Report No. 20754-BR Brazil Growth and Poverty Reduction in Pernambuco Developing a Sustainable Policy Framework January 28, 2002 Brazil Country Management Unit PREM Sector Management Unit Latin America and the Caribbean Region Document of the World Bank States of Brazil AC Acre PB Paraiba AP Amapa PE Pernambuco AL Alagoas PI Piaui AM Amazonas PR Parana BA Bahia RJ Rio de Janeiro CE Ceara RN Rio Grande do Norte DF Distrito Federal RO Rondonia ES Espirito Santo RS Rio Grande do Sul GO Goias RR Roraima MA Maranhao SC Santa Catarina MG Minas Gerais SE Sergipe MT Mato Grosso SP Sao Paulo MS Mato Grosso do Sul TO Tocantins PA Para Vice President David de Ferranti Director Vinod Thomas Lead Economist Joachim Von Amsberg Task Managers Mauricio Carrizosa CURRENCY EQUIVALENTS Currency Unit - Real (Rs$) EXCHANGE RATE 1998 Rs$1.21 = US$1 1999 Rs$1.79 = US$1 August2000 Rs$1.82 = US$1 WEIGHTS AND MEASURES Metric System FISCAL YEAR January I - December 31 ABBREVIATIONS AND ACRONYMS AHSFRA Administra,ao da Hydrozoa de Sao Francisco Sao Francisco Waterway Administration AIDS Auto Immune Deficiency Syndrome BANDEPE Banco Estadual de Pemambuco Pemambuco's State Bank BNDES Banco nacional de Desenvolvimento Economico e National Development Bank Social CELPE Companhia Energetica de Pernambuco Pemambuco Energy Company CES Conselbo Estadual de Sadde State Health Council CFN Companhia Ferroviaria do Nordeste Northeast Railway Company CHESF Companhia Hidroeldtrica do Sao Francisco Sao Francisco Hydroelectric Company CIB Comissao Intergestores Bipartite Two-Party Management Commission CIODS Centro Integrado de Operacoes de Defesa Social Integrated Centerfor Social Defense Operations CLT Consolida,ao de Leis Trabalhistas Consolidated Labor Code COMPESA Companhia Pemambucana de Saneamento Pemambucan Water Company CONDEPE Instituto de Planejamento de Pemambuco Pernambuco Planning Institute COSEMS Consellho Estadual de Secretarios Municipais de State Council of Municipal Health Saude Secretaries EMBRATEL Empresa Brasileira de Telecomunica,oes Brazilian Telecommunications Enterprise ENEM Exame Nacional do Ensino Medio Secondary School Test FCVS Fundo de Compensacao de Varia,ces Salariais Wage Change Compensation Fund FES Fundo Estadual de Sauide FGTS Fundo de Garantia do Tempo de Servi,o Time in Service Guarantee Fund FINOR Fundo de Investimentos do Nordeste Northeast Investment Fund FMS Fundo Municipal de Sauide Municipal Health Fund FNS Fundo Nacional de Saude National Health Fund FPE Fundo de Participa,ao dos Estados State Revenue Sharing Fund FPM Fundo de Participa,ao dos Municipios Municipal Revenue Sharing Fund FNDE Fundo Nacional de Desenvolvimento da Educacao National Education Development Fund FNE Fundo Constitucional do Nordeste Northeast Constitutional Fund FRANAVE Companhia de Navegacao do Sao Francisco Sao Francisco Navigation Company FUNAP Fundo de Aposentaduria de Pemambuco Pernambuco Retirement Fund FUNDEF Fundo de Manutencao e desenvolvimento do Basic Education Development and Ensino Fundamental e de Valoriza,co de Teacher's Incentive Fund Magisterio FUJNDESCOLA Fundo de Fortalecimento da Escola School Development Fund GDP Produto Intemo Bruto Gross Domestic Product GOP Govemo de Pernambuco Government of Pemnambuco HCR Proporc,ao de Pobres Head Count Ratio HH Domicilio Household IBGE Instituto Brasileiro de Geografia e Estatistica Brazilian Institute of Statistics and Geography ICMS Imposto Sobre Operacoes relativas a Circula,cao de Tax on the Circulation of Goods, on Mercadorias e sobre Prestaffes de Servi,os de Interstate and Intermunicipal Transportation Transporte Interestadual Intermunicipal e de Services, and on Communication services. Comunica,ao. ICRG Guia Intemacional de Risco Pais International Country Risk Guide IDB Banco Interawmericano de Desenvolcimento Inter-American Development Bank IGP-DI Indice Geral de Pre,os - Disponibilidade Intema General Price Index - Domestic Supply IMR Taxa de Mortalidade Infantil Infant Mortality Rate INSS Instituto Nacional de Seguridade Social National Social Security Institute IPEA Instituto de Pesquisa Econ6mica Aplicada Applied Economic Research Institute IPI Imposto sobre Produtos Industrializados Industrial Products Tax IPSEP Instituto de Previdencia Social do Estado de Pemambuco Social Security Institute Pernambuco IR Imposto sobre Renda e Provendos de Qualquer Income Tax Natureza ISS Imposto sobre Servi,os Tax on Services LO Land Owner MEC. Ministeriode Educa,ao Ministry of Education INEP Instituto Nacional de Estudos e Pesquisa National Education Research Institute Educacionais MOH Ministerio de Saude Ministry of Health NE Nordeste Northeast NGO Organi,acoes Nao-Gubernamentais Non Government Organization NOB Norma Operacional Basica Basic Operational Norm PACS Programa de Agentes Comunitarios de Saude Community Health Agent Program PAB Piso Assistencial Basico Basic Assistance Floor PDV Program de Demissao Voluntaria Voluntary Departure Program PETI Programa de Erradica,cao de Trabalho Infantil Child-Labor Eradication Program PNAD Pesquisa Nacional por Amostra de Domicilios National Household Survey PNAE Programa Nacional de Alimenta,ao escolar National School Lunch Program PNBE Programa Nacional Biblioteca da Escola National School Library Program PNLD Programa Nacional do Livro Didatico National Textbook Program PPA Plano Plurianual Multi-Annual Plan PRODEPE Programa de Desenvolvimento de Pemambuco Pernambuco Development Program PROFORMA(CAO Programa de Forma,cao de Professores em Teacher Training Program Exercicio PSF Programa de Saude Familiar Family Health Program PTA Parent Teacher Association RMR Regiao Metropolitana de Recife Recife Metropolitan Region SAEB Sistema Nacional de Avalia,ao da Educacao National Primary Education Evaluation Basica System SECTMA Secretaria de Ciencia, Tecnologia e Meio State Secretariat of Science, Technology Ambiente and the Environment SRH Secretaria de Recursos Hidricos State Secretariat of Water Resources SUDENE Superintendencia do Desenvolvimento do Northeast Development Superintendency Nordeste SUS Sistema Nacional de Saude National Health System TFP Produtividade Total dos Fatores Total Factor Productivity UN Na,oes Unidas United Nations VAT Imposto de Valor Agregado Value Added Tax WHO Organiza,ao Mondial da Saude World Health Organization Brazil Growth and Poverty Reduction in Pernambuco: Developing a Sustainable Policy Framework Table of Contents Page No. EXECUTIVE SUMMARY ............................................................................; Growth and Poverty Reduction Performance ............................................................................i Policies to Encourage Growth and Poverty Reduction ........................................................................... iv Possible Impact of Improved Policies ........................................................................... xiv 1. BACKGROUND ............................................................................1 INTRODUCTION ...........................................................................1I BRIEF HISTORICAL AND GEOGRAPHICAL BACKGROUND ............................................................................1 PERNAMBUCO TODAY ............................................................................3 2. PERNAMBUCO'S ECONOMIC PERFORMANCE .......................................................................7 THE GROWTH RECORD ............................................................................7 GROWTH FUNDAMENTALS ........................................................................... 19 THE POVERTY REDUCTION RECORD ........................................................................... 26 POVERTY FUNDAMENTALS AND OTHER CORRELATES OF POVERTY ......................................................... 28 CRIME AND VIOLENCE ........................................................................... 37 VOICES OF THE POOR ........................................................................... 37 PERFORMANCE OF EDUCATION SERVICES ........................................................................... 39 PERFORMANCE OF HEALTH SERVICES ........................................................................... 43 DELIVERY OF OTHER SOCIAL PROTECTION SERVICES ........................................................................... 44 3. GROWTH AND POVERTY REDUCTION POLICIES ............................................................... 49 A POLICY FRAMEWORK FOR IMPROVED PERFORMANCE ........................................................................... 49 THE GOVERNMENT'S FISCAL CONSTRAINTS ........................................................................... 52 THE GOVERNMENT'S DEVELOPMENT POLICIES ........................................................................... 65 INFRASTRUCTURE DEVELOPMENT ........................................................................... 65 OTHER PRIVATE INVESTMENT POLICIES ......................... .................................................. 74 EDUCATION POLICIES ........................................................................... 78 HEALTH POLICIES ........................................................................... 88 MUNICIPAL DEVELOPMENT ........................................................................9......................2............... ..... 92 LABOR POLICIES ..... 94 SOCIAL PROTECTION POLICIES ..... 95 SAFETY POLICIES ..... 95 REFERENCES ..... 98 APPENDIX A: ALTERNATIVE POVERTY MEASURES ................................... 102 APPENDIX B: ILLITERACY RATES ................................... 107 This report was written by Mauricio Carrizosa (Task Manager), based on the findings of a mission that visited Pernambuco during the months of December 1999 and January 2000, on additional research undertaken during January-July 2000, including background reports by team members, and on additional discussions with the Government in October, 2000. More recent developments and policy changes have not been reflected in this report. Mr. Gobind Nankani provided advice at various stages of the project. Background reports were prepared by Luiz Gabriel Azevedo (Water Infrastructure), Williarn Dillinger (Fiscal Performance and finalization of the report), Norbert Fiess (Growth), Yoni Sampaio (Education), Alberto Valdes (Rural Poverty), and Dorte Verner (Poverty, Labor Incomes). Anne Pillay provided final editing of the report. While this report has been discussed with institutions and individuals of the Brazilian Government and the State of Pernambuco, the views expressed in this report are exclusively those of the World Bank. BRAZIL GROWTH AND POVERTY REDUCTION IN PERNAMBUCO: DEVELOPING A SUSTAINABLE POLICY FRAMEWORK EXECUTIVE SUMMARY ES1. Despite its origin as one of the pioneering and richest states of Brazil, Pernambuco today has a history of slower economic growth than Brazil as a whole. Pemnambuco was born as a sugar producing state and the expansion of sugar production led its economic development until the mid-1600s, when the Dutch broke relations with Portuguese sugar producers and took their business, including their monopoly in international trade, to the Caribbean. Then Pernambuco declined, as sugar production became more profitable elsewhere in the world, with the volume of sugar exports declining sharply in the second half of the seventeenth century. It is estimated that, at least since 1939, the first year for which regionalized GDP data is available, Pernambuco's average growth rate has been lower than Brazil's, despite the fact that its per capita GDP is well below Brazil's average. Poverty data starts much more recently than GDP data. The story provided by the available data is that since the early 1980s, Pernambuco has made little or no headway in reducing absolute poverty. There was probably poverty reduction prior to the 1980s, when growth rates were high. The recent record of low growth and no poverty reduction motivates the present report. It focuses on (1) analyzing, in collaboration with the Government of Pemambuco, economic growth, and poverty in the State; and (2) identifying policies that can enhance economic growth and reduce poverty while improving fiscal performance. These two themes govern the structure of the report. The first chapter focuses on growth and poverty reduction performance and the second chapter focuses on policies to improve those performances. The findings are summarized below. GROWTH AND PO VERTY REDUCTION PERFORMANCE ES2. As with Brazil as a whole, Pernambuco's growth declined after Chart I - Pernambuco: Per Capita GDP: 1980. On an absolute basis, the Selected 1939-1998 Years state's per-capita GDP doubled 2000.0 1800.0 between 1960 and 1980, whilst it 1600°0 __ - only increased by 3.1% (see Chart en 1400.0 I) from 1980 to 1998. Moreover, 1200.0 , long term per-capita GDP growth in - 800.0 Pernambuco has been below the @ 600.0 average for Brazil and the 400.0 o- 200.0 I Northeast. Pernambuco's average 00 -, annual per-capita GDP growth from 1 GDPPE - 8 8 - ui 8 8 Year 1939 to 1997 (2.4%) was lower than Brazil's (2.9%) and the . Northeast's (3.0%). The modest economic expansion that has occurred since 1980 originates chiefly in the services sector. Both industrial and agricultural value added have declined. ES3. The lower growthI performance is also described by a Chart II - Pernambuco - Relative GDP Per perfong anerm decine ao d bedy stats Capita: Selected Years from 1939-98 Period long term decline in the state's 150.0% relative economic well-being (see 140.0% Chart II, which plots the ratio of 130.0% Pemambuco's per capita GDP to 130.0% the Northeast's per capita GDP). A E 120.0% simple convergence analysis, ' 110.0% indicating the relationship between .0I the growth rate and the initial level 100.0% of per-capita GDP, shows that 90.0% ' . . . . . . Pemnambuco does not appear to be O O C O co converging to relatively higher - PENE levels of GDP per head in Brazil; rather, other poorer Northeast Chart IlIl - Brazilian States - Growth economies appear to be converging Convergence: 1985-97 to Pernambuco's per-capita GDP. 6.0% The convergence analysis (Chart .5 5.0% III) shows that (i) Pernambuco , , . =-0.l2IL(x) +0.114| (indicated by a + sign in Chart III) - 4.0% is well below aggregate 'L 3.0% convergence lines (i.e. the lines ' 2.0% _ _ __ _ showing the relationship between < 0 y = -1 EO5x +00 growth and initial per-capita GDP c 1.0% for Brazil); (ii) other states in the . o.o% . Northeast have been converging to -1.0% 2000 4000 M0 80 lO0 1 )0 Pernambuco; (iii) the convergence 1985 Per Capita GDP line for the Northeast appears to be - Lnear (Northeast)-Log. (Other) | lower (i.e. lower growth at each I_I initial per-capita GDP) than the convergence line for the rest of Brazil, possibly indicating the adverse influence of geographical factors (chiefly climate). ES4. Many of the factors that help explain growth in international comparisons may Chart IV - Pernambuco - Headcount Ratio have negatively affected Pemambuco's by Poverty Line (L in 1996 Reais): Selected Years 1981-98 growth: macroeconomic imbalances; low 100.0% -._____ levels of human capital; trade, investment, Per 80.0% and labor market restrictions; policy ce 60.0% discontinuity; weak institutions; and adverse nt 40.0% j geographical factors (notably lack of water 20.0% and the prevalence of infectious diseases). 0.0% Some of these factors have been federally 1981 1985 1988 1990 1993 1995 1997 1998 determined (macroeconomic imbalances; *EL1=R$65 OL2=R$132UL3=R$205 Year iii trade and labor restrictions) but others may be subject to state policies (e.g. better conditions for private investment in water supply, more policy continuity, better police institutions). The next section in this summary indicates the policies that may address these factors. ES5. With low average growth, poverty in Pernambuco has not declined during the last two decades (Chart IV). The relationship between growth and poverty levels should be highlighted. This relationship is appreciated in Chart 19 (p.39), where states have been ranked according to their per-capita GDP. There, higher levels of GDP per capita tend to be associated with lower levels of indigence. Note, however, that Pemambuco, Sergipe and Bahia are examples of states that probably have higher than expected poverty levels. Other factors are at work, possibly low education levels. ES6. The standard human capital model provides a good explanation of relative individual incomes (Table 9, p.30) in Pernambuco, indicating that the poor are poor largely because they lack earning power from education. Completing 4, 8, 12 years of schooling, and university education in Pernambuco raises per capita household income respectively by 33%, 55%, 106%, and 219%. Chart 17 (p.30) shows the very close relationship that exists between average years of education and average wages across states in Brazil. Nevertheless, quantile regressions (Table 10, p.3 1) suggest that the poor get less out of education than the rich. Whether this is the result of lower quality of education, a less favorable family background, less health, or some other factor will need to be investigated further. A look at other correlates of poverty shows that: * household size is inversely related to per-capita household income; * poverty is largely rural (Chart 18, p.32), with evidence that proximity to urban areas brings to poor rural families income opportunities that provide higher returns, particularly in the non-farm sector; * rural poverty is partly associated with the small size of farms. To generate sufficient value added to guarantee a standard of living over and above the poverty line, specialists in Pernambuco recommend minimum non-irrigated farm sizes that range from 30-50 hectares in semi-arid regions. In Pernambuco, 72% of all farms are below 10 hectares in size; * formal sector jobs command much higher wages than informal sector jobs, even after accounting for experience and schooling in conventional regression equations. However, this may reflect "cherry picking" by formal sector employers or non-linearities in the return to education; * the poor have inadequate access to public services, thereby exacerbating their poverty; * Pemrnambuco's health outcomes are, in general, lower than those of the rest of the country (Northeast's and Brazil's averages). People live on average 4 to 6 years less than the average Brazilian. Infant mortality remains high (62% compared to 58% for the Northeast and 36% for Brazil in 1998) despite recent imp1ovements. These conditions are likely to affect mostly, but not only, the poor. * Crime appears to be correlated with poverty, although there is no adequate data or a diagnosis of victimization. iv ES7. Poverty also depends on the quality of social services. A review of the education sector shows that, despite recent improvements, there are large imbalances in performance, with particularly low performance of municipal schools (which account for 50% of enrollment). Outcomes in health have also improved, but they remain undermined by limited quality, efficiency, and financing of service provision, and by low levels of education and water supply, which have an adverse bearing on health. There are several social protection services, but their budget is limited and their impact needs assessment. POLICIES TO ENCOURAGE GROWTHAND POVERTYREDUCTiON ES8. In a growth and poverty reduction policy framework, investment in physical capital, which helps determiine the rate of growth, is affected by trade and investment openness, by taxes and public expenditures, and by institutional quality and political stability. Growth reduces poverty through its impact on employment and wages. Investment in human capital is partially determined by education and health policies. The level of human capital affects fertility, population growth and the dependency ratio, which also help determine the rate of growth. Labor market flexibility has a positive bearing on investment, employment and wages. Transfers to the poor (social protection services) and infrastructure services for the poor have a direct bearing on poverty levels. Figure I (p.50) summarizes these relationships. ES9. The experience of some of the high growth performers illustrates the framework. As examples, in Chile and Ireland, high growth has been attributed to (i) integration with international markets, through liberalization or multilateral agreements; (ii) investment openness; (iii) low labor costs/high labor quality; (iv) a supportive regulatory framework; and (v) fiscal sustainability coupled with low taxation. These examples are relevant for Pernambuco. First, more trade openness with other countries will benefit Pernambuco, which presently trades mostly with other states. Second, investment openness can also become a positive factor, for example in the water sector, where a major gap in supply exists. Third, Pernambuco will benefit from improving the quantity and quality of education, where it lags with respect to other states outside of the Northeast. And fourth, Pernambuco will profit from a more flexible labor market that allows higher investment to translate into a higher demand for labor and human capital. In addition, based on the endogenous growth literature, Pernambuco's growth performance could be improved by actions on other fronts. These include actions to raise the efficiency of public expenditures, to reduce the taxation of labor, and to improve institutional quality. These are discussed in Chapter II and are summarized further below. ES10. Pernambuco's present government, which took office in 1999, is undertaking important efforts to have the state depart from its past poor performance in terms of growth and poverty reduction. The Government feels that this poor performance is largely attributable to years of policy drift and is aiming at developing a stronger continuity in policies, which has benefited states like Bahia and Ceara. This will require a policy framework that is sustainable, i.e., one that succeeds in improving performance and thus in achieving a consensus on fundamentals among succeeding governments. The highlights of the Government's policies are: v * On fiscal policy, there have been successful actions to increase the primary balance, by containing wage and employment growth in the government. A higher primary balance will pave the way towards better access to credit and open more space for higher rates of public investment. * On public expenditure, the government has undertaken a major redesign of multi-annual planning, to make it more participatory and responsive to local needs; this could result in an improvement of the allocation and technical efficiency of expenditures. * On investment openness, the state has privatized its commercial bank and the electricity company, has developed private port services under the landlord model, and is considering privatization of its water company. This will allow to bring much needed financing for meeting the demand for infrastructure services, * On education, the state is developing collaboration with municipalities to transfer primary education to the municipalities in an efficient manner; aiming to improve school autonomy, with measures such as election of school principals; introducing testing; and improving teacher training. ES 11. The policies listed above are in the right direction, although the authorities recognize that much more needs to be done in these and other areas. The following paragraphs discuss these and other policies, including policies with respect to infrastructure investment, other private investment, education, labor market, social protection, and security. These policies are summarized in Table II (p. xv). ES 12. Fiscal Sustainability. Fiscal sustainability, requiring the ability to service debt without compromising essential public services or investments, is a fundamnental policy to sustain growth. Excessive debt will either lead to excessive debt service payments that reduce essential service provision or to default, which undernines borrowing for critical investments. Pemambuco's fiscal performance over recent years was undermined by weak revenue growth due to slow growth and tax Table I: Pernambuco: Upper subsidies and by excessive spending due to fast growth and Lower Paths of in personnel expenditures. As a consequence, the DebtlRevenue Ratio government's debt/ revenue ratio, which had declined Year Upper Lower from 1.30 in 1993 to 1.08 in 1995, increased to 1.40 in Limit Limit 1998 and to about 1.70-1.80 in 1999. If guarantees are 1999 1.70 1.70 included, the debt/revenue ratio would have been about 2000 1.77 1.75 2.00-2.10 in 1999. In 1999, the Government began an 2001 1 .67 1.57 adjustment program that decreased the state's primary . deficit from 8.4% of net revenues in 1998 to 2.6% of 2002 1.61 1.42 revenues in 1999. Pernambuco has signed a debt 2003 1.56 1.29 refinancing agreement with the federal government. 2004 1.49 1.15 Under this agreement, the Government is committed to 2005 1.38 1.02 reducing its debt revenue ratio over the next five years within the time paths indicated in Table I. This could bring Pernambuco to fiscal sustainability (defined as a debt revenue ratio of unity) by 2005. This adjustment would be faster than those of many of the other states (e.g. Rio, Sao Paulo, Rio Grande do Sul and Minas) that have signed debt refinancing agreements. Gross debt of the central state vi government increased from R$3,777 in December 1999 million to R$3,962 in September 2000, an increase of about R$200 million, much smaller increase than in 1998 (R$1.0 billion) or 1999 (R$500 million). Total gross debt (including debt of the central government and of decentralized entities) declined from its peak of R$5,257 million in May 1999 to R$4591 in July 2000. ES 13. To strengthen sustainability, Pernambuco's Government has successfully contained the growth of personnel expenditures, through reduction of special appointments (cargos de confianca) and restraint in wage and employment growth. With increased tax growth triggered by faster GDP growth in 2000, the Government had reduced annual personnel expenditures to 65% of net revenues by mid-2000. A reasonable target would be to reduce these costs to about 55% of net revenues (the level proposed for the 2001 budget), or about a 15% further reduction in the ratio. This will require both increased efforts to raise ICMS revenues, for example by 10% of total net revenues (or about 1% of the state's GDP), as well as efforts to deepen administrative reformn and reduce personnel costs, by say 5%. With 55% of net revenues in personnel costs and a primary surplus of 15%-20% of net revenues to help meet debt service obligations, the Government could possibly allocate around 15% of net revenues to investment, higher than the 2% in recent years. A sustainable effort to reduce personnel expenditures will need to address the imbalance between civil service retirement benefits and retirement contributions. The new fund that is being created does not fully address this imbalance, as it only suffices to pay for the future benefits of those that are about five years away from retirement. The resolution of the imbalance lies in new social security reforms that need to be undertaken at the federal level. ES14. Sustainability will also improve with a better allocation of public moneys. The state govermnent has undertaken a major effort to improve expenditure planning and implementation through its 2000-03 participatory PPA, which is being implemented in an increasingly decentralized fashion. The new PPA was prepared in close consultation with municipalities and civil society, in a process that yielded regionalized budget priorities as well as partnership agreements between the state and the municipalities. The new challenge for the Government is to introduce evaluation of programs on a systematic fashion to improve the technical efficiency of expenditures in future budgets. ES 15. Infrastructure Investment Policies. Pemambuco suffers from important shortfalls in the availability of infrastructure. A key policy to increase investment in infrastructure is openness to private sector investment and management in this area. This is so because the public sector is constrained in its capacity for direct investment and because efficient management of infrastructure facilities is best achieved by a private sector operation. ES 16. Transportation is an important area for Pernambuco's development prospects (see maps on pages 5 and 71) because trade is one of its key economic activities. Pernambuco will benefit from policies that open up trade. These policies are largely federal and include import restrictions and trade infrastructure. Import restriction levels are deternined federally and in the context of the Mercosul agreements. In this area, Pernambuco should provide support for faster trade liberalization, which will increase vii the amount of trade that goes through Brazil. To benefit from trade liberalization, Pemambuco must further develop its trade infrastructure, which includes telecommunications and transportation infrastructure. In telecommunications, Pemambuco is served by Telemar, a private fixed telephony company that replaced the state-owned TELPE and by two mobile companies. Pemambuco has 20 internet service providers connected through EMBRATEL and the National Research Network (Rede Nacional de Pesquisa). The Government is developing a state computing network to interconnect the state cities to each other and to the national backbone. ES 17. In ports, Pernambuco has made much headway. Pernambuco has two sea ports, the Federal Port of Recife and the State Port of Suape. Both ports follow the landlord model, whereby public facilities are rented to private operators on a long term basis. In 1998, these two ports moved a total of 6.7 million tons. This is 1.5% of Brazil's total of 443 million tons of cargo, less than Pernambuco's share of Brazil's GDP (2.7% in 1997). In contrast, the mostly private ports of Espirito Santo (with 1.7% of Brazil's GDP) moved 22% of Brazil's total cargo in 1998 (chiefly due to Vale de Rio Doce exports through the Port of Tubardo). Pernambuco, which arguably has an advantageous location for moving and trading cargo, should probably have a higher share in Brazil's total. To achieve this aim, it will need to become even more competitive in ports than it is today. This means increasing the capacity and efficiency of its ports and roads further. Although the ports of Pernambuco (Recife and Suape) are reportedly efficient by Brazilian standards, efficiency remains limited by legal constraints on the hiring of labor for port operation. Not only are these ports required to hire more expensive labor from restricted unions linked to the port authority, they must also abide by operational rules set by those unions. In contrast, ports owned by the private operators can hire labor freely and set rules under labor code constraints. Improved policy in the ports area requires federal action to improve the federal legal framework (i.e. the law setting the rules for employment in public ports'). In this regard, federal action to improve the federal legal framework would be advisable. ES 18. In roads (see Map on page 71), Pernambuco needs improvement in federal road BR-232, which crosses the state from east to west and connects to other federal roads to the north and south as well as to the state's feeder road network; and BR-101, which connects Recife to most of the other Northeast capital cities. The GOP, with the Federal Government and other states, are devoting resources to upgrade these roads. Other federal roads (e.g. BR-316 to Araripina and BR-110 (crossing through Pernambuco from Salvador in Bahia to Areia Branca in Rio Grande do Norte) also require repair. Several state roads also need work. Acceleration of public investment in roads is limited by present budget constraints on public sector investment, although the Federal Government is gradually upgrading the federal roads through five year maintenance contracts with private contractors. Financing through toll roads appears to be ruled out due to road users reluctance to pay tolls and to the high interest costs that toll concessionaires must pay. As long as these conditions remain, investment in roads is likely to remain slow. Eventually, Pernambuco could consider introducing a toll road program, like those that have been ' This issue is discussed at length in World Bank (1 999B). viii introduced in some of the other states. While these may result in increased fees paid by the consumer, those fees will bring forth better service and help overcome the financial constraint. This would increase investments in and maintenance of roads and reduce the pressure of these on the public budget. A toll program could begin by initially covering only part of the costs and gradually increasing to full cost recovery. ES 19. Complementing the State's road network is the federal rail from Recife to the city of Salgueiro in the west. Pernambuco is seeking the recovery of that stretch, construction of additional rail from Salgueiro to Petrolina, from Salgueiro to Missao Velha in Ceara, and from Salgueiro to the Gypsum pole in Araripe, and recovery of the line from Recife. It forms part of the long-standing Transnordestina rail project. The listed investments would cost some $716 million. The rail is in concession to Companhia Ferroviaria do Nordeste (CFN). The State, the Federal Government (including BNDES) are presently working with CFN to accelerate the execution of this project, which is regarded as a priority project by the GOP. ES20. Waterways can be another important area for transportation in Pernambuco. The main waterway relevant for Pemambuco is the Sdo Francisco river. It offers an inexpensive mode of transport between the Center-South and the Northeast of Brazil, with significantly lower fuel and labor requirements. However, to improve navigability, its waters need to be regulated and the riverbed needs to be dredged in some areas where the impact of erosion has made it difficult to navigate. The river is managed by the Administra,cio da Hidrovia de Sao Francisco (AHSFRA), an agency of the Federal Companhia Docas do Estado da Bahia. Given the importance of internal and international trade for Pernambuco, it would be important for the State to get more involved in improving the federal arrangements for assessing, financing, and undertaking needed works, including possible private sector participation. ES21. In addition to transportation, the other key infrastructure area for Pernambuco is water supply, both for irrigation and for human consumption. The GOP has been undertaking a number of policies to improve the water supply framnework, but more needs to be done. A key action is acceleration of the privatization process of COMPESA, the State's water company, to bring in needed investments that improve supply conditions of water for human consumption and for sanitation services, and to improve the operational efficiency of the company. In the meantime, the Government is obtaining federal assistance (R$183 million in 2001-02) to alleviate critical shortfalls. At the Government's request, the Bank is discussing possible Bank support in developing a broader medium- term water sector strategy for water and sanitation. In addition, more effort needs to be made to determnine the best use of the waters from the Sdo Francisco river for irrigated agriculture. While the old project which would deflect water from the river to the North needs to be better assessed, other options need to be considered as well. Pernambuco can also improve the use of water resources by beginning to charge for the use of bulk water and/or introducing a system of assignment of water rights. ES22. Other Private Investment Policies. Firms make their location decisions based on a number of criteria, including the size of the market and the availability of infrastructure, human resources, fiscal subsidies, and property rights. To attract investment, state policy Ix should focus on improving the openness for infrastructure investment, as discussed above, and to improve the state's human resources, through improved coverage and quality of its education and health services, discussed further below. Investment could also be attracted by actions to reduce crime rates, which would improve the enforcement of property rights. ES23. States in Brazil, have made a rather frustrating effort to attract private investment through tax exonerations. It is an effort where all states give up tax resources to attract investment, but few states achieve net increases in investment as states bid each other out. This results in a loss of public services or of public investment. A better approach would be to retain the taxes for public services and improve the allocation of expenditures on public services that will help attract investment. Hence it is suggested that Pernambuco stop granting tax subsidies to firms and use those resources in the other areas that will also attract investment, namely education, health, and improvements in security. ES24. This recommendation is made on four grounds. One, the size of those programs (about 3.5% of total investment) is likely to make the impact on growth rather limited. Two, the design of these programs typically involves picking winners, a very risky procedure in economic development policy. Three, it is far from obvious that industrial location has followed fiscal incentives; some of the other considerations relevant to industrial location and subject to public policy are equally or more important. Fourth, expenditure policies can attract investment to Pernambuco in more effective ways. Thus, a strategy that reallocates expenditures on tax subsidies to improvements in the state's human resources, on protection of property rights, and on other public goods could generate net gains to the state: public expenditures could be more effective and investors would still come in because the exemptions that they do not receive would be replaced by more effective public services. This would of course require a capacity to implement efficiently the alternative expenditures. Finally, it should be added, that the sustainability of investments attracted by subsidies is always in question, implying the likelihood of a permanent fiscal burden. If, despite the above reasons, it is felt that subsidies need to be in place to offset those of other states, a better approach would be to design a simpler subsidy that targets net additions in employment. An example of this approach, using an auctioned subsidy to minimize cost or maximize outcomes, is described in the text (para. 156, p.77). ES25. Another development policy has been to support synergies and economies of scale in selected clusters of private economic activity, including medical care and informatics in Recife, poultry in the Agreste region, mining (gypsum, marble, and granite), tourism in several places in the state, and fruits in the Sao Francisco Valley. These are sectors or clusters where growth has been high. As with tax subsidies to specific sectors, these strategies also run the risk of distorting economic activity (by assisting some sectors more than others) or of a poor return on public investments (by investing in a failing sector). An example of a cluster (informatics) is provided in Box 5 (p.79). ES26. Education Policies. Investments in education will be critical to determine the state's success in accelerating growth and poverty reduction. A diagnosis of the education x sector in Pernambuco shows that, while indicators of educational outcomes (literacy, average years of education, coverage, approval rates, drop out rates, test outcomes) have improved, quality and coverage still need additional improvement, with quality a major challenge in municipal schools and coverage a major challenge for secondary education. The GOP is undertaking important initiatives. These include municipalization of the first four years of education, with a children's schools format (i.e. one teacher responsible for all subjects); implementation of autonomous management in state schools; development of a training program; improvement of quality through evaluation; and allowing monetary incentives as a reward for improved efficiency. These initiatives are a step in the right direction, although care should be taken that municipalization of primary education be accompanied by better teacher quality at the municipal level. Furthermore, it is recommended that, (i) in improving quality through evaluation, Pernambuco start linking the evaluation process to wage policies and to administrative reform provisions regarding the performance of the civil service; and (ii) that Brazil start thinking of options to allow private schools to deliver public education. ES27. With regard to (i) above, it should be noted that the l9'h amendment to the Constitution allows the public sector to dismiss poorly performing employees. Regulation of this provision is being presently considered by Congress. It is unlikely that under this provision, many public employees will be fired. What may happen instead is that the threat of losing employment or earning low salary increases will help improve performance. Nevertheless, evaluation of teachers -- a controversial procedure even in the US -- will require adequate measurement of performance, including measurement relative to peer groups. But the challenges of adequate measurement of performance will need to be overcome in order to improve the strength of public schools and make them competitive with private schools in providing quality education. This is only fair if it is noted that it is public schools that provide education to the poor. ES28. With regard to (ii) above, its should be noted that the Constitution allows private provision of public education, but only on a rather exceptional basis. Article 213 indicates: * that public moneys may be used for community, religious, or philanthropic schools, provided that these schools (i) be non-profit (with financial surpluses spent on education); (ii ) commit their net worth to another eligible school in case of closure; * public moneys can also be used for scholarships based on need, when there are no available places in public schools and with the obligation that the public sector invest in expanding the public school system where there are insufficient places. These restrictions do not benefit the cause of education. Instead, the state should allow, as in health, the use of private facilities in providing public education. This would strengthen public schools by subjecting them to stronger competition and would also promote investment in private schools. Public moneys could then be used more freely to provide scholarships or to finance charter schools. The Constitutional constraints on the use of private schools for public education are helping undermine the access of the poor to better education. xi ES29. Health Policies. Pernambuco's health policy objectives most probably include further reductions in infant, maternal, and child mortality; improved nutrition; reduction of infectious, communicable and vector diseases; and reduction of chronic diseases and injuries (with victims of violence being the chief cause of hospital admissions). In some of these areas, Pernambuco has made much progress. For example, the program of Community Health Agents (initiated in Pemambuco) is believed to have contributed to the decline in child mortality. Policies to address Pernambuco's health challenges will need to focus on education, access to water and sanitation (where low access is a chief cause of transmitted diseases), adequate decentralization, intergovernmental definition of roles (who does what), improved institutional, managerial and administrative capacitv, improved relations between players, and improved contracts with service providers. Education and access to water and sanitation issues were discussed above. The following paragraphs focus on the other challenges listed here. ES30. As in the case of education, decentralization in health must proceed with caution. The pace and scale of decentralization in this sector has raised several concerns among some policy makers. Some doubt the technical and administrative capacity of municipalities. A recent report by the MOH's Auditing Department denounces municipal deficiencies in the areas of planning and programming; health counsel's operations; accounting and financial control; general management and management information systems; availability of municipal counterpart funding; and misuse of public funds2. Concerns have also been raised about high administrative costs, losses of economies of scale, and fragmentation that may arise from the small size of many municipalities. ES3 1. Excessive decentralization in health facilities may prevent municipalities benefiting from economies of scale. In Pernambuco, considerable investment in private sector health facilities, particularly in Recife, has taken place. These facilities are being used to service not only Pernambuco's population but also to export health services to residents from other states. As with other services, it makes sense to locate certain facilities in centralized areas and not expect every municipality to have access to all medical services. Hence, municipalities should be not be encouraged to establish self- sufficient facilities to service their populations but rather to make the best possible use of existing facilities in the state through adequate contracting. ES32. The assignment of roles to different government levels refers mainly to the assignment of management to different levels in the health sector. The transfer to municipalities of financial resources, of the public sector's infrastructure and human resources, and of resources for contracting with the private sector depends on the municipalities managerial capacity. Assignment of management to very small municipalities may also prevent economies of scale in administration. Hence, the Government's policy of establishing consortia of municipalities to administer certain health services may help make decentralization more effective. ES33. Much can be done with regard to institutions, management, administration and frnance. Institutional/regulatory areas that need attention include strategic planning and 2 Ministry of Health (1998). xii programming; auditing, licensing and accreditation of health facilities; and development of norms and protocols for service provision supported by evidence-based medicine. Technical assistance, training of health personnel, as well as improving the compensation system should also be emphasized. In the financing area, four areas merit attention: (i) improving the costing and cost-effectiveness of interventions (with an emphasis on prevention and a focus on risk groups); (ii) exploring new alternatives for financing and paying health services; (iii) developing adequate formulae for allocating financial resources within the state in view of improving access with equity; and (iv) developing the instruments and systems required to charge privately insured patients for services rendered by the SUS. Privatization or corporatization of publicly owned facilities, through legal arrangements such as organizaq6es sociais3 could also be explored. ES34. Improvement of relations between players, including contracts with service providers is a central challenge of Brazil's health system. Brazil needs to move from a system based on Government payments for specific medical services to Government payments for medical insurance coverage. Under an insurance-based system, the government does not need to determine or pay prices for specific interventions. These prices would be determined by the market, through the interactions between health insurers and service providers. When the government sets prices, providers are encouraged to make incorrect claims and the prices do not reflect a competitive mechanism that brings them to the lowest possible level. In a health insurance approach, health insurers would bid to provide services to municipalities. Implementation of this approach, however, needs to address a number of challenges, including the use of existing public sector providers (which could become organizaq5es sociais and work under contract), the lack of profitable basic facilities in many small municipalities, and provisions to contain the high costs of medical services that have resulted from high demand triggered by unrestricted insurance (as in the US). ES35. Labor Policies. Labor policies are primarily federal policies. The chief labor market issues include: (i) the heavy burden of non-wage costs, particularly the impact of taxes, including the income tax (15%), the Salario-Educaqdo (2.5%), social security contributions (31%), and contributions to FGTS (8%);4 (ii) constitutional constraints on the organization of labor unions (the unicidade sindical issue), which disallows labor choice with regard to the particular form and level of union organization; (iii) the weakness of collective bargaining; (iv) the juridical weakness of labor contracts, both individual and collective; and (v) the excessive role of labor judges (the poder normativo issue), which prevents negotiation and mediation between the parts in conflict. Resolution of these issues is being sought through labor reform. Measures that have been adopted to improve flexibility include the removal of wage indexation and the introduction of: profit sharing by labor, fixed term contracts, seasonal and cyclical allocation of work hours, part-time labor, and mediation in collective bargaining. 