheve announced fiscal targets puts downward pressure on actual expenditure execution. Moreover, inefficiencies include difficulties inprocurement and contracting processes, and weakness in financial planning by sector agencies (such as the absence o f reliable revenue flow estimates dunng the year and uncertainties about the completion o f physical execution). These factors also explain large differences between budget allocations and expenditure execution, and contribute to lower than budgeted expenditure execution. 4.33 "Reversal of priorities" results from budget reductions and restrictions on the release of financial flows ("contingenciamento'? during the execution phase. The budgetary and financial constraints faced by sector agencies associated with achieving overall fiscal targets (in addition to spending inefficiencies) give the authorizing central Ministry an opportunity to justify changes in resource allocations set in the LOA. A mnistry's budget proposal is altered several times dunngthe budget process: first in the intra-executive negotiations with the Mimstry o f Plamng, then by Congress inthe LOA approval process, and finally through the discretionary control o f spending l i m t s during budget execution. As a result, the priorities set in the PPA, (which are barely reflected dung the budget approval process) are frequently ignored dunngthe expenditure executionprocess. 4.34 The execution of spending i s not well monitored and evaluated. While momtonng o f cash flows follows formal norms and rules, the monitonng o f physical capital remains weak. Evaluations are not systematically conducted and they focus pnmarily on cash spending, not on outcomes and results. Evaluations that are conducted do not feed into planning and budgetary processes that follow in future penods. Thus, budgets are not formulated with a view to outputs or results. 17 PUBLICSPENDINGIS NOTACHIEVING THE DESIREDRESULTS 4.35 The quality of spending is poor relative to spending levels. A World Bank report on traclung health pend ding'^ and the example of roadtransportation (see below) illustratethis point. 4.36 Health indicators in Brazil are well below the average for middle-high income countries, even though per capita health spending is above average (US$631 versus an average of US$585 for other mddle-high income countnes). Health expenditure in Brazil has grown consistently, putting severe pressure on already scarce public resources. International expenence shows that the critena and ways inwhch the resources are allocated, transferred from central authonties to provider units, and how they are used thereafter, significantly affect the impact that financial resources can have on health services and outcomes. 4.37 Given the country's federal structure and the decentralized nature of the health management system, the financial flows are extremely complex. The Umfied Health System (SUS) covers the entire Brazilian population and offers a complete range of services free o f charge, although the quality of service provision vanes widely across junsdictions. Under ths system, the federal government transfers resources for health activities to the states and municipalities inover 70 different ways. 4.38 Health planning, budgeting and execution are disjoint. The plamng and budgeting processes inthe SUS are fairly well structured and formal; however, complexity and bureaucratic formalism l i m t the usefulness of the SUS as an effective 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. The inflexibility of this legislation, designed to l i m t the msuse of public funds, also greatly limits the autonomy of local managers and thus their capacity to effectively manage resources. In addition, poor managerial capacity in local umts, scarce information, and a msaligned incentive structure are other factors affecting the quality of health service provisionunder the SUS. l4 See "Brazil: More Efficiency for Better Quality. Resource Management inBrazil's UnifiedHealth System (SUS)" Report No 36601-BR. 18 5. THE EXAMPLE OF ROADTRANSPORTATION 5.1 Assuming existing budget rigidities can be diminished, more funds could become available for public investment, but this would not resolve the infrastructure bottleneck. Currently over 80 percent of total central government investment i s dedicated to the transportation sector. The example in the report concentrates on road transportation because it represents the largest component of transportation investment. The lack o f adequate road transport infrastructure i s a critical bottleneck to sustained economc growth. The report shows inefficiencies inpublic spending on roads. A simple increase inbudget allocation may not lead to immediate expansion in the formation of public capital. Weaknesses mplwng, budgeting, and expenditure execution make production of infrastructure highly inefficient. In spite of their scarcity, expenditures in the transportation sub-sector are not well-planned and prioritized, and their execution suffers from delays and cost escalation. The extent and the quality of road infrastructure are insufficient to support the country's economc activities (especially exports). Thus, greater public sector efficiency combined with increased public investments in roads and other transport infrastructure should enhanceBrazil's growth prospects. 