3These are private, not-for-profit organizations, operating under management contracts, which manage public infrastructure using government funds. 4 See Rodriguez (1998). xiii ES36. Reforms to improve labor market flexibility will help Brazil grow faster and will improve labor incomes, through increased wages and employment. Therefore, Pernambuco should lobby and support labor reforms that make it easier for employees and employers to agree on labor contracts. For Pemnambuco, a flexible and dynamic labor market will be the key to translate investment and education efforts into poverty reduction, as wages are the chief income of the poor. ES37. Social Protection Policies. Social protection programs target child labor, youth exposure to drugs and violence, old age, AIDS infection, homelessness, unemployment, training, rural community investments, distribution of seeds, agricultural infrastructure, technical assistance, and facilitation of land ownership. It was argued that the limited resources devoted to these programs (about R$100 million in 1995) are incapable of raising the welfare of the population of poor in any significant amount. Although some of these programs have been positively evaluated in their design, an impact assessment of the different programs would be useful to determine how these programs should be extended or reduced, depending on their effectiveness. For example, the program targeting child labor through improved incentives for education appears to be an important candidate for expansion, but an assessment of its impact should complement the positive assessment of its design. With an impact evaluation of these programs, the Government could consider some tradeoffs among public expenditures. One such trade- off would be between expenditures on the PRODEPE program (amounting to R$666 million per year), which are probably not a good use of public resources, and expenditures on the most effective social programs. ES38. Crime. Chapter 1 cites evidence that suggests that crime levels in Recife may be higher than in the other main metropolitan areas of Brazil. This is an area that demands far more attention than it has received thus far, although recent developments have brought the issue to the forefront of national public policy attention again. Attention to crime is important from the point of view of growth and poverty reduction. High crime levels deter investment. High crime levels are often linked to poverty, with the poor most often being victims of violent crime. ES39. Research shows that crime is often linked to poverty particularly under conditions of income inequality, so that low wages and unemployment are important factors at work. As discussed before, Pernambuco's poverty levels are among the highest in Brazil. Starting in 2001, the GoP launched a major program to address crime and violence in the State. About R$1 million is being invested in the computerization of crime statistics and in the training of law enforcement officers on the uses of information technology for crime prevention. Additional resources will be mobilized to fund surveillance equipment in crime-prone areas. Recife will be divided into different zones based on the severity of crime rates, and will involve the public in an overall campaign to fight crime. In addition, crime prevention operations will be centralized at a newly created agency-Centro Integrado de Opera,ces de Defesa Social (Ciods). The responsibility for ensuring public safety is presently being shared by three agencies: the Fire Department, Military Police and Civilian Police. The entire program is expected to be fully operational by the end of 2001, starting with a pilot in two of the highest crime areas. The GoP hopes to reduce crime statistics by 30% within two years. Nevertheless, Pemambuco needs to conduct a xiv victimization survey, collect data on causes of death and illness from the health system, and undertake prison surveys, in order to improve the policies to combat crime. ES40. The above approach can be combined with other policies to increase growth and reduce poverty that have been discussed in this study. The overall impact would be to improve the effectiveness of crime prevention and repression. In the latter area, the functioning of the newly created Ciods should be monitored to assess its impact in improving the crime statistics and containing the risks of victimization by police (which is high in Rio and Sao Paulo). POSSIBLE IMPA CT OF IMPRO VED POLICIES ES4 1. How much additional growth can be obtained from better policies? There is little empirical basis in Brazil to answer this question. One can observe, however, that the average expected yearly growth in per-capita GDP for Brazil's non-NE states is, at the same per-capita GDP level as Pernambuco, about 1 percentage point per year higher than Pemambuco's (Chart III, p. ii) This difference, if attributable to factors (like poor health and lack of water) which could be addressed by better policies, provides a base case order of magnitude for the possible impact of the improved policies, resulting in a base case enhanced per-capita GDP growth of 1.7% per year. Given the dispersion of growth rates across states, an extra percentage point due to good policies would also appear to be possible. This would give us a growth range of 1.7%-2.5% per year. Obtaining these growth rates during five years would raise per-capita GDP by 9%-13%. Based on the cross-state relationship between indigence and per-capita GDP, this growth range could imply a reduction in the ratio of indigent of 3% to 4% percentage points during a five- year period, reducing the indigence ratio from about 37% to 32%. xv Table II - Summary of Policy Actions and Recommendations Policy Area Government Actions Recommendations Infrastructure Energy and Privatized Telecom. Water Company GOP is considering privatization of COMPESA, based Accelerate privatization process. on strategy designed by BNDES. See. paras. 142-43. Water Resources Construction of adutoras, wells, channels and dams. See paras. 142-43; Transposition of the Sao Francisco River Undertake review of assessment of the transposition project. Ports Develop Port of Suape as Landlord Port Support change in Federal Ports Law to allow free hiring of labor by private operators Roads Upgrade key roads, including BR-232 and BR-101 Eventually introduce toll road program; Waterways Improve navigation conditions of the Sao Francisco Support development of framework for River, with possible federal PPA support. private sector participation in maintenance and investment. Private Investment Reform of PRODEPE Program First best: Abolish PRODEPE Program. If not possible, replace by simpler program focused on employment creation. Education Municipalization Municipalization of First Four Years of Schooling Improve Municipal Teacher Training School Implement autonomous management in state schools to Encourage and support development of Autonomy improve efficiency PTAs to increase impact of autonomy. Teacher Training Achieve federal goal of at least secondary education; Emphasize training of municipal teachers. Develop "competence and ability" training program. Testing Introduce evaluation in pilot group of schools Make testing, evaluation and feedback mandatory. Teacher Develop monetary incentives to teachers as reward for Link evaluation to wages; Incentives improved efficiency Implement Administrative Reform of Civil Service Performance. Private Schools Allow contracting with private schools (requires change in Federal Constitution) Health Decentralization Expand municipalization process, except for highly Avoid excessive decentralization that complex services, which would remain in the state. would result in loss of economies of scale in service delivery and in management. Management Seek improvements Contracting Introduce insurance-based contracts Prevention Emphasize basic attention. Social Protection Implement programs targeting specific vulnerable Undertake impact assessment of programs groups and extent successful programs. Crime Developed an information-based integrated police Undertake diagnostic study to formulate action strategy. crime prevention policy Fiscal Sustainability Undertake fiscal adjustment as per Debt Refinancing Meet agreed fiscal and debt targets Agreement. Introduce Participatory PPA Establish evaluation framework to improve delivery Brazil Growth and Poverty Reduction in Pernambuco: Developing a Sustainable Policy Framework 1. BACKGROUND INTRODUCTION 1. Despite its origin as one of the pioneering and richest states of Brazil, Pernambuco today has a history of slower economic growth than Brazil as a whole. Pernambuco began as a sugar producing state and the expansion of sugar production led its economic development until the mid 1600s. Then Pernambuco declined, as sugar production became more profitable elsewhere in the world. It is estimated that, since 1939, the first year for which we have regionalized GDP data, Pernambuco's growth rate has been slower than Brazil's growth, despite the fact that its per capita GDP is well below Brazil's average. Although poverty data is much more recent, the story provided by the available data is that since the early 1980s, Pernambuco has made little or no headway in reducing absolute poverty. These facts motivate the present report, which focuses on analyzing, in collaboration with the Government of Pernambuco, (1) economic growth, and poverty in the State; and (2) identifying policies that can enhance economic growth and reduce poverty while improving fiscal performance.5 These two themes govern the structure of the report, with the first chapter focusing on growth and poverty reduction performance and the second chapter focusing on policies to improve those performances. The remainder of this introductory section provides some historical and geographical background of the state. BRIEF HISTORICAL AND GEOGRAPHICAL BACKGROUND 2. Wood, Sugar, Coffee, and Cotton. Pernambuco (the name derived from the indigenous Tupi word Paranampuka, or "sea that hits the rocks", in one translation) is located at the Northeast tip of Brazil (see Map in p.5). After Brazil's discovery, foreign occupation of the Pernambuco area began in the early 1500s by Portuguese, Spanish and French traders, chiefly of Pau-Brasil,7 a valuable tree from which a red dye was extracted 5 This report complements a broader prior Bank report on poverty alleviation in the Northeast, as well as other state economic memoranda on Ceara. See The World Bank (1999), (2000A) and (2000B) 6 The Tupis, the best known indigenous group in Brasil, were the first to enter into contact with European settlers. The name of Pernambuco (also interpreted as buraco do mar [sea hole] from the Tupi word) was actually first given to the mouth of the Santa Cruz river, which separates the Island of Itamaraca (which was part of Pernambuco between 1763 and 1824) from the continent. See Correia de Andrade (1997). 7 Pernambuco's coast was first visited at the beginning of 1500 (January 26), before Pedro Alvarez Cabral's arrival in Porto Seguro, by the Spaniard Vicente Yanez Pinzon. (See Correia de Andrade [1997]). He probably disembarked at the Cabo de Santo Agostino (so named by the Portuguese when they arrived there in August, 1500, but initially named Rostro Hermoso, by a second Spanish visitor, Diego de 2 and used for fabrics in Europe. In 1528, the first permanent population nucleus, afeitoria, or trading post where traders bought the pau-brasil in exchange for other merchandise, was established by a Portuguese sea captain (Crist6vao Jacques) who commanded a Portuguese Crown expedition to discourage French pirates from engaging in trade with the colony. These early events marked the start of Pernambuco's long lived economic structure of primary production for trade as well as of commerce sector activities. 3. In 1532, Pemambuco became one of the fifteen Capitanias, or horizontally arranged stretches of land into which King Joao III divided Brazil to promote Portuguese occupation. These were donated with hereditary rights to those interested in exploiting the new lands. As the Pau-Brasil tree became progressively scarce in the coastal region, the Portuguese introduced sugar cane, the cultivation of which they had developed in the Azores.8 In 1535, Duarte Coelho, the beneficiary of the Pemambuco Capitania, arrived in Pernambuco to establish sugar mills. Slave-based sugar production progressively became the basis of a strong export-led economic expansion in Pernambuco (and of Sergipe, Alagoas and Bahia in the Northeast) until the mid-17th century. The number of sugar mills increased from five in 1550 to 144 in 1630.9 Prosperity was later contained by cheaper sugar production in the Caribbean, financed by the Dutch, after they lost control of and were expelled from Pernambuco in 1654.10 With the gold rush of the seventeenth century, economic development shifted to the state of Minas Gerais. By 1800, Pernambuco and the rest of the Northeast had become impoverished with respect to Brazil's now dynamic south. Nevertheless, sugar production continued to be prominent in Pernambuco (with 287 mills in 1774). It survives today as Pernambuco's main, albeit declining, agricultural product. For example, the quantity of sugar produced declined from 25 million tons in 1992 to 12 million tons in 1999. Yet sugar products still account for over 50% of Pemambuco's exports. 4. Other important products in Pernambuco's development were cotton and coffee. Cotton production was next in importance to sugar. An important autochthonous crop in the 16th century, it was initially overshadowed by sugar in the seventeenth century. Cotton became important again in the mid-i 700s, aided by stagnation in the sugar sector, and even more important in the mid-1800s, when "seedlings of herbaceous cotton from North America were introduced into Pemambuco.." and adapted well to diverse climatic conditions in the state.11 Although cotton production was largely slave-based by large landowners, it was also produced by small landowners, renters and sharecroppers. The expansion of cotton was triggered by increased population and domestic markets, as well as by access to booming export markets through the opening of Brazil's ports to friendly nations in 1808. The cotton boom was reached as a result of the US Civil War, which closed foreign markets to the US southern states. It remained an important product until Lepe). The Cabo de Santo Agostino belongs to the municipality where Pemambuco's Suape Industrial and Port Complex development enclave is located. See Gazeta Mercantil (1998), pp 16-20. 8 The Portuguese Crown established the first sugar mill in the Island of Itamaraca in 1516. See Correia de Andrade (1997). 9 See Correia de Andrade (1980). 10 The Dutch had invaded Pernambuco in 1630. The price of sugar declined by 1/2 between 1640 and 1660. " Correia de Andrade (1980), p.1 18. 3 about 1940. The two products competed with each other for available land. According to Correia de Andrade (1980), "(a)nyone studying the economic history of the Northeast will see that it can be summed up, in the last two centuries, as the struggle between sugarcane and cotton." 12 The demise of cotton in Pernambuco and the Northeast in general is attributed to the scourge of the bollworm and to expanded production in Sao Paulo. Introduced in the late 18th century, coffee production became important in Pernambuco in the mid-i 800s. It began to decline in 1995, when the Government paid coffee growers to remove coffee trees. Production of both cotton and coffee has virtually disappeared. PERNAMBUCO TODAY 5. Pernambuco today is a low middle-income state with a GDP of about R$25 billion in 1998, or 2.7 percent of Brazil's GDP and 21.0 percent of the Northeast's GDP.13 Pernambuco's per capita income (R$3,196 in 1998) is the highest in the Northeast region of Brazil, followed by Bahia (R$2,971) and Sergipe (R$2,941). Pernambuco's total population is around 7.5 million, or 4.7 percent of Brazil's total, and the annual population growth rate (for people older than 5 years) is low (0.5%). Pernambuco's population is larger than or equal to several other Latin American Countries (e.g. Bolivia has a similar population and about a third of Pernambuco's GDP, despite its much larger area). The economically active population accounts for 58% of the population in the State and has increased 2.7% per year in the 1990s. Forty-one percent of the State's population and 60% of the state's workers are in the Recife Metropolitan Area, which accounts for 70% of industrial value added. The population of Recife grows at 1.1% per year. Most of the state's GDP originates in services (about 49%, including the Government); industry accounts for 330% and agriculture for 9%. 6. Pernambuco's area today is 98,938 square km, or about the size of Guatemala, Honduras, or South Korea. 4 Throughout its history, Pernambuco's area was progressively reduced. By the 17th century, Pernambuco became Brazil's most extended general Capitania (with jurisdiction over the dependent Capitanias of Ceara, Rio Grande do Norte, and Paraiba as well as the Comarcas of Alagoas and Rio San Francisco). However, by 1817, it had lost jurisdiction over its dependent Capitanias. It also lost the Piaui area in 1715 as well as the area between the San Francisco river and its tributary Carinhanha River in 1721. In 1824, Pernambuco lost the Sao Francisco River Comarca. Much of this reduction in Pernambuco's prominence is attributed to retribution for its revolt against the Portuguese in 1817. 7. The state is divided into five regions (see Map in p. 6, which also shows the 10 subregions, as used in the decentralization of the state's budget; see para. 119): 12 See Correia de Andrade (1980), p.58. Calculation based on the newly released Conta Regionais do Brasil (1999) for 1997 and extrapolated for 1998 using annual GDP growth rates fiom SUDENE and price deflators from IBGE. According to Contas Regionais do Brasil (1999), the per capita incomes in Pernambuco, Sergipe and Bahia in 1997 are as follows: 3,145, 2,900 and 2,890. The corresponding figures for Brazil and NE are 5413 and 2494. 4 Pernambuco's population size of 7.5 million is similar to Bolivia's and the Dominican Republic's. 4 * The Recife metropolitan area (RMR) region (2.8% of the state's area; 41.8% of population), at the east tip of Brazil's coast, accounts for most production, with a diversified industrial structure, a modem service sector (health and software production), commerce, tourism, and a major regional port (SUAPE). Poultry, honey, flowers and sugar cane are the main agricultural activities. * Surrounding Recife is the forested Mata region (8.6% of area; 15.5% of population; divided into the Mata Norte and Mata Sul subregions), a sugar cane plantation economy with beverage, alcohol, sugar, shoe and ceramics industries and diversified retail trade activities. Poultry is another important agricultural activity. * West of the Mata region is the intermediate Agreste region (24.7% of area; 25.3% of population; divided into the Agreste Setentrional, Agreste Central and Agreste Meridional subregions), with sufficient rainfall for most crops (not for sugarcane); it includes the largest milk basin of the state (to the south); production of fruits, vegetables, poultry and brown sugar; manufacturing (textile, apparel, furniture, paper, and ceramics); mining (granite) and diversified wholesale and retail trade. * West of the Agreste the dry semi-arid Sertdo region (39% of area; 11.7% of population; divided into the Pajeu/Maxot6 and Araripe subregions), with mostly agriculture, extensive-based sheep raising, poultry, apiculture, and mining (marble, iron, gypsum), and pisciculture (by the San Francisco River). * South of the Agreste region is the Sao Francisco region (24.5% of area; 5.7% of population; divided into the Itaparica and Sao Francisco subregions), with large scale irrigated agriculture (chiefly fruits and horticulture), agri-industry, pisciculture, apiculture and textile/apparel manufacturing. 8. As is typical of the Northeast, the regions within Pernambuco that are most dynamic economically are the urban areas (most specifically Recife) and the irrigated Sdo Francisco area. Growth in the Mata and semi-dry regions has been much slower. 15 9. As suggested by the description above, Pernambuco has come a long way from being a sugar cane economy. Whereas in 1800 Pemambuco was primarily a sugar producer, today most economic activity is in the service sector (66.8%), and agriculture accounts for only 8.4% of the State's GDP. The following chapter analyzes the State's economic growth and poverty trends under the state's more diversified production structure. 15 See Maia and Vergolino (1995). | Ministerio \ ~~Sao Luis T_ _sportes \ 4- F dirC,°r ra ~PRINCIPAIS RODOOVAS FEDERAIS Sar ; 2i>' > 2 Fortaleortalza REGiAO NORDESrE T., ; ; = - r,;,y6;.,t, Duir34 | 1 i l t u s 22;, | - 2 ' 1 '5' \ dG T pt-- *HOO 3 V6US 3-Rr~ tt>. '~ l~ O_N3WIA1OANi5Q 3U3 S30IO3d 3 S3OIOUOS3W 0 flnSWWNZIN 7 2. PERNAMBUCO'S ECONOMIC PERFORMANCE 10. Growth and poverty reduction are closely related. A high rate of growth greatly improves the odds for poverty reduction. This chapter analyzes first Pernambuco's growth performance and then discusses the state's poverty reduction record. The overall aim of the chapter is to identify the factors that have been key in determining Pemambuco's rather disappointing economic performance as well as those that may help to improve this performance. THE GROWTH RECORD 11. This section describes Pernambuco's long term growth record, in terms of both absolute and relative (to other states) progress in GDP. The section also covers GDP's sectoral performance, as well as the demand and supply sources of growth. It concludes with a discussion of output convergence models and their implications for understanding growth in Pernambuco. 12. Pernambuco's long term growth record. In 1939, the first Chart 1 - Pernambuco: Per Capita GDP: year with available information on Selected 1939-1998 Years economic activity at the state level, 2000.0 Pernambuco's per capita GDP in 1600.0 _ * real terms was about US$735 at the u, 1400.0 ;--- ; : * ' '-| - 1995 exchange rate, or about 63% ¢ 100.0 -t- --; -r.j- o~1000.0 of Brazil's per capita GDP at the - 800.0 time. Chart 1 shows that per capita a: 600.0 GDP in Pernambuco (i) increased 2000 prior to Brazil's "lost decade" - the 0.0 X 1980s; (ii) decreased during the *GDPPE CD 8 a a Year 1980s; and (iii) recovered after D 0 0 0 C C - 1990. 16 Therefore Pernambuco's GDP followed Brazil's overall trends. Indeed, as discussed later in this section, there is a close statistical relationship between Pemambuco's growth and Brazil's aggregate growth, reflecting the response to common macroeconomic forces. 13. Despite Pernambuco's progress in the level of its GDP, its relative position within Brazil has deteriorated. Given the historical background discussed above, Pernambuco's long term progress before 1939 (and probably since the seventeenth century) was probably slower than Brazil's average. Since 1939 Pernambuco's economic progress, while positive, has also been relatively poor compared to Brazil's overall growth. For the observed 51-year period (1947-98) Brazil's per capita average GDP growth averaged 16 The GDP series that is the basis of Chart I is a spliced series from Azzoni (1997) through 1985; the recently published by IBGE (2000) for 1985-97; and SUDENE (2000) for 1998. In turn, Azzoni's series was assembled using the state shares in total GDP from 1939-68 data prepared by the Funda ao Getulio Vargas; and 1970, 75, 80-85 data prepared by IBGE. Extrapolations to 1990, 1995 and 1997. 8 3.5% while Pemambuco's per-capita growth averaged 2.9%. Recent performance (1985- 98) maintains the gap, with Brazil's per capita GDP growth at 2.6% and Pernambuco's at 2.0%. The difference between Brazil's total average growth (6.0%) and Pernambuco's (4.6%) has been even larger. Pernambuco's relative GDP position is depicted in Chart 2, which shows the ratio of Pernambuco's per-capita GDP to per capita GDPs in other aggregates and states. Back in 1939, Pernambuco's per capita GDP was 63% of Brazil's per-capita GDP; today this percentage has declined to 47%. Its position with respect to the Northeast and center west aggregates has also declined. Ceara provides a dramatic comparison: whereas, in 1947, Pernambuco's per capita GDP was estimated to be 80% above that of Ceara, today Ceara has almost caught up with Pernambuco. Similarly, the comparison with the Center-West is also Chart 2 - Pernambuco - Relative GDP Per Capita: impressive, with Selected Years from 1939-98 Period Pemambuco falling behind 200.0% by about 50% between the 180.0% - - 1950s and today. 140.0% _ .. . - Understanding 120.0% 7 - Pemambuco's relatively C 1000% s_ . 80.0% 7 slow performance is an 60.0% important challenge that 40.0% :, m:. hopefully will identify 20.0% ; ; Year policies that can improve 0.0% that performance. One U a ¢ Ln ) 0 - CO co . (D CD (D can't be too sanguine -CPE/Brasil- *PEINE PE/CW PE/BA ---- PE/CE about the data that is generating these results (although the state level data has improved much recently at least in terms of consistency with overall National Accounts), but the results confirm at least the perception that Pemambuco has not been a star performer in terms of economic growth. The results are also consistent with forces and fundamentals behind economic growth, such as the decline in the role of sugar. 14. Sectoral Performance. A discussion of Pernambuco's structure of output sheds light on the State's sectoral weaknesses and strengths: declining agriculture and industry and strong growth of services, much like the average for Brazil. While sugar, cotton, and other agricultural products were important components of GDP in the past, today the service sector has become the most important in the state' s GDP. Nevertheless, agriculture remains important in that most poverty is concentrated in that sector. Also, the sugar sector remains important in that it is the main Pernambucan export (53.1% of the total exports). 15. As the GDP data recently calculated by IGBE shows (Table 1), Pernambuco's tertiary sector (services) accounts for over one half of the state's value added. Moreover, the share of services has increased through time, while industry and agriculture have declined. These trends are confirmed by state level GDP data calculated by SUDENE (Table 2), which provides a longer time span (1970-95) than the new IBGE data (1985- 9 Table 1 - Pernambuco - Structure of GDP: 1985, 1997 (IBGEj Pemambuco Absolute Share PE Share/Brazil Share Sector/Year 1985 1997 1985 1997 Agriculture 14.0% 9.2% 1.24 1.20 Manufacturing 25.4% 16.7% 0.79 0.71 Mining 0.1% 0.1% 0.02 0.12 Electricity, Gas and Water 1.5% 2.3% 0.56 0.66 Construction 8.5% 13.8% 1.52 1.42 Commerce 14.2% 12.1% 1.59 1.54 Lodging and Food services 2.7% 3.9% 1.65 2.14 Transportation and Storage 3.2% 2.0% 1.12 1.00 Communications 0.6% 2.00/0 0.58 0.98 Financial Intermediation 7.3% 2.8% 0.55 0.44 Real Estate, Rentals, and Services to businesses 5.8% 11.2% 1.10 0.78 Public Administration 13.3% 17.7% 1.47 1.19 Education 2.0% 3.0% 1.17 0.93 Other Collective, Social, and Personal Services 1.2% 2.7% 1.34 1.83 Domestic Services 0.5% 0.5% 0.89 1.00 Agriculture 14.0% 9.2% 1.24 1.20 Industry (Manuf., Mining, Construction, Utils.) 36.0% 34.9% 0.80 0.88 Services 50.1% 55.9% 1.14 1.06 Total 100.0%° 100.0% 1.00 1.00 Source: IBGE, Contas Regionais do Brasil, 1985-1997 Table 2 - Pernambuco - Structure of GDP: 1970, 1980, 1990, 1995 (SUDENE) Year 1970 1980 1990 1995 Agriculture 14.1% 13.2% 10.1% 6.6% Mining 0.0% 0.1% 0.0% 0.0% Manufacturing 20.4% 22.0% 16.9% 12.8% Energy/Water 2.1% 2.5% 3.4% 3.0% Construction 6.3% 3.6% 5.2% 4.7% Commerce 17.3% 20.1% 17.2% 22.9% Transport and Communications 3.7% 4.7% 3.1% 3.0% Finance, Real Estate and Business Services 18.3% 20.7% 23.3% 27.0% Community, Social and Individual Services 17.8% 13.2% 20.8% 20.0% Total 100.0% 100.0% 100.0% 100.0% Of which: Government Consumption n.d. 12.8% 15.4% 14.6% Agriculture 14.1% 13.2% 10.1% 6.6% Industry (Mining, Manuf., Energy/water, Const.) 28.8% 28.2% 25.6% 20.5% Services 57.1% 58.6% 64.4% 72.9% Source: SUDENE Website Data 97). Under this calculation, the service sector is even larger, accounting for almost three quarters of GDP.17 Note also, from the last two columns of Table 1 that Pernambuco has 7Communications is included in the services aggregate calculated from the SUDENE data, as it comes together with transportation. However, this overestimates services by little (maybe 1% or 2%), judging from the share of communications in the IBGE data. The discrepancy on the share of services is not surprising, for it is acknowledged to be the sector for which it is most difficult to obtain accurate estimates. 10 relatively more agriculture, less industry (except for construction, where it has more) and more services than the average for Brazil. Indeed, the Government feels that PE's comparative advantage lies largely in services, and this is reflected in several of the development poles (medical, informatics, and tourism) that the Government is promoting through a cluster approach. The following paragraphs describe the agricultural, industrial, and services sectors in more detail. Agriculture 16. The declining share of agriculture in Chart 3 - Pernambuco - Agriculture GDP: the state's GDP noted above reflects 1970-1999 Penambuco s stagnant agricultural 2000 production. The growth of Pernambuco's 1500 agricultural sector ceased in 1985 and, looo since 1992, agricultural GDP has been 500 gradually declining. By 1998, the 0 production of all major grain and industrial _- crops had fallen to unprecedented lows. O w aM D N cJ CD There are two principal types of agriculture IAgricuKureXP(SUXNP -AgrkuKureG[(IBl in Pernambuco: subsistence and commercial. Collapses in the production of two sectors in particular, cotton and sugarcane, have affected both subsistence and commercial farming. This crisis in the two sectors triggered far reaching regional and state-wide consequences in terms of both rural employment and poverty. The crisis is reflected in the SUDENE estimates of agricultural GDP, which are depicted in Chart 3*18 17. The analysis shows declining land use (e.g. a decline of 33.4% in crop area), with some of the more drastic declines occurring in industrial crops (e.g. cotton, coffee, castor oil beans, and cassava). The analysis also shows declining yields in cotton, with the collapse of the cotton sector linked to (a) the incidence of the bicudo pest (cotton weevil) on the cotton arboreo, (b) a fall in the world price of cotton and, (c) according to some experts, changes in the labor code (which resulted in a sharp rise in the payroll tax increasing the cost of labor). Industrial crops with stable yield include coffee, sugar cane, and cassava, while castor oil beans, corn, sorghum, garlic, and pineapple show unstable yields. Yields have risen in onion, grapes, tomato, and rice. 18. A detailed examination of the grape sector shows a considerable expansion linked to the development of large scale canals that enabled wider irrigation from the Sao Francisco River, although the expansion is somewhat threatened by declining prices and by the strong competition of Bahia's higher yields. In contrast, a detailed examination of the sugar sector shows a boom that is linked initially to an expanding exports market (responding to a sugar price boom; see Chart 4) and then to the introduction of the Federal PROALCOOL program, a program of alcohol production subsidies that Brazil introduced in 1975 in response to higher oil prices. Sugar prices then declined (Chart 4) 18 As a matter of interest, it is noted that IBGE estimates of agriculture GDP (also depicted in Chart 5) differ substantially from SUDENE's, particularly after 1993. I1 and Proalcool program was progressively withdrawn during 1986-95. The decline in production has resulted in a major decline of agricultural employment. Since the hilly sugar region in Pernambuco is not easily mechanized, it has not benefited as much from mechanization as other regions in Brazil. Hence Pernambuco has lost its Chart 4 - Real International Free Nbrket competitive edge in sugar, and land Pnce of Sugar owners are beginning to respond by 2o co diversifying to other crops. 9j I _)----, Industry 19. There have been close links ,, '_ i- between the development of u.w I Pernambuco's manufacturing sector and (D C(D the development of its agriculture. The closest are those between sugarcane and _ _ l sugar, and cotton agriculture and textiles. Indeed, it can be said that industrial development began in the 16"' century with the development of the sugar "industry", which was triggered chiefly by world demand. The cotton industry, after an early attempt in 1826, really took off at the end of the 19th century, triggered also by world demand for cotton, specially from England, and bv local demand for textiles. 20. Important Pernambucan manufacturing today includes, in addition to sugar and textiles, chemicals and petrochemicals (with gas and alcohol as the main products). pharmaceuticals, wood products (including a paper factory), food products, non-metallic products (mainly cement, glass, and plastics), beverages (two soft drink competitors) and tobacco, autoparts (mainly a car batteries factory) and transport materials, and appliances. Although sugar and textiles remain at the top in terms of assets or gross revenues, investment levels are highest in the chemicals and petrochemical sectors. Other important industrial sectors include construction and utilities, which responded to the growth of Pernambuco's domestic demand. Mining has never been important (less than 0.1% of GDP), except for today's growing Gypsum development pole, 312 Chart 5 - Pernambuco - Agricultural and firms employing 12,000 people that Manufacturing Production (1991=100) meet most of Brazil's demand for the 140 product. 120 _ 100 21. Chart 5 shows the 80 ----- - -- - performance of agricultural, 60 _ _ _ -__ -__.*- -_- manufacturing, mining and service 40 _- _-_."' -- sector production in Pernambuco 20 _ ____ since 1981. Both agriculture and 0 Year manufacturing have declined since C c ( ( (1 1987. Mining also declined during v _ . __ - - @ _ the 1990s (data for the 1980s is __services n Agriculture unavailable). In contrast, the other _ _ _ _ _ _-Xx- 12 industrial sectors, including construction and utilities services (except water), and the services sector continued to expand. These developments reflect; first, the impact of declining subsidies and competitiveness in sugar, which helps explain the decline in sugar production and part of the decline in the chemical sector; second, the impact of the disappearance of cotton cultivation due to pests, which may explain why the textile industry declined by about 1/2 between 1990 and 1998; and, third, more generally, the real currency appreciation that discouraged the production of tradeables during the period. In contrast, the sectors that expanded (construction, utilities, and services) correspond more to non-tradeable goods and services that would have been encouraged by the currency appreciation. Services 22. The tertiary sector plays a prominent role in Pernambuco, accounting for about 56% of the state's GDP (IBGE), slightly higher than Brazil's average share (53%). Table 3 - Pernambuco - The Tertiary Subsector As discussed before, the share of the state's Tertiary Sub-Sector 1985 1997 tertiary sector in the State's total GDP has Commerce 28.3% 21.7% been expanding, from 50% in 1985 to 57% Lodging and Food 5.4% 6.9% in 1997. Based on IBGE's regional Transportation and Storage 6.3% 3.6% accounts, the largest sub-sector share is the Financial Intermediation 14.5% 5.1% public sector (Table 3), which expanded Real Estate 11.6% 20.0% through 1996 and accounted for about a Public Sector 26.6% 31.6% third of the total tertiary sector in 1997. Private Health/Edu 3.9%i 5.4% The commerce sector follows in Other Social/Personal Services 2.4% 4.8% importance (22%), although its share has Domestic Services 1.0% 0.8% declined. By 1997, the share of the Total 100.0% 100.0% expanding real estate services subsector Source: Contas Regionais do Brasil, 1985-1997 (20%), which includes real estate brokerage, management, and rental, had almost caught up with the share of commerce. This expansion mirrors the expansion of construction noted before. As with the rest of Brazil, the share of the financial services sub-sector declined with stabilization, and accounts for only 5% of the total tertiary sector. The share of lodging and food (6.9% in 1997), which captures tourism, has expanded only mildly; it is, nevertheless, a more important sub-sector than the average for Brazil. The share of private education/health expanded somewhat in 1996 and 1997, possibly reflecting the growth of the private polo medico (with 350 hospitals and clinics in 1999). 