5.2 Prioritizationandresourceallocationsdefinedinthe multi-yearplans(PPAs) do not guide annual budget formulation even for critical sectors such as transportation. 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 LOAs approved by Congress. This inconsistency has several roots including: (i) technical imperfections in medium-term revenue forecasting and program costing; (ii)particular vulnerability of the transport budget to Congressional amendments and discretionary management of expenditure execution by the Executive (e.g., so-called "contingenczumento", or partial freeze on spending authonzations); and (iii) frequent delays inphysical execution o f public works due to irregulanty in resource flows, adverse weather (whch tends to coincide with the time when the budgeted funds are releasedduringthe fiscal year), and sometimes detailed external oversight for fear of corruption, 5.3 The government introduced the PPI to improve the efficacy of public investment execution. The objective is to strengthenbudget execution of selected infrastructureprojects by protecting them from funding freezes and closely monitoring their execution by central economic agencies. The implementation expenence of the PPI 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 of institutional capacity for appraisal, implementation, and evaluation of public investment projects continues to be a major obstacle for the enhancement of the efficiency inpublic infrastructure projects. 5.4 Other government initiatives that aim to increase the amount of road transport infrastructureinvestmentwill still needto address weaknesses in expenditure management. Some of these imtiatives include Public-Private Partnerships, decentralization of federal road projects to state governments, and results-based contracts and concessions. A deeper set of reforms would be needed, in conjunction with these initiatives to improve both the quantity and quality of infrastructure investment (inroads and other forms of assets). 19 20 6. RECOMMENDATIONS 6.1 In sum, the size andcompositionof the public sector, inthe contextof the Brazilian fiscal target, hasledto a fiscal conundrumand lower than desiredeconomicgrowth.The public sector i s large and growing, due to legally mandated, as well as policy driven, increases in current spending (especially on social secunty and social assistance). Moreover, fiscal adjustment has aggravated the composition o f spending away frompublic investment and has resulted in increased taxation. Budgetary rigidities explain, inlarge part, the size and composition o f the public sector. To address the fiscal conundrum, this report recommends the following measures. REDUCE SIZEOFTHE PUBLICSECTOR (AND CONTROL EXPLOSIVESPENDING) THE 6.2 Changesto legal mandatesare requirednot only to restrainthe growth of the public sector, but to reduce its size. This will require Constitutional and other legal changes to break the ngidity in the social security scheme and to add flexibility insocial sector spending. Inshort, nondiscretionary expenditures needto be controlled and their nsing trend reversed. 6.3 Measuresshouldfocus on: a. reducing current spending on the pension system (specific parametnc recommendations especially for the RGPS are contained in a separate World Bank study); b. reducing certain specific subsidies, especially on credit through public financial entities; C. increasing revenue by selectively applying fees for services and elimnating tax expenditures; d. revitalizingthe privatizationprocessto add efficiency and limtquasi-fiscal deficits; e. improving the targeting of spending in social assistance programs and in health and education; t h s could provide some efficiencies thereby augmenting discretionary spending; and f. takinginto accountthefullfiscal impactsoffurther increasesinthe minimumwage, including indirect impacts on social protection spending. Ideally, spending offsets should be contemplated for such increases. T o make these adjustments more persuasive, a certain share o f the fiscal saving should be directed to increase public investment (see below). INCREASE PUBLIC (AND RELATEDPRIVATE) SECTOR INVESTMENT 6.4 To enhance economic growth Brazil should increase its public investment by reallocating expenditure from other less productive areas, while keeping the overall tax burden and indebtednessunder control. The econometnc analysis and the simulation exercise in the report support t h s conclusion. Expenditure reallocations, however, are a difficult propositionfrom both political andtechmcal points o f view. The instruments at the government's disposal, at least in the short run, are limted, especially due to existing budget ngidities. Nonetheless, over the long run Brazil must gain more effective control over growing current expenditure in order to arrest decades o f fiscal expansion and to reallocate more resources to public investment. 6.5 Gradual increases in public sector investment spending would be desirable, but insufficient.Merely increasing the budget for public sector investment, given weakness inpublic 21 expenditure management, would not lead to substantial creation of capital assets. Systematic efforts will be required to strengthen institutional capacities for plamng, budgeting, and executing public investments. Such efforts could only be implemented over the medium term. A clear strategy for improving infrastructure mnistnes' capabilities to manage these limted resources efficiently and effectively i s laclung. Such a strategy could involve the followmg elements: (i)a medium-term infrastructure investment strategy; (ii)a set of performance indicators to monitor efficiency and effectiveness of the infrastructure 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 mmstries' business processes and orgamzational structures, as necessary 6.6 Given the constraints on the size of public sector investment, especially over the mediumterm, the privatesector will needto playa more active role if economic growthis to become more robust. The World Bank infrastructure report15 highlights specific recommendations that could be implemented to promote more and better private sector participation in infrastructure investment. It notes that: public-private-partnerships are a useful - but not the only - mechanism to address the lack of infrastructure. It underscores the importance of designingsolid and reliable contracts (public and pnvate); and points to public sector reforms that would strengthen the regulatory environment to create more stable and reliable conditions for private sector investment. Moreover, getting the size of the public sector under control and continuously worlung to reduce indebtedness would set enabling conditions for continued reductions ininterest rates, leading to more pnvate sector investment. MORE AGGRESSIVELYDE-EARMARK REVENUEINCOMBINATION WITH REDUCED MANDATORY SPENDINGTO ALLOW A BETTER COMPOSITION OF PUBLIC SECTOR EXPENDITURE 6.7 Revenue earmarking contributes to budget rigidity that can complicate day-to-day management of budgets, and is ineffective in protecting funding for specific purposes, as shownby the evidenceinBrazil and internationally.Inthis context, the federal de-earmarlung mechanism (DRU) has beneficial, albeit limted, effects to add flexibility to the budget and contributes to acheving fiscal targets. 6.8 Brazil should pursue a more aggressive approach to implement revenue de- earmarking, but not mainly through the DRU.As a first step, all earmarked fundingshould be set in terms of reazs. Linkages to revenue collection or any other form o f indexation should be resisted. If this i s not politically feasible, then earmarlung could be set in terms of ream in real terms (constant pnces), that i s indexedto inflation. Ths should not prevent additional funds from being dedicated to prionty areas over and above the earmarked amount that is budgeted. Moreover, earmarked funds that are not used in a given year should become available for domestic debt reduction in that year, or saved to fund additional infrastructure spending in the following year. In this way, over time, public resources would s h f t to where they are most needed and effective, while simultaneously encouraging further investment and a more sustainable fiscal path. 6.9 However, it is important to prevent additional de-earmarked revenues from being absorbed by rapidly rising non-discretionary expenditure. De-earmarlung should be l5See "Brazil: How to Revitalize Infrastructure Investments inBrazil Public Policies for Better Private Participation" Report No 36624-BR. 22 accompanied by reductions inmandatory spending. Only under conditions of hgher flexibility to allocate public spending would de-earmarhng efficiently create fiscal space and lead to further public investment and debt reduction. IMPROVE PUBLIC SECTOR MANAGEMENT THROUGH STRENGTHENINGPLANNING, BUDGETING, EXECUTION, AND EVALUATION TO ENHANCE THE QUALITY OF SPENDING 6.10 Reformsto improvethe coherenceof planning, budgeting, execution, and evaluation should be implemented across all levels of the public sector. Specific efforts could focus on improving the budget cycle and developing a consistent medium-term macroeconomc framework withn which the budget would be developed. Other actions could be directed to improve the quality of public spendingwithn t h s framework. 6.11 The LDO is the most appropriate vehicle to develop a medium-term framework. Ths improvement inBrazil's institutional arrangement for planning andbudgeting mght help the government to control the size and composition o f expenditure over the medium term. The most common instrument that serves this purpose i s a medium-term framework, combined with politically well supported budget rules to guarantee fiscal equilibnum and to control the growth in the size of government. A medium-term framework would keep government expenditure and revenue on target because it makes medium-term fiscal impacts of policy decisions explicit and makes this information available duringthe budget planningprocess. 