23. Sectoral Prospects. In broad terms, Pernambuco's economy is likely to continue to be a largely services based economy. Agriculture will continue to suffer from diminished competitiveness in its traditional crops (sugar and cotton) and from the state's limited endowment of land.. Nevertheless, new products are diversifying the sector and tempering the decline in its share in economic activity. Fruits (grapes, mangoes, melons) in the Sao Francisco Valley and cattle, coffee, and rubber in areas formerly dedicated to sugar, are examples of diversification, partly in response to the adverse shocks suffered by traditional crops. Some investments in manufacturing are taking place, for example, to 13 take advantage of the Suape complex, Table 4 - Pernambuco which provides firms with easy access to Estimated Investments: 1998-2005 (US$Million) foreign and domestic markets, or to General Services 997 develop the wine industry from the Sdo Chemicals and Petrochemicals 892 Francisco grape sector. Of some Transport and Storage 548 importance for agricultural and industrial Public Services 447 development may also be the reversal in the Food 338 real exchange rate, which until 1998 Beverages and Tobacco 163 discouraged the production of tradeable Textiles and Leather 126 goods and favored the expansion of the Pharmaceuticals 122 services sector. With the devaluation in Wood Products 64 1999, production of tradeables for export or Non-Metal Products 61 domestic demand may have become more Metallurgy 50 attractive. Yet services will continue to be Plasticas and Rubber 44 at the forefront. An indication of this Construction 29 prominence is that more than half of Machinery 25 expected major investments in Electric and Electronics 25 Pernambuco, as reported by Gazeta Finance 17 Mercantil (Table 4), are in services. The Transport Material 12 remainder is in industry, with most Communications 10 explained by one large investment in the Informatics 5 petrochemical sector. Vehicle Distribution 4 Total 3979 24. Capital Per Worker. To some Source: Gazeta Mercantil, Balan~o Annual Pernambuco extent. Pernambuco's relatively low per- (1999) capita GDP growth is expected to reflect relatively low rates of per capita capital Chart 6 - Northeast - accumulation. Although time series data Investment/GDP and GDP Growth: does not suggest a strong impact of capital 1970-75 accumulation, cross section comparisons I 3 bring out the effect quite strongly. There h 2 5 . 0 - * are no available statistics on physical I -. investments for Pernambuco alone. X r no, SUDENE data on total investment in the o. Northeast throws little light on the role that D investment plays in explaining year to year growsth in per-capita GDP (Chart 6).20 Also, o _____._L human capital accumulation in Pernambuco -5.0% 0 0% 5.0% 10.0% 15.0% during 1981-1998, as measured by the Investment/GDP change in average years of education of the r Investment/GDP(5 Year Moving Average) labor force has been similar to the average - Linear Trend for Brazil (i.e. a change of 1.8 years) and 9 Maia Gomes and Vergolino (1995) argue that lower growth in the Northeast after 1986 is linked to lower investment rates. 20 A regression of per-capita GDP growth on log investment per-capita and lagged per-capita GDP growth does not show investment as having a significant effect. 14 somewhat higher than the average change in the Northeast (1.6 years). Again, this does not suggest a strong impact that would explain Pernambuco's relatively low growth. Using cross section data, however, that states with higher product per worker have higher levels of both physical and human capital. A statistical comparison indicates that I % increase in average years of education is associated with a 0.3% increase in output per worker and that a 1% increase in the physical capital proxy (electric energy use) is associated with a 0.1% increase in output per worker. In contrast to the time series regression, the cross-state comparison strongly suggests that Pernambuco will need to accelerate investment in human and physical capital and encourage productivity growth in order to accelerate GDP growth and catch up with other states. Chart 7 shows an index of human and physical capital per worker based on the regression coefficients. It provides an indication of the extent of "catching up" that Pernambuco must undertake with respect to other states of Brazil. Chart 7 - Brazil States Index of Capital Per Worker: 1997 70 ;, >. i 50 x 40 l X 301 1 *E0,, 201 __. l ____ 10 0 1- - - - - - - UrWx -1 r;;9 t mC- rS co ;u002 g s.>~ X n m 3 t r z > > w mrn - State U Index of Capital Per VVorker 25. As expected, higher capital per worker leads to increased average wages. This relationship is discussed further ahead (para. 51, p.30, and Chart 17). Pernambuco's average wage has about the level expected from the cross sectional relationship between average wage and average years of education. This means that attaining higher output per-capita and higher average wages will require higher capital per-capita including a higher level of education. This is the reason why improved educational performance is emphasized as a key requirement to increase production and wages and, given the link of wages to the income of the poor, to reduce poverty. 26. An important component of capital formation is basic infrastructure, including transportation, energy, telecommunications and water. Given the importance of trade with other states and countries, transportation infrastructure plays a strong role in Pernambuco's economic development. Despite its limited endowment of fertile land, Pernambuco's location is a favorable natural resource, providing close access to the large Northeast and to developed foreign markets.21 There are three key development 21 Fortaleza, Natal, Joao Pessoa, Macei6, Aracaju and Salvador are within a 500 mile radius from Recife. The USA and Europe are 3500 and 4300 miles away by sea from Recife. 15 investment areas that focus on this role: the federal highways, the railway network, and the Recife Ports. With regard to highways, the state presently has 4.500 km of paved roads, of which about half are state roads and the other half are federal roads. The state's federal highways include federal highway BR-232, which crosses the state from east to west and the BR-101, which connects north and south Pernambuco along the coast. A major priority for the state has been the upgrading of BR- 101. This is one of the projects included in the Federal Governnent's Avanqa Brasil program for the Transnordestino Axis to which Pernambuco belongs. This program also provides for the upgrading of BR- 101 from Natal to Aracaju. 27. With regard to the rail network, Pernambuco presently has an old railway from Recife to the western town of Salgueiro that needs to be recovered. The Government's aim is to recover the rail and to integrate it with a new "Gypsum" railway to Ararapina, the northwestern town around which the Gypsum mines are located. Furthermore, and again with the support of Avan,a Brasil, the existing railway would be extended to the southwestern city of Petrolina, where it would connect to existing railway to Salvador and then south to Minas Gerais, and to the state of Ceara, where it would connect to the existing rail network in the Northeast. This would complete the Transnordestino Network, a project that has been in the books for very long and that needs much investment both in the new sections as well as in the recovery of the majority of the existing rail. 28. With regard to the ports infrastructure, Pernambuco is developing the Industrial and Port complex of Suape, initiated in the early 1970s. Because of its 16 meter natural depths, Suape is envisaged as a "hub" port that redistributes cargo to shallower ports and makes multimodal transshipment operations such as the transshipment of liquid gas for consumption in the Northeast. Suape's area (135 square Km.) will also host a number of industries in Pernambuco that want to take advantage of the transportation infrastructure that is linked to the complex. Finally, mention should be made of rural infrastructure, including access roads, and most importantly irrigation, discussed further below. 29. Pernambuco's electric energy demand is met by the distribution company CELPE, a former state company that was privatized in March, 2000. CELPE sells to six million consumers in Pernambuco through a 76,166 Km distribution network. Annual growth of energy demand is about 8%-9%. CELPE purchases its energy from the Sao Francisco Hydroelectric (CHESF), a federal generation company. To meet increased energy demand, Petrobras has engaged a private sector partner (Asea Brown Boveri) to establish a 260MW thermoelectric generating plant in Suape that they will sell to CHESF. It will use gas obtained from a Petrobras-Shell regasification terminal, also to be established in SUAPE. 30. Pernambuco's telecommunications services are provided by Telemar, a private fixed telephony company that replaced the state-owned TELPE and by two mobile companies: TIM in Band A, which provides 800,000 lines to the states of Pernambuco, Alagoas, Paraiba, Rio Grande do Norte, Ceara and Piaui, 300,000 of which are to Pernambucan clients; and BCP in Band B, which provides 260,000 lines to the same states. Pernambuco has 20 internet service providers connected through EMBRATEL and 16 the National Research Network (Rede Nacional de Pesquisa). The Government is developing a state computing network to interconnect the state cities to each other and to the national backbone. 31. Water supply is Pernambuco's most challenging infrastructure bottleneck. Water effectively available (2,13 billion m3/ year) does not meet current demand (let alone current needs) of 2,3 billion m3/ year, which is comprised of 66% irrigation (of which 70% is met by the Sao Francisco River), 23% for non-industrial, urban use, and 7% for industrial use (40% of which is absorbed by the sugar-alcohol industry). 32. Eighty percent of the state's territory is in the so-called drought polygon of the Northeast, where rain is below 800mm. As a result, Pernambuco is the Brazilian state with Chart 8 - Percent of Drought Area in the lowest per-capita water supply, with a Selected Northeast States drought area comprising almost 90% of its 120.0% territory. Its virtual water availability (i.e. the 100.0% maximum attainable supply) of 976 = 80.0% X m3/hab/year does not meet the UN's "water @ 60.0% stress" benchmark. The amount of water a 40.0% _ e * effectively available (288 m3/hab/year) is 20.0% , ; | well below the virtual level. In addition to 0.0% - .4,i being scarce, available water is poorly used, State with excessive perforation of urban wells in l Percent of DroughtArea Recife (leading to contamination) and underutilization of water sources in the interior of the state e.g. only 14.7% of the virtual underground supply is being utilized). Pollution also originates in low sewerage facilities (covering only 19.9% of the population). Main problems in the different regions are summarized in Table 5, with insufficient water supply emerging as a problem in all areas except the Sdo Francisco River. Only to illustrate, in Recife in April-May 1999, many areas had rationing of one day with water for every ten days. But poor administration of water resources is clearly another main issue that manifests itself in the other various problems identified in Table 5. 33. Productivity Growth. Total factor productivity growth (TFP growth) can explain a significant percentage of output growth. The measurement of that contribution, however, is fraught with problems. In the case of Brazil, for example, measurements of productivity with respect to the US benchmark (100) have been estimated at 100.2 and 42 under two different studies.22. These measurement problems make it very difficult to estimate Pernambuco's productivity gap. Nevertheless, for policy purposes, one can work on the hypothesis, given Pernambuco's development level, that the state has opportunities to exploit the conventional sources of "productivity growth": technological progress, improved allocation of resources and economies of scale. Technological progress requires federal provision of an open trade and investment environment that facilitates diffusion of technology. Improvements in resource allocation require reductions in price distortions and improvements in the supply of public goods. Economies of scale also 22 See Islam (1999), p. 506. 17 requires federal provision of an open trade and investment environment that facilitate specialization. Table 5 - Pernambuco - MIain Problems of Water Supply and Nlanagement Systems Reoion Chief Problems V Insufficient production of treated water and high indices of loss and waste; / Deficiency in sewerage system; / Poor urban cleaning and waste disposal services; Recife v/ Drainage problems and inadequate use and occupation of urban soil; / Acute water pollution; V Uncontrolled perforation of wells; V Absence of conservation management of water basins. v/ Pollution of water resources due to dejetos and industrial waste; Mata V Inadequate water supply/distribution systems; / Absence or inadequacy of sewer systems; v/ Limited outflow and poor quality of underground waters; V Poor quality of superficial water (with soils that encourage salinization and an inadequate operation of water reservatorios). Agreste V Low irrigation potential (absence of significant areas of adequate soils); V Water pollution due to industrial and sewer waste; Inadequate water supply/distribution systems. V Inadequate operation and use of available superficial water sources; Serto Underutilization of sedimentary underground water sources; Sertao Vt Inadequate water supply/distribution systems; v Absence of basic sanitation services. v/ Increased salinization of soils; Sao Francisco V Inadequate water storage and lack of pasture, during the dry season, for cattle at the sequeiro area; V Absence of basic sanitation services. 34. Demand Factors. Information on the demand sources of growth in Pernambuco is also very limited. It can be conjectured that the investment ratio should not be too different, although perhaps lower (given Pernambuco's lower growth), than the Northeast's overall average investment ratio of 19% of GDP, with public investment accounting for about 4% of GDP (1996). Government consumption accounts for about the same share of GDP (15%) as the average share for the northeast. Hence, public expenditure is almost one fifth of total GDP, and an important source of demand. Pernambuco's trade deficit (including trade with other states) varies a lot from year to year (10.4%, 13.9%, and 6.4% respectively in 1996-98), with no information on the service balance. Assuming a 10% current account deficit, private consumption in Pernambuco would be of the order of 76% (100-19-15+10) of GDP. With a balanced current account, private consumption would be 66%, closer to Brazil's average of 63%. A 18 current account deficit is the more likely condition, given debt accumulation (i.e. a current account deficit) vis-a-vis the rest of the world and the federal government. 35. The above figures hide the enormous importance of trade. Imports from and exports to other states and the rest of the world averaged respectively 40% and 29% during 1996-98. Most trade is interstate trade. Since sugar has played an important role in exports, and exports account for such a large percentage of GDP, it is clear that changes in the price of sugar have been an important source of change in income and demand. The long-term decline in the price of sugar could be one important explanation for Pernambuco's modest growth. 36. Influence of Macro Factors. Chart 9 Chart 9 - GDP Growth in Brazil shows that there is indeed a positive and Pernambuco: 1947-1998 relationship between Pernambuco's growth and Brazil's overall growth. Increases in Brazil's growth are associated with lower increases in 23~~~~~~ Pernambuco's growth .3 This association ,.,?0, ,',.: suggests that forces behind Brazil's aggregate 2 growth have a strong bearing on Pernambuco's I -0 O/4fi M .0% 10.0% 15.. individual growth. Lessons from Brazil's 20th E century history can be summarized as follows: E . . -r * External shocks are widely believed to X. . have had a strong impact on Brazil's '. . -. GDP growth. These may include weak export markets, higher import prices, Brazil Growth and international financial shocks. . Pernan-buco - Linear (Pernanbuco) Examples are numerous: the Great Depression, which reduced Brazil's GDP, albeit mildly, between 1928 and 193224; the 1937 US recession, which contributed to slower growth in Brazil from 1936 to 1942; a prosperous world economy, which helped Brazil's economic expansion from 1968 to 1973; the 1978-79 oil price shock and the increase in world interest rates, which triggered the 1981-83 recession after; the Asia and Russia crises, which decreased growth from 1997 to 1999. The shorter available evidence for Pernambuco suggest that these external events also had an impact on the State's performance, with relative prosperity during 1968-73 and recession in the early 23 The regression equation between the two per capita GDP series is: PEgdp = 0.006 + 0.65xBRgdp (0.79) (4.77), where PEgdp is Pernambuco's per capital GDP growth and BRgdp is Brazil's per capita GDP growth. t ratios are in parentheses. The R square is 0.32. 24 Some authors suggest that the impact of the recession was stronger than official data would lead us to believe. See, for example, Haber, Stephen, "Lucratividade Industrial e a Grande Depressao no Brasil: Evidencias da Industria Textil de Algodao", in Estudos Economicos, Vol. 21, pp. 241-70, May-August 1991. The author reviews the performance of textile firms in Sao Paulo, Rio de Janeoiro and Minas Gerais, and shows that the Great Depression had a significant adverse impact on the performance of these firms. As in the rest of Brazil, the textile industry was also important in Pernambuco. 19 1980s and over 1997-99. Some external shocks can have a more pronounced impact on Pernambuco than in Brazil in general. The most important of these is fluctuation in international sugar prices, which have been wide, as shown in Chart 4 (p. 11). Furthermore, the secular decline in real price (together with the federal government's withdrawal of alcohol subsidies) helps explain Pernambuco' s reduced dedication to sugar indicated before. In the past, coffee and cotton price fluctuations were also inmportant, not only for allocation of land across different crops but also for Pernambuco's general economic progress. Macroeconomic policy shocks have also played a role in explaining aggregate output fluctuations. It is believed, for example, that excessively restrictive policies in early 1963 contributed to the 1963-65 aggregate recession, which also shows up in Pernambuco's slow growth during those years. Also, restrictive monetary policies after 1995 possibly contributed to declining output performance in 1998. On the other hand, a credible fiscal policy beginning in 1964 helped explain aggregate economic recovery in 1966-67 that shows up as a major increase in Pernambucan output in 1967. Similarly, a credible fiscal policy in 1999 is contributing to renewed growth in 2000 in Brazil in general as well as in Pernambuco. 37. In addition to external factors and macroeconomic policies, local shocks and policies also affect Pernambuco's GDP performance. A chief example is climate shocks. which can cause large agricultural output fluctuations. In Pernambuco, the standard deviation of annual agricultural value added during 1970-98 was 20.5% while the standard deviation of the services sector was only 5.3%. Some of the greater fluctuations in agricultural output result from climatic changes. As discussed above, agricultural price volatility also plays a role. And in output from agriculture, pests have played a role. Cotton disappeared from Pernambuco due to the ballworm scourge. Chapter II includes a fuller discussion of the impact of local policies. GROWTH FUNDAMENTALS 38. Sectoral and demand decompositions of output provide chiefly "accounting" explanations of the growth process. They indicate, for example, whether growth has originated in the sugar sector, or whether growth is accounted for by export or consumption growth. These may help in identifying possible explanatory source of growth such as the sugar export prices or the relative price of tradeables. Closer to the realm of explanation is the decomposition in terms of capital accumulation labor growth and, productivity growth, which is based on production theory. But the most fundamental inquiry is to ask which are the forces behind sectoral or demand growth and behind investment and productivity growth. The latter is the question that has been addressed by endogenous growth theory, which ultimately asks what it would take for poor countries or states to converge to higher levels of income. 39. The simplest approach is to investigate whether the rate of growth depends negatively only on the initial level of per capita GDP (the so-called absolute i convergence). This inverse relationship originates in a production function with 20 diminishing returns: states with lower per capita output would grow faster than states with higher per capita outputs because the marginal product of capital would be higher in the former. A study of absolute convergence for Brazilian states concludes that, on the average, there is convergence, albeit very slow .25 In the case of Pernambuco, this conclusion is not of much help. As discussed before, Pernambuco's per capita GDP has lost ground with respect to the averages for Brazil and the Northeast. Instead of Pernambuco converging to higher incomes, some poorer states have converged to Pemambuco and above. Indeed, in the statistical analyses of absolute convergence, Pernambuco's lower than average performance is in evidence. Chart 10 relates the 1939- 97 growth record to 1939 per capita GDP. The black dot indicates Pernambuco. The Chart shows how weak the absolute convergence relationship is, with wide variation from a simple linear relationship, with Pernambuco well below the trend line. The relationship between 1985-97 growth and 1985 per-capita income is better (Chart 11), possibly benefiting from the more consistent data recently published by IBGE, and suggests the path that growth rates would follow as the state GDPs approach the highest per-capita incomes (Brasilia and Sao Paulo). Still, Pemambuco is again well below the trend line. 40. Some further light is shed on convergence when the focus is restricted to the Northeast states. This is done in Charts 12 and 13. Chart 12 depicts the relationship between 1939 per capita incomes and 1939-97 growth rates. Again the relationship is very weak, with only Pernambuco providing some sense of where the other poorer states (in 1939) might go. Chart 13, which benefits from the better IBGE data shows a clearer picture. It indicates convergence from 1985 initial incomes, with the poorer states growing faster. In both cases, Pernambuco, the richest state in the Northeast, appears to indicate the level per-capita GDP to which some of the other Northeast states are converging. 41. The graphs above indicate some weak absolute convergence. Charts 11 and 13 suggest that those states with lower initial per capita GDPs grow faster. But Pernambuco is not one of those. In the Charts with all the states, Pemambuco's growth is well below the trend lines. In the Northeast Charts, Pernambuco already has the highest per capita GDP and the other states seem to converge to it. All of this leads to the question of why Pernambuco has not performed as well as would be expected by low income relative to Brazil as a whole. Conditional convergence investigates the negative relationship between growth and initial per-capita GDP when other variables are taken into account in order to represent the "steady state" level of per capita GDP. That is, the convergence hypothesis postulates that growth is inversely related to the gap between existing per- capita GDP and the steady state level of per-capita GDP. The main variables that have been considered in the literature, and applied in cross-country research are discussed below (the sign in parenthesis indicates the a priori impact on growth) as to their possible 25 See, for example, Cavalcanti and de Goes (1996). Their finding is that it would take about 90 years for Brazilian states to achieve their steady state level of GDP, longer than what has been found, for example, for US states. The authors also conclude that in the long run the dispersion of income diminishes through time (c convergence). 21 impact in Pernambuco. The discussion is largely guided by the work of Barro and Sala-i- Martin (1995). Chart 1 0 - Brazilian States - Growth Chart 11 - Brazilian States - Growth 5.0% Convergence: 1939-97 6.0% Convergence: 1985-97 45.% 6 1 .0% ,. 4.0% - _ ____ _ 50 ____ 4.035% y OO O° -3EO0xtO.333 ' 0 y=-0 0102Ln(x) +0.0967 X 3.5% -- 0_ _ X u_ _ ,_ _ _ _ ---- 3. C925 -- r ,33- _ .. 3.0% 2.5% _ S a 1. 0% _ S: , __.0 02.5% ) _ c 2.0 %4 0 0.0% -1.0% 0 500 1000 1500 2000 2500 1985PerCapitaGDP(1997F) o Grth39-97 -Linear (Grth39-97 - Log. (Average Annual GDP'Grow th 1985-97) Chart 1 2 - Brazil Northeast States - Gowth . .Chart 1 3 - Brazil Northeast States - Gonvergence: 1939-97 ..Convergence: 1985-97 4.0% - -- - -- - - __- -_ 3.0% ,,, 3.5% --___ O _ _- _--- __ 2.5% o-- _-- __ d0 3.0% __ _ _ ___.- ---. - X. e 2.5% __% _20 40 MM_ 2 .0 ^ 2.0%!- _________15%-.______ c l1.5% -y --2E05x '00282 e____--c_____ 1.0%O _ _ - __ _ _-s y=-1E0O5x4 00363 4:0.5% _. -- - __ __ _ ________ 4 0.5% - o____________ _ _ 0.0% 0 200 400 600 800 10 00 0.0%- 500 1000 1500 2000 2500 3000 1939 Per Capita 1985 Per Capita GW (1997 R$) Grth39-s7 - Linear (Gcth3g97) o * Human capital (+), including both education and health, enhances economic growth both directly via the increased productivity and creativity and indirectly via reduced fertility rate. However, the effect of education has been controversiaL with some studies claiming that the impact of education is insignificant.26 For Pernambuco, 26 A recent study persuasively argues that the result that educational attainment is unrelated to economic growth has been caused by high measurement error in first-differenced cross country data and that "(a)fter accounting for measurement error, the effect of changes in educational attainmnent on income growth in cross-country data is at least as great as microeconometric estimates of the rate of return to years of schooling. See Krueger and Lindahl (2000). 22 insufficient human capital has probably been an important constraint to growth. Although no studies are available on the role of education in growth in Pemambuco, the cross state comparisons discussed further below (para. 51, p.30 and Chart 17) as well as micro estimates presented in the next section show that education may be a significant explanatory factor determining incomes. In 1998, Pernambuco's active population, defined as people between 15 and 65 years of age, had the highest average level of completed education (5.28 years) in the Northeast (e.g. the labor forces in Bahia and Ceara were on average less educated, with 4.59 and 4.7 years respectively). However, average education in Brazil in 1998 was higher (6.22 years), 1.5 years more than in the Northeast and 1 year more than Pernambuco. Low quality of education (discussed below) may also have been an important constraint to growth, as would be surmised from international research that finds a significant impact of proxies for education quality on growth. Poor health conditions may also have been a factor undermining growth. Life expectancy in Pernambuco (58 years for men and 67 for women), an indicator of health conditions, is lower than the average for Brazil (64 years for men and 72 years for women) and the average for men in the Northeast (61 years; the NE average for women is also 67 years). Population Growth (-). In the Swann-Solow model of economic growth, population growth requires that a fraction of investment be used to provide capital to new workers in order to maintain the level of capital per worker. Hence the impact on per capita GDP of population growth is negative. Since the population growth in Pernambuco has been lower than in other states in the Northeast and Brazil as a whole (Table 6), this factor would contribute positively to per capita GDP. Alternatively, however, the relationship can be interpreted inversely, i.e. that low output performance and wages, and possibly adverse health conditions in Pernambuco have led to low population growth through mortality and emigration. For example, poor health conditions in Pernambuco are indicated by high infant mortality in 1998 (60 per 1000 births). Although infant mortality declined from 129 in 1980, it is still higher than Brazil's (35) and than the Northeast average (56). Between 1991 and 1996 Pernambuco gained 116000 immigrants from other states and lost 273000. As argued by Ramos and Araujo (1999), the loss is consistent with a measure relative income expectations (average income corrected by the unemployment rate), in the context of a Harris-Todaro model. Much of the net loss was to Sao Paulo (90000). Overall, the share of Pernambuco in Brazil's total population declined from 5.8% in 1960 to 4.7 in 1996. In general, however, it has been found that migrations have little or no effect on convergence.27 A critical demographic variable is fertility. Brazil reported a fertility rate of 2.4 in 1997, lower than the Northeast's (2.7). In cross-country growth regressions, population growth and fertility (the typical woman's expected number of live births over life) have significant coefficients, with fertility having a negative impact (presumably due to its adverse impact on the steady state per-capita GDP) and population growth having a positive effect (presumably indicating higher immigration 27 Barro and Sala-i-Martin (1995) arrive at this conclusion after tests conducted for the USA, the EU, and Japan. Pitella Cancado (1999) arrives at the same conclusion with tests for Brazil. 23 or lower mortality as a result of higher growth). In Brazil, fertility has declined since the 1970s. In the case of Pernambuco, the decline was from 7.0 in 1970 to 2.3 in 1997. Another important development is the decline in the fertility differential between the more educated (eight years and above) and the less educated (less than four years) mothers, from around 4 children in the Northeast in 1980 to 1.9 children in 1998. This differential declined to 1.3 children in Brazil's richer Southern states. Table 6: Total Population and Population Growth in various regions of Brazil Population in 1996 Geometric annual growth rates: Percentages 1970/80 1980/91 1991/96 Years of Age <14 15 to 64 >65 157 070 163 31.6 63.0 5.4 2.48 1.93 1.36 Northeast 44 766 851 35.6 59.0 5.5 2.16 1.83 1.06 Pernambuco 7 399 071 33.2 60.9 5.9 1.76 1.36 0.76 Ceara 6 809 290 35.7 58.4 5.9 1.95 1.7 1.38 Bahia 12 541 675 35.0 59.8 5.3 2.35 2.09 1.13 Source: Anuario Estatistico do Brasil 1997, IBGE * Government taxes and expenditures (?). Governments can increase growth by raising the marginal product of capital through expenditure on public services and reduce growth through the effect of taxation on the after tax marginal product of capital. This is the thrust of the model by Barro (1998). It has been investigated for Brazil using 1970-90 growth data.28 The tests suggest the postulated u-inverted relationship between taxation and growth, with low levels of taxation insufficient to provide adequate public services for growth, excessive distortions from high levels of taxation more than offsetting the impact of public services on growth, and an intermediate level of taxation that "maximize growth", with the size of Pernambuco's public sector somewhat beyond the point of maximum contribution to growth. Compared to other states, Pernambuco's public sector (18% of total value added) is larger than the average for Brazil (15%) but lower than other states in the Northeast except Bahia (13%). As discussed later, the policy implication of this relationship is that a state like Pernambuco should improve the effectiveness of expenditures (e.g. better quality education) while meeting fiscal constraints. Other research (Monteiro Neto and Maia Gomes (2000); de Oliveira Cruz and de Albuquerque Oliveira [1999]) points to the strong role of net federal transfers in explaining the strong growth of the Center West and of Amapa and Roraima. This is most obvious for the Distrito Federal, which has the highestper capita GDP of all Brazilian states thanks to large federal net transfers since the time the capital was transferred to Brasilia. * Trade and investment Restrictions (+). Governments can increase growth by opening up to trade and investment. Trade openness encourages exports and opens foreign market opportunities. At the level of Brazil as a whole, trade liberalization in the early 1990s probably was a major factor explaining the resumption of growth. Investment openness helps bring in capital, a key supply factor of growth. As discussed before, trade plays an important role in Pemambuco, with total exports to 21 See Duarte Lled6 and Cavalcanti G. Ferriera (1997). 24 other states and countries accounting for about 29% of the state's GDP. It will be argued that faster growth in Pernambuco is likely to involve an even more open state economy. * Other Market Distortions (-). More generally, international studies find that proxies for market distortions (e.g. the black-market foreign exchange premium) are inversely related to growth. In addition to foreign trade and investment barriers, other key relevant market distortions may include excessive taxation of labor and limited flexibility of labor contracts and constraints on private investment in infrastructure (e.g. water and sanitation). * Political Instability (-). Political instability threatens property rights and thus discourages investment and growth. Pernambuco's history is checkered by much political turmoil. The work by Correia de Andrade (1997) describes this instability from the disputes between the initial French and Portuguese visitors in the sixteenth century to the suspension of democracy in 1964. Although much of the turmoil that affected Pernambuco has been national in scope, the state has probably been more polarized than others (perhaps with the exception of Rio Grande do Sul). * Institutional Quality (+). Following Barro and Sala-i-Martin (1995), description of institutional quality can be based on the ICRG index, which includes: rule of law ("the extent to which institutions provide effectively for implementation of laws, adjudication of disputes, and orderly succession of power"); corruption in Government ("related to the frequency of bribes in areas such as international trade, taxation, and police protection"); quality of the bureaucracy ("including the degree of autonomy from political pressure); expropriation risk (assessment of risks of outright confiscation and force nationalization); and repudiation of contracts (including risks of repudiation or modification due to changes in Government). In Barro and Sala-i- Martin's work, the rule of law index is likely to capture the "overall effect of the legal and political framework on the growth rate", as none of the other variables are significant when added to the rule of law variable. The ICRG index provides numbers for Brazil, but not for the individual states. It shows some Chart 14 - Brazilian States - Growth deterioration of the rule of law Convergence: 1985-97 index since 1992. This 6.0% -.'. - deterioration may have had an a 5.0% adverse effect on the overall rate 1. - rx21+X )0+o.114 of investment and growth. * Geographical Factors (+1-.). <,. 3.0% Some writers, notably Gallup ' 2 2.0% M _ and Sachs (1999), argue that , f 1 0. S 1 ' +*.036 geography (location and 5 1.0% climate) is a fundamental 0 00% determinant of economic growth _1.0% 8000 10000 12 and of the quality of economic 1985 Per Capita GDP policies. Transportation costs, * Other ( Northeast tropical climate, and prevalence (her 25 of infectious diseases have an adverse effect on growth.29 Furthermore, adverse geography may affect policies; for example, countries that are proximate to markets may choose more open trade policies. While Pernambuco does not have an abundance of natural resources, including water, its location provides the state with closer access to trade with developed countries than the southern states have, although the latter have closer access to Mercosul trade. Also, the small size of the state could imply diseconomies of scale that hinder growth, although the estimated impact of size on growth has been very small in international regressions (e.g. small- sized economies like Hong-Kong, Singapore, and South Korea have grown very fast). In their review of the literature on Geography and Development, Henderson, Shalizi, and Venables (2000) summarize by saying that "geography matters for development, but that economic growth is not governed by a geographical determinism."30 This is because, for example, the costs of remoteness can be reduced. Important policies that emerge in this context are trade liberalization, infrastructure improvements, and regional integration (of the north-south type). More than by geography, Pernambuco is constrained by related infrastructure bottlenecks, including transportation to other Brazilian markets and inadequate water supply.3 l The experience of Brasilia, the territorial division with the highest per capita GDP, shows how an economic factor (a strong source of demand -- public expenditure) may overcome the impact of adverse geography (disadvantaged 32 location). There is, nevertheless, some appeal to the hypothesis that geography matters. The convergence lines for Northeast and other states are put together in Chart 14, which suggests that the Northeast could well have a lower long term or steady state level of per-capita GDP. This possibility, however, does not rule that policy can improve that level. * Growth in Terms of Trade (+). The terms of trade is a statistically significant determinant of growth. In the case of Pernambuco, the most relevant price in terms of trade has been the price of sugar, which was discussed above. * Physical Capital Stock and Investment (+). Investment was discussed before as having a positive impact on growth (para. 24) from the point of view of supply, where rates of growth are "accounted for" by changes in factor inputs. In growth convergence equations, where the impact of growth fundamentals are explored, the positive impact of investment will appear significant when some of the other variables discussed above are excluded. Furthermore, prior research also suggests that an "impact" of investment may reflect reverse causation from growth to investment.33 29 For example, these authors explain 69% of national per capita income variation across 83 countries with four variables: (i) endowment of hydro-carbons per capita; (ii) incidence of malaria; (iii) percent of population within 1OOKM of the coast; (iv) and international transport costs. 30 See Henderson, Shalizi, and Venables (2000), p. 27. 3' A recent firm survey by Vasconcelos and Vergolino (1999) ranks insufficient infrastructure as the main factor discouraging investment in Pernambuco. 32 This is not to recommend public expenditure as the engine for growth. This cannot really happen, as more transfers to one state inevitable imply less transfers to others. 33 See Barro and Sala-i-Martin (1995), p 433. 26 We conclude that investment goes hand in hand with growth and convergence, and that it is, as growth is, determined by the other fundamentals discussed before. 42. This concludes the discussion on the record and possible causes of growth. This section has shown that (i) per-capita GDP growth in Pernambuco has been below the average for Brazil and the Northeast; Pernambuco's per-capita GDP in 1998 was only 3.2% above per-capita GDP in 1980 (as compared with 8.4% for Brazil as a whole); (ii) Pernambuco's most dynamic activities have been, by and large, in the tertiary sector; (iii) low physical and human capital per head are closely linked to Pernambuco's relatively low output per head performance; (iv) a corollary of (i) - Pernambuco does not appear to be converging to higher levels of GDP per head in Brazil; rather, other Northeast economies appear to be converging to Pemambuco's GDP per head level; and (vii) based on the endogenous growth literature, Pernambuco's growth performance could be improved by actions on a diverse number of fronts. These include actions to invest in human capital, to reduce population growth, to raise the efficiency of public expenditures, to deepen trade and investment openness, to reduce the taxation of labor, and to improve institutional quality. These are discussed in Chapter II. The next section turns to a discussion of poverty trends and its causes. As it turns out, sustained poverty reduction performance depends critically on aggregate growth and on the education of the poor. However, actions on these two fronts need to be complemented by effective social protection activities. The latter will both support the effectiveness of education expenditures as well as alleviate the poverty of those that cannot benefit from education. THE POVERTY REDUCTION RECORD 43. Changes in Poverty Levels. Three measures are used in this report to assess poverty: the headcount ratio (PO), poverty gap (Pl), and the squared poverty gap (p2). The first measures the magnitude of poverty and the latter two measure both magnitude and intensity. Annex A discusses the definition of these poverty measures. Three poverty lines are used to calculate the three measures of poverty: (i) an indigence line (L0) based on the cost of a minimum food basket;34 (ii) a lower-bound poverty line (L2), which 'scales up the cost of the minimum food basket to take into account the non-food expenditures of those people whose total incomes would just allow them to purchase that minimum food basket".35 It corresponds to twice the lowest poverty line for each region and to R$132 in 1996 (R$148.98 in June, 1999); and (iii), an upper-bound poverty line (L3) "scales up the cost of the minimum food basket to take into account the non-food expenditures of those people whose actual food expenditures equal the cost of the minimum food basket; this line is equivalent to R$205 in 1996. 34 This measure is based on the cost of a "minimum food-basket" equal to the FAO minimum caloric intake of 2288 per day. The poverty line is expressed in the same currency unit as the income vector (that is, the 1998 prices ruling in Sao Paulo). It corresponds to around R$65 monthly per capita (R$73.36 in June, 1999) for the metropolitan area of Sao Paulo in 1996 prices. The food-only poverty line is given by the local value of a food basket that yields an internationally accepted caloric intake. The values for 1999 are calculated in Neri (1999). The value for other metropolitan areas is adjusted by the local cost of living in accordance with the estimations of Ferreira, Lanjouw and Neri (1999). 35 Ferreira, Lanjouw and Neri (1999), p.7. 27 44. The results with regard to the headcount ratio, depicted in Chart 15, indicate that 36 while poverty declined after 1993 , the 1998 headcount level of indigence was fundamentally the same as back in 1981, with the two headcount poverty measures only slightly lower.37 The poverty gap measures indicates some increase in indigence between 1981 and 1998 (Chart 16), suggesting a lower average income of the indigent in 1998 than in 1981. This is confirmed in Table 7 below, which shows that average incomes of the indigent were indeed lower in 1998 than in 1981. The path of rising poverty through the early 1990s and of declining poverty thereafter is not too surprising. The "lost decade" of the 1980s was a decade of increasing poverty due to slow growth, while reduced poverty after 1994s was, in good measure, the result of Brazil's lower inflation, a shift towards higher relative prices of tradeable goods, higher real wages, and better growth performance during the decade, which also included better growth performance in Pernambuco. Chart 15 - Pernambuco - Headcount Ratio Chart 16 - Pernambuco - Indigence by Poverty Line (L in 1996 Reais ): Measures: 1981, 1993,1998 Selected Years 1981-98 50.0% - 100.0% - . 40.0%__- 80.0% ~ ~ ~ ~ ~ ~ 4.0 ----- ____ Per 8.%30.0%- ce 60.0% nt 40.0%200 - 2&0% ~~~~~~~~~~10.0% 20.0%~~~~~~~~~~00 0.0% 1981 1993__ __ _ __ _ __ _ __ .0 19-rea 1981 1985 1988 1990 1993 1995 1997 1998 1981 1993 1998 Year * Ll =R$65 c L2=R$132i L3=R$205 Year D Headcount Ratio C Poverty Gp Squared Poverty 45. Pernambuco's poverty reduction performance is below average performance for Brazil, for the Northeast as a whole, and for Ceara specifically. It is, however, similar to Bahia's, where the level of poverty in 1998 was similar to that in 1981. All calculations show an increase in poverty through the early 1990s and a decline thereafter. The detailed numbers are provided in Annex A (Tables A1-A3). 46. It may be noted also that the average income of the population in the lowest 50% of the income distribution declined. In 1998, the average income of this population was 36 Actually the decline in Pemambuco possibly started after 1994, the year that average poverty in Brazil poverty peaked. Chart 14 provides data only for selected years. 37 It should be noted that by 1981 the strong poverty reduction trend, which lowered Brazil's overall headcount ratio from 68% in 1970 to 35% in 1980, has been interrupted. See Rocha (2000), who reports that the headcount ratio in 1980 was about one half of the poverty ratio in 1970. This decline must have taken place early in the decade, for Ferreira and Paes de Barros (1999), report that the headcount ratio (urban only) was about equal in 1976, 1981, 1985, and 1996. Presumably Pernambuco shared in this reduction., as GDP growth in the state, as in the rest of Brazil was also high. It should also be noted that the measurement of the poverty ratios is very sensitive to the dates and definitions of the deflators used. Alternative calculations have shown some decline in poverty between 1980 and 1998. 28 10.1% lower than in 1981. In contrast, the average income of the richest 50% was 15.0% higher in 1998. Thus relative poverty increased, with the average income of the poorest declining from 18% of the average income of the richest in 1981 to 14.1% in 1998. Clearly, the fruits of the very limited overall economic progress during 1981-1998 (per capita GDP increased by only 3.2% between 1981 and 1998; see Table 1, p.9) did not trickle down to the poor in Pemambuco. Table 7 summarizes the mean incomes, including the mean incomes of the indigent, which also declined (by 10.4%), as indicated above. Appendix A Tables A4-A6 provide the breakdown between urban and rural sectors. Both sectors exhibit similar changes between 1981 and 1998. Table 7: Pernambuco - Mean Individual Monthly Income in 1996 Rs$ Mean Income of: 1981 1985 1990 1995 1998 Total 138.52 132.24 127.59 145.23 153.79 Indigent 38.4 36.1 34.8 35.73 34.38 Poorest 50% 42.48 37.93 32.64 40.22 38.08 Richest 50% 235.53 226.84 219.15 251.06 270.69 Poorest/Richest 50% 18.0% 16.7% 14.9% 16.0% 14.1% Source: Staff Estimates from PNAD Household Surveys 47. The 20-year period comparison allowed by the available poverty data discussed above provides only a brief glimpse into Pernambuco's 500 year history. Less numerical assessments by historians provide more information about what has actually been a long- standing plight in Pernambuco, despite improvements such as the possible decline in poverty during the high growth 1970s period. Pernambuco's chronic poverty problem can be surmised from historical accounts of the sugar economy and of the city of Recife. These accounts make frequent reference to pervasive conditions of poverty. POVERTY FUNDAMENTALS AND OTHER CORRELATES OF POVERTY 48. What distinguishes the poor in Pernambuco? This question is addressed by analyzing the headcount ratio for different groups of people, and in terms of variables indicating: (i) household "causes" of income (i.e. household factors that could affect income, including education, age/experience, health, race, gender, physical assets; and factors that affect expenditure requirements, basically family size); (ii) urban-rural location, which may affect price levels, income measurement, relative strength of labor markets, or other sources of urban-rural wage differentials; (iii) work status (employer, employed, self-employed), which may, for example, approximate risk-taking by the employed or reflect formal/informal sector income differentials; (v) lack of access to public services, which lowers welfare; (iv) economic sector, which indicates where the poor tend to be employed. All of these factors are important from the point of view of understanding poverty levels and perhaps of drawing policy implications for poverty reduction. This section discusses them in turn. 49. Household Causes of Poverty. Table 8 indicates headcount ratios (HCRs) for different groupings of households according to household characteristics. These characteristics cover gender, race, age, years of experience and education of the household head, the dependency ratio of the household, and durable/non-durable house 29 condition (as a [poor] proxy for household wealth). Column (1) of the table shows the HCRs for 1981 and 1998, the end-years of the period covered by the PNAD sample. Column (2) divides the HCRs in Column (2) by the overall PE HCR to indicate the specific groups that are relatively more (when >1) or less (when <1) poor than the overall PE HCR. Column (3) divides column (2) by the same calculation for Brazil, to indicate how different the profile is in Pemambuco as compared with Brazil. Column (4) divides column (2) by the same calculation for the Northeast to indicate how different is Pernambuco when compared with the Brazil average. Table 8: Pernambuco (PE) - Headcount Ratio (HCR) of the Idi ent Household Characteristic Headcount Ratio PE Relative HCR = PE Relative PE Relative (HCR) of Indigent PE HCR for Specific HCRlBrazil HCR/Northeast Characteristic/ Relative HCR Relative HCR Average PE HCR (1) (2) (3) (4) Year 1981 1998 1998 1998 1998 Total 36.9% 36.7% 1.00 1.00 1.00 Gender: Male 36.1% 37.3% 1.02 1.01 0.98 female 40.0% 35.3% 0.96 0.97 1.06 Race: white 33.5%* 27.2% 0.74 1.23 9.84 black 57.4%* 44.8% 1.22 0.95 1.09 mulatto 48.6%* 43.6% 1.19 0.78 1.07 indig 66.7%* 20.0% 0.54 0.43 0.57 Age: <25 30.6% 43.6% 1.19 0.89 0.99 25 to 45 39.4% 40.8% 1.11 0.99 0.99 45 to 65 36.0% 34.6% 0.94 1.00 0.96 >65 33.8%0 26.6% 0.72 1.15 1.20 Dependency 1 10.0% 15.2% 0.41 1.02 0.91 Ratio I4 55.8% 46.1% 1.26 0.97 1.11 Literate? yes 27.1% 30.7% 0.84 1.08 1.05 no 52.5% 54.0% 1.47 0.71 1.02 Schooling no education or 51.8% 51.8% 1.41 0.73 1.01 less than I year I to 4 45.9% 46.2% 1.26 1.04 1.06 4 to 8 29.2% 39.1% 1.07 1.20 1.16 8 to 12 10.7% 17.7% 0.48 1.16 1.15 more than 12 0.5% 2.1% 0.06 0.90 1.08 NA 12.9% 33.3% 0.91 1.44 0.87 Durable yes 34.5 35.45 0.97 1.05 1.04 House? no 57.3 66.67 1.82 0.60 0.97 Source: PNAD Surveys (1993) 50. The columns in (1) confirm the expected inverse relationships between poverty and: the level of education or literacy, age or experience, and house condition; and (2) the expected positive relationship between poverty and the dependency ratio (the ratio of household size to employed household members). The columns also show that families 30 headed by blacks and mulattos are poorer. Column (2) shows, for example, that young households (head less than 25 years old) have an HCR 19% higher than average; that households with 3 to 4 people per employed member have an HCR 29 % higher than the average; that households where the head has received no education have an HCR 41% higher than average; and that households without durable housing have an HCR 81% higher than average. Columns (3) and (4) show some surprises. Whites and the old in Pemambuco do less well than in the rest of Brazil or the Northeast. Also, those with any years of education do relatively less well in Pernambuco than in the rest of Brazil or the Northeast. 51. The role of education appears very clearly when average wages, as calculated Chart 17 - Brazil States - Years of from the PNAD 1998 survey, are related to Education and Average Wages average education levels at the state level. 10800 -, ;. y84.905e | Chart 17 shows the resulting relationship, = 600 , 0 which is statistically very strong. The data a 600 ' fits best into a standard exponential @ 400 function. The function indicates that, across ? 200 states, an additional year of average 0 education yields 30% in additional average 3 5 7 x =Average Years of Education wage incomes. This time, Pernambuco (the o LnAvrgWage98(PNAD) - Exponential Function black dot) does not appear as an outlier. Its average wage is the level expected from its average education.38 The chart suggests that Pernambuco's labor is not relatively cheap or expensive relative to labor in other states. 52. The above relationships are Table 9: Pernambuco - Determinants of Per assessed jointly in a multivariate Capita Family Incomes: 1998 regression analysis of per capita (Dependent Variable = Log of Per Capita Family household income as a function of Income education level, age (and age2), and Independent Variables Coefficient t ratio family size (and its square). The Constant 3.6122 37.4 results, reported in Table 9 show, the Age of Head in years 0.0319 8.6 right signs and the statistical Age of Head Squared -0.0002 -4.8 significance of all variables. The Family Size -0.2591 -18.2 coefficients indicate that: Family Size Squared 0.0073 6.1 * Completing 4, 8, 12 years, and 4 years of education 0.3344 12.3 university education in 8 years of education 0.5495 15.7 Pernambuco raises per capita 12 years of education 1.0657 31.7 household income respectively University education 2.1944 44.2 by 3%,55% 10%, nd Durable housing 0.29 12 6.6 by 33%, 55%, 106%, and Source: Staff estimates from PNAD Survey ( 998) 219%. The effects with regard R2= 0.49; F = 9419.4 to the school levels of 38 If the average wage of registered workers is used in lieu of the average population wage, the best fit is a cubed exponential function, indicating increasing returns to average education. This could reflect the positive impact of lesser informality on wages, as states with higher average wages have less informality. 31 education are lower than those in similar regressions for Brazil (48%, 76%, 125%, and 227% respectively) as a whole. The effects also lower than those in similar regressions for the Northeast (33%, 63%, 116%, 228% respectively).39 * One additional year in the age of the household head increases per capita household income by about 3.2%, an effect that is lower than in Brazil (4.9%) or the Northeast (3.6%). * A unitary increase in family size reduces per capita household income by about 26%, a slightly lower effect than in Brazil (28%) or the Northeast (27%) as a whole. 53. The results above are derived by applying a standard human capital model. A more sophisticated exercise was also conducted to ascertain the impact of education (and other variables) at different levels of income. The technique - quantile regressions - estimates the impact of, say, 12 years or more of education on the income of those in the tenth quantile and on the income of those in the goth quantile. The respective impacts are of course different and capture the effect of other variables, including unobserved variables, on incomes. Generally, the impact of education is higher for higher quantiles. In the case of urban areas in Pernambuco, 12 or more years of education raises monthly wages by 28% in the 10th quantile and by 180% in the 90th quantile One may hypothesize, for example, that many of the poor that have 12 years of education received poor education and thus did not benefit as much from the twelve years as those that Table 10: Pernambuco - Impact of Education on attended the better schools. Wages at Different Income Quintiles in Urban Areas The quality of education in Education _25_7 9 Pernambuco is discussed 1-4 Years -0.196 0.062 -0.118 0.068 0.456 further below. Table 10 5-8 Years -0.023 0.130 -0.060 0.164 0.571 presents the result for urban 9-11 Years -0.023 0.315 0.211 0.400 0.630 areas in Pernambuco. Chapter >12 Years 0.279 0.843 0.814 1.026 1.802 5 of Vol. II discusses the full Literate 0.280 0.172 0.292 0.110 -0.386 results for Brazil, Source: Staff estimates from IBGE surveys. Pernambuco and other states. 54. There is an important caveat regarding the finding of high returns to schooling. In Brazil as well as in six other Latin American countries the effect of education for rural households is surprisingly small, for those that remain in the sector. For Brazil the measure is the effect of education on farm revenue; for the other countries, it is the effect of education on household income. The higher returns to education accrue to those who migrate to urban areas or, to a lesser extent, to the rural non-farm sector. These results are based on econometrically estimated income equations; they show consistently low elasticities of education on income (and, for Colombia, relatively higher effect of income on education). The contribution of one additional year of schooling to household per 39 One possible concern is that the yield of education (and/or experience) may have fallen through time, as Ferreira and Paes de Barros (1999) found for Brazil as a whole. For Pernambuco, the same relationship of Table 9 was estimated for 1993. A decline was found for completion of 8 and 12 years of education. The changes in the yield of primary or university were not significant. 32 capita income ranging from US$14 to US$40 per person (only one country with $60). In Brazil, the elasticity (farm revenue to schooling) is negligible for small farmers. In making these estimates, schooling of the head of household was used, and alternatively average schooling of the family. No adjustment was made for quality of education, which is typically very low in rural areas, and in most cases these reflect returns to primary education. 55. Urban-Rural Location. The urban or rural location of the household may also have an impact on poverty. For example, rural labor markets could be depressed due to poor product market conditions, triggering salary differentials Chart 18 - Pernambuco - Indigence by between the urban and rural sector. Rural/Urban Area: 1996 300 -.*120.0% Price levels and consumption 20 100% baskets could be different, leading 2 to rural/urban differences in the real g200 . .80% value of money. Coverage of @ 150 60.0% 0. income could also be different if, I 100 40.0% for example, rural income has 50 _ 20.0% unreported income in kind. Whatever the case, there are large o .3 0.0% differentials in the incidence of Z 3 _ a C poverty across urban/rural areas. Chart 18 shows that the HCR rises Rural/Urban Area as one moves from large urban areas, including metropolitan FG*Per capita Income (Rs$) O3HCR (%) *% of Poor Population Recife to the small towns and the rural sector. Despite the large concentration of the total population in the urban areas, the concentration of the indigent of the poor remains in the rural sector. 56. The smooth rise in poverty as one moves from more closely urban to more closely rural locations tones down the sharp distinction that is frequently made in dual labor market models. Nevertheless, the urban rural distinction has been Table 11: Pernambuco - Population (lOOOs): shown to be important in that the Rural Urban Total % Rural urban-rural wage differentials are 1940 1900 788 2688 70.7% not fully explained by individual or 1950 2228 1167 3395 65.6% household characteristics. These 1950 2270 1825 4095 55.4% differentials are consistent with the 1970 2350 2811 5161 45.5% considerable amount of migration 1980 2359 3783 6142 38.4% that has taken place from rural to 1991 2076 5051 7127 29.1% urban locations, that has contributed 199 1922 5477 7399 26.0% to the stagnation and decline of Source: IBGE rural population (Table 11). 40 See, for example Tannen (1991). 33 57. The statistical analyses in that discussion show that, whatever the location, what is important is the availability of urban employment. Proximity to urban areas brings to poor rural families income opportunities that provide higher returns, particularly in the non-farm sector. Moreover, low as their wages might be in the non-farm employment sector for the poorest in isolated areas, these do constitute an increase in their household income over their best rural alternatives. 58. Rural poverty is partly associated with the small size of farms. To generate sufficient value added to guarantee a standard of living over and above the poverty line, specialists in Pernambuco recommend minimum non-irrigated farm sizes that range from 30-50 hectares in semi-arid regions. In Pernambuco, there are 186,699 farms (i.e. 72.2% of all farms) that fall into the less than 10 hectares category. According to the IBGE agricultural census, of these smallest of farms: (i) only about 13,354 (7.1%) have some form of irrigation, (ii) only 6,310 farms (3.3%) receive some form of extension services, and (iii) only 22,893 (12.2%) use chemical fertilizers. What might in part help to compensate these farm households is non-farm income that can be earned nearby. However, these opportunities are highly unlikely in isolated and low productivity rural areas. 59. Compared to Ceara, Pernambuco has less tenants and squatters, and more owner operated farms. In general, farm managers/decision makers are generally old and poorly educated, raising the question of whether farmers have an adequate ability of many farmers to adjust to changing market conditions. Not surprisingly, a large percentage of family income (33%) is derived form non-farm employment 60. An analysis of cash and non-cash (and farm and non-farm) income in a sample of farmer household incomes reveals important findings. Approximately 42% of the HH in the sample are land-owners (LO), which means they are operating their own farms. The remaining 56% either cultivate land as tenants, sharecroppers, or squatters, or they are landless. The pattern that emerges from a comparison of the two groups suggests that on average the category of households that own land are generally poorer than non-land owners, probably because they are less mobile and in part because they are older. The median income of LO households - both in terms of total and per capita income is about 30% lower than that of Non-LO. LO obtain about 49% of their total household income from agriculture (including food produced and consumed at the farm level), compared to 37% for Non-LO households. LO households are headed by older individuals, averaging about 53 years for LO vs. 42 years for Non- LO. LO households are poorer probably because they are less mobile and because LOs are older than non-LOs. Non-farm incomes, particularly income from wages and public transfer payments, are much more important in low-income than high-income HH's. Demographics also matter. Wealthier HH's are significantly smaller and more likely to be female-headed than poorer HH's. The average size of households declines continuously from 5.9 members in the poorest 20 percent to 3.1 members in the best-off 20 percent of households. At the same time, the proportion of women heading rural households increases significantly from 8.2% in the poorest, to 31.6% in the best-off IlH's. 34 61. Work Status. Another distinction of interest is work status. This refers to working in formal or informal jobs in the urban sector and to whether the worker is employed, self-employed, or an employer (i.e. work position) with the later distinction perhaps interacting with the first, as the self-employed are typically viewed as belonging to the "informal" sector. Defined as jobs for which social security payments are made, formal sector jobs command much higher wages than informal sector jobs, even after accounting for experience and schooling in conventional regression equations. Nevertheless, once account is taken of "cherry picking" by formal sector employers (i.e. employing those with the most favorable characteristics over and above their schooling and experience), the positive impact on earnings of the formal sector variable disappears.4' On the other hand, differences between earnings by employers and employees, which are significant, may reflect the market reward to employers of the larger income risks of their employer condition. 62 Adding the rural/urban and worker position variable reduces the impact of education. Completing 4, 8, 12 years, and university education raises per capita household income respectively by 30%, 49%, 98%, and 204%, lower than those in Table 9. The effect of age declines to 2.8% and the effect of family size remains unchanged at 26%. Table 12 indicates the poverty ratios for urban-rural groupings as well as for work status and work position groups. It shows that the incidence of poverty is higher in the rural sector and in the group of informal employees. Table 12: Pernambuco (PE) - Headcount Ratio (HCR) of the Indigent Household Characteristic Headcount Ratio of PE Relative HCR - PE Relative PE Relative Indigent PE HCR for Specific HCR/Brazil HCR/Northeast Characteristic/ Relative Relative HCR Average PE HCR HCR (1 ) (2) (3) (4) Year 1991 1998 1998 1998 1998 Total 36.9% 36.7% 1.00 1.00 1.00 Location _ Urban 32.9% 33.5% 0.91 1.11 1.08 Rural 51.4% 54.8% 1.49 0.82 1.04 Working Informal Employee 46.8% 51.0% 1.39 0.88 1.02 Status Forrnal Employee 27.0% 31.0% 0.85 1.48 1.16 Inactive 0.0% 36.9% 1.00 1.06 1.13 not worked in ref. Week 42.1% 41.3% 1.13 0.95 1.11 Work Employee 33.4% 34.2% 0.93 1.11 1.01 Position Self-employed 43.7% 39.8% 1.08 0.88 0.92 Employer 5.2% 9.2% 0.25 1.39 1.05 63. Other variables of the poverty profiles tell us: (i) as expected, the poor have inadequate access to public services, thereby exacerbating their poverty; and (ii) again, the economic sector where poverty concentrates most is agriculture, followed distantly by 41 See Tannen (1991). In Reis and Gonzaga (1997) provide an alternative formulation where the assumption of a non-linear return to education performs better than the dual labor-markets model. 35 industry. In those profiles, public services include water, sanitation and electricity. As argued below, poor (low quality) access to education services must be added to the list. 64. Health Conditions. The tables above do not provide an indication of health conditions. Brazil's health outcomes ranked low in a recent international comparison.42 Poor health conditions are both a central consequence (as a result of the impact of low income and education on the demand for health) and cause (as a result of the impact of low health on the educational performance of the poor) of poverty. As indicated in Table 13, Pernambuco's health outcomes are in general lower than those of the rest of the country (Northeast's and Brazil's averages). People live on average 4 to 6 years less than the average Brazilian. Infant mortality remains high (62%, compared to 58% for the Northeast and 36% for Brazil in 1998) despite recent improvements. 65. Pernambuco has made considerable progress in reducing infant mortality (IMR) in recent decades. Infant Mortality decreased from 185 deaths in 1960 to 62 deaths per 1,000 live births in 1997 (although there is current under-reporting of infant deaths). Much of this improvement has been attributed to the implementation of the Community Health Agent Program. Mortality is due in large measure to Perinatal causes (58% of total deaths before 1 year of age). Infectious and Post-neonatal mortality, which is largely attributable to adverse environmental conditions such as water and sanitation, is responsible for 47% of all infant deaths. Table 13: Pernambuco: Health Outcomes Indicators Indicator [Init Pemambuco Northeast Brazil Total fertility rate ( 1998) Average no. of children 2.32 2.79 2.44 Life expectancy at birth (men) (1999) Years 60.74 62.41 64.33 Li.fe expectancy at birth (women) (1999) Years 66.08 68.53 71.4 Source: Ministry of Health. DATASUS Table 14: Brazil: Infant Mortality State/Region: Est. infant mortality rate (no. of deaths Post neo-natal infant mortality rate (no. of deaths of of infants per 1.000 live births) (1998) infants (28-364 days) per 1,000 live births) (1998) North Region 34.6 13.13 Northeast Region 53.5 26.43 Bahia 46.3 20.75 Pernambuco 59.0 30.78 Ceara 53.3 28.03 Southeast Region 22.1 7.46 South Region 18.7 7.58 Center-West Region 25.6 8.9 Brazil 33.1 13.33 Source Ministrv of Health. DATASUS 66. Child mortality is an central concern of the health system. There are an estimated 762,000 children less than five years old living in Pernambuco (about 10% of the total population). The group from 1 to 4 years old accounts for about 612,000 children. The major causes of death in the former group are respiratory, infectious, parasitic and 42 See World Health Organization (2000), Statistical Annex Table 2. The validity of the WHO ratings has been contested by the Government. 36 external diseases. Most of these conditions Table 15 Brazil: Child Mortality 1998 are largely preventable or avoidable with State/Region: Diarrhea Respiratory adequate health interventions. Diarrhea (%) (%) accounts for 12.57% of children's deaths North Region 7.59 5.88 (compared to the country average of Northeast Region 11.88 6.30 6.81%). Bahia 9.28 6.12 Pernambuco 12.57 6.01 67. Infectious, communicable and Ceara 16.97 7.64 vector transmitted diseases are still Southeast Region 3.32 7.89 important. Infectious and communicable South Region 4.20 8.17 diseases still account for a large proportion Center-West Region 4.95 6.07 of the population's health problems. They Brazil 6.81 7.08 also contribute to worsening the nutritional Source: Ministry of Health DATASUS status of poor children, which in turn, makes them more susceptible to contracting diseases by debilitating their immune system. Dengue and Tuberculosis are also important sources of illness and death in Pernambuco. Pernambuco has one of the highest incidence of cholera (29.81%) compared to the national average of 2.52% in 1999. The state is also facing the challenges of emerging non-communicable and chronic diseases. Cardiovascular diseases, injuries and cancer have become major health problems. Like in most of Brazil, the group of external causes occupy the 2nd place of total deaths (14%), mainly due to road accidents and homicides. In sum, while there have been some important improvements in health indicators in recent years, the current health situation needs to be improved much further. Table 16: Brazil: Selected Infectious and Communicable Diseases (1999) State/Region: Malaria Dengue(incidence Cholera(incidence Tuber-culosis (incidence rate rate per 100,000) rate per 100,000) (incidence rate per per 1000) 100,000) North Region 46.97 121.88 0.0 53.40 Northeast Region 1.19 200.84 7.89 50.46 Bahia 0.0 60.33 2.46 59.99 Pemambuco 0.01 381.25 29.81 47.50 Ceara 0.01 222.81 1.34 50.35 Southeast Region 0.01 72.36 0.01 51.40 South Region 0.02 5.91 1.92 37.24 Center-West Region 0.6 102.89 0 33.72 Brazil 3.36 104.48 2.52 47.96 Source: Ministry of Health DATASUS 37 Table 17 - Brazil - Exposure to Violence - Ten CRIME AND VIOLENCE Capital Cities (% of People Knowing Persons that Watched 68. At least a brief reference Cases of Violence should be made to crime and Total Recife Recife/ violence in Pernambuco, which has Total an adverse impact on both growth Armed Robbery 24 30 1.25 and on the welfare of the poor. A House Break-In 18 21 1.17 recent survey of ten capital cities43 Murder 13 28 2.15 in Brazil reveals Recife's population Shooting 12 27 2.25 as perhaps the most exposed to Assasination 12 33 2.75 violence. Recife has the highest Physical Agression 11 28 2.55 exposure in nine of ten indicators of Taking a weapon from 9 27 3.00 violence (in house break-ins, Recife someone was exceeded by five other cities). Drug Use 8 23 2.88 This is summarized in Table 17. Drug Purchase/sale 8 24 3.00 Recife dominates also in several Imprisonment 8 27 3.38 other indicators of violence in the Knived 8 28 3.50 survey. High levels of violence in Abduction 5 19 3.80 Recife is matched by a perception Source: Secretaria de Estado de Direitos Humanos (2000) of poor efficiency of the relevant institutions, including a perception of very poor efficiency of the police. However, the Government has recently launched a major program to strengthen its crime prevention program. VOICES OF THE POOR 69. After looking at all the numbers, it is revelatory to hear on the condition of poverty directly from the poor. This is described in Box 1, which draws from the Bank's study on Voices of the Poor, which included Pernambuco. There, the impact of unemployment on poverty comes out more strongly as a cause of poverty, followed by the role of education. Lack of empowerment comes out as a fundamental factor behind the persistence of poverty. Also, poverty is seen as a cause of violence and crime. 70. This concludes the discussion of the record and possible causes of poverty levels in Pernambuco. This section has shown that: (i) poverty in Pernambuco has not declined during the last two decades; (ii) the standard human capital model provides a good explanation of income differentials; the poor are poor chiefly because of their low levels of human capital; the yield of investments in human capital appears to be lower for the poor; (iii) household size has an adverse effect on per-capita household incomes; (iv) poverty is largely a rural phenomena in that most of the poor are in the rural sector; incomes from urban sources provide an important complement to income from rural sources for rural dwellers; (v) the quality of education delivery is lower for the poor due to lower levels of parent education and lower quality of teachers (specially at the 43 See Secretaria de Estado de Direitos Humanos (2000). The cities investigated include Porto Alegre, Sao Paulo, Rio de Janeiro, Belo Horizonte, Salvador, Recife, Belem, Manaus, Porto Velho, and Goiania. 38 municipal level); there is some scope for improved efficiency of education delivery without increased wages, in that class sizes are still not too large and teacher's salaries in state schools are higher than teachers' salaries in private schools. Box 1: Voices of the Poor In 1999, in preparation for the Bank's World Development Report 2000, the Bank undertook direct consultations with the poor in twenty three countries (including Brazil).' The report of these consultations throws valuable light on perceptions by the poor of their well-being and social mobility, priorities, institutions, and gender developments in Recife. The Recife survey covered consultations in Padre Jordanio, a new and very poor shanty town, but with a fairly organized communiy; Bode and Morro da Conceiqdo, both with good urban infrastructure and a highly organized communities; Vila Unido, with a fairly organized community; Borborema, one of the poorest communities and very poorly organized; and Entra a Pulso, with a very active community. With regard to well-being and social mobility, important findings include (i) the perception, with some exceptions, that poverty has declined, but that insecurity has increased (due to increased unemployment and violence) (ii) the equation of poverty to powerlessness and impotence and the relating of well-being to security (including steady incomes, employment, land tenure, home ownership, and access to health services and education); (iii) the perception that poverty is caused chiefly by unemployment, followed by lack of schooling and sanitation. and that popular mobillization is a key mechanism for achieving better living conditions; (iv) the feeling that education is the key to social mobility, but that mobility has been undermined by rising unemployment; (v) the perception that the rise in unemployment has also contributed to increased violence and reduced safety (a first priority problem in the view of those interviewed in Recife) and to declining social cohesion; (vi) at the same time reduced employment opportunities are perceived to lead to illegal activities (e.g., drug dealing) in male youngsters and prostitution in youn females; and (vii) the perception that the poor suffer from social exclusion, in the sense that discriminated against and harassed. Perceptions with regard to priority problems mirror perceptions on poverty and its causes. Unemployment, housing and sanitation, and public safety are indicated as the problems with highest priority. The poor feel unable to solve problems by themselves, particularly with regard to unemployment and safety. In the view of those interviewed, the Catholic Church is the most important institution and has the highest approval rating, as a result of its being a key community center and its providing spiritual and financial assistance. Neighborhood associations were ranked second, inasmuch as they provide an accessible channel for expression of demands and for assistance during times of crises. Schools were also highly rated, because they are viewed as a venue for social mobility. Utility companies are also viewed as important institutions, with eletirciyty and phone companies reagrded as the more professional and efficienct, and water and sewerage as the worst companies. The police was cited as the worst institution, extensively cited as ansebt when needed and brutal with the poor. With regard to gender, main perceptions are that women have become more independent and overloaded, increased their education and their eaming capacity (through incorporation into more jobs in the labor market), and increased their empowerment within the household and in the community, although men are percieved to remain as responsible to be the family provider. Nevertheless, poverty is not directly associated with gender, with extended family relations remaining strong. 1. See The World Bank (1 999C). 39 71. The relationship poverty levels and growth should be highlighted. This relationship is appreciated in Chart 19, where states have been ranked according to their per-capita GDP. There, higher levels of GP per capita tend to be associated with lower levels of indigence. On a time series basis, it is likely that Pernambuco's poverty declined prior to 1980, when growth was fast. International evidence also indicates a strong link between growth and poverty reduction.44 Care should be taken not to overstate the argument. Pernambuco (and Sergipe and Bahia) is an example of a state that probably has higher than expected poverty levels. Other factors are probably at work, mainly low education levels. Chart 19 - Brazil States - GDP Per Capita and Indigence - 1997 12000 - - - 80.0% 70~0% 60 0% 50 0% 8f 6000 ___ _____0% 30 0% 4000200 2000 0~~~~~~~~~~~~~~~~~~~~~~~10 Cl '' D I O P1 Dmi > O II X~ > m a > Z > m - - 0 > State -GDP Per Capita IBGE 1997 * Indigence 72. The discussion above and in the previous section on growth suggests strongly that poverty reduction efforts must focus on growth fundamentals, education, health, and social protection. The next chapter discusses this focus in more depth. To conclude this chapter, though, discussions of the performance of education, health, and social protection delivery services are appropriate, as they are critical to the discussion of policies that follows after. These follow in turn. PERFORMANCE OF EDtJCATION SERVICES 73. Education is regarded as a key variable explaining individual and average wage incomes (see paras. 51-54). A higher level of education is also associated with higher per- capita GDPs. Consider, however, the following. Between 1981 and 1998, the education of the indigent in Pemambuco increased from 2.0 to 3.6. Yet, the average income of the indigent in Pernambuco remained flat or declined between 1981 and 1998. This means that while education may be a critical variable to raise incomes and reduce poverty, it is 44 See Roemer and Gugerty (1997), who find that, for the poorest 20% of the population, a rise of 1% in GDP increases incomes by 9.2% (i.e. and elasticity of 0.9; Gallup, Radelet, and Warner [1998], who find an elasticity of 1.0; and Dollar and Kraay [2000], who also find an elasticity of 1.0). 40 not the only variable that matters. Clearly, nil growth during the period did not allow increased education to translate into higher incomes for the poor, despite the strong relationship between incomes and education in cross sections studies of state and individual incomes. Other variables may have also kept a significant fraction of population in indigence, despite increased education levels. These conclusions appear to extend to the poorest 50% of the population, whose average income in 1998 was lower than in 1981, despite a higher average level of education. In contrast, the average income of the richest 50% in 1998 was well above that on 1981. It was also noted that a quantile analysis of labor incomes in Pernambuco confirms that the gain from low levels of education (perhaps through the first eight years) is weak for low-income persons. One - but not the only - possibility that explains this is that education quality is poor. Accordingly, the following paragraphs take a look at education outcomes and policies. 74. Education Level attained by Pernambuco's Population. The educational attainment of Pernambuco's population has improved in Brazil in recent years. Illiteracy rates decreased from 25,4% in 1980 to 14,7% in 1996; the average number of years of education increased, and net enrollments reached 91% in 1996 and are predicated to have reached 95% in 1998. But regional inequalities are large, with performance in the Northeast below those of other regions. The illiteracy rate in the Northeast is 28,7%, the highest in the country. Between 1991 and 1995 the South and Southeast regions reduced illiteracy by 41,6% and 38,4% respectively while in the Northeast illiteracy reduced by only 26,3%. Average schooling is 6.6 years of education in the Southeast but only 4.4 years in the Northeast, again the lowest in the whole Brazil. 75. In Pernambuco, indicators of outcomes have also improved. As discussed before, average years of education increased, from 2.7 in 1980 to 4.4 in 1997. Illiteracy has been declining, from 42.4%in 1980 to 26.2%. in 1996. The decline has been larger for the youngest group. In fact, the illiteracy of those 15-19 years of age in 1995 (11.8%) was much lower than the illiteracy of those 50 and older (53%). Average illiteracy remains higher in rural (52%) than in urban (23.4%) areas, and this is explained by lower coverage in the former areas. Average illiteracy is also higher for males (35.9% than for females (28%), Table 18: Pernambuco - possibly because males enter the labor market sooner. Basic and Secondary Education - Net Covergae 76. Present Education Efforts. Future Rates (%): 1990, 1998 educational attainment of Pernambuco's population Year Basic Secondary will depend on the coverage and quality of current I Education Education and future education efforts. In general, progress has 1990 83.3 14.4 been made. Between 1994 and 1998, school 1991 84.0 15.0 attendance of the 7-14 and 15-17 age groups 1992 81.6 15.9 increased respectively from 92.5% to 94.3% and from 1993 80.8 15.5 63.6% to 76.4%. Net coverage has been increasing 1994 82.7 17.0 strongly, but remains quite low for secondary 1995 83.0 18.0 education (Table 18). The decline in coverage as the 1996 91.5 18.3 level increases is clearly appreciated in standardized 1997 92.8 20.2 cohort flow profiles, which indicate the percentage of 1998 93.4 22.9 qualified students (i.e. students that have attained the Source: SEC PE, Situacao Educacional de Pernambuco, 1990-1998. 41 required prior level), adjusted by age group, that enroll in a given level of education. The profile declines steeply by level. It indicates that of the population that begins first grade of basic education (i.e. of the first eight years of education), 55% enroll in the 2nd year, 27% complete the eighth year, 6% enters secondary education (the next four years of education after basic), and only 3% enters the university level. 77. As one would suspect, the decline in enrollment with age applies fundamentally to the poor. This is so in two ways. First, more of the poor never get to enroll: more of the poor never enter primary school (29% of children belonging to indigent families; 19% of children belonging to families between indigence and poverty). Second, the poor repeat more or enter later: of those children aged 11-14, a higher percentage of the indigent (71%) or poor (53% were still enrolled in the first four years of education (typically for children aged 7-11) than of the richer households. Presumably there has been some progress in increasing the coverage of the poor, but the existing rates still remain too low. 78. There is also some indication that quality has improved. Available data for 1991- Table 19 - Pernambuco - 98 shows an increase in approval rates, a Approval, Failure and Drop Out decline in failing rates, and a decline in drop- Rates (') out rates (Table 19). This follows trends for 1991 1998 Brazil as a whole, although approval rates Approval 53 71 remain lower and failing and drop out rates Failure 24 14 are higher. Data from the National System of Drop Out 18 15 Basic Education Evaluation (SAEB) for 1995 and 1997 also show improvements in Portuguese and mathematics proficiency, except that Portuguese proficiency declined for the first four year cycle of basic education. Chapter 6 of Vol. II provides more details on this data. There is very little research on the determinants of quality. Analysis of the secondary school exam (ENEM), shows that proficiency increases with the schooling of parents, decreases with the age of the student, increases for private school attendance, and for daytime attendance (as opposed to classes at night). 79. School Systems. As in the rest of Brazil, education in Pernambuco is delivered by all public sector levels (federal, state, and municipal) as well as by the private sector. In Pernambuco, most basic education is delivered by municipal schools (51 % of enrollments in 1999, up from 41% in 1980), followed by state (37%, down from 44%) and private schools (12%, down from 15%). There is virtually no federal basic education delivery (except that the federal government finances some of it through transfers, particularly through FUNDEF). In contrast, most secondary education is delivered by state schools (71%, up from 40%), followed by private schools (17%, down from 41%), municipal schools (9%, down from 24), and federal schools (2%, down from 5%). 80. The source of education delivery has a bearing on education quality. In general, private schools have done the best in SAEB tests, followed by state and municipal schools. This results either from the higher parent education of children in private schools or from the better training of teachers in private schools (compared with municipal schools). Thus, teacher training remains and important issue for the delivery of better 42 education to the poor. For example, teachers at state schools are better trained than teachers at municipal schools45. 