6.12 A medium-term framework could also .facilitatereallocations of budget resources over time, providedthat the structural causes of budget rigidityare lifted. Without changing the expenditure composition in favor of investments in economc infrastructure, the budget's contribution to long-term growth will remain limted. Even in the case of successful constitutional reforms to bring greater budget flexibility, however, it is not clear that there is sufficient information to detemne precisely where the budget should be cut and where it should be preserved or increased in the absence of a medium-term framework. An institutionalized process of expenditure reviews for generating mcro-level information about saving options would therefore be auseful addition to Brazil's planningandbudget process. 6.13 To strengthen executing agencies, measures include improving financial management, increasing flexibility regarding human resource management, and implementing a solid set of monitoringand evaluation systems. Capacity buildingand inter- agency exchanges would be useful to develop the skills required for financial management purposes. In addition, giving managers greater freedom to find and keep quality staff for the provision of public services would be desirable. There i s a need for integration of financial execution and physical execution of projects. Eventually, a medium-term framework and the monitoring and evaluation system could be developed ina harmonized manner to support results- basedbudgeting for the entire general government. 23 24 ANNEX 1:BRAZIL'SFEDERALBUDGETRIGIDITIESINA NUTSHELL Budget rigidity in Brazil arises from both sides o f the federal budget. On the revenue side, constitutional or other legal revenue earmarlung mechanisms guarantee that certain types o f government revenue finance specific expenditure. On the expenditure side, there are expenditure items which are considered mandatory and include constitutional and other obligations o f the government such as interest payments, personnel expenditures, social security benefits, and social assistance entitlements. There are some overlaps between the revenue and the expenditure rigidities: there i s a portion o f mandatory expenditure that i s financed with earmarked revenue. The degree o f budget ngidity i s extremely high. In 2004, combining revenue earmarlung and mandatory expenditures, the "fully free" portion o f the executed federal budget was just 2.2% o f GDP (14 % o f primary expenditures). FEDERAL REVENUE EARMARKINGMECHANISMS(2004) Transfers to states and municiualities. There are two types o f transfers from the federal government to subnational governments: 1) general transfers for equity and regional redistribution purposes, such as the Fundo de Particzpaca"o dos Estados (FPE) for states and the Fundo de Participaqiio dos Munickios (FPM) for municipalities; and 2) targeted transfers for specific types o f spending (such as for education and health). On general transfers, the 1988 Constitution established that 21.5 percent o f both the income tax (IR - Imposto de Renda) and the industrial products tax (IPI - Imposto sobre Produtos Industrzalizados) would be transferred to states (as the FPE - Fundo de Participaqiio dos Estados) and 22.5 percent each tax would be transferred to municipalities (as the FPM - Fundo development funds (FCO - Fundo do Financiamento do Centro Oeste, FNE - Fundo do de Participapio dos Munickios). In addition, 3 percent o f both taxes finance regional Financiamento do Nordeste and FNO -Fundo de Financiamento do Norte). In addition, 10 percent o f the IPI i s transferred to states to compensate states that export goods for the loss o f revenue associated with the state value-added tax (ICMS) deductions. Also, 50 percent o f the rural property tax (ITR - Imposto Territorial Rural) i s transferred to municipalities. Moreover, about 90 percent o f special levies on natural resource exploitation for water, electncity, and mneral resources are transferred to states and municipalities. Since 2004, 29 percent o f the government established a tax on gasoline and oil denvatives (Contribuz@io de Intewenpio no Dominzo Econc?mico, CIDE) i s transferred to states and municipalities (see below). Finally, the government distributes 100percent o f oil royalties to different units o f government. Five percent o f on-shore oil revenue i s transferred to states and municipalities (100 percent), while 80 percent o f off-shore oil revenue goes to these governments and the remaimng 20 percent goes to the Navy "Above normal" on-shore oil revenue i s transferred to states and municipalities (75 percent) and to the Fundfor Technology and Science Development (FNDCT) (25 percent), while above normal" off-shore oil revenue is shared as follows: 60 percent to states and municipalities; 15 percent to the Navy and 25 percent to the FNDCT In addition, a special royalty from oil revenue i s shared as follows: 50 percent to states and municipalities; 40 percent to the National RegulatoryOil Agency (ANP)and 10 percent to the Mimstryo f Environment (MMA). Specific Duruose transfers mainly aim at education. The federal Fund for the Development o f Fundamental Education and Teacher Valonzation (FUNDEF) i s legally mandated to receive 18 percent o f all tax revenue net o f transfers to states and municipalities. Thus, 9.5 percent o f the IR and 7 7 percent o f the PI i s directed to the FUNDEF. Moreover, 18 percent o f the financial operations tax (IOF) and 9 percent o f the ITR i s given to the FUNDEF In addition, as another 25 earmarlung mechanism for education expenditures, one third o f the salhio-educar$io contribution i s earmarked to the National Development Educational Fund (FNDE) and the remainder i s transferred to state education secretanats16 Other targeted transfers. In2001, the government established a tax on gasoline and oil denvatives (Contribuiqa`o de Intewenqa`o no Domini0 Econdmico, CIDE). Since 2004, 29% o f CIDE collection i s transferred to states and municipalities, and the rest (71 percent) i s directed to finance transport infrastructure investment, oil price subsidies, and environment projects related to oil exploration. Social contributions. The 1988 Constitution created the "Social Security Budget" to finance expenditures for social security, social assistance, and health. Ths budget i s financed by the following contributions: 100 percent of the social security contributions paid by private sector employees and employers to the Social Secunty General System (RGPS); 100 percent of the contributions of public servants to the Public Sector Social Secunty System (RPPS), 100 percent of the federal value added tax (COFINS - Contribuiqa`o Para o Financiamento da Seguridade Social), 100 percent of the tax on profits (CSLL - Contribuiqa`o Social sobre o Lucro Liquido, and 100 percent of the revenue from lottenes and other mnor revenue sources. Inaddition, 100 percent of financial transactions tax (CPMF - Contribuiqa`o Provisdria sobre Movimentaqa`o Financeira) is used to finance the Poverty Reduction Fund (25 percent), the Ministry of Health (50 percent) and to cover social security benefits (25%). Another value added tax (PIS-PASEP- Programa de Integraqa'o Social / Programa de Formaqiio do Patrimho do Sewidor Ptiblico) is usedto fund the FAT (Fundo de Amparo ao Trabalhador), unemployment insurance (Seguro Desemprego) and the salary bonus program (Abono Salarial) (60 percent of the PISRASEP), as well as to fund the BNDES (Banco Nacional de Desenvolvimento Econdmico e Social) (40 percent of the PISRASEP) which uses these finds to finance economc development projects, often lendingthe funds at below market rates. Other revenue collected bv federal agencies. Revenues collected directly by different agencies and entities are 100 earmarked to expenditures o f the collecting agency or entity (e.g. environmental licenses to IBAMA (the Brazilian Environment Agency), judiciary charges to the Justice Branch, etc). De-earmarlunn mechanisms - FSE/FEF/DRU. With the aim o f decreasing the level of earmarlung of the federal budget, in 1994 the government created a mechanism (called the Social Emergency Fund (FSE) to "de-earmarked" 20 percent of earmarked taxes and contrib~tions.'~ Initially the fund had two implications: 1) an increase in federal government resources due to a corresponding decrease intransfers to states and mumcipalities, and 2) increased flexibility of the federal budget. In 2000, when the mechanism was renamed the De-earmarking o f Federal Government's Revenues DRU (Desvinculuqa`o de Receitas da Unia`o), the new regulation excluded intergovernmental transfers (or general revenue shanng) from the eligible finds under the de-earmarlung process. Inthe event, the DRUhas not had much impact, because mandatory expenditures limted the flexibility the DRUwas supposedto provide (see below). l6FUNDEFis actually hndedby all three levels o fgovernment through the allocation ofresources according to a specific formula. 17The 1994FSE(Fundo de Social de Emerge^ncia)was renamed in 1996to the FEF(Fundo de EstabilizaGZo Fiscal) and to the DRU(DesvinculaGEo de Receitas da UniZo) m2000. In2003, the DRU was extended to end-2007 Discussions are continuing on the possibility to raise the DRUabove 20 percent. 