81. The source of education delivery also has a bearing on the average relationship between outcomes and costs, or average efficiency. Information is very limited, but a number of points can be made with respect to efficiency: With regard to outcomes: * Many municipalities still have low coverage (in 1998, 41 municipalities had coverage below 80% of the population aged 7-14. Instances of low coverage occur in all sub-regions (Mata, Agreste, and Sertao). This low coverage can be linked to (i) lack of schools; (ii) enrollment in better schools in neighboring municipalities; (iii) low demand for education. To the extent that it reflects (ii), low coverage in selected municipalities could be an efficient outcome. * The approval rates are highest for private schools (94%) and lowest for municipal schools (66%), with state schools also low (66%). This ranking applies to Brazil and to the Northeast in general. * The graduation rate for basic education (eighth grade graduates/total enrollment) in 1998 was highest for private schools (7.9%), lowest for municipal schools (2.4%), with state schools in the middle (6.9%). The same ordering of school systems is given by approval rates. The graduation rate has increased since 1990 for all systems. With regard to inputs and costs: * The present federal policy of increased municipalization of education could reduce quality if not offset by increased quality of teachers at municipal schools, who at present are less trained than teachers at state schools. For exarnple, the relative quality of Sao Paulo's schools is believed to be better partly because of the larger prominence of state schools, with better teachers. * Average student/teacher ratios in Pernambuco (below 30%) leave room for increases that improve cost effectiveness. These increases could be implemented through a school bus system (today students use normal public transportation, with a 1/2 ticket subsidy). * Based on rather outdated information (1987), costs per student in state schools are higher than costs in municipal schools, but at some price, for quality is better in state schools. Recent data on state schools indicates that costs increased from 1997 to 1998 as a result of FUNDEF, and that teachers accounted for 66% of total expenses. Costs in private schools are higher (with a lower share accounted by teachers compensation) but quality is better. * Teacher salaries for basic education in 1997 were highest in state schools and lowest in municipal schools. Salaries for secondary education were highest in private schools (lowest in municipal schools). Average wages, for given 45 A measure of qua]ity reported in Chapter 6 of Vol. 11 is based on degrees held. It varies from a minimum of zero (incomplete basic) to a maximum of 500 (at least a baccalaureate). State schools seldom fall below 300, the minimum accepted. In several municipalities, the index for municipal schools fall below 300. 43 qualifications, differ sharply across schools systems, with state schools paying the most, followed again by private schools and municipal schools. From the above it can be concluded that there is room for improvement of outcomes for a given cost. First, increased coverage can presumably be raised through a larger teacher student ratio. Second, increased coverage could be possibly linked with fewer but better municipal teachers, who have today the lowest qualifications. Third, wage growth for state schools can be contained, as it appears to be comfortably above that of private schools, where teachers have good quality and outcomes are good. PERFORMANCE OF HEALTH SERVICES 82. Despite lower health outcomes in Pernambuco, indicators of access/utilization of services in Pemambuco are similar to the country's average (Table 20). 46 In fact, public sector health expenditures in Pernambuco (3.5% of the state's GDP in 1995) are equivalent to the national average (3.4% of GDP in 1995). This disconnect between similar health attention and dissimilar health outcomes could have at least three possible reasons: * Improvements on the supply side are relatively new and could explain only recent improvements on health outcomes. This is unlikely. Health outcomes, as measured by life expectancy and child mortality, have been improving for several decades, but the initial health outcome gap between Pernambuco and Brazil's average was significant. Pernambuco's life expectancy was 15 years below Brazil's life expectancy in 1960. The gap had declined to 5 years by 1997. If anything, this is evidence of gradual improvements in health delivery. Table 20: Pernambuco - Indicators of Health Services Indicator Unit Pemambuco Northeast Brazil Hospital beds Beds!l0,000 inhabitants 34.4 30.4 36.4 Doctor visits Visits/inhabitants/year 2.4 2.21 2.3 Hospital admissions Admissions/ 100 inhabitants/year 8.0 7.7 7.4 Vaccine coverage DPT % of children under one year old vaccinated 74.8 69.4 75.5 Vaccine coverage Measles % of children under one year old vaccinated 87.2 76.6 78.5 Source: Ministry of Health * There are serious shortages of other social services that affect health outcomes, such as levels of education, water, and sanitation. As discussed before, education levels in Pernambuco as well as access to water and sanitation are below Brazil's average. 46 As in the case of health outcomes, the performance of Brazil's health system also ranks low in international comparisons. See World Health Organization (2000), Statistical Annex Table l. 44 * There are serious challenges on the supply side in terms of quality, efficiency, financing of services. These challenges include decentralization,47 lack of clarity in state/municipal roles, weak institutional capacity for decentralized health service provision, inadequate instruments to regulate the relationships among the key players of the system48, and poor technical and administrative capacity at the municipal level49 In any case, the policy implications are that further efforts need to be made to improve the delivery of health services. This is particularly so as adverse health conditions (particularly the prevalence of infectious diseases) are often quoted as one of the "geographical" reasons for poor growth and poverty reduction performance. DELIVERY OF OTHER SOCIAL PROTECTION SERVICES 83. Pernambuco's Government offers several programs that are targeted to specific manifestations of poverty. These are summarized in Table 2 1. Targeting criteria/assistance include child labor, youth exposure to drugs and violence, old age, AIDS infection, homelessness, unemployment, training, rural community investments, distribution of seeds, agricultural infrastructure, technical assistance, and facilitation of land ownership. Most programs are federal in design. Some of these programs have been positively evaluated by the Bank (in the context of the Social Protection Special Sector Adjustment Loan, Approved in December 1998). 84. The overall size of these programs is very limited. They can only have a minor impact on overall poverty reduction. IPEA estimates that public sector expenditures on social assistance in Pernambuco amounted to R$103.5 million in 1995, or about 0.6% of the State's GDP. The sources of funding were 28.3% federal, 38.4% state, and 33.2% municipal. If we assume that the income value to beneficiaries of the programs is equal to the cost of the programs (i.e. 0.6% of GDP) and that the beneficiaries are indigent, the programs would lift the income of the indi ent (who account for about 38% of population and receive about 6% percent of GDPs ) by 10%. The increase in average income required to bring the indigent alone to a monthly poverty line of R132 would be 266% of 47 The extreme level of decentralization that the Brazilian process entails may generate high administrative costs, loss of economies of scale, fragmentation of the system and economic inefficiencies. 48 For example, SUS transfers or payments are made without a clear linkage to performance and outcome indicators, nor are they well controlled and evaluated. Similarly, most "contracting" to the private sector occurs without formal contracts or with obsolete contracts 49 There are several reports pointing out this problem. One of the most compelling ones is the report from the MOH's Auditing Department which denounces municipal deficiencies in the areas of planning and programming; health counsel's operations; accounting and financial control; general management and MIS; availability of municipal counterpart funding; and misuse of public funds. See Ministry of Health (1998), 50 This is calculated on the basis of 1996 data: the average monthly income of the indigent was R$36. Pemambuco's population in 1996 was 7.4 million, of which 2.8 were indigent and received RI.2 billion, or 6% of Pernambuco's R$2 1.4 billion GDP. A 0.6% of GDP transfer 45 the average income of the indigent ($R36). Nevertheless, the programs have a more significant impact on the specific populations that they are targeting. Table 21 - Pernambuco - Social Protection Programs Program Description Number of 1999 Budget Assessment Beneficiaries (R$ 1,000) PETI/Bolsa R$150 grant to families that maintain their 73.126 children. 3,928 Program is effective and well Escola: children at school. Targeting includes 49.193 grants targeted to reduce child labor, in Eradication of children working in sugarcane, manioc, particular in unhealthy or Child Labor fruit, and vegetable production otherwise harmful conditions. Brasil Crian,a Provides education, training and other 9865 youngsters 2,088 Cidada services to youngsters exposed to drugs and violence below 24 years of age Atencao a Pessoa Finances NGO-provided services for the 13.091 1,769 Program provides services for a Idosa elderly poor, such as old-age care. vulnerable group and is targeted at families with a per capita income less than 1/2 of a minimum salary. Apoio a Pessoa Finances NGO-provided services for the 4766 3,387 Provides services for a vulnerable Portadora de disabled, such as care homes and group and is targeted at families Deficiencia rehabilitation with per-capita income less tan 1/2 of a minimum salary. Programa Creche/ Provides assistance for Child Care 46143 6,353 Manutencao _ Table 21 - Pernambuco - Social Protection Programs (cont.) Program Description Number of 1999 Budget Assessment Beneficiaries (R$ 1,000) Programa de FAT funded training program 11 .780 trainees 14,041 Qualifica,ao e in Mata Norte. Requalifi,co 15.720 in Mata Profissional Sul zone. Abrigo Provides Shelter 771 324 Combate a Finances Productive Community 7181 families in 5,551 Pobreza Rural Investments and resettlement benefiting Mata zone small farmers Distribu9ao de Distributes Seeds 31,000 small n.d. Sementes farmers Fortalecimento da Finances Infrastructure and Services 20730 persons 854 Agricultura at the Mata Familiar zone Genoma da Cana- Finances development of genetic 250 de A,ucar sequencing of sugarcane cultivation at the Mata Zone Assistencia e Finances Technical Assistance 6049 n.d Estensao Rural Table 21 - Pernambuco - Social Protection Programs (cont.) Program Description Number of 1999 Budget Assessment Beneficiaries (R$ 1,000) A,ao Fundiaria Finances resettlement, regularization of 2453 711 properties and project preparation Maxoto/Pajeul Finances Infrastructure, Technology and 335000 people 22625 Project Training in poor Sertao ___________________ _______________________________Municipalities Programa Finances minimum income to workers and n.d. n.d. Emergencial De basic consumption baskets to families Combate Aos affected by drought. Also finances water Efeitos Da Seca distribution; perforation and recovery of wells and desalinization; access to water in rural schools, and training for coping with droughts. Programa de Provides training, employment, and an 10000 Ocupa,co income of R$80 per month Temporaria e Capacita,ao de Mao-de-Obra .. 49 3. GROWTH AND POVERTY REDUCTION POLICIES A POLICY FRAMEWORK FOR IMPROVED PERFORMANCE 85. Over the last twenty years, income per capita in Pernambuco increased only 3.4%. This slow increase did not allow any poverty reduction. Poverty reduction will require both faster growth of per-capita GDP as well as faster and better investment in human capital by the poor. Growth theory postulates that a higher gap between the long term (steady state) level of per-capita GDP and existing GDP will trigger higher growth. Policy can affect growth if it has an impact on the variables that determine the long term level of per-capita GDP. As discussed in the previous chapter, the variables that may determine long term per-capita income include human capital (i.e. the stock of educational attainment) and the quality of schooling, population growth, taxes and public expenditures, market distortions (trade, investment, labor), the degree of political stability, institutional quality (including governance, or the ability to enforce laws and to implement programs), geographical factors, and growth of the terms of trade. 86. Given a constant income distribution, growth will reduce the poverty headcount ratio, as evidenced in available cross-country studies. In addition, increased educational attainment of the labor force belonging to lower income households will reduce income dispersion and the poverty headcount ratio. Finally, the system of public sector taxes and transfers will also change distribution and poverty, with the direction of the change depending on the nature of the taxes and transfers. 87. Analvzing Poverty Reduction Policies. The analytical framework for growth and poverty reduction policies discussed below is derived from growth theory, as discussed in the previous chapter, and our knowledge of the key factors behind poverty reduction. It can be explained with the help of Figure 1 below, which indicates linkages between public policies and other exogenous variables, investment, growth and poverty reduction. The linkages are described as follows: * Investment and Productivity Growth. Investment (in physical and human capital) and productivity growth, the key supply or production function factors behind output growth, provide the mechanics through which economic growth occurs. In Brazil, richer states have higher capital intensities and, specially, higher physical capital intensities. Raising the capital intensity and reducing the productivity gaps that exist with respect to other states and countries are the two processes whereby convergence of average income to possible higher levels can happen.5' Hence, policies and other exogenous variables that affect investment or productivity also have an impact on growth. This is represented in Figure 1 by arrows from policies (indicated by grayed areas) and other exogenous variables (e.g. growth in terms of trade) to investment and productivity growth and to output growth. These policies and exogenous variables reflect the prior discussion on growth determinants (para. 41). 51 See Islam (1999), p 496. 50 Figure 1: Conceptual Framework for Growth and Poverty Reduction Policies Initial Level Growth of Geography: of Per-Capita Terms of Trade Location, GDP Resources and Size Trade and Taxes and Institutional Investient Public Quality and Openness Expenditures Political Stabiliry _______________________________ __________________________ Fertility, population Growth, and dependency Human Capital Investment and p- Productivity Growth 4 Schocling and Per Capita Health: QuaniitN GDP Growth *_ and Qaality Employment/ * Labor Market F |Wages Fleubilitl Poverty * InfTastructure Reduction Ser ices for Ihe Poor _ Transtfers to the Poor Population Growth. As discussed before (para. 41), high fertility has a negative effect on income and high population growth may reflect the positive impact of higher growth on immigration and the negative impact on mortality. Fertility could also depend on education and income. Hence, these demographic variables are partly endogenous. This is captured in Figure 1 by an arrow from education to the demographic variables, a two way arrow between these variables and growth, and an arrow from demographic variables to poverty, as the dependency ratio has a direct relationship with poverty (see para. 50). * A government (federal or state) can increase profitability of investment by improving the supply of the key public goods with a bearing on profitability (e.g. property rights, security, regulatory framework) or by providing subsidies or tax exemptions on investments. Typically, state governments subsidize investments in education and in selected industries. The level of investment also depends on expected profitability. Therefore, an arrow also goes in the other direction: an increase in (expected) growth raises the demand for investment because higher growth raises the expected profitability of investment. Also, since profitability is linked to the marginal return on capital and capital exhibits diminishing returns, investment and growth in lagging states are expected to be larger (under the unconditional convergence hypothesis discussed above in paras. 39-40). * Incomes of the Poor. The incomes of the poor are determined by the level and yield of their assets. The main relevant asset is human capital, although others (e.g. land) can also be important. Investment in human capital (education/health) of the poor is viewed as the chief policy to achieve a sustained reduction in their poverty, by raising productivity, employability and wages. * Labor Markets and Policies. Given the central role of income from human capital, labor markets are the key markets for poverty reduction policies. Labor market and other policies have a bearing on employment and wages, which are the chief foundation of the income of the poor. For example, when labor markets become more flexible employment increases and, hence, poverty can decline. The chief policy to improve employment and wages must be to reduce the taxation of labor incomes, including the taxation implicit in excessive regulatory rigidities. Investment policies are also critical for employment creation because firms need to be established and invest to create a demand for employment. In turn, labor policies influence growth by providing one of the fundamental elements in the framework for private sector activity and investment. Figure 1 captures these relationships with arrows from labor market flexibility to employment and wages and to investment an productivity growth. * Access to Infrastructure Services. As indicated before (para. 63) the poor have limited access to infrastructure services. This may reflect a policy-induced market failure. For example, poor access to sanitation services reflects the public monopoly of those services, which has not invested to provide these services. Investment in infrastructure services for the poor (water and sanitation/transportation) reduces poverty by, for example, improving public health or raising the time available for income generating activities. These relationships are captured in Figure 1 by arrows from investment to provision of infrastructure services for the poor, and to poverty reduction. * Transfers to the Poor. Social protection policies are mainly transfer programs to the poor. The main transfers are education and health. But there are several other programs, which are discussed below. These programs are aimed at tailoring poverty reduction transfers for specific groups of the poor. 88. The Role of the Public Sector. Several of the factors discussed in paras. 41 and 87 above define the quality of the environment for businesses, which are fundamental engines of economic growth. That is, a propitious business environment will include favorable geography (access to markets, access to water, and climate), an adequate supply of human capital; a level of taxation that is balanced with the quality of public sector services delivered; access to input, output, and capital markets through liberalized trade and investment markets (including public infrastructure) and flexible labor markets and contracts; and quality public institutions or governance (rule of law [including contract law], transparency, quality bureaucracy, low expropriation risk, and low risks of repudiation of contracts). In this context, the role of the public sector is fundamental in ensuring an adequate business environment, as most of the factors listed above are in fact policy determined. 52 89. In addition to the policies implied by those factors, which mainly provide and environment for the private sector to flourish spontaneously, there may be a more pro- active public sector role in helping materialize synergies within the private sector through coordination of private sector agents. The public sector can perhaps play a positive role in materializing economies of scale that, because they are external to each firm, would not occur by private sector actions alone, although spontaneous collective action is often strong. The areas where the public sector may play a role in this regard is in the provision of such public goods as informnation, research and development, and openness of foreign markets. 90. The framework has been used to understand differences in growth across nations. Box 2 illustrates with two good performers, Ireland and Chile, and indicates the possible relevance of those performances to Pernambuco. The remainder of this chapter discusses Pernambuco's policies. Policies are discussed in terms of three possible dimensions of public policy: (i) rules; (ii) the role of taxes and public expenditures, and (iii) the quality of implementing institutions. These are linked to various proximate targets of policy: infrastructure investment, other investment, labor markets, human capital (education, health), and social protection transfers to the poor. Thus, in Figure 1 (p.50) trade and investment openness, taxes and public expenditures, and institutional quality help determine physical investment. Similarly, it is argued that openness of education delivery options, impact of taxes and public expenditures on education, and the quality of educational institutions help determine the success of education and health delivery policies. The policy discussion below begins with an analysis of the Government's fiscal constraints, which provide a fundamental framework to public policy. THE GOVERNMENT'S FISCAL CONSTRAINTS 91. Fiscal Performance and Debt Outcomes. Pernambuco has led a precarious fiscal life over the last decade. The state's current account surplus (the difference between current revenues and current expenditures) has been close to zero since 1995. Because the state has generated no savings from current revenue, it has had to finance what little capital spending it has done through borrowing or asset sales. As a consequence, the government's debt/ revenue ratio, which had declined from 1.30 in 1993 to 1.08 in 1995, increased to 1.40 in 1998 and to about 1.70-1.80 in 1999. If guarantees are included, the debt/revenue ratio would have been about 2.00-2.10 in 1999. With a new administration taking office in January 1999, fiscal balance has improved considerably, from a deficit of 8.4% of net revenues in 1998 to a deficit of 2.6% on net revenues in 1999. Under the Government's fiscal program, the debt/revenue ratio is expected to decline after 2000 to about 1 (i.e. debt = annual net revenues) by about 2007. 53 Box 2: Two Tales of Growth International comparisons indicate that growth performance depends on fundamental variables such as initial income, initial human capital, fertility, taxes and expenditures, market distorsions, political instability, institutional quality, geographical factors, investment, and growth of the terms of trade. The differences in growth performance across geographical areas are very significant. During 1985-98, annual average per capita GDP growth in the states of Brazil ranged from -0.3 (Rondonia) to 4.3% in Mato Grosso, with Pemambuco at 0.7%. Differences in growth across countries are even more striking. During 1980-98, annual average per capita GDP in 125 countries with available data ranged from -6.0% (Georgia) to 8.4% (China). Among growth success stories, one can point to countries as different as the Republic of Korea (6.2%), Mauritius (4.6%), Ireland (4.0%) and Chile (3.9%). Two tales of successsful growth can illustrate the workings of the fundamental factors at work behind such successes. In Ireland, per capita GDP in 1987 was 63% of Great Britain's. Rapid growth during the late 1980s and 1990s, coupled with declining growth in the rest of Europe, allowed Ireland to catch up with Great Britain. In accounting for Ireland's growth, all supply sources have been important. High employment growth resulted from baby-boom demographics, higher labor participation, lower unemployment and reverse migration. To this, an upgrading of human capital must be added, including third level education that feeded, for example, into the development of the information technology sector. The flow of foreign investment to Ireland has also been remarkable, attracted chiefly by Ireland's low profit tax rates on manufacturing and internationally traded services,and by Ireland's attractive policy and institutional setting. On the demand side, Ireland benefited from trasfers from the European Union's Community Support Program and from increased integration with the other economies of the European union. Trade and investment liberalization contributed static and dynamic gains in output and productivity.' Chile has enjoyed continued high growth since 1984, after the strong recession of 1982. In the aftermath of the recession, through 1986, the key source of growth was employment growth linked to declining real wages that had resulted from high prior unmployment rates. After 1986 and through 1997, employment growth was constrained by the growth of the labor growth and higg output growth was then linked to investment and productivity growth, with the later explained largely by trade openness.2 Barro (1999) has identified six fundamentals that help forecast continued above average growth. Since Chile's GGP is above average today, the convergence force has a negative impact on future growth. Variables with a positive effect include Chile's better rule of law, lower fertility, higher democracy index, higher investment ratio, and lower inflation. Both education and inflation variables are close to the world averages and have a negligible effect on above-average growth How is all of this this relevant for Pemambuco? First, trade openess, beyond Brazil's federation of states, will benefit Pemambuco, who presently trades mostly with other states. Second, investment openess can also become a positive factor, particulalrly in the water sector, where a major gap in supply exists. Fourth, Pemambuco will benefit from improving the quantity and quality of education, where it lags with respect to other states outside of the Northeast. And fourth, Pernambuco will profit from a more flexible labor market that allows higher investment to translate into ahigher demand for labor and human capital. I. See OECD (1999) 2. See Coeymans (1999A) and (1999B) 54 92. Revenue Performance. Pemambuco derives 60 percent of its gross revenues52 from taxes, nearly all of it from the state value added tax (ICMS) of 17% on interstate sales of goods and selected services (mainly telecommunications), with 7% charged on interstate sales to Pernambuco by the more developed "producer" states of the southeast).53 Transfers from the federal government account for most of the remainder (35%.). The largest such transfer is the FPE, a formula-based revenue sharing of the federal income and industrial product taxes. In the years just preceding the Plano Real, the state also derived a considerable proportion of its revenue from interest. With the collapse of inflation, this source of revenue has largely disappeared. 93. The Value Added Tax. ICMS revenues fell in real terms between 1990 and 1994. They then recouped in 1995 after the introduction of the Plano Real, growing 20% in 1995. Growth has subsequently slacked off. ICMS revenues grew 10% in 1996, but only 2% in 1997, and 3% in 1998. In 1999, real ICMS revenues declined by 6.4% and in 2000 (through September) they recovered by 7.8%. Part of this variation is explained by trends in the state economy. During the 1990s, ICMS revenues have Table 22 roughly paralleled changes in state Impact of Lei Kandir on ICMS Revenues GDP. Tanzi effects also appear to Rs$ Mn have a major impact on revenues. ________________ 1997 1998 1999 2000* haveeanuesolaImpact onPrevenues. Exports 24 27 26.3 24.2 Revenues lagged GDP growth Electric energy 23 23 5.7 5.7 during the period of high inflation Capital goods 46 46 51.9 41.9 in the early 1990's, exceeded it in Subtotal 93 96 83.9 71.8 the post-inflationary period Less compensation 0 36 10.8 39.3 following the introduction of the Net impact 93 60 73.1 32.5 Plano Real, again lagged low GDP % gross ICMs 5.8% 3.5% 4.1% 1.5% growth in 1999, and probably Will Source: State Secretary of Finance. exceed faster GDP growth during 2000. 94. Two Explicit tax policiesalso appear to explain the weak performance of the ICMS in the last several years. The first is the state's PRODEPE program discussed further below. (paras. 148-56). A recent state report estimates that PRODEPE tax deferrals reduced ICMS revenues by 5.5% in 1997.54 The second is the 12997 federal Lei Kandir, requiring states to exempt most exports from the ICMS and to grant tax credits for investments in capital goods. The exemption on exports had only a minor impact in Pernambuco. The state's principal exports-sugar and tropical fruits-were already exempt. The tax credit for capital goods did have a significant impact, however. 52 Gross of mandatory state tax sharing with local governments (municipios). 53 Within Pemambuco, the largest shares of ICMS revenues are derived from gasoline (15%); telecommunications (10%) and electric energy (5%). As a sector, manufacturing accounts nearly forty percent of ICMS revenues. Most of this is derived from domestic consumption, however: beverages, food products and tobacco. Unlike the neighboring state of Bahia, Pernambuco has no preeminent export sector that would permit the state to export part of its tax burden onto other states. 54 "Program de Reestructuracao e Ajuste Fiscal do Estado de Pernambuco, 1999-200 1" January 1999. 55 The Lei Kandir cost the state 5.8% of ICMS revenues in 1997. In 1998, a compensation formula went into effect, reducing the net impact of the Lei to 3.5% of revenues. 95. Growth in ICMS revenues will largely depend upon the state's GDP growth, improvements in compliance, and the impact of possible tax reform. The ICMS is a broad based tax, able to capture much of the economy, with the important exception of services (other than telecommunications). GDP growth could explain an average real ICMS of about 4% per year, given constant compliance. There is also some prospect for increasing ICMS revenues through improvements in administration, particularly at the retail level. Retail sales are notoriously difficult to tax in developing countries. But, given comparisons with other states, there would seem to be scope for increased collections. Chart 20 indicates that Pernambuco ranks low (18th out of 27 states) in terms of the ratio of ICMS collection to GDP. It ranks sixth (together with Sergipe) in the ten Northeastern states. Furthermore, collection has not improved during recent years. The ratio remains at around 7%. Perhaps a ratio of about 8% could be achieved with stronger enforcement. Finally, under the tax reform now under debate, taxes on goods and services at the state, municipal, and federal level would be combined into a single value added tax, imposed by both federal and state governments. The rate on interstate sales would be reduced to zero (except for an interim federal rate, intended to discourage evasion). Other things being equal, this reform would, following an adjustment period, favor Pernambuco at the expense of major producer states such as Sao Paulo. By reducing the interstate rate to zero, it would permit the state to tax 100% of the value added incorporated in products sold within its boundaries. 96. Revenue sharing . Pernambuco's other principal source of revenue consists of transfers from the federal government. By far the largest of these transfers-accounting for 71% of the total-is the fundo de participacao dos estados, the FPE. FPE resources are derived from fixed shares of the federal income and industrial products taxes. These resources then are distributed among the states and the federal district according to a formula based on population and the inverse of GDP per capita. Chart 20 - Brazil - Ratio of Value Added Tax (ICMS) to GDP by State - 1997 14.0% 12.0% 6.0% - _ _ _ _ _ _ _ _ 4.0% -- . 2.0%…S III 0.0% state FE CMS/G7DP 56 97. Transfers have grown briskly in the post Real period. In the year immediately following the plano real, total revenues from transfers increased 28% in real terms. In the following three years they increased by 8%, 10% and 14% respectively. The outlook for transfers largely depends upon growth in federal tax revenues. Changes in the tax structure under the tax reform would be accompanied by changes in the revenue sharing system. 98. Personnel Expenditures. Much of the state's fiscal difficulties are due to Table 23 - Pernambuco State Government - large and growing recurrent expenditure Number and Wage Bill of Personnel Number Wage avg. expenditures. Personnel expenditures- ** bill wage by far the largest item-peaked in 1995 Civilian personnel 159203 151.5 952 at 83% of revenues. Although the Inactive 50590 40.6 803 proportion dropped to 71% by 1998, Legisl.& judicial 21132 23.6 1117 personnel spending still constitutes by Active civilian 87481 62.2 711 far the largest claim on state resources Direct adminis 62838 43.6 694 (63% for the 12-month period ending in Education 39184 20.3 518 August 2000). Pernambuco has about Health 4900 6 9 1474 160,000 people on its payroll.55 As Others 7782 11.2 1439 shown in Table 23, roughly one-third of Indirect admins 23621 18.6 787 these are retired. Teachers (and other Active militiary 16.5 education personnel) constitute the *excludes staff not paid by treasury largest block of active staff, accounting **excludes military for about one-third of the total. Staff in the legislative and judicial branches ("outros poderes") and in indirect administration (including the highway department) also account for a significant proportion of staff. Wages vary considerably across sectors. As shown in Table 2, the average wage in the education secretariat is less than half that of the legislative and judicial branches. As a result, Pernambuco spends a larger proportion of its wage bill on the judicial and legislative branches than it does in education. Box 3: Cost Implications of FUNDEF Yet another recent claim on the state's resources is a federally mandated program aimed at increasing spending on primary education. The FUNDEFE program, which went into effect in 1998, requires each state to divert 15% of its ICMS and FPE revenues to a special state fund earmarked for primary education. These state contributions, along with funds from the municipal and federal government are divided between the state and its municipalities on the basis of primary school enrollment. Because the majority of primary education in Pemambuco is provided by municipalities (only 44% of primary school students are enrolled in state schools) the state contributes more to the fund than it receives back. In 1998, the state's net loss totaled Rs$85 million. In 1999, the state expects to contribute Rs$ 322 million to the fund and receive only Rs$ 220 back, for a net loss of Rs$ 101 million. 55 This figure excludes military police. 57 99. Salary increases granted by the former administration explain much of Pemambuco's recent financial difficulties. In 1995, the incoming administration granted a 50% increase in salaries, propelling personnel costs to 83% of net current revenues. In the following year, the state attempted to retrench, by introducing a voluntary departure program (PDV). Although the state hoped to reduce staff by 11,800, only 1,332 staff were ultimately let go. A second attempt (PRODESI II) produced only 258 participants. In 1998, the state returned to its policy of increasing salaries, with concessions to education, health, and security, as well as a 10% increase for judges and a 100% bonus for state attorneys. These measures increased the wage bill by 11% in nominal terms. 100. The current administration, which took office in January 1999, has taken some steps to rein in personnel costs. It began by eliminating 2,029 senior political appointments (cargos de confianza), 20% of the total. It also cut certain non-salary benefits--such as the licensapremia, which permitted staff to take a six month paid leave after ten years of service-and instituted a permanent PDV program. In the near term, it intends to impose a Rs$ 7000 ceiling on all staff in the executive branch. 101. However, like all Brazilian states, Pernambuco confronts a set of Constitutional constraints that make it difficult to control personnel spending. Under current legislation (to be changed shortly) staff cannot be dismissed except for cause (except under certain conditions described below). Salaries cannot be reduced in nominal terms. The constitution also grants generous retirement benefits: civil servants may retire after 35 years of service (5 less for women, and for teachers of either sex) with benefits based on 100% of the retiree's exit salary, indexed to wages in the position formerly occupied. 102. Many of these stipulations pre-date the current Constitution. Civil servants have long been protected from arbitrary dismissal, for example. State government traditionally accommodated the special benefits and protections granted to civil servants by hiring staff under private sector labor law (CLT). Under the CLT, staff can be dismissed without cause. The CLT also provides for a pension scheme, funded by employer contributions, with benefits considerably smaller than those granted to civil servants. The most striking effect of the 1988 Constitution was to end this practice. Until recently, the Constitution required all government bodies to adopt a single employment regime (regime unico). In effect, this forced states to extend civil service benefits to all staff, including former 56 CLTistas. As a result, virtually all Pernambuco's staff enjoy the protection and benefits of civil servant status. 103. Recent amendments to the federal Constitution will slightly increase the state's discretion over personnel costs. Under the I gth Amendment, government bodies (including states) are granted a temporary authority to dismiss civil servants provided (1) personnel costs exceed 60% of revenues, a threshold established in complementary legislation (the Fiscal Responsibility Law); (2) at least 20% of cargos de confianza have been eliminated and (3) non-confirmed civil servants have been dismissed. 56 While the 1988 Constitution did not explicitly require the conversion of all CLTistas to estatutarios, states chose to comply rather than convert estatutarios to CLTistas. 58 104. Pernambuco is moving forward to meet these requirements. With personnel costs at 71% of net current revenues, it far exceeds the Camata threshold. There is scope for further personnel cost reductions. Pernambuco's ratio of employee costs-i.e. net of payments to retirees-to net revenues is about 50%, or about the average that Brazilian states spend on employed personnel (ativos). It is widely believed that this ratio can be reduced. Some states report ratios as low as 40% of net current revenues. 105. The state has already eliminated 20% of cargos de confianza. It is now in the process of dismissing unconfirmed staff. The state's first priority is the 1,370 staff who entered service after 1988 without undergoing the constitutionally mandated public competition (concurso publico.) It then intends to pursue the some of the 17,968 staff who entered service between 1983 and 1988 without a concurso publico. Those staff who are employed in education and health sectors will be exempted. This will leave 5,151 staff vulnerable to dismissal. While the state is on its way to acquiring the power to dismiss civil servants, it is not clear that it intends to exercise it. In a recent press conference, the State's governor announced that he no longer sees the need for wholesale dismissals of unconfirmed staff.57 106. The state is taking more decisive steps to address its retirement obligations. At present, the state has 51,000 retirees. Payments to retirees constitute about 27% of total personnel costs. Virtually all benefits are paid out of the general revenues of the treasury.58 Like many Brazilian states, Pernambuco is concerned that the aging of its civil servants, combined with the generous retirement benefits mandated by the Constitution, will bankrupt the state in the near future. In an attempt to forestall this, the state is in the process of establishing a retirement fund. The retirement fund proposal is closely entwined with planned changes in an existing fund which finances state employees' medical benefits and payments to dependent survivors (IPSEP). At present, IPSEP is financed from employees deductions and contributions from the state. The employee contribution is equal to 10% of wages; the state's contribution, five percent. As this is insufficient to cover the fund's cash flow, the state also contributes an open ended subsidy (currently equal to Rs$ Smillion per month). 107. Under the state's proposal, a new fund (FUNAP) would be established, which would assume responsibility for the state's retirement obligations. IPSEP would be abolished. Its dependent survivor program would be transferred to FUNAP. (Several proposals for handling IPSEP's medical benefits program are under discussion. See box.) Funding for FUNAP would come from two major sources. The first is employee contributions. FUNAP is to receive the entire 10% employee contribution now paid to IPSEP, plus an additional 4% surcharge to be imposed this year. FUNAP is also expected to derive significant income from returns on assets. The state plans to capitalize FUNAP to the tune of Rs$ 575 million, consisting of (1) Rs$ 300 million from sale of state power company (to be sold in February, 2000); (2) Rs$220 million from the former state bank's 57 Jornal de Commercio, 2 January 2000 58 The state retains a small number staff in indirect units administration who are employed under the CLT. Their retirement benefits are paid by the federal INSS. 59 real estate portfolio59; (3) Rs$ 200 million from the repayment of a debt owed by a federal housing fund (FCVS); (4) Rs$ 117 million from state government real estate assets, and (5) Rs$ 54 million from the real estate portfolio of IPSEP. In addition, FUNAP will be allowed to accumulate cash in its initial years of operation. For the first five years, the state government intends to pay all retirement and dependent survivor benefits from general revenues, while allowing employee contributions to build up in the fund. 108. This arrangement will not be sufficient to finance the full costs of Pernambuco's existing and future retirement obligations. The state expects that FUNAP's annual income will be only sufficient to pay the benefits of currently active staff who are at least 5 years away from retirement. As a result, the state will continue to pay the pensions of existing and pending retirees (and their dependent survivors) as long as they live. 109. Is this plausible? In a study of this kind, it is not possible to perform an adequate actuarial analysis. Back-of-the-envelope calculations suggest that the figures are of the right order of magnitude. If, for example, the rate of return on assets accumulated in the first five years is six percent in real terms, the fund will have accumulated capital of Rs$ 1.7 billion before it begins to make payments to beneficiaries. If the return on assets continues to be six percent, this capital will generate about Rs$ 103 per year, or about 20% of the state's current annual spending on retirement. Box 4: Financing Medical Benefits With 100% of employee contributions allocated to FUJNAP, the state's reforn proposal would leave medical benefits bereft of funding. The state initially considered this option. In effect, it would transfer the obligation for medical insurance onto the federally funded universal health care system, SUS. SUS funds, however, are allocated among states on the basis of population. Transferring state staff would increase the number of SUS participants by 400,000 with no increase in funding. This option was therefore rejected. Another option is to reduce benefits. Under this option, the state would continue providing services in state-run facilities, but would no longer reimburse treatments by private providers. The state has rejected this option on the grounds that existing state hospitals lack the specialized facilities available through private providers. The state briefly considered expanding specialist services in its own facilities, but rejected the idea due to the heavy volume of investment it would require. The proposal currently under consideration is to maintain the existing system of benefits, but split the costs between the state and employee contributions. 110. Whether the fund is desirable is a different question. If the return on capital is only six percent, the state would be equally well off using its assets to reduce its stock of debt. (As discussed below, the average interest rate on state debt is about six percent.) Enthusiasm for the fund may come from the expectation that returns will more closely approximate the yields on federal bonds in the post Real period. Real interest rates have 59 BANDEPE was sold in 1998, net of these assets. 60 fallen sharply recently, however, and this expectation may not be justified. The real key to Pernambuco's retirement obligations lies not in increasing funding but in reducing benefits. This key lies in Brasilia, not in Recife. 111. Debt Service. Debt service is not yet a major claim on Pernambuco's cash flow. The state's debt stock/revenue ratio has, however, increased sharply over the past decade. In 1992, contractual and bond debt was equal to 73% of net current revenues. By 1998, that ratio had increased to 129%. As of September 1999, it stood at 177%.6° 112. In the early 1990's, virtually all of Pemambuco's debt consisted of debt to the federal government arising from previous rescheduling. This included debt formerly owed to federal financial intermediaries (rescheduled under Law 8727/1993); former foreign debt (rescheduled under Law 7976/1989 and related legislation) and debt to the federal unemployment insurance scheme (FGTS). A fiscal crisis in 1996 prompted a new foray into the capital market. In that year, the state issued Rs$622 million in bonds to cover general expenditures. It also borrowed Rs$ 75 million from the federal treasury to pay arrears to personnel, and another $17 million from the Brazilian American Merchant Bank, an private bank headquartered in the Cayman Islands. The latter two loans were rescheduled in 1997. Under the terms of Lei 9496/1997, they were transformed into long term (30 year) debt to the federal treasury at a real interest rate of 6%. 113. The stock of debt jumped again in 1998, when the state borrowed Rs$ 1.245 billion from the federal government to finance the privatization of the state bank BANDEPE. (BANDEPE was sold in November, 1998). This debt was financed on Lei 9496/1997 terms. 114. Pemambuco's debt stock continues to grow under the current administration. Little of this represents explicit new borrowing, however. To date, the largest debt contracted by the new administration has been a Rs$70 million loan from the federal government to compensate for the costs of increased spending on primary education. (See the discussion of FUNDEFE, below.) Most of the increase in debt is instead due to (1) the state's assumption of the debt of the state housing company (Rs$505 million)6" and (2) the revaluation, in Real terms, of the state's foreign debt, due to a devaluation in early 1999; and 3%, the indexation of refinanced debt at the IGP-DI rate, which far exceeded general inflation in 1999, due to the high weight of tradeables in the IGP-DI. 115. The most pressing debt problem facing the new administration was the bonds floated in 1996. They were due to mature in 1999-2001, with fairly substantial cash flow implications. 62 In June, agreement was reached with the federal government to refinance 60 Based on the actual stock of debt on September 30, 1999 and an estimate of annual revenues based on receipts through September. 61 Note that the housing company debt does not represent an increase in the state's total indebtedness. Before 1999, it was carried on the books as a loan guarantee. Since the state has assumed responsibility for servicing the debt, the current administration has chosen to classify it as an obligation of the treasury. Legally, however, it remains a guarantee. 62 In prices of December 1998, Rs$ 242 million was due in 1999, Rs$ 296 million in 2000, and Rs$ 332 million is due in 2001. (As of September 1999, the 1999 block had been yet amortized.) 61 the bonds on Lei 9496/97 terms. Amortization was therefore halted, and the debt added to the stock of the state's long term debt. 116. In fiscal terms, the stock of debt remains manageable. The average interest rates on the state's debt are low and maturities are long. As shown in Table 24, the largest single block of debt consists of a package now incorporating the BANEPE loan, and the rescheduled BAMB and Resolution 162 loans. It bears a real interest rate of six percent and a maturity of 30 years. Other federal debt that is denominated in Reais (about 36% of the stock) carries similar terms. Interest rates on debt denominated in dollars (about 14% of the stock) is subject to nominal interest rates ranging from 4 to 7 percent (but also carries a foreign exchange risk.). Rough calculations put the state's current interest obligations at about 13% of net current revenues.63 Table 24 - Pernambuco State Government - Trends in Stock of Debt Mn Rs (prices of Dec. 1999) 1992 1993 1994 1995 | 1996 1997 1998 1999 Sep-00 Total 1,435 1,519 1,346 1,504 2,167 2,268 3,276 3,777 3,962 Internal Debt 1,249 1,299 1,155 1,374 2,032 2,143 3,123 3,549 3,737 7976,8727, 538 1,214 1,015 1,207 1,129 1,095 1,115 1,022 1,008 Other contractual 711 85 140 167 281 166 180 329 343 Bonds 622 722 686 133 141 9496 (including BANEPE) 160 1,142 2,065 2,245 COHAB 579 589 Foreign debt 186 220 191 130 135 125 153 228 225 Guarantees 611 422 103 599 607 618 651 689 701 Source: State Secretary of Finance 117. Capital Spending. Capital spending in Pernambuco is virtually non-existent. As shown in Chart 6, state spending on investment steadily declined throughout the 1990s, reaching its present level of about 2% of net current revenue under the current administration. As in many places, capital spending in Pernambuco is a function of resource availability. Steadily declining current account surpluses during the 1 990s reduced the amount of internal savings available for investment. (The spurt in capital spending in 1996 was in a sense financed from the proceeds of the state bond issue, since the bonds funds were fungible.) 118. The state engages in virtually no project financing. Since 1994, the only project financing of any consequence has been a small rural development loan from the World Bank and a loan for tourism development from the Banco do Nordeste. From a fiscal standpoint, this was probably desirable. With virtually no current account savings, the state had very little margin to take on additional debt service obligations. t3 Much of Pernambuco's debt to the federal government is subject to a debt service ceiling. If debt service obligations exceed 11.5% of net current revenue, the state is allowed to defer the excess, capitalizing it into the stock and amortizing it once the debt service ratio falls below the threshold level. Prior to 1988, the state's debt service obligations were far below the threshold level. The BANEPE loan has pushed the state close to it, however. 62 119. The State's PPA. Since 1993, Brazil's federal Government has been introducing improvements in its public expenditures, through a better design of multiannual plans, a closer link of these plans to the budget, and stronger implementation. In Pernambuco, the PPA process is also being strengthened, including a participatory process that is appropriate for the close contact that sub-national governments have with the beneficiaries of public expenditure. Pernambuco's 2000-03 PPA, entitled Change and Development (Mudanqa e Desenvolvimento) represents a major change in this regard with respect to the 1996-99. PPA. Under the new approach (Programa Governo nos Municipios - OrCamento Participativo), Pernambuco's municipalities were grouped in 10 regions to undertake participatory discussions/negotiations of development priorities and partnerships. From those discussions, the Government developed a participatory and regionalized budget as well as partnership agreements between the state government and the municipalities. The budget is increasingly being implemented through a network management system (based on a matrix where implementing actors are classified as strategic or programatic). Implementation emphasizes decentralization (implementation at the most local level and by society, not the public sector, where possible); flexihility, with the State playing its role of negotiator/articulator/coordinator; transparency, with mechanisms of fiscal control and access to information; IT/communications technology; and management by results or products. 120. Fiscal Sustainabilitv. The public sector's capacity to fund investment and investment promotion programs is limited by fiscal conditions, which require governments to generate primary surpluses to contain debt levels. At the federal level, expenditures on these programs will be limited by the need to generate primary surpluses of about 2.5% of GDP in order to meet debt targets. As in most states in Brazil, Pernambuco's fiscal performance became unsustainable. Although debt levels were moderate for a Brazilian state (the debt/revenue ratio was 1 in 1997, low for Brazilian standards), excessive expenditures increased debt levels to 1.7 times the level of net revenues. This is now requiring fiscal adjustment, including the recent dismissal of 608 non-tenured employees. It also covers administrative reform, including the merging of several state secretaries and other public agencies, and the extinction of several public enterprises. The Government reduced its primary deficit from 8.4% of net revenues in 1998 to 2.6% of net revenues in 1999. Still, the wage bill remains at 63.5% of revenues (August 2000) as defined by the Law of Fiscal Responsibility. As a result of excessive expenditures, interest obligations have exceeded the primary balance. Therefore, Pernambuco continues to accumulate debt, albeit at a smaller pace. Pernambuco's present debt of about R$4.0 billion (September 2000) is 1.7 times its level of net current revenues of R$2.2 billion. Pernambuco is covered by a debt restructuring agreement that will require the state to generate primary surpluses. The Government expects to generate a primary surplus in 1999. There appears to be little or no scope for further net borrowing. 121. What is the overall prognosis? Under the debt refinancing agreement that the Government of Pernambuco signed with the Federal Government, Pernambuco's state debt must begin declining this year. The path of the debt as a ratio of net real revenue is indicated by Chart 21. Three alternative paths are in this Chart. All assume a 6.6% real interest rate (the average rate of the states' debt in Brazil) and a 4% revenue growth rate. The paths are as follows: 63 * The highest path shown maintais 1999 debt (1.7 Chart 21 - Pernambuco - Projected Debt of the times the level of net real States: 2000-2010 revenues). This path requires 1 8 - a perpetual primary surplus 1.6 of 4.3% of revenues a14 beginning in 2000. Z 1.2 0> * The higher of the two debt D f' 1.0 agreement paths reflects new o 0° 8 borrowing, including 'O 0.6 possible Bank support for an t . urban infrastructure project. 0.2 7 This path achieves a - N N N N debt/revenue ratio of 1.0 in N o ° ° C ° ° 2008, requiring an average LUpper Debt Agreement Path primary surplus of 13.4% of [ --- 1999 Level Path net revenues (about R$400 million in 2001) during 2000-08. This effort could be reduced to the extent that privatization revenues are used to reduce debt. To maintain the revenue ratio at 1.0 beyond 2003, the state must then generate a perpetual primary surplus of 2.4% of revenues. * The lower path assumes that there would be no new borrowing. This path achieves a debt/revenues ratio of 1.0 in 2006 and requires an average primary surplus of 16% of revenues (about R$500 million in 2001) during 2000-06. Again, this effort could be reduced to the extent that privatization revenues are used to reduce the stock of debt. Again, to maintain the revenue ratio of 1 beyond 2003, the state must then generate a perpetual primary surplus of 2.4% of revenues. Hence, moving from the 1999 level of debt to the higher of the two debt agreement paths requires an additional average primary surplus of about 9.1 percentage points of net revenues during 2000-08 (i.e., above the primary surplus required to maintain the debt/revenue ratio constant), but achieves a reduction of 1.9 percentage points of net revenues in the perpetual primary surplus required to maintain debt at the lower level. Moving from the 1999 level of debt to the lower of the two debt agreement paths requires an additional average primary surplus of 11.7 percentage points of net real revenues during 2000-06 to achieves the reduction of 1.9% of revenues in the perpetual primary surplus required to maintain debt at the lower level. 122. Although any of these debt/primary surplus paths is sustainable (in the sense that the present value of the prescribed perpetual primary surplus exceeds the value debt in the initial period), the lowest path is clearly the less risky to growth and interest rate shocks. For example, a decline in the average rate of growth from 4% to 3% requires a smaller adjustment in the primary surplus of lowest path (0.8 of a percentage point) than in the primary surplus of the highest path (1.5 percentage points) in order to reestablish the desired debt levels. Similarly, an increase in the average real rate of interest (say, because of higher indexation than general inflation) also requires a smaller correction of 64 the primary surplus in the lowest path than in the highest path. In fact, lower debt levels, as compared with the highly indebted states of the south, will allow Pernambuco to achieve a debt revenue ratio of 1 much more easily than in those other states. These lower risks suggest that Pemambuco should opt for a low debt path to be made binding for all governments. The new Fiscal Responsibility Law provides a framework to establish lower state debt targets. These would become part of the Budget Directives Laws, that would be binding for the next Government. 123. Will adjustment to lower debt levels be possible? In 2000, the State's fiscal situation has been largely determined by privatization proceeds from the sale of CELPE, the power company. The sale, undertaken in February 2000, generated $Rs 1.1 billion in net proceeds. However, this sale is unlikely to contribute to debt reduction. As noted earlier, roughly fifteen percent of this has been pledged to capitalize the state's proposed retirement fund. Much of the rest will probably be spent on long-delayed capital works. 124. Once this windfall is exhausted, the immediate prospects of the state are mixed. Growth in revenues will largely depend upon growth in the economy, which has probably picked up since late 1999, although some increases in the ICMS/GDP ratio can be obtained by improving administration at the retail level and by scaling back tax incentives. In its present form, the proposed tax reform would benefit Pernambuco considerably (given the destination-based nature of the new VAT). The ultimate wording and timing of the refonn is, however, still open to debate. 125. In terms of reducing expenditu:res, the state continues to confront a large personnel bill. To date, the administration has taken the more expedient measures to control it-reducing special appointments (cargos de confianqa) and dismissing staff who had been hired illegally. The state now has the Constitutional authority to undertake the more difficult task of dismissing statutory employees. It remains to be seen whether it will use it. With the measures taken thus far and the recent revenue growth due to the economic recovery, the state had reduced the ratio of personnel expenditures to revenues to 63% (for the 12-month period ending in August 2000). The state expects to further reduce this ratio, to about 55% in 2001. 126. There are of course positive aspects to the state's situation. Debt service costs- other than the immediate amortization of bonds-are reasonably low. The state's decision to increase employee contributions to 14% also promises to reduce payroll costs, regardless of the merits of the other aspects of the FUNAP plan. 127. The most encouraging aspect of the state's situation may be the management approach of the present administration. As indicated the beginning of this section, the first nine months of this administration saw reversal in previous downward trends. The state is now running a current surplus (of about four percent of revenues.)64 Even the overall balance is slightly positive (1% of net current revenues), due to a virtual shut down in capital spending. 64 This was largely due to cuts in "other current expenditures". On an annualized basis, ICMS revenues continued to fall, and personnel costs, to rise. 65 128. As indicated, should Pernambuco aim for the lower debt path, it would need to generate surpluses of the order of 16% of net revenues until 2006. This assumes that the Government spends the proceeds from the privatization of the energy company, in lieu of using these proceeds for debt reduction. It will not be easy to generate the required surplus with current personnel costs, which amount to about 63% of net revenues and leave only 19% of net revenues (about 1% of the state's GDP) for public investment (which has been about 2% of net revenues) and other costs. The slower path of debt decline would require surpluses of the order of 13% of net revenues, leaving 24% of revenues for investments and other costs. To recover the state's investment, a further reduction in the ratio of personnel costs to net revenues is called for. The Government's target of reducing these costs to about 55% of net revenues (a 15% reduction in the ratio) is reasonable. This will require both increased efforts to raise ICMS revenues, for example by 10% of total net revenues (or about 1% of the state's GDP), as well as efforts to deepen administrative reform and reduce personnel costs, by say 5%. THE GOVERNMENT'S DEVELOPMENT POLICIES 129. The Government of Pernambuco's formulation of its development policy strategy is organized in four areas: * economic development, focused on infrastructure, technology, education, and employment: * economic diversity, focused on industrial and tertiary development, small businesses, tourism, and agriculture recovery; * welfare of the population at large, focused on social exclusion, water and sanitation, health, education, culture, safety, and justice. * participation and transparency, focused on improved public administration and fiscal policy; 130. In essence, the Government's growth vision (the economic development and economic diversity components above) comprises infrastructure development that will help improve the state's access to domestic and foreign markets as well as energy supply and private development - including education, technology, and employment - that will strengthen the state's private production. Table 25 summarizes the Government's more specific growth policy actions or targets. INFRASTRUCTURE DEVELOPMENT 131. With regard to infrastructure development, the State's policy depends partly on federal priorities. In 1996, the Federal Government introduced its Brasil em Acao program consisting of 60 development actions in Brasil. Investment projects are defined for various development poles. Pernambuco is covered by the Polo Petrolina/Juazeiro (in Bahia), which will help develop the San Francisco waterway. Brasil em Acdo programs are funded by various sources, including federal and state budgets, external loans, private financing, and internal funding. 66 Table 25: Pernambuco's Development Policy Framework: Growth Areas Actions/targets Infrastructure * Upgrading of road and rail network *Improve management of ports *Complete investments in Suape complex * Upgrade air transport infrastructure * Extend rural access to electric power * Regulate energy use and improve quality * Increase quantity, diversity and quality of telecommunication services * develop supply of natural gas * develop renewable energy sources Industrial * attract private investments, with a focus on state-of-the-art technology and Development exports * support industrial recovery in the textile, food, metal-mechanics, and chemical sectors * support the development of mining and gypsum * small business credit and training Agricultural * Support production of fruits and vegetables; cattle, cotton, coffee, and sugar- Development alcohol * Strengthen agricultural research * Promote development of water reserves * Improve marketing, storage, and inspection * Implement diversification at the Zona da Mata Tourism * help identify best tourism products * help develop service capacity * invest through PRODETUR Technology * Specialized knowledge through informatics * Promote science and technology education in schools * Support business research * Develop resources for use of high speed information networks Promote research in strategic areas Training and * Raise training opportunities, specially for first job seekers employment * Increase training for priority areas (e.g. irrigated agriculture) * Disseminate employment opportunities * Establish mechanisms to improve negotiations between labor and businesses Source: Summarized from Governo do Estado (1999) 132. More recently, the Government introduced a broader development program, Avan,a Brasil. This is part of the Federal Government's 3-year budget plan (PPA). The program is framed in terms of nine development axes. Pernambuco belongs chiefly to the Transnordestino Axis but is also covered by some actions in the Sao Francisco Axis. In the Avan,a Brasil program, the Federal Government has identified a number of development actions for this axis. These cover chiefly investments in infrastructure and expenditures on education and health. The activities can be summarized as follows: 67 Transportation * Further works in the San Francisco Waterway; * Construction of the Trans-Northeast Railway from Petrolina in the southwest tip of the state to Salgueiro (NE of Petrolina) and to Missao Velha in Ceara. The state intends to recover its existing railway running from Recife to Salgueiro. * Upgrading (doubling) of the Recife-Cuararu Road (BR-232) * Upgrading of the Natal-Aracaju road (BR-101); Energy * 240MW Thermal Generation Plant near the Port of Suape; Telecommunications * Several investments in the regional telecommunications network; Water Management * Water management investments: (i) water channels in: Jucazinho, Pajeui, Carpina, Sertao Pernambucano, Oeste; (ii) dams and channels in Pirapama, Pontal, Pontal Sobradinho, Brejo de Santa Maria, Serra Talhada, (iii) underground water development in Jatoba; (iv) deflecting water from Rio San Francisco to PE (and CE, RN, and PB); Education * Bolsa Escola, remedial education, increased coverage of primary education, teacher training; Health * Increased Health coverage (aiming at 100% coverage by 2007; control of communicable diseases, increased hospital beds and physicians; Water and Sanitation * Water and sanitation investments (increased coverage of sewerage to 25% and of waste collection to 60%); * Several studies on restructuring/development of production. Financing sources for these programs include taxes/public borrowing at the federal, state, and municipal levels. 133. Not surprisingly, many of the investments in the federal PPA also appear in the state PPA, as both are possible sources of funding. The 2000-03 State PPA, submitted last year, envisages investments in irrigation (R$760 million), municipal water supply (R$138 million), the Recife-Cuaruaru road (R$160 Million), ports (container and grain terminals, R$210 million), the thermal generation plant near the Port of Suape (R$270 million), upgrading of airport (R$196 million), tourism (R$594 million), basic education (R$1.3 billion), health (R$95 million), training of the unemployed (R$220 million), security (R$35 million), and technology (R$30 million).65 134. As indicated above, infrastructure investment plays a central role in the GOP's development plan. The Government is aiming to increase investment in transportation 65 See Govemo de Pernambuco (1999), pp. 48-59. 68 (ports, roads, railways, waterways, and airports), irrigation, water for human consumption, energy, and telecommunications. Perceived investment requirements in these areas are probably beyond Pernambuco's public sector financial means. Goodfiscal performance, obtention of required funding, and efficiency in investment and in the delivery of infrastructure services will probably benefit from strong private sector involvement in infrastructure investment and delivery. This is already happening in some areas like ports (under the landlord model), telecommunications (with Telemar the private company that replaced the state-owned TELPE) and energy (with the privatization of CELPE). Nevertheless, more can be done. 135. Even in ports, an area where Pernambuco has been successful, there is room for improvement. Pernambuco's trade and investment openness, a growth determinant, depends on Brazil's import restrictions (chiefly tariffs), on transportation infrastructure, and on investment restrictions. Import restrictions are primarily determined by federal trade policies, on which Pernambuco has little say. However, the State has influence on port costs, inasmuch as it helps determine the efficiency of seaport operations, which depend on the federal Port of Recife and the State Port of Suape. 136. Both ports follow the model of landlord ports, with all port facilities delegated (rented) to firms. Suape's port authority delegated all port operations to firms (including Petrobras, the Brazilian public sector firm that owns the largest number of port terminals (8) in Brazil) in 1991, and Suape was the first Brazilian port to have all of its port operations managed by independent firms. Similarly, the Port of Recife delegated all of its operations to firms in 1996. Nevertheless, the impact of privatization on port costs has been contained by the excessive costs of unionized port labor. Landlord port operators rent a space (i.e. they are arrendatirios) from a public port authority, must use the port's union labor, and cannot reject freight. In the alternative case of new privately owned ports, operators established after 1993 can employ any labor they choose and decide which clients they serve. In Brazil, of 443 million tons of cargo that went through sea ports in 1998, only 127 million tons went through public piers; in Pernambuco, all cargo (6.7 million tons in 1998) went through the public rented piers, amounting to 1.5% of Brazil's total cargo.66 Fees in the Port of Recife are relatively low (R$158 per container in 1997)67 in the Brazil context; in the Port of Suape, fees have also been lowered. Pernambuco could perhaps benefit from opening competitive labor markets to port operators. This could occur by changes in the 1993 Federal Port Law (Lei de Modernizaqdo dos Portos), which presently restricts the use of non-registered labor by facilities rented from the port authority. Surprisingly, there are no existing or requested privately owned terminals in Pernambuco. This could reflect the fact that the existing 66 In contrast, the Ports of Espirito Santo (with 1.7% of Brazil's GDP) moved 22% of Brazil's total cargo in 1998. 67 Recife was cited as the Brazilian port with the lowest fees in a Survey conducted by the Empresda Brasileira de Planejamento de Transporte, cited in Gazeta Mercantil (1998). World Bank (1999B) reports the Port of Recife as the lowest in cost compared to the ports of Fortaleza, Salvador, Santos, Suape, Rio de Janeiro, and Vitoria. To achieve low costs, the Port of Recife had to reduce the number of employees from 1300 to 190, while privatizing the operation (through a landlord arrangement). Furthermore, the Port of Recife has an automatized graill terminal that does not require stevedores. 69 arrangement of the existing landlord ports is working well68 In principle, all ports, whether rented from the public sector or owned by the operators, should have competitive access to the labor market and also be subject to a framework that regulates the operators to ensure adequate provision of this important public service. 137. A related issue that needs to be resolved, at the federal level, is the regulation of multimodal transport, where (i) present insurance policies place the burden of responsibility for losses and damages on the owner of the merchandise or on the specific transport operators, and not on the multimodal transport operators, and (ii) the ICMS is collected in a cascading fashion by all states through which the merchandise passes. 138. Roads also appear to be critical for the State's development vision. Two federal roads are important for Pernambuco's trade with the rest of the world, including trade with other Brazilian states (see Map on p.5). These are federal roads BR-232, which crosses the state from east to west and connects with other federal roads to the north and south as well as to the state's feeder road network; and BR-101, which connects Recife to the most of the other Northeast capital cities (Natal, Joao Pessoa, Recife, Maceio, Aracaju and Salvador) and to Fortaleza and Sao Luis through BR-304. As indicated before, the GOP, with the Federal Government and other states, are devoting resources to upgrade these roads. Other federal roads (e.g. BR-316 to Araripina and BR-1 10) crossing through Pernambuco from Salvador in Bahia to Areia Branca in Rio Grande do Norte also require repair. Several state roads also need work. Acceleration of public investment in roads is limited by present budget constraints on public sector investment, although the Federal Government is gradually upgrading the federal roads through five year maintenance contracts with private contractors. Financing through toll roads appears to be ruled out due to road users lack of willingness to pay and to the high interest costs that toll concessionaires must pay. As long as these conditions remain, investment in roads is likely to remain slow. Eventually, Pernambuco could consider introducing a toll road program, like those that have been introduced in some of the other states. While these may result in increased fees paid by the consumer, those fees will bring forth better service and help overcome the financial constraint. This would increase investments in and maintenance of roads and reduce the pressure of these on the public budget. 139. Complementing the State's road network is the federal railway from Recife to the city of Salgueiro in the west. Pernambuco is seeking the recovery of that stretch, construction of additional rail from Salgueiro to Petrolina, from Salgueiro to Missao Velha in Ceara, and from Salgueiro to the Gypsum pole in Araripe, and recovery of the line from Recife. It forms part of the old Transnordestino project discussed before. The listed investments would cost some $716 million. The rail is in concession to Companhia Ferroviaria do Nordeste (CFN)69. The technical projects are close to being completed, 68 It could also result from the fact that port costs of sugar shipments are low because the grain terminal is automatized and does not require stevedores, 69 Until recentty (September 2000), CFN ownership was equally shared by Companhia Vale do Rio Doce (CVRD), Companhia SiderOrgica Nacional (CSN), and Taquari Participacoes S.A (whose shareholder also owns CSN). In September 2000, ownership of CFN by CVRD and CSN was consolidated and BNDES (which is providing R$65 million in funding), SUDENE, and the GOP are trying to work with CFN to have the projects implemented. 70 but CFN has not yet made most investments to improve the rail. This could indicate that the expected payoff from potential users may well not justify those investments. The State, the Federal Government and the concessionaire need to work on the project evaluation of these investments to ascertain whether they are worth pursuing. 7 1 ,<1 1 6 a l\J/\\~Aifi IL~~~~~~~~~~~~ *4 j! y~~~~~~~~~~~I - f gil <,~~ ' P'' zX i F, O | _ \t E X * - 72 140. Waterways can be another important area for transportation in Pernambuco. The main waterway relevant for Pemambuco is the Sao Francisco river. It offers an inexpensive mode of transport between the Center-South and the Northeast of Brazil, with significantly lower fuel and labor requirements. The river offers natural navigation capacity from Pirapora in Minas Gerais to Petrolina-Juazeiro in Pernambuco-Bahia (see map p.73), with some variation in depth that makes navigation difficult in some months. This navigability to Petrolina makes the river an important option for multimodal transport to Pernambuco and the rest of the world. In 1996, the river carried 2.4 million tons, but the Companhia de Navegaqdo do Sao Francisco - (FRANAVE) estimates additional demand of 2.5 million tons per year if improvements were made to transportation conditions. However, to improve navigability, its waters need to be regulated and the riverbed needs to be dredged in some areas where the impact of erosion has made it difficult to navigate. The river is managed by the Administraqao da Hidrovia de Sao Francisco (AHSFRA), an agency of the Federal Companhia Docas do Estado da Bahia. Given the importance of internal and international trade for Pernambuco, it would be important for the State to get more involved in improving the federal arrangements for assessing, financing, and undertaking needed works, including possible private sector involvement. 141. The further development of transportation infrastructure (including the Port of Suape) and the opening of public utilities to privatization are aimed at opening trade and investment in the State. Tax expenditures (subsidies) are also aimed at attracting private investment, particularly in the industrial sector. The following paragraphs discuss opportunities for progress in these and the other areas; rule of law ("the extent to which institutions provide effectively for implementation of laws, adjudication of disputes, and orderly succession of power"); corruption in Government ("related to the frequency of bribes in areas such as international trade, taxation, and police protection"); quality of the bureaucracy ("including the degree of autonomy from political pressure); expropriation risk (assessment of risks of outright confiscation and force nationalization); and repudiation of contracts. 142. Water supply, for irrigation and for human consumption, is perhaps Pernarnbuco's main infrastructure problem. To improve water supply the state has taken actions to: * Deflect water management from COMPESA, which has a vested interest in using water as an input, to a Secretary of Water Resources (SRH) and the state's environmental agency (SECTMA) * Obtain approval of State Law 11426 and 11427 on January 1997, which has a sufficient scope to allow effective water management * Initiate water basin management, albeit with inadequate resources; * Initiate assignment of water rights, albeit at a slow rate (meeting less than 50% of requests); Accelerate the privatization process of COMPESA, the State Water Company, which has a record of low efficiency, poor financial performance, and inadequate investments. 73 Sar49Kana V m ;raer; I bc'Ciria i Lav/ U Z j 5 a a Sglrfflirll r lS 7 n oi on: d Jeu dI lsap 3 - iharil+'C' JS(ie So tradin io C J~~ ~~ 4,rm .1i1je'? que a Sarata i(G quRlgf u ll i Jesu ; da Lar,.q y F.~~~~~~~~~~~~~~~~~~~~~1 A,~~~~~~f-41js C$C Key recommendations include: * Accelerate the privatization process of COMPESA. Pernambuco urgently needs an improvement ini supply conditions of water for human consumnption and of saniitation services. The Bank has recently completed a study that covers recomm-ended privatization practices in the water sector. 70 * Begin collecting charges for bulk water: charges for bulk water (perhaps thlrough comipetitive a, 2ii:iicnl of water rights) are a clear prerequisite for effective water management. all the more so in areas where water is relatively scarce. * Increase investments in water supply. As discussed before, Pernaatbuico is one of the slates where water supply is lowest, Increased investments will requ'ire bringing in private finance through privatizationi. * tnicrease investment in mainitenance of existing systems. 0 The Wold Batik ( i999A). 74 * Improve metering of water consumption (today only 34.1% of produced water is metered). Again, private sector involvement is advisable to ensure adequate metering. * Protect and recover water sources through improved sanitation (sewerage); * Improve operation of all water sources, which requires better information on those sources and batter rules of use. * Optimization of Sao Francisco irrigated agriculture; * Introduction of more efficient irrigation techniques * Speed up assignment of water rights, including collection of fees. 144. Furthermore, there is a need for water market reforms that create incentives for private investment in water infrastructure. Irrigation around the San Francisco river, such as in Petrolina, seems to be contributing to create many new jobs in the fruit sector - presumably a positive externality in a condition of labor surplus in the region. A policy issues is how to encourage an efficient market driven expansion of irrigation. An approach could be for the government co-finance part of the investment, and private users pay the full cost of maintenance of the system. There are relevant examples in some countries regarding irrigation investment and tradable water rights. Chile, for example, has both. On investments, the government offers small and medium scale irrigation subsidies under a competitive biding system, where beneficiaries have to "offer" how much they will contribute to the total investment, that is, a variable subsidy scheme. 145. As expected, a poverty reduction issue is providing infrastructure for the poor. Those without access to infrastructure are poorer. This is evident from household surveys, which reveal a higher percentage of poor among those that: burn household waste because there is no collection; do not have access to water; use septic Tank, rudimental cesspit, drain, river or lake disposal; or other rudimentary disposal systems because there is no sewage collection; and have no access to electricity. The Government is presently developing an infrastructure project for poor areas in Recife. The Bank is presently discussing with the GOP possible Bank support for this project. Furthermore, at the Government's request, the Bank is also discussing possible Bank support in developing a medium-term water sector strategy for water and sanitation. OTHER PRIVATE INVESTMENT POLICIES 146. Private investment promotion policies include subsidized credit, tax subsidies, and cluster strategies. Federal credit programs (FINOR, BNDES, FNE) provided loans amounting to 2% of GDP in 1995 and 1996. FINOR, the chief regional economic incentive offered by SUDENE, offers financial assistance for new or expanding firms in the Northeast71. BNDES also provides specially priced investment loans to northeast firms, and the Northeast Constitutional Fund provides low priced loans, preferably to 71 FINOR has financed many blue chip industries, including White Martins, Companhia Vale do Rio Doce, Baterias Moura, Gerdau, Carrefour, Votorantim, Nestle, Gessy Lever, Alcoa, Samello, Grendene, Azaleia, Dakota, Bauducco, Monsanto, Brahma, Antarctica, Kaiser, Schincariol, Coteminas, Vicunha, Azulejos Eliane. 75 small business and rural producers, to activities using local labor or intermediate inputs, and for production of basic foods. 147. There are two sources of tax exemptions. First, there are income tax exemptions through SUDENE for new, expanding, or modernizing enterprises (75% reduction from 1998 to 2003 in industry or agriculture; and for existing enterprises (37.5% reduction from 1998 to 2003). SUDENE also offers to exempt half of the income tax, if reinvested and complemented by own resources (amounting to half of the amount exempted). 148. Second, there is the Programa de Desenvolvimento de Pernambuco (PRODEPE), which approved tax exemptions on investments amounting to about 2% of GDP in 1995 and 1996. In its present version, enacted in October 1999, PRODEPE offers the following incentives: * a 12-year (extendable to 15) 75% ICMS credit (85% during the first four years if the firm is located outside of Recife or Suape) on sales of incremental production to expanding firms in priority industries (productive chains in agri-industry, metal-mechanics and transport materials, electroelectronics, pharmaceuticals, beverages, and non-metal minerals); * a 15-year deferment of payment of up to 5% (3%) of the ICMS on interstate sales by firms in priority industries (by distributions firms); * a 12-year loan of 60% of the ICMS due on sales of incremental production of products not formerly produced in the state (30% for products already produced), with a 75% forgiveness of amortization amounts upon payment, to expanding firms in eligible non-priority industries. * a 15-year deferment of payment of up to 5% of the ICMS on interstate sales of production by firms in eligible non-priority industries. * a 5-year loan of 10% of imports by wholesale importers * an ICMS tax credit of 3% of transfers from industrial firms in other states to distribution firms in Pernambuco. 