26 Table A.l: FederalRevenueEarmarking (2004) Yoof Tax and other Revenue 2004 Earmarking Revenue Imposto de Renda (IR) 26.6 21.5% to states; 22.5% to municipalities; 3% to Fundo do Financiamento do Centro Oeste (FCO), Fundo do Financiamento do Nordeste (FNE), Fundo de Financiamento do Norte (FNO); and 9.5 to Fundo Nacional de Desenvolvimento e Manutenqiio da Educaqa'oFundamental (FUNDEF)* Contributions to the General 22.7 100% to National Institute of Social Secunty System Social Secunty System (RGPS) (100%9 I N S Contribuiqa'o Para 0 18.8 to Social Security and Social Assistance Financiamento da Seguridade System Social (COFINS) Imposto sobre Produtos 5.5 21-5%to states; 22.5% to mumcipalities; Industrializados (IPI) 10% to exporter states; 3% to FCO, FNE, FNO; and 7.7 FUNDEF Contribuiqa'o Provisdria sobre 6.4 25% to the Poverty Fund; 50% to the Health Mimstry Movimentapio Financeira and 25% Social Security and Social Assistance (CPMF) System Contribuiqa'o Social sobre o 4.7 100% to Social Secunty and Social Assistance System Lucro Liquid0 (CSLL) - Imposto sobre Operapjes 1.2 18% to FUNDEF; 30% states 70% municipalities Financeiras (IOF/IOF-Ouro) Programa de Integraqa'o Social / 4.7 60% to Fundo de Amparo ao Trabalhador (FAT), Programa de Formaqa'o do unemployment insurance, and the Abono Salarial; Patrimdnio do Sewidor Ptiblico 40% to Banco Nacional de Desenvolvimento (PISRASEP) Econdmico e Social (BNDES) Contributions of Public servants 1.8 100%to the Public Servant's Social Secunty System to the Public Sector Social Secunty System (RPPS) Contribuica'o de Intewenqiio no 1.9 71% to road transportation investments and Domini0 Econdmico (CIDE) environment programs related to oil exploration; 29% to states and municipalities Salhrio Educaqa'o 1.2 33% to the FNDE; 66% to states education secretanes Other (lottenes and others) 0.6 100% Social Secunty and Social Assistance System Imposto Territorial Rural (ITR) 0.1 50% to municipalities; 9% to FUNDEF Charges, licenses, etc 0.6 100% to the collecting agencies Oil royalties 1 100 percent to: states and municipalities; National Fund for Technology and Science Development (FNDCT); National Oil Agency (ANP);Mimstry o f Environment (MMA); and to the Navy Special levies on natural resource 0.0 90% transfers to states and municipalities exploitation (water, electncity, mneral resources) ' The earmarlung mechanism for educa 3n establishes that 18% o f the tax revenues net o f transfers to states and municipalities should finance the FUNDEF 27 MANDATORY EXPENDITURES The distinction between discretionary and non-discretionary expenditures is not obvious since some expenditure items may be discretionary in the short run but closer to non-discretionary in the medium or long run (for example, Bolsa Familia is currently classified as discretionary, however, these benefits become acquired nghts and over time permanent entitlements). In2004, federal discretionary expenditure was only 14 percent of pnmary (non-interest) spending. The most important mandatory expenditure items are: Public sector personnel expenditures and pensions. Since the late 1980s, there has been an increase inthe share of social secunty pensions intotal personnel expenditure from 26.8 percent in 1987 to 48 percent in 2004. From 1987 to 2004, expenditure on employees increased 74 percent inconstant pnces, while expenditure onpensioners rose 300 percent. Private sector social secunty - RGPS. Private sector social security expenditure increased from 2.5 percent of GDP in 1988 to 8 percent of GDP in 2005. Ths increase was due to: (i) increases inthe number of beneficiaries; and (ii) increases in benefit levels. The number of beneficianes more than doubled from 11.1 million in 1988 to 23.9 million in 2005, a 115 percent increase. Duringt h s period, the population increasedby 27 percent. The increase in coverage was due to bothaging and the adoption of more generous benefits since 1988. Constitutional requirement for a budget "floor" on health mending;. Constitutional Amendment No. 29 of 2000 established a mnimumamount of health expenditure for the federal government, to be increased each year inline with nomnal GDP growth through 2004. This amount i s subject to be revaluedthrough a Complementary Law. Unemplownent Insurance. The 1988 Constitution created an unemployment insurance program and the salary bonus program, Abono Salarial (with benefits linkedto the mnimumwage). Both are financed by the PISPASEP Beneficianes are formal sector workers with a monthly income equal or less than 2 mnimumwages; the benefiti s 1mnimumwage. Other mandatorv expenditures. Other mandatory expenditures comprise about 10 percent of total da Assut2ncza Social), the lifetime monthly income program (RMV - Renda Mensa1 Vitalicia) mandatory expenditures. Of these, the Organic Law of Social Assistance (LOAS-Lei Orgcinzca and the unemployment insurance program are the most important. The LOAS and the RMV are federal government social assistance programs. LOAS, approved in 1993, regulated the constitutional mandate for aged and disabled people with family income lower than one mnimum wage. The benefit i s equal to one mmmum wage and beneficiaries do not need to have made contributions to be eligible. The RMV existed before the LOAS with the same objective, and the LOAS program i s supposedto progressively replace the RMV The amount of bothprograms has steadily increased over time, from 0.27 percent of GDP in 1995 to 0.5 percent o f GDP in 2004. Another mandatory expenditure was established in the Lei Kandir (LC87/1996), which requires mandatory compensation to states for the loss inrevenues denved from the deductions they give on ICMS (state VAT) for exports. 28