149. It is strongly recommended that the Government suspend its PRODEPE program of investment subsidies to firms. This recommendation is made on four grounds. One, the feasible size of those programs is likely to make the impact on growth rather limited. Two, the design of these programs typically involves picking winners, a very risky procedure in economic development policy. Three, it is far from obvious that industrial location has followed fiscal incentives. Fourth, expenditure policies can potentially attract investment to Pernambuco in more effective ways. These reasons are discussed in turn. 150. During 1996-99, PRODEPE-supported investments amounted to US$2.0 billion, or about US$666 million per year. Pernambuco's annual GDP during those years was of the order of US$20. Hence, PRODEPE supported investments amounted to about 0.7% of GDP. Assuming that this support was definitive for bringing in those investments, the program would explain only about 3.5% of total investment. This would assume, in addition, total investment in Pernambuco is around the Northeast average of 20% of GDP. 76 151. But one cannot comfortably assume that the Table 26: Pernambuco - Main incentives were necessary to attract investment to Reason for Loaction in Pemambuco. This is so because firms locate in a given Pernambuco of Firms place for many different reasons and because other Receiving Fiscal Incentives policies also have an impact on the firms decisions.72 Percent Consider first the reasons why firms would invest in Fiscal Incentives 20.5% Pernambuco. These have been investigated by Market 22.7% Vasconcelos and Vergolino (1999) and are indicated in Pemambucan Firm 11.4% Table 26. The 44 firms surveyed are firms that received Lack of Competition 6.8% fiscal incentives. The table indicates that fiscal incentives Raw Materials 6.8% were the most important reason in 20% of the firms. Pre-existing Plant 6.8% Overall, the most important reason was the size of the Other 15.9% market and many other reasons were also important (e.g. None 2.3% Pernambucan firm, a proxy for networking).In another No answer 6.8% survey, fiscal incentives does not even appear as a major Total 100.0% Source: Vaconcelos and Vergolino reason for location. Instead, the market remains as the (1999), P. 27. predominating motivation, followed by the availability of human resources, the geographical location, diversification of supply, and supply of infrastructure.73 In fact, the response to a slightly different question (Obstacles to Investment in Pernambuco, Table 27) highlights infrastructure as the main constraint to investment in Pernambuco74. 152. But perhaps the most important reason to suspend the PRODEPE program is that it competes with better uses of the resources, some of which also have a bearing on attracting investment, and with fiscal stability, an important condition for the overall quality of public policy. The many reasons cited for location of immigrating firms in Pernambuco include infrastructure supply and human resources, on which public policy has an important bearing. But more broadly, public policy has an impact on other factors that determine a firm's competitiveness in Pernambuco, including, for example, the extent of trade and investment openness, the extent of crime, and the quality of Government. 153. Moreover, the PRODEPE program is designed to support selected industries where it is believed that there is development potential. This is a costly, difficult, and 72 International evidence indicates that investment subsidies, in the form of grants, soft loans or tax concessions, indeed have an effect on location. However, this does not imply that these incentives are necessary or the best way to attract investment. A policy maker can choose alternative ways of spending the subsidy money, for example in better training of the labor force or in transport infrastructure. 7 See TGI Consultoria em Gestao (1998), p. 85. Market size is also a leading force in the literature on the concentration of economic activity, with economies of scale (internal or external to the firm) forcing agglomeration in large markets, and congestion, high input prices, and limits to market size encouraging dispersion to other markets. See Henderson, Shalizi, and Venables (2000), pp. 2-4. 74 Pacheco (1999) investigates the patterns of industrial location in Brazil. He finds a loss of participation by Rio and Sao Paulo; the emergence of "productivity islands" in all regions; and stronger growth in old urban peripheries and middle-sized cities. The shares of the Northeast has increased since 1970. However, Pernambuco's share has probably declined as its share in industrial employment has declined and its share in total industrial investments has been below its share in GDP. 77 risky task. Costly, because it demands resources for detailed analysis of sectors and supervision of the program; difficult, because of the inherent difficulty of picking winners and, more importantly, picking winners who need the subsidy to be attracted to locate in Pernambuco; risky, because the chances of missing the target are enormous, leading to a waste of resources that could other wise be used in better areas. Furthermore, the opportunities that the program creates for corruption cannot be dismissed. 154. To these arguments, it may be added that the sustainability of investments attracted by tax subsidies is in question, as the investments may lack a foundation based on the fundamental merits that the state has to offer. If so, they would put a permanent burden on the government's resources. 155. To conclude, incoming tax reform will disallow tax competition, for long a pervasive feature of Brazil's ICMS tax system.75 This and the reasons above, will require Pernambuco to find better ways to attract investment. The focus of this study on investment openness, human capital formation, public expenditures and other actions to improve the supply of public goods, and fiscal balance provides a better paradigm than a program with very uncertain prospects of success that provides subsidies to selected firms. 156. If, despite the reasons indicated above, it were felt that subsidies need to remain to offset those of other Table 27 - Pernambuco: Main states, the government needs to consider a simple and Obstacles to Investment in transparent design that promotes employment. One Pernambuco possible approach is to focus on net increases in Lac oe 15.1% employment. For example, an auctioned ICMS subsidy Incentives linked to employment created in connection with the Fiscal Incentives 9.6% investment could be considered. Under this method, the Tax Burden 6.8% Government would award a tax credit of, for example, Lack of Infrastructure 42.5% ten years, on the ICMS to be applied to the value added Political Inefficiency in 12.3% by those investments with the lowest bid on the size of Attracting Businesses the tax credit per employment created and sustained. Bureaucratic Burden 4.1% This would maximize the "bang for the buck" in terms Other External Factors 9.6% of employment creation. To implement this approach (or Total 100.0% any other approach), the Government (at the Finance Source: Vasconcelos and Vergolino (1999) Secretariat) must have satisfactory data on the employment history of beneficiary firms, which would be furnished and certified in the bidding documents, as well as an adequate capacity to assess that the additional employment is not fictitious. 157. In addition to the subsidies described above, Pernambuco and other Northeast states are implementing cluster development strategies. In Pernambuco, clusters include medical care and informatics in Recife, poultry in the Agreste region, mining (gypsum, marble, and granite), tourism in several places in the state, and fruits in the Sao Francisco Valley. These appear to be high growth sectors. The cluster strategy for the informatics 75 See Piancastelli and Perobelli (1996). 78 cluster is described Box 5. As discussed there, these strategies involve the use of public resources in direct support of a sector. In principle, they run the risks of distorting resource allocation (because some sectors are more supported than others) and of supporting the wrong sectors (sectors that will not survive) and thus of a poor return on public investments. 158. Growth policies are a fundamental part of the Government's development program to reduce poverty. This is because growth is likely to help reduce poverty through increased demand for labor but also because growth policies include education and health policies, which include building the human capital of the poor. As indicated before, the GOP policy program includes policies to reduce social exclusion, to improve the infrastructure of the poor, and to improve education and health delivery. These are discussed in what follows. EDUCATION POLICIES 159. Pernambuco's educational policies depend partly on federal programs, partly the State's programs, and partly on municipal implementation. The following paragraphs discuss the level of resources that are earmarked for education in Brazil and Pernambuco, the federal and state programs to improve coverage and quality, the possible impact of those prograrns, and recommendations to improve this impact. 160. Public Expenditure in Education. Brazil has a vast number of Federal Government programs that are aimed at increasing schooling for the poor. To begin with, Brazil's public sector budget assigns a fairly strong priority to education. The Brazilian Constitution (Art. 212) requires that the Federal Governmnent spend at least 18% of its tax revenues on education and that states spend at least 25% of their tax revenues, including the proceeds from inter-governmental transfers, also on education. Sixty percent of these resources must be used for elementary education through the year 2006. In addition, the Constitution (Art. 212) provides for earmarking of a tax of labor earnings (the salario- educaqdo, set by Decree 87043/1982 at 2.5% of wages) for expenditures on primary education. The tax is collected by the Fundo Nacional de Desenvolvimento da Educav ao (FNDE), with 1/3 distributed to Federal Government programs to attain universal access to primary education and 2/3 distributed to state secretaries of education (prorated according to tax collection amounts).76 76 Prior to December 1996, when Law 9424 was enacted, businesses could use the proceeds of the tax to pay for the education of their employees' children, through a established school at the firm (some 60 still exist in Brazil, many in sugar mills), through tuition at a private school, or through compensation to those employees that pay private schools for their children's education. 79 Box 5: Pernambuco's Clusters The GOP's development vision espouses the idea of development clusters, or sets of inter-related and interdependent firms. In a cluster, firms are related by their geographical proximity and their input- output relationships. In clusters, well-developed and diversified supply chains, adequate access to critical inputs (human resources, infrastructure, financing, technology, business climate), access to information and markets, and collaboration between the private sector and the public sector enhance competitiveness. In economic terms, clustering may generate economies of scale (including economies of scale of public goods, e.g., information), reduce information costs and improve communications and information across firms. These generate an agglomeration of economic activities, often sectoral. Examples are Silicon Valley in the USA and the "Silicon Triangle" (with Chennai, Bangalore, and Hyderabadas as its vertices) in South India, where the concentration of skill have facilitated the development of competitive IT industries. The GOP has identified a number of possible clusters in Pernambuco. These include medical care and informatics in Recife, poultry in the Agreste region, mining (gypsum, marble, and granite), tourism in several places in the state, and fruits in the Sao Francisco Valley. The World Bank is assisting the Northeast's efforts to diagnose and implement development clusters (the Pro-Northeast Initiative launched by Pernambuco, Bahia and Ceara). 'The Northeast informatics cluster (where Pernambuco may be favored by its high human capital endowment relative to the other NE states), with 257 IT companies in Recife alone, illustrates the approach. The cluster's relative competitive position is characterized by the large domestic market (the largest in Latin America), its low-end market niche (integration services, customized programming, or internet related services) compared to the market niche of firms in the south of Brazil, and the relatively small size of NE informatics firms. To advance its development, the cluster is seeking to reduce fragmentation, develop a "best practice" business model, increase access to equity capital, increase the supply of skilled labor, increase access to basic and advanced applied research, increase knowledge about the cluster, develop a public sector strategy, and improve the cluster's national and international visibility. Initiatives being developed to achieve these aims include internet-based management training (through a possible partership between Softex, the State Government's software business and export promotion agency, and a British Training Institute that specializes in web-based training and SMEs) and information sharing among firms (through a digital platform established by PortoDigital2), closer linkages between sofware and harware firms, implementation (in Recife) of an Information Technology and Business (ITBC)3, public events to attract equity capital, international graduate and business informatics programs, sofware application districts that develop specialized solutions to the informatics needs of NE's major industries (e.g., in the fruit growing Sao Francisco region. NE, states' export promotion efforts (through integration of the states' export promotion agencies, Softex), quality assurance through development of a Software Quality Institute, and software publishing. The Bank is presently following up on these actions with Norheast state governments (and with Pernambuco in particular) and supporting efforts to identify further bottlenecks to NE competitiveness. The Government is investing R$33 million to develop the informatics cluster, including R$10 million for the venture and human capital funds; R$10 million for specialized informatics knowledge institutions (CtN (Centro de Informadtica) at UFPE (Universidade Federal de Pernambuco) and CESAR (Centro de Estudos e Sistemas Avanqados do Recife), a private non-profit institution; $R6 million in Softex/ITBC; R$ 2 million in INCUBATEP , a government agency that supports the development of informatics businesses, R$2 million in the Technology Secretariat (SECTMA), and $R3 millon in refurbishing. I. See The World Bank (2000C). 2. Porto Digital is Recife's Informatics Center, comprising financing services for venture capital and human captal, and providing a venue for interaction and cooperation in informatics among governments, companies and universities. 3. To be financed by the PortoDigital, the ITBC Center will increase assistance to businesses, centralize knowledge and experience, and serve as a showcase for the cluster. The ITBC will be in place next year. 80 161. The Bank's education tables report that Brazil's public sector spent 5.1% of GDP in education in 1997, equal to the average for upper middle income countries. In 1995, the only year for which a comprehensive intergovernmental consolidation of expenditures is available, Brazil's public sector spent R$28 billion (4.3% of GDP) on education77 out of total public sector tax revenues of R$192 billion, or 15.6%. Of those R$30 billion, R$0.7 (2.3% of the total and 4.0% of the state's GDP) were spend on Pernambuco's public education, with the federal, state, and municipal governments contributing respectively, 28%, 43%, and 29% of the State's total. 162. As discussed, the distribution of education resources is constrained by a floor and an earmarking provision that favor primary education. In addition, the distribution between states and municipalities of part of the minimum 60% that the federal government must spend on education is ruled by FUNDEF (Fundo de Manutenqdo e Desenvolvimento do Ensino Fundamental e de Valorizagdo do Magisterio, established by the 14th Constitutional Amendment to the Constitution ). This fumd (of some R$15 billion per year today), created in 1996, ensures that financing of primary education per student attain a minimum value of $R333 per student in the first four years of the elementary education cycle and R$350 per student in years fourth to eighth (values for the year 2000), with a minimum of 60% of the Fund allocations used to pay teacher compensation.78 163. Federal Proerams. A number of federal programs are targeted to improving the quality of primary education These include: School Development: FUNDESCOLA (Fundo de Fortalecimento da Escola), a $R1.3 billion fund (with World Bank Finance) aims to improve access, attendance, quality and administration of primary education in several municipalities in the North, the Northeast, and the Center-West. The program aims to provide schools and their parent associations with more autonomy over financial resources; to make sure that schools meet minimum operation standards with regard to teacher qualification, pedagogical inputs, and equipment and physical conditions; to design and implement schools development plans; to implement school improvement projects, which are comprised of actions and measurable learning improvement targets; and to develop, test and disseminate various pedagogical and organizational frameworks that could improve learning. The fund finances inputs, development plans, and technical assistance to primary schools. See IPEA (1999). The amount reported in the IPEA study is far higher than that reported in IBGE's national accounts ($R20 and R$24 billion respectively in 1997 and 1998, or 2.3% and 2.9% of GDP, more in the ballpark of international experience). See IBGE (1998), p.72 and IBGE (2000) p. 68. 78 The Fund has the following resources: 15% of the FPM/FPE constitutional transfers (the 44% of the IR and IPI that the Federal Government must transfer to states (21.5%) and municipalities (22.5%); 15% of the IPI-EXP (the 10% of the IPI that the Federal Government must transfer to states on a state export pro-rata basis); 15% of the ICMS Lei Kandir (LC No. 87/96) federal compensation to states, as mandated by Law 9424/1986; 15% of the ICMS; and federal resources as required to meet the minimum expenditures per student. 81 * Direct Financing of Schools: The Direct Cash at Schools Program (Programa Dinheiro Direto Na Escola) provides financing for meeting basic school needs. Schools with more than 100 students that have an established executing unit (a PTA at public schools or an NGO) and a bank account are eligible for this program, which transfers cash directly to the school for use by the executing unit on the basis of the number of students. The program is funded by FNDE. * Teacher Training: The Government's Teacher Training Program (PROFORMA(AO - Programa de Formaqdo de Professores em Exercicio) is targeted to the North, Northeast, and Center-West. It is financed by FUNDESCOLA and implemented through the Government's Distance Education Program (both computer-based (PROINFO and TV-based [TV-Escola]). * Curricular Benchmarks (Parametros Curriculares Nacionais): comprises development and training with respect to six flexible modules of curricula covering pre-school, reading and writing, first four years of formal schooling, years 5" to 8th of formal schooling, young and adult education, and education for indigenous populations. * Textbooks: includes a continuous assessment of existing textbooks (Avalia9do do Livro Didatico) for the first four-year series of the primary education cycle as well as a program (Programa Nacional do Livro Didtitico [PNLD]) of free distribution of primary education textbooks of Portuguese, mathematics, science, social studies, history, and geography. * Libraries: The National School Library Program (Programa Nacional Biblioteca da Escola - PNBE), introduced in 1997, distributes reference and literature books to schools, with complementary teaching and library administration materials. A total of eight million books gave been distributed to some 56.000 schools. * Evaluation of Results: SAEB (Sistema Nacional de Avaliaqao da Educaqdo Bcasica) is evaluating educational outcomes at the end of the schooling cycles (fourth, eighth and eleventh year). Surveys for 1995 and 1997 have been completed. * Eradication of Child Labor: The primarily rural PETI (Programa de Erradicaydo de Trabalho Infantil), managed by the Social Security and Social Assistance Ministry, provides transfers to local schools per every child in the program (mostly to establish and extended school day, or Jornada Ampliada) and income to selected poor families (Bolsa Crianqa Cidadd). The program is targeted to families with an income below 1/2 the minimum wage with school age children (7-14). The program began in 1996 in Mato Grosso do Sul and covered eight states in 1999, with 125.000 student beneficiaries (of which 73.000 were in Pernambuco). * Student Health: The Student Health National Program focuses on ensuring good sight and hearing. It includes a sight rehabilitation campaign ("Olho no Olho" and "Quem Ouve Bem, Aprende Melhor!" "Quem Ouve Bem, aprende Melhor'); distribution of training and evaluation materials; medical treatment to those students with sight problems; and distribution of glasses and hearing aids. The prograrns are funded with FNDE and Ministry of Health resources. 82 * Student Nutrition: The well-known program -merenda escolar" (or Programa Nacional de Alimentaqao Escolar - PNAE, introduced in 1955) is aimed at providing 15% of students' daily caloric and protein needs. It is funded by FNDE. * Student Transportation: FNDE finances municipal acquisition of vehicles for transporting students that live far away from schools. * Education Statistics: A national school census is conducted every year. It is comprised of three surveys that cover respectively school education, university education, and education financing. In addition , a census of teachers was conducted in 1997. 164. In addition, SUDENE (which benefits from FNE transfers), the northeast regional development federal agency, in partnership with various entities (Alfabetizaqdo Soliddria, Servi o Nacional de Aprendizagem Rural - Senar, Comunidade Soliddria, Xingo Educaqdo, and the Fundaqdo Apol6nio Sales de Desenvolvimento Educacional from the Universidade Federal Rural de Pernambuco - Fadurpe/Chesj), supports education and training programs (including the Casas Familiares Rurais program, which alternates training of farmers with work at their farms) for young and adults in the Northeast. Through these programs, 386.677 beneficiaries became literate and 210.880 people were trained in 1999, with a budget of R$ 31,6. 165. Secondary education also benefits from federal programs to improve access and quality, including the School for Youth Project (Projeto Escola Jovem), which seeks improvements in curricula, teacher training, libraries and laboratories, distance education, expansion of capacity, and administration of school systems. The project is financed by IDB with cofinancing from the Federal and State Governments. Distance education for teachers is also available through TV-Escola. A specialized test for secondary education (ENEM - Exame Nacional do Ensino Medio) was introduced in 1998. This is in addition to the SAEB test, indicated before, which covers all basic education including secondary education. 166. State Policies. Pernambuco's Secretary of Education has been pursuing three major policy objectives: (a) increased coverage, to meet an additional demand of about 53 thousand new students every year during this decade, with changing emphasis from pre-school and basic education (prominent in the 1980's) to secondary education; (b) improve the efficiency of the educational system by increasing the approval rate and decreasing the failure and drop out rates, thus increasing the flow of students from the first year of basic education to the third grade of secondary education; and (c) improve the quality of education (See Secretaria de Educa,co de Pernambuco (1998)). To achieve these goals the main actions have been to increase the number of schools at all levels, to increase the number of teachers, to train these teachers and to create programs such as "acceleration" courses (aceleragao) that received an increased flow of resources from FUNDEF. 167. For 1999-2003, the State Government has defined four main priorities: (i) give autonomy to state schools as a means of improving efficiency, under the hypothesis is autonomy will help schools to better adjust their means to their goals and save resources: (ii) improve quality and establish evaluation as a basic tool to improve efficiency and efficacy, working with municipalities to improve quality in basic education: (iii) improve teachers' qualifications, particularly for municipal teachers, while focusing qualification on "competency and ability", rather than formal education beyond a bachelor's degree: (iv) a wage policy that links teachers' wages to competence and efficiency. 168. The Government's policies to assure universal coverage, eradicate illiteracy and improve the quality of education include: * transfer all enrollments in the first four grades to municipalities, within a Children's School format, justified on pedagogical grounds, whereby one single teacher is responsible for all subjects; state efforts would focus on youngsters' schools, including the last four grades of primary school, with teachers specialized by subject, and on secondary education; * implement autonomous management in state schools as a way of increasing efficiency; * develop a "competence and ability" training program leading and replace teachers without a secondary degree (pro fessores leigos); * improve quality through evaluation in a pilot group of reference schools; * define a new teachers career that allows monetary incentives as a reward for improved efficiency. * eradicate child labor 169. Transfer of Primary Education to Municipalities and of Secondary Education to States. Transfer of responsibility for primary education (Grades 1 to 4) to the municipalities has been occurring at a rapid rate. However, the state is still responsible for 22% of enrollments in the first four grades and municipalities are responsible for 28% of enrollments in grades 4th to 8th and 9% of enrollments in the secondary. In addition participation in lower secondary education (grades 4th to 8th ) has actually decreased in state schools and increased in municipal schools. In numbers, municipal assumption of primary education and state assumption of secondary education represents a transfer of 250,000 primary education students from state to municipal schools and of 200,000 lower secondary students and 31,000 upper secondary students from municipal to state schools. 170. A detailed analyses of locations, classes and buildings is required to define how many new professors will be necessary. The state is prepared to maintain the payment of all teachers transferred to municipal schools. This makes the proposal attractive to municipalities but very costly to the state. As a maximum, it will be necessary to hire 6,000 new teachers for basic school and 1,220 for secondary school in addition to the 3,000 to 4,000 that will be required because of enrollment expansion in secondary school. This represents an additional cost of R$78 million or 12% of the total budget of the State Secretary of Education.79 The viability of this proposal can be questioned. 171. On the other hand, transferring all enrollments in basic education to the Municipios can be very deleterious to quality. Teachers' qualifications, which is already low, would be worsened because almost 9,000 new teachers will be required in municipal 79 Total cost will reach up to RS 95 million. 84 schools and it is doubtful that all or most municipalities can attract and hire the best qualified candidates. In sum caution will be necessary in effecting the transfer of primary education to municipalities. This transfer should be coupled with programs for municipal development and, specifically, a strong training program for municipal teachers. 172. Autonomous Management. Two paradigms of autonomous management have been discussed in Brazil recently. First, the model of university autonomy which has been in discussion in Brazil for the last 12 years. And second, the development based on local communities, which has been proposed by international agencies. University autonomy is a success in the State of Sao Paulo but remains a dream in Federal Universities. The success in Sao Paulo is due to complete autonomy to hire and dismiss, to define wages and rewards. The doubt in relation to Federal universities refers to a strong corporativism if favor of stability and equal wages. Without flexibility for dismissal and unequal payments for unequal performance there is little room to exercise autonomy. 173. Autonomy based on local communities or local development requires the development of intangible capital in the way of local associations, as Putnam (1993) correctly pointed out. In the case of a school it can improve if local communities "adopt" the school, maintaining its physical condition and stimulating students. Autonomy can give managers an extra reward to look for support from local communities. But local development in Pernambuco is rather weak. The inclination to form associations is scarce, particularly in the countryside. But autonomy can be a step in the right direction. Quality changes probably will not show up in a few years as it takes longer to develop the necessary institutional expertise. Thus, no major short run changes in quality improvement can be expected from autonomous management. Recent assessments indicate that improvements eventually occur (see Jimenez and Sawada [1999] and Tan et al. [1999]). A recent study finds that although autonomy (as encapsulated by financial autonomy, a more competitive and democratic system of electing school principals, and deliberative associations comprising teachers, parents, students anrd other community members) has had some effect in improving educational performance, the impact has been rather modest.80 174. Trainin2 Programs. Training is almost compulsory by law because secondary studies or a higher degree will be required of all teachers (secondary only for teachers in the first four grades). All studies show that there is a strong correlation between proficiency and teachers' schooling achievement up to a university diploma. An emphasis on "competence and ability" is appropriate (s6me studies do not probe the role of teacher activity and of the required competence in classroom). 175. In Pernambuco, as indicated before, an emphasis should be given to training of municipal teachers, particularly as municipalization of primary education proceeds. The actual teacher profile shows that municipalities are more adapted to the Children's School (i.e. first four years of formal schooling) while the profile of state teachers is better fitted to Youngsters' Schools, thus achieving some consistency with the law. But a major effort so See Paes de Barros and Mnedoga (1998). 85 will be required to improve the competence and ability of the over 3,000 teachers presently without a university degree. This proposal should receive high priority. New instruments such as distant learning, indicated above, should be more extensively applied. 176. Evaluation and Oualitv. Evaluation should be a mandatory procedure. A correct assessment of difficulties and achievements is essential. But evaluation is only justified if the necessary feed back mechanisms are developed. Thus, evaluation should receive high priority as a way of improvement of educational efficiency and effectiveness. A selected group of the 1,000 state schools and 8,400 municipal schools should be chosen to serve as a pilot experiment in evaluation. Creating a culture of quality evaluation is essential, particularly as it bring local communities into the evaluation process. 177. Efficiency and Incentives. A policy of wage incentives can be very important to stimulate efficiency. Nevertheless, it is a very controversial policy, with strong opposition from unions. A permanent change in teachers' career, payments and rewards needs to be discussed but will take time to get approved by the State Congress. In the meantime it may be possible to design creative incentives such as study trips and computer facilities to performing teachers and schools. 178. Eradication of Child Labor. A major concern in Pernambuco is the fact that many children do not attend schools because they are put to work. As in other states (e.g. DF, BH, MG), Pernambuco has a program of income support to help poor families send their children to school (PETI - Bolsa Escola). The program comprises a guarantee of access and permanence in primary school, supplementary family income, and extended school day. The program is presently targeted to 65 municipalities, mostly in the sugarcane sector. Children must attend school for at least 80% of the school days. The cost of the program in 1999 was R$30 million. The Bolsa Escola programs have generally received positive reviews, although more formal evaluations would be appropriate.8 ' Pernambuco's combination of PETI, which explicitly supports schools, with Bolsa Escola, which supports families is likely to improve the overall effectiveness of the program in raising the education of the poor and in reducing child labor. 179. Teacher efficiency can only be assessed at school level, based on student performance. Thus all teachers and other personnel in a efficient school should be rewarded. A state competition including all public schools can considered. Increasing Coverage 180. Increasing the coverage of education faces a demographic challenge. The state will need only a small increase in the number of basic schools (if any) because population in the age group is decreasing (Table 28) and coverage is relatively high. However, for secondary schooling, increasing today's low coverage will require increased resources. It can be anticipated that low population growth will make it easy to achieve universal coverage in basic education and that the State can stimulate Municipalities to assume higher responsibilities for basic education while the State concentrates its efforts 81 See Sedlacek, lIahi, and Gustafsson-Wright (2000). 86 in the secondary level. In addition, urbanization will lead to a decrease in rural schools if at the same time the system of school transportation is improved. Table 28 - Population Projections (000) by Age Group - Pernambuco: 1998-2010 Year Age Group 1998 1999 2000 2005 2010 7 to 14 years 1335 1311 1288 1191 1184 l5 to l7 years 517 512 508 486 433 l 8 to 24 years 1142 1156 1164 1145 1074 Source: Population Projections by IBGE (Preliminary). 181. Simple projections can highlight the challenges for the future. As the absolute number of students attending basic schools will decrease,82 there will be no pressure to increase the number of schools (Table 29). On the contrary, the total number of schools can decrease, particularly the rural schools. This results from demographic changes in the Brazilian age structure and from urbanization and also from improvement in system efficiency. But there will be an increasing pressure in secondary level schools. Enrollments in secondary school are expected to increase 20%, from 332.000 to almost 400.000. As the state's participation in secondary education is increasing, it is projected that in 2005 the net increase in state school students will be around 81 thousand to 109 thousand (Chapter 6 of Vol. II provides a simulation exercise; it is assumed that 80% of all secondary enrollments will occur in State schools). This represents an additional of about 3,000 to 4,000 teachers. The additional 81,000 students will cost R$20 million a year, a 3% increase in the total budget. Table 29 - Projected Population and Enrollments (000) Year 1998 1999 2000 2001 2002 2003 2004 2005 Population 7 to 14 years 1335 1311 1288 1265 1245 1226 1209 1194 Enrollment in Primary 1818 1824 1824 1828 1808 1790 1773 1758 Net schooling(%) 0 0 0 0 0 0 0 0 Population 15 to 17 years 517 512 508 503 499 494 490 486 Enrollment in Secondary 303 333 342 353 364 375 386 396 Net schooling (%) 19.4 25.4 27.5 30.0 32.5 35.0 37.5 40.0 Source: MEC. INEP - Sinopse Estatistica da Educacao Basica, 1998. Enrollment - projections based on an increase of net coverage and maintenance of absolute numbers of inadequate - out of the age group. Improving Quality 182. But the State's major challenge is improvement in quality. If universal coverage of basic education is to be attained early in the next decade and secondary education is progressively offered, improvement in student performance presents a real and difficult challenge. It will require that improvements of teacher qualification that have occurred in the state and private systems also extend to the municipal system. The improvement of 82 The number of students in grades I to 4 are already decreasing; the number of students in grades 5 to 8 is expected to increase through 2003 and start to decline in 2004. See Vol. II, Chapter 6, Table 48. 87 teacher qualification remains a major challenge. About 4,300 teachers (4% of the total), mainly at the municipal level, need formal higher degrees or need to be replaced, including 1,500 teachers with only an incomplete first grade. In addition, a much greater number needs improvement in its "competency and ability" requiring specific training programs. And the development of an evaluation program is a must if monitoring and incentives are to be used efficiently, to improve the quality of the system, decreasing drop out and increasing approval rates. 183. With the provisions discussed above, a few goals can be set in relation to quality improvement for the next 5 years. The approval rate should increase at least 15% (it already increased 34% from 1991 to 1998). Most of the increase shall occur in municipal schools and would be obtained if drop-out rates decrease to the Brazil level. This will require a close monitoring of the system, and the adoption of incentives to stay at school. Permanent evaluation of student performance by system of administration is essential - with an emphasis on municipal schools - that can detect where the system is not responding to incentives. Proficiency is expected to increase in the SAEB and ENEM exams. 184. It is interesting to remember that school facilities have been evaluated as adequate in general. But a close follow up can detect where conditions are still deficient and provide them where absent. Necessary physical investments should not be of great magnitude. 185. Recommendations. The GOP 's education policies are aimed at improving coverage, efficiency and quality. The actions to achieve these aims -- municipalization, autonomous school management, teacher training, tasting, and teacher performance incentives -- are all steps in the right direction, although care should be taken that municipalization of primary education be accompanied by improved teacher quality at the municipal level. Furthermore, it is recommended that, (i) in improving quality through evaluation, Pernambuco start linking the evaluation process to administrative reform provisions regarding the performance of public sector employees; and (ii) that Brazil start thinking of options to allow private schools to deliver public education. 186. With regard to (i) above, it should be noted that the 19th amendment to the Constitution allows the public sector to dismiss poorly performing employees. Regulations of this provision is being presently regulated by Law. It is unlikely that under this provision, many public employees will be fired. What may happen instead is that the threat of losing employment coupled with a link of wages to performance will help improve performance. Nevertheless, evaluation of teachers -- a controversial procedure even in the US -- will require adequate measurements of performance, including measurement relative to peer groups. But the challenges of adequate measurement will need to be overcome in order to improve the strength of public schools and make them competitive with private schools in providing quality education. This is only fair if it is noted that it is public schools that provide education to the poor. 88 187. With regard to (ii) above, its should be noted that the Constitution allows private provision of public education, but only on a rather exceptional basis. Article 213 indicates: * that public moneys may be used for community, religious, or philanthropic schools, provided that these schools (i) be non-profit (with financial surpluses spent on education); (ii) commit their net worth to another eligible school in case of closure; * public moneys can also be used for scholarships based on need, when there are no available places in public schools and with the obligation that the public sector invest in expanding the public school system where there are insufficient places. developing a system of application training programs (now mandatory by Law)To increase growth and reduce poverty, the focus of policy should mostly be on encouraging investment in physical and human capital. These restrictions do not benefit the cause of education or the cause of the poor. Instead, the state should allow, as in health, the free use of private schools to provides schooling. This would strengthen public schools by subjecting them to stronger competition and would also promote investment in private schools. Public moneys could then be used more freely to provide scholarships or to finance charter schools. The Constitutional constraints on the use of private schools for public education are helping undermine the access of the poor to better education. HEALTH POLICIES 188. The policy framework (Figure 1, p.50) also indicates that improvements in health delivery must be an important component of a poverty reduction strategy. Broadly speaking, improved health is required to replace the vicious circle of poverty, high fertility, poor health, and low economic growth by, replacing by a virtuous circle of greater productivity, low fertility, better health, and rising incomes. Good health has a direct bearing on access to productive employment and on the effectiveness of education. Better nutrition, for example, helps children lead healthier lives and improves their capacity to learn-and their subsequent performance in the labor market. This is so, in particular, because dietary deficiencies of vitamins and minerals cause learning disabilities, mental retardation, poor health, low work capacity, blindness, and premature death. 189. Policy Framework. The Brazilian Constitution of 1988 established a universal right to all-inclusive and free-of-charge health services to be provided by the public sector. To translate this constitutional principle into reality, Law 8080 of 1990 created the Unified Health System (Sistema Unico de Saude - SUS). In practice, the health system consists of both a substantial private sector and the government's Unified Health System. Although SUS services are both publicly and privately provided, SUS financing is entirely public (a mix of social security contributions and taxes). In 1995, the only year for which a consolidated review of health expenditures has been conducted, Brazil's public health expenditures (R$21.7 billion) accounted for 3.4% of GDP, about 1/2 of total public and private expenditure on health. Public sector (federal, state and municipal) 89 expenditures on health in Pernambuco (R$613 million) accounted for 3.5% of the state's GDP. 83 190. Since the early 90's, the financing and provision of SUS services is being decentralized from the federal to the state and municipal levels. Federal contributions may be transferred from the National Health Fund (Fundo Nacional de Sauide - FNS) directly to either a State Health Fund (Fundo Estadual de Saude - FES) or a Municipal Health Fund (Fundo Municipal de Saude - FMS). Similarly, a state can make transfers from its FES to an FMS. There are two main federal transfer mechanisms for financing ambulatory and hospital care, which are provided by either the state or the municipal levels. The first one is an automatic transfer from fund to fund (from FNS to either FES or FMS), and the second one, is the direct payment of public and private providers for services rendered. One example of first kind of transfer is the per-capita transfer for primary care (Piso Assistencial Basico - PAB). The PAB financed a list of cost-effective basic interventions at a level that equaled $10 per-capita in 1998. Other automatic transfers are financial incentives given in addition to the per-capita transfer for implementing special programs. Some of them are the Community Health Agent Program (PACS)84 , Family Health Program (PSF)85, Basic Drug Program (PFB) and Nutrition Program. These programs provide preventive and primary care services to the poorest segments of the population, especially to women and children. 191. The management and provision of health services are being decentralized to the municipal level. Decentralization entails the transfer of financial resources, the public sector's infrastructure and human resources, and resources for contracting with the private sector. Two stages of decentralization are authorized, depending on the municipality's ability to manage the system: Full Management of Basic Care86 (Gestdo Plena da Atenqdo Basica) and Full Management of the Municipal Health System87 (Gestao Plena do Sistema Municipal). In order to be decentralized, municipalities have to comply with specific requirements (Norma Operacional Basica NOB 01/96). Health services in municipalities that have not been decentralized are managed by the state level. he NOB 01/96 also regulates the level of decentralization to the State Health Secretariats. There are two degrees of decentralization at the state level: Advanced Management of the 83 See IPEA (1999), Table 13. 84 The PACS is a strategy for expanding primary care based on a team of community health workers, each of whom typically resides in rural areas and serves an average of 150 families. 85 The PSF is a strategy for expanding primary care based on a team of health professionals (a doctor, a nurse, health assistants, and community health workers) who provide services to an average of 766 families within a determined geographical area. 86 This decentralization level allows for the management of primary care facilities and for the receipt of transfers fund to fund. 87 This decentralization level allows for the same as in footnote 5, but also for the management of secondary and tertiary facilities. 90 State System (Gestdo Avan9ada do Sistema Estadual) and Full Management of the State Health System (Gestao Plena do Sistema Estadual)88. 192. Recent changes in the health system have also increased social participation in health affairs. The state health system is governed by the Pernambuco State Health Secretariat, which follows the policy and normative orientations generated at the federal level by the Ministry of Health. Social participation at the state level is guaranteed by the State Health Counsel (Conselho Estadual de Saude - CES), which has representatives from the users, health workers, providers, and government. The CES has deliberative powers. Coordination and negotiation among key sector players is conducted in the Two Party Management Commission (Comissdo Intergestores Bipartite - CIB), composed of representatives of the State Health Secretariat and the State Municipal Health Secretariats (Conselho Estadual de Secretarios Municipais de Saude - COSEMS). 193. At the federal level, the government has established the following priorities for the health sector89: (i) Stabilize financing of the health sector90; (ii) Consolidate the decentralization of health services; (iii) Reorient the health system's responsiveness towards the demand of services rather than the supply of services91; iv) Give priority to prevention and primary care92; (v) Evaluate the impact of current health programs; and vi) Adopt new practices in health policy aimed at improving the quality, transparency and social control of health services. 194. At the state's level, GOP's health policy strategy envisages (i) increased and improved decentralization to municipalities, including consolidation of the decentralized emergency attention service, through municipal consortia; (ii) definition of an integration model of health actions; (iii) definition of a health service network profile; (iv) improved State-Municipal SUS management relations; (v) strengthen prevention through basic health services; (vi) strengthen and extend state production and distribution of medicines; 8S The two levels are differentiated by the percentage of municipalities decentralized, the percentage of federal funds transferred automatically fund to fund, and the authority to manage and control the state's service provider network. 89 Cardoso, Fernando H. (1998) 90 The federal government is seeking approval of a Constitutional Amendment (Proposta de Emenda Constitucional No. 82) to increase health financing in the next several years. The amendment proposes to increase SUS revenues starting in CYOO with an additional 5% over the level of the budget committed in CY99. For the years 2001 and 2002, it increases the budget in the same amount that the nominal GDP. The amendment also increases contributions from state and municipal revenues in the following manner: states, from 7% in 2000 to 12% in 2004; and municipalities, from 7% in 2000 to 15% in 2004. The PEC 82 has been voted and approved twice at the house of representatives and it is now being debated in the Senate. 9' Financial resources for health have been unequally distributed across stets and municipalities because intergovernmental transfers follow the supply of services rather than the demands or needs of the population. Since 1996, the government is addressing this problem with the introduction of the per- capita transfer for primary care (PAB). 92 Most public spending goes to financing hospital care and complex procedures rather than cost-effective basic interventions. In 1997, average public spending per hospital admission was SR 250 compared to R$ 3.5 per ambulatory care. 91 (vii) strengthen and improve hemotherapy systems; (viii) strengthen community development systems. 93 195. Health Policy Recommendations. Pernambuco's health policy objectives most probably include further reductions in infant, maternal, and child mortality; improved nutrition; reduction of infectious, communicable and vector diseases; and reduction of chronic diseases and injuries. Policies to address these and other challenges will need to focus on education, access to water and sanitation, adequate decentralization, intergovernmental definition of roles, improved institutional, managerial and administrative capacity, improved relations between players, and improved contracts with service providers. Education and access to water and sanitation issues are addressed elsewhere in this report. The following paragraphs focus on the other challenges listed here. 196. As in the case of education, decentralization in health must proceed with caution. The pace and scale of decentralization has raised several concerns among some policy makers. Some doubt the technical and administrative capacity of municipalities. A recent report by the MOH's Auditing Department denounces municipal deficiencies in the areas of planning and programming; health counsel's operations; accounting and financial control; general management and management information systems; availability of municipal counterpart funding; and misuse of public funds94. Concerns have also been raised about high administrative costs, losses of economies of scale, and fragmentation that may arise from the small size of many municipalities. 197. Excessive decentralization in health facilities may prevent that municipalities benefit from economies of scale. In Pernambuco, considerable investment in private sector health facilities, particularly in Recife, has taken place. These facilities are being used to service not only Pernambuco's population but also to export health services to residents from other states. As with other services, it makes sense to locate certain facilities in centralized areas and not expect that every municipality to have access in the municipality to all medical services. Hence, municipalities should be not be encouraged to establish self-sufficient facilities to service their populations but rather to make the best possible use of existing facilities in the state. 198. The assignment of roles to different government levels refers mainly to the assignment of management to different levels. As discussed before, the transfer to municipalities of financial resources, of the public sector's infrastructure and human resources, and of resources for contracting with the private sector depends on the municipalities managerial capacity. Assignment of management to very small municipalities may also prevent economies of scale in administration. Hence, the Government's policy of establishing consortia of municipalities to administer certain health services will help make decentralization more effective. 9£ See Governo de Pernambuco (1999), p.32. 94 Ministry of Health (1998). 92 199. Much can be done with regard to institutions, management, administration and finance. Institutional/regulatory areas that need attention include strategic planning and programming; auditing, licensing and accreditation of health facilities; and development of norms and protocols for service provision supported by evidence-based medicine. Technical assistance, training of health personnel, as well as improving the compensation system should also be emphasized. In the financing area, four areas merits attention: (i) improving the costing and cost-effectiveness of interventions (with an emphasis on prevention and a focus on risk groups); (ii) exploring new alternatives for financing and paying health services; (iii) developing adequate fonnulae for allocating financial resources within the state in view of improving access with equity; and (iv) developing the instruments and systems required to charge privately insured patients for services rendered by the SUS. Privatization or corporatization of publicly owned facilities, through legal arrangements such as organizaq6es sociais95 could also be explored. 200. Improvement of relations between players, including contracts with service providers is a central challenge of the health system. Brazil needs to move from a system based on payments for specific services to an insurance-based system. Under an insurance-based system, the government does not need to determine prices for specific interventions. These prices would be determined by the market, through the interactions between health insurers and service providers. When the government sets prices, providers are encouraged to make incorrect claims and the prices do not reflect a competitive mechanism that brings them to the lowest possible level. In a health insurance approach, health insurers would bid to provide services to municipalities. Implementation of this approach, however, needs to address a number of challenges, including the use of existing public sector providers (which could become organizaqoes sociais and work under contract) as well as the lack of profitable basic facilities in many municipalities. MUNICIPAL DEVELOPMENT 201. With decentralization of such key areas as education and health, as discussed above, development of implementation capacity at the municipal level becomes essential. In general, Brazilian municipalities have responsibility for the provision of usual "household" services - solid waste, police, water and sanitation, some public transit. They also share responsibility for education and health with the states and, in recent years, have progressively gained more responsibility for investing in the infrastructure required for economic and social growth of local communities. 202. Recent experience with municipal capacity-building programs is encouraging. For example, within Bank-supported municipal development programs, institutional development goes in parallel with the financing of municipal infrastructure. These programs focus on fiscal management, tax collection, project preparation and selection, 95 These are private, not-for-profit organizations, operating under management contracts, which manage public infrastructure using government funds. 93 municipal associations, training of municipal officials, developing local development strategies with the input of participatory methods. Box 6: Brazil: Health Service Institutions and Delivery In the World Health Organization's year 2000 review of the overall performance of health systems in its 189 member countries, Brazil ranked 125, above Honduras (131), Peru (129), Guyana (128), and Bolivia (126), but below all other Latin American Countries.' The measure relates, through a frontier production model, overall health system achievement (a composite measure of the level and distribution of health and of system responsiveness, and of the fairness of financial contributions) to health system expenditure and educational attainment. Brazil's low overall ranking reflects, particularly, the lowest ranking on the fairness of financial contributions (189) and a low ranking (130) on the system's level of responsiveness (defined by dignity, autonomy, confidentiality, promptness of attention, quality of amenities, support services, and choice of provider), despite a relatively higher ranking of per capita expenditures (54). The validity of the ranking has been strongly contested by the federal government, citing the sparcity of the data that it is based on. A recent analysis of four health system's in Latin America (Argentina, Brazil, Chile, Colombia) underscores the weak attention that has been paid in Brazil to the demand side of health, including exposure of the public sector to competition and consumer protection through regulation of private insurance.2 It points out specifically that, in Brazil's system, in constrast to the other system's surveyed, individuals do not take their financing with them when they opt out of the public system. Brazil's public system of payment to providers (on the basis of utilization), does not represent the contracting out of management or insurance, resulting in a weak focus of providers on outcomes as opposed to inputs. It also notes that private health insurance is marred by exclusions and restrictions, unsound finances, and possibly fradulenit practices that need to be addressed through better regulation and monitoring of the system. A change from paying providers on a utilization basis to contracting out insurance, as suggested in the text, would strengthen incentives to quality and cost effectiveness. These contracts probably make it easier to address the difficult assymetric information problems (moral hazard and adverse selection) of health insurance. They can also be set in a more competitive environment where many providers compete for the insurance contracts and individuals are enabled to choose among providers, bothl public and private. The use of the social organizations framework could be used to promote efficiency in public providers of health (which may be needed in remote regions). 1. See World Health Organization (2000), Statistical Appendix. 2. See Jack (2000). 203. Three key issues need to be addressed if one wants to push forward an effective program of technical capacity at municipal level. The first issue is the challenge of the very small municipalities. While a recently approved law forbids the creation of new municipalities below 5,000 people, many of the poorest municipalities are badly managed due to lack of resources, information and general knowledge. Institution programs for those municipalities need to be prepared at the state level using all the schemes that may be in existence. The reluctance that many small municipalities have against municipal associations as well as the enormous distances between them are additional obstacles to be faced. 94 204. The second challenge is the need to widen/open the ongoing programs to include training on the management of tl'e social sectors for wvbich muinicipalities have a large share in responsibility. Capacity building in managerr.er! f health and education services would benefit from coordination at federal level. The programs could be carried out in association with the states and municipalities and fiv-riced preferably by a system of matching grants. 205. The third issue is the need to develop a coherent strategy/approach to municipal capacity building in Brazil. This would call for better coordination and integration among ongoing capacity-building programs taking advantage of the lessons of each program, in particular by the municipal development funds financed by the Bank. The impetus to improve municipal capacity brought by the requirements of the LRF should be used to consolidate and enlarge existing fiscal and financial management programs. The main pillars of such strategy are presently being explored by the Bank and Brazilian authorities. It calls for an agreement on the scope of the program, identification of the needs or capacity gap, how to close the capacity gap, managing implementation and the process of building such strategy. LABOR POLICIES 206. Labor policies are primarily federal policies. The chief labor market issues include: (i) the heavy burden of non-wage costs, particularly the impact of taxes, including the income tax (15%), the Salario-Educa,co (2.5%), social security contributions (31%), and contributions to FGTS (S%)*96 (ii) constitutional constraints on the organization of labor unions (the znicidade sindical issue), which disallows labor choice with regard to the particular form and level of union organization; (iii) the weakness of collective bargaining; (iv) the juridical weakness of labor contracts, both individual and collective; and (iv) the excessive role of labor judges (the poder normativo issue), which prevents negotiation and mediation between the parts in conflict. Resolution of these issues is being sought through labor reform. Measures that have been adopted to improve flexibility include the removal of wage indexation and the introduction of: profit sharing by labor, fixed term contracts, seasonal and cyclical allocation of work hours, part-time labor, and mediation in collective bargaining.97 207. Reforms to improve labor market flexibilityfwill help Brazil grow faster and will improve labor incomes, through increased wages and employment. Therefore, Pemambuco should lobby and support labor reforms that make it easier for employees and employers to agree on labor cortracts. For Pernambuco, a flexible and dynamic labor market will be the key to translate investment and education efforts into poverty reduction, as wages are the chief income of the poor. 96 See Rodriguez (1998). 97 See Amadeo (1998). 95 SOCIAL PROTECTION POLICIES 208. As indicated in Chapter 1, social protection programs target child labor, youth exposure to drugs and violence, old age, AIDS infection, homelessness, unemployment, training, distribution of seeds, agricultural infrastructure, and technical assistance. It was argued that the limited resources devoted to these programs (about R$100 million) are incapable of raising the welfare of the population of poor in any significant amount. Although some of these programns have been positively evaluated in their design, an impact assessment of the different programs would be useful to determine how these programs should be extended or reduced, depending on their effectiveness. For example, the program targeting child labor through improved incentives for education appears to be an important candidate for expansion, but an assessment of its impact should complement the positive assessment of its design. With an impact evaluation of these programs, the Government could consider some tradeoffs among public expenditures. One such trade-off would be between expenditures on the PRODEPE program (amounting to R$666 million per year), which are probably not a good use of public resources, and expenditures on the most effective social programs. 209. In addition, the federal and state governments, in collaboration with the World Bank, have an important program on the ground on land reform (Cdlula da Terra). Furthermore, the State of Pernambuco Northeast Rural Poverty Alleviation Program, also with support from the World Bank, provides an effective mechanism to channel more funds for public investment in rural areas, particularly in infrastructure, implemented directly by the beneficiaries and local governments, which is leading to more effective targeting towards the needs of the rural poor. 210. A general point to be made regards the question of whether, at the margin, it is worthwhile to increase rural social spending more than urban social spending. This is a point related to a finding in the Bank's ongoing Brazil Rural Poverty study, where its is concluded that rural social spending is very progressive and has a larger poverty reducing effect than social spending overall. Under rural social spending in the Northeast one would include education and basic health, transfer programs, drought workfare, rural pensions, some rural infrastructure (rural electrification, roads), the RPAPs (Northeast Rural Poverty Alleviation Project) discussed above, and some land reform. SAFETY POLICIES 211. Chapter 1 indicated evidence that suggested that crime levels in Recife may be higher than in the other main metropolitan areas of Brazil. This is an area that demands far more attention than it has received thus far, although recent developments have brought the issue again to the forefront of national public policy attention. Attention to crime is important form the point of view of growth and poverty reduction. High crime levels deter investment. And high crime levels are often linked to poverty, with the poor most often being victims of violent crime. 212. Research shows that crime is often linked to poverty, so that low wages and unemployment are important factors at work. As discussed before, Pernambuco's poverty 96 levels are among the highest in Brazil. But information currently available probably does not suffice to make an adequate analysis of crime levels in the state. Pernambuco needs to conduct a thorough diagnosis (such as those that have been conducted in Rio de Janeiro and Sao Paulo), with a well-designed survey of victimization, police statistics, data on causes of death and illness from the health system, and prison surveys. Such a diagnosis would identify crime rates, different types of crime, spatial distribution of crime, and several possible correlates of crime. 213. With such a diagnosis, Pemambuco could develop a policy approach to reducing crime. Such a approach would surely include the other policies to increase growth and reduce poverty that have been discussed in this study as well as policies to improve the effectiveness of crime prevention and repression. In the latter area, a review of police institutions would trigger actions to improve the effectiveness of the police corps and contain risks of victimization by police (which is high in Rio and Sao Paulo). 97 Box 7: Pernambuco: Community-based rural poverty reduction The State of Pernambuco has been implementing since late 1993, with World Bank support, a community-based development approach to rural poverty alleviation, through the reformulated Northeast Rural Development Project (R-NRDP), which was followed by the current Rural Poverty Alleviation Project (RPAP), now under the umbrella of the newly created Projeto Renascer. Through this decentralized approach, communities select their own investment priorities, prepare their proposals, which are analyzed by project municipal councils (with 20% participation from local authorities and 80% from potential project beneficiaries and civil society) and approved by the project Technical Unit. The community associations then execute themselves the approved subprojects. For this purpose, funds are released directly to the community associations. Infrastructure subprojects (e.g. water supply, rural electrification) have been most demanded by the communities. Since the beginning of the R-NRDP, approximately 2,200 community subprojects have been approved and funded by the two projects, 63% of them infrastructure, directly benefiting some 92,000 poor rural families. These beneficiaries amount to roughly 47% of Pernambuco's poor rural families. Rural water supply subprojects (some 40% of total subprojects) proved crucial to rural families in confronting the persistent drought in the semi-arid regions of the state. Rural electrification represents another 20% of subprojects and has made night schooling, home lighting and refrigeration available to these families. Throughout the state, 168 municipalities (93 % of total) have participated with at least one subproject. Since the 1993 reformulation, a total investment of approximately US$80 million has been made by the State and beneficiaries through this Project. A recent assessment made by independent consultants, demonstrated the positive impacts of the two projects on many accounts, among them development of social capital at community and municipal levels, increased well-being and family income. On the former, the projects have stimulated the formation of 75 project municipal councils, through which not only the project's but also other investment priorities are debated. The independent evaluation also found that, from the 2,200 community subprojects, some US$14 million of incremental income/savings are being generated annually. These projects have led to the creation of nearly 6,800 permanent jobs and brought an additional 5,700 hectares under cultivation in the state. Stimulated by the overall success of these projects, the State of Pernambuco is seeking to develop a new project, with total additional investments of about US$80 million. On another front, the State of Pernambuco, with assistance from the Federal government and the World Bank, is implementing a land-based rural poverty alleviation project (Cedula da Terra). This pilot project has been implemented since 1997 and uses the same community-based development approach. As of May 2000, some 1,435 families of landless workers had benefited from the project. A total of about 26,000 hectares had been acquired at a total cost of about R$9.5 million. Additional investments executed by the beneficiaries on the acquired land, totaling about R$24 million, were financed on a matching grant basis. Currently, new farms are in the process of negotiation between new beneficiaries and landowners. Evaluations conducted by independent researchers have shown that the new farms created through the project are deeply transforming the social and economic conditions of the beneficiaries, even under the unfavorable conditions of the semi-arid region. The Federal government and the Bank are currently finalizing a new, larger project, which will succeed the current Cedula da Terra, incorporating lessons learned from it. Pernambuco will be a participating state under this new project. 98 REFERENCES Amadeo, E. (1998), "A Reforma Trabalhista Brasileira", IPEA, Mercado de Trabalho, Cojuntura e Analise, No. 9, November. Barro, R. and Sala-i-Martin, X. (1995), Economic Growth, McGraw Hill Advanced Series in Economics. Barro, Robert J. (1999), "Determninants of Economic Growth: Implications of the Global Evidence for Chile", Cuadernos de Economia, Pontificia Universidad Catolica de Chile, No. 107, pp. 444-78. Cardoso, Fernando H. (1998) Avanqa Brasil, Mais 4 Anos de Desenvolvimento Para Todos, proposta de Governo. Brasilia. Cavalcanti and de Goes (1996) "Convergencia entre a Renda Per-Capita dos Estados Brasileiros", Revista de Econometria, April 1996 Coeymans, Juan Eduardo (1999A), "Ciclos y Crecimentinro Sostenible a Mediano Plazo en la Economia Chilena", Cuadernos de Economia, Pontificia Universidad Catolica de Chile, No. 107, pp. 545-88. Coeymans, Juan Eduardo (1999A), "Determinantes de la Productividad en Chile: 1961- 1997", Cuadernos de Economia, Pontificia Universidad Catolica de Chile, No. 107, pp. 597-637. De Oliveira, B. and de Albuquerque, C.W. (1999), "Federalismo, Repasses Federais e Crescimento Econ6mico: Um Estudo sobre Amapa e Roraima. Dollar, D., and Kraay, A. (2000), "Growth is Good for the Poor", The World Bank, March. Duarte Lled6 and Cavalcanti G. Ferriera (1997), "Crescimento End6geno, Distribucao de Renda e Politica Fiscal: Uma Analise Cross-Section para os Estados Brasileiros", Escola De Pos-Graduacao Em Economia Da Fundacao Getulio Vargas. Ensaios Economicos (Brazil); No. 300:1-32, February. Ferreira, F.H.G., and Paes de Barros, R. (1999), "The Slippery Slope: Explaining the Increase in Extreme Poverty in Urban Brazil, 1976-1996", mimeo, April. Ferreira, F., P. Lanjouw and M. Neri (1999), "The Urban Poor in Brazil: A New Poverty Profile Using PVV, PNAD and Census Data." Mimeo. Gallup, J.L. and Sachs, J.D., (1999), "Geography and Economic Development", Harvard Institute for International Development, Consulting Assistance on Economic Reform II, Discussion Paper No. 39, March. 99 Gallup, J.L., Radelet, S., and Warner, A., (1998), "Economic Growth and the Income of the Poor", Discussion Paper No. 36, Consulting Assistance on Economic Reform, Harvard Institute for International Development, November. Gazeta Mercantil (1998), Balan,o Anual 98 Pernambuco, August Governo de Pernambuco (1999), Mudan,a e Desenvolvimento, Recife, September. Henderson, Shalizi, and Venables (2000), "Geography and Development", Policy Research Working Paper Series, No. 2456, The World Bank. IBGE (1998), Contas Nacionais, 1997. IBGE (2000), Contas Nacionais, 1997. IPEA (1999), "Gastos Sociais das Tres Esferas de Governo - 1995", unpublished manuscript. Islam, Nazrul (1999), "International Comparison of Total Factor Productivity: A Review", Review of Income and Wealth, Vol. 45 (4), December, pp.493-518) Jack, William (2000), "The Evolution of Health Insurance Institutions: Theory and Four Examples from Latin America. Jimenez, Emmanuel and Sawada, Yasuyuki (1999). "Do Community Managed Schools Work ? An Evaluation of El Salvador's EDUCO Program", the World Bank Economic Review, 13 (3). Kuerger, A.B. and Lindahl, M. (2000), "Education for Growth: Why and for Whom?", Working Paper 7591, National Bureau of Economic Research, March. Monteiro Neto, A. and Maia Gomes, G. (2000), "Quatro Decadas de Crescimento Econ6mico no Centro-Oeste Brasileiro: Recursos Puiblicos em Acao", IPEA, Taxto Para Discussao No. 712, March. Maia Gomez, G. and Vergolino, J.R. (195) "A Macroeconomia do Desenvolvimento Norderstino: 1960/1994", IPEA, Texto Para Discussao No.372, May. Ministry of Health (1998), Secretaria Executiva, Departamento de Controle, Avaliacao e Auditoria. Sistema Nacional de Auditoria - SNA Componente Federal, Atividades - lo. Semestre/98. Brasilia, 1998. Neri, Marcelo, (1999). "Uma fotografia recente da pobreza Brazileira", IPEA OECD (1999), Economic Survey of Ireland, OECD Economic Surveys, May. Paes de Barros and Mendo,a (1998), "O Impacto de Tres Inova,oes Institucionais Na Educao Brasileira", IPEA, Textp para Discussao No. 566, June. 100 Pitella Cancado, Raquel, "Migrayoes e Convergencia no Brasil: 1960-91" (1999), in Revista Brasileira de Economia, 53(2), April, pp. 211-36. Rocha, Sonia (2000), "Pobreza e Desigualdade no Brasil: 0 Esgotamento doe Efeitos Distributivos do Plano Real", IPEA Texto Para Discussao, No. 721, April. Nunberg, B. (1986), "The Politics of Sugar in Brazil since 1964", in Latin Amrican Research Review, Vol. 21(2), pp. 53-92. Pacheco, C. A. (1999), "Novos Padroes de Localizacao Industrial? Tendencias Recentes dos Indicadores da Produ,ao e do Investimento Industrial", IPEA, Texto para Discussao No. 633. Piancastelli, M. and Perobelli, F. (1996), "ICMS: Evolucao Recente e Guerra Fiscal", IPEA, Texto para Discussao No. 402, February. Putnam, Robert (1993), Making Democracy Work, Princeton University Press, New Jersey. Ramos, C.A., and Araujo, H., (1999), "Fluxos Migratorios, desemprego e Diferenciais de Renda", Texto para Discussao No. 657, July. Reis Soares, R. and Gonzaga, G., (1997) "Determinacao de Salarios no Brasil: Dualidade ou Nao Linearidade no Retrono a Educacao", Texto para Discussao No 380, Departamento de Economia, PUC-Rio Rodriguez, J.J. (1998), "Carga Tributaria Sobre os Salarios", Ministerio da Fazenda, Coordenacao-Geral de Estudos Econ6mico-Tributartios, Texto para Discussao 01. Roemer, M. and Gugerty, M.K. (1997), "Does Economic Growth Reduce Poverty", Technical Paper, Harvard Institute for International Development, March. Schwartsman, A. (1996), "Convergence Across Brazilian States", umpublished mimeo. Secretaria de Educacao de Pernambuco (1998), "Situacao Educacional de Pernambuco 1990 a 1998", Recife, 1998. Secretaria de Estado de Direitos Humanos (2000), "Pesquisa sobre Atitudes, Normas Culturais e Valores em Relacao a Violencia em 10 Capitais Brasileiras". Sedlacek, G., Ilahi, N., and Gustafsson-Wright E., (2000), "Targeted Conditional Transfer Programs in Latin America: An Early Survey", The World Bank, unpublished manuscript. Shikida, F.A., and Bacha, C.J.C. (1999), "Evolucao da Agroindustria Canavieira Brasileira de 1975 a 1995" in Revista Brasileira de Economia, Vol. 53 (1), pp. 69-89. 101 Tannen, Michael B. (1991), "Labor Markets in Northeast Brazil: Does the Dual Market Model Apply?", Economic Development and Cultural Change. Tan, Jee-Pen, Lane, Julia and Lassibille, Gerard (1999), "Outcomes in Philippine Elementary Schools: An Evaluation of Four Experiments", the World Bank Economic Review, 13 (3), 1999. Vasconcelos, J.R. and Vergolino, J. R., (1999), "Pernambuco: Economia, Financas Publicas e Investimentos non Anos de 1985 a 1996", IPEA, Texto para Discussao No. 628, February. TGI, Consultoria em Gestao (1998), Empresas & Empresarios, Recife The World Bank (1 999A), Brazil, Regulations for Better Water and Sewerage Services, Report No. 19568-BR. The World Bank (1999B), Brasil, A Reforma Portuaria Brasileira, Report No. 19569- BR. The World Bank (1999C), Brazil, Consultations with the Poor, May The World Bank (2000A), Brazil: Poverty Reduction, Growth and Fiscal Stability in the State of Ceara, Report No. 19217-BR. The World Bank (2000B), Brazil: Broadening the Base for Growth, A Report on the State of Bahia, Report No. 21377-BR. The World Bank (2000C), "The Pro Northeast Initiative: Final Report", unpublished memo, March. World Health Organization (2000), The World Health Report 2000, Health Systems: Improving Performance, downloaded from www.who.org. Zacksesky, N.R. and Octaviano, M.L. (1999), "Uma Aproximacao a Atual Politica Regional no Brasil", IPEA, Texto Para Discussao No. 694, December. 102 APPENDIX A: ALTERNATIVE POVERTY MEASURES The headcount ratio is defined as the proportion of people with income below a predefined level or "poverty line". One concern applying the Po measure is that each individual below the poverty line is weighted equally. Therefore, the principle of transfers is violated: it is possible to increase social welfare by transferring money from the very poor to lift some richer poor above the poverty line. This is because P0 takes no account of the degree of poverty and is unaltered by policies that lead to the poor becoming even poorer. One measure of poverty that takes this latter point into account (at least in weak form) is the poverty gap measure. Pl is the product of incidence and the average distance between the incomes of the poor and the poverty line. It can be interpreted as a per capita measure of the total economic shortfall relative to the population. P' distinguishes the poor from the not-so-poor and corresponds to the average distance to the poverty line of the poor. One problem with the poverty gap measure is that it will increase by transfers of money from poor to less poor (who become nonpoor), and from poor to nonpoor. Furthermore, transfers among the poor have no effect on the poverty gap measure. The p2 measure of poverty is sensitive to the distribution among the poor as more weight is given to the poorest below the poverty line. p2 corresponds to the squared average distance of income of the poor to the poverty line. Hence, moving from P0 towards p2, more weight is given to the poorest in the population. 103 Table Al: Brazil, the Northeast and Pernambuco - Headcount Ratio: 1981-1998 P0 1981 1985 1988 1990 1993 1995 1997 1998 Brasil _ 1_1111 65 Reais 24.5 25.9 27.7 30.8 26.2 21.8 22.1 22.1 132 Reais 50.0 51.2 51.7 55.3 50.4 44.8 44.6 44.9 205 Reais 65.7 66.6 66.2 68.7 67.6 62.5 60.6 61.6 NE 65 Reais 45.2 46.3 48.8 53.6 46.3 40.3 41.7 39.8 132 Reais 73.3 72.7 74.1 75.9 70.6 68.5 69.0 67.2 205 Reais 83.9 83.7 83.7 84.6 83.7 81.0 79.9 80.3 Pernambuco 65 Reais 36.9 39.9 42.5 46.7 43.3 35.1 38.1 36.7 132 Reais 68.2 69.3 69.7 71.4 68.2 66.3 66.9 64.6 205 Reais 80.9 82.0 80.9 81.7 82.2 79.1 78.9 78.0 Ceara 65 Reais 49.5 47.7 48.5 56.4 45.2 41.6 42.4 40.5 132 Reais 75.2 73.4 74.5 77.2 71.5 68.1 67.8 67.7 205 Reais 84.8 83.4 83.4 85.1 83.8 80.1 79.1 80.2 Bahia 65 Reais 35.9 38.0 41.6 46.5 43.2 37.7 40.4 36.7 132 Reais 65.4 65.0 66.8 69.6 67.2 65.6 68.1 64.3 205 Reais 78.2 78.3 78.3 79.5 80.5 79.3 79.6 78.5 Source: Author's own calculations from PNAD data set. Note: Figures are percentages 104 Table A2:Brazil, the Northeast and Pernambuco - Poverty Gap:1981-1998 P1 1981 1985 1988 1990 1993 1995 1997 1998 Brasil 65 Reais 9.6 10.4 11.7 13.3 11.9 9.6 10.0 9.7 132 Reais 24.3 25.5 26.5 29.0 26.0 22.3 22.2 22.2 205 Reais 36.5 37.7 38.3 41.0 38.2 34.0 33.5 33.8 NE 65 Reais 19.8 20.6 22.7 25.6 23.2 17.8 18.9 17.7 132 Reais 41.1 42.0 43.6 46.6 42.4 36.9 37.5 36.5 205 Reais 54.7 55.3 56.5 58.8 55.5 50.7 50.9 50.3 Pernbc 65 Reais 14.5 16.6 18.5 20.3 21.6 14.3 17.1 15.7 132 Reais 35.1 37.3 38.7 41.0 40.3 33.3 35.3 33.9 205 Reais 49.4 51.5 52.0 53.9 53.5 47.8 48.9 47.6 Ceara 65 Reais 22.4 21.4 22.4 27.2 22.5 18.3 18.8 17.9 132 Reais 44.0 42.9 43.3 48.3 42.2 37.5 37.6 36.9 205 Reais 57.0 55.9 56.3 60.2 55.6 50.9 50.6 50.6 Bahia 65 Reais 14.3 15.6 17.8 21.5 21.0 16.6 17.6 16.2 132 Reais 34.0 35.4 37.4 41.0 39.6 34.9 36.2 34.2 205 Reais 47.7 48.7 50.3 53.1 52.5 48.5 49.9 48.0 Source: Author's own calculations from PNAD data set. Note: Figures are percentages 105 Table A3:Brazil, the Northeast and Pernambuco - Poverty Gap:1981-1998 P2 1981 1985 1988 1990 1993 1995 1997 1998 Brazil 65 Reais 5.1 5.6 6.7 7.5 7.4 5.9 6.3 6.0 132 Reais 14.9 15.8 17.0 18.8 16.9 14.1 14.4 14.2 205 Reais 24.4 25.5 26.5 28.8 26.2 22.7 22.7 22.7 NE 65 Reais 11.2 11.8 13.6 15.5 14.9 10.7 11.6 10.7 132 Reais 27.3 28.2 30.0 32.9 29.9 24.6 25.4 24.4 205 Reais 40.0 40.8 42.3 45.0 41.6 36.5 37.0 36.2 Pernbc _ 65 Reais 7.7 9.1 10.6 11.4 14.1 8.2 10.6 9.5 132 Reais 21.9 24.1 25.7 27.6 28.2 21.1 23.5 22.3 205 Reais 34.5 36.6 37.8 39.8 39.7 33.1 35.0 33.7 Ceara 65 Reais 12.9 12.1 13.2 16.4 14.5 11.0 11.4 10.8 132 Reais 30.0 29.0 29.7 34.4 29.5 25.2 25.5 24.7 205 Reais 42.6 41.6 42.1 46.6 41.5 36.9 37.0 36.5 Bahia 65 Reais 7.5 8.4 10.1 12.5 13.3 10.1 10.5 9.9 132 Reais 21.4 22.8 24.8 28.3 27.6 23.1 24.2 22.7 205 Reais 33.4 34.6 36.5 39.8 38.9 34.6 35.8 34.1 Source: Author's own calculations from PNAD data set. Note: Figures are percentages 106 Table A4 - Pernambuco - Mean Education and Income for the Total Population: Selected 1981-98 Years 1981 1985 1990 1995 1998 Total Average Years of Education 3.47 3.84 4.29 4.74 5.28 Population Mean Income. 138.52 132.24 127.59 145.23 153.79 Indigent Average Years of Education 2.05 2.29 2.88 3.06 3.6 Population Mean Income 38.4 36.1 34.8 35.73 34.38 Poorest Average Years of Education 2.11 2.34 2.82 3.15 3.71 50% Mean Income 42.48 37.93 32.64 40.22 38.08 Richest Average Years of Education 4.83 5.2 5.73 6.3 6.87 50% Mean Income 235.53 226.84 219.15 251.06 270.69 Source: Staff Estimates from IPEA PNAD Data Table A5 - Pernambuco - Mean Education and Income for the Urban Population: Selected 1981-98 Years 1981 1985 1990 1995 1998 Total Average Years of Education 4.39 4.69 5.04 5.46 5.99 Population Mean Income. 168.21 158.46 148.59 165.32 174.93 Indigent Average Years of Education 2.72 2.94 3.46 3.58 4.21 Population Mean Income 39.85 38.3 36.94 36.83 35.47 Poorest Average Years of Education 2.9 3.1 3.49 3.76 4.36 50% Mean Income 49.92 44.99 38.57 45.37 43.57 Richest Average Years of Education 5.89 6.16 6.6 7.05 7.57 50% Mean Income 286.69 271.76 258.68 283.77 305.46 Table A6 - Pernambuco - Mean Education and Income for the Rural Population: Selected 1981-98 Years _ 1981 1985 1990 1995 1998 Total Average Years of Education 1.49 1.85 2.24 2.23 2.7 Population Mean Income. 74.87 72.58 69.11 73.94 85.16 Indigent Average Years of Education 1.16 1.34 1.75 1.91 2.23 Population Mean Income 36.49 33.11 30.67 33.35 31.93 Poorest Average Years of Education 1.13 1.3 1.65 1.85 2.09 50% Mean Income 32.44 27.65 22.83 26.41 25.28 Richest Average Years of Education 1.85 2.33 2.79 2.59 3.36 50% Mean Income 91.38 119.33 115.16 1 17.47 125.09 107 APPENDIX B: ILLITERACY RATES Table B1 - Illiteracy Rate of Population of Fifteen Years of Age and Older by Age Group, Pernambuco, Northeast and Brazil, 1980-95 Illiteracy Rate (%) Region/Year 15 year 15 to 19 20 to 24 25 to 29 30 to 39 40 to 49 50 years or more years years years years years or more Brazil 1980 25.5 16.5 15.6 18.0 21.9 26.5 44.0 1991 20.1 12.1 12.2 12.7 15.3 23.8 38.3 1995* 15.6 6.8 7.5 9.3 11.0 16.7 32.7 Northeast 1980 45.5 34.3 33.4 37.6 42.8 48.0 65.2 1991 37.6 25.6 26.5 28.2 33.2 45.2 38.3 1995 30.5 16.3 18.1 21.9 25.5 35.0 54.6 Pernambuco 1980 42.4 32.7 30.1 33.4 40.8 48.3 61.8 1991 43.3 23.1 24.5 26.1 29.1 39.5 55.8 1995 29.8 14.5 18.5 22.5 25.1 32.6 53.8 1996 26.2 11.8 13.5 17.0 n.a n.a n.a 108 Table 2 Table B2 - Illiteracy Rate of Population Fifteen years of age and Older, by Sex and Residence Pernambuco, 1991 and 1995 Residence/Sex/year Illiteracy Rate (%) 15 year 15 to 19 20 to 24 25 to 29 30 to 39 40 to 49 50 years or more years years years years years or more Urban 1991 25.5 14.9 16.5 17.9 20.8 29.6 46.5 1995 23.4 9.7 12.7 16.7 19.6 26.0 45.7 Rural 1991 58.7 42.6 47.8 51.7 56.6 66.6 78.4 1996 52.0 29.7 40.1 43.0 46.8 57.5 77.9 Male 1991 35.9 29.5 28.8 29.3 30.2 38.7 53.5 1995 31.8 19.9 23.0 28.7 28.1 32.5 52.1 Female 1991 33.0 16.9 20.6 23.1 28.2 40.1 57.6 1995 28.0 9.0 14.3 17.1 22.5 32.7 55.3