Report No. 24500-BR Brazil Strategies for Poverty Reduction in Ceara' The Challenge of Inclusive Modernization (In Two Volumes) Volume II: Technical Paper April 10, 2003 Brazil Country Management Unit Latin America and the Caribbean Region Document of the World Bank CURRENCY EUDVALoETS AND EXH A EATES (R$/UI$) Currency Unit - Real (R$) December 2000: R$1.95 December 2001: R$2.31 December 2002: R$2.65 WEilHTS ANW D EASUR The Metric System is used throughout the report. FD;CAL YIEAR January 1 to December 31 ABREITO SAND ACDMM CREDE : Centro Regional de Desenvolvimento da Educa-ao FUNCEME : Fundaq&o Cearense de Meteorologia FUNDEF : Fundo de Manutengo e Desenvolvimento do Ensino Fundamental e de Valoriza,co do Magisterio IBGE : Instituto Brasileiro de Geografia e Estatfstica ICT Information and Communication Technology IPLANCE : Fundaq&o Instituto de Pesquisa e Mnformagao do Ceara LDB : Lei da Diretrizes Bdsicas MEC : Minist6rio da Educa,co PETI : Programa de Erradicagao do Trabalho Infantil PIB : Produto Interno Bruto PNAD : Pesquisa Nacional por Amostra de Domicilios PROGERIRH : Programa de Gerenciamento Dntegrado de Recursos Hidricos PSJ : Programa Sao Jose RMF : Regiao Metropolitana de Fortaleza SDR : Secretaria de Desenvolvimento Rural SEBRAE : Servi,o de Apoio as Micro e Pequenas Empresas SEDUC : Secretaria de Educa9ao Basica SENAI : Servi9o Nacional de Aprendizagem Industrial SEPLAN : Secretaria de Planejamento e CoordenaqAo TVC : TV Ceard UFC : Universidade Federal do Ceara UFPE : Universidade Federal de Pernambuco Vice President LCR : David de Ferranti Director LCC5C : Vinod Thomas Lead Economist : Joachim von Amsberg Task Manager : Michael Walton Brazil: Strategies for Poverty Reduction in Ceara TABLE OF CONTENTS ACKNOWLEDGMENTS ................................................ I I. INTRODUCTION .................................................. 1 II. POVERTY, INEQUALITY AND THE PATTERN OF GROWTH IN CEARA . ................................... 3 11.1 Introduction. 3 11.2 Overview of poverty and inequality trends. 3 11.3 Patterns of growth in incomes, Ceara in the context of Brazil. 7 11.4 Growth performance and patterns in Cear .11 11.5 The structure of poverty in Cear1 44 11.6 Conclusion .20 Annex to Chapter 2. How multiple dimensions of poverty affect poverty trends- preliminary results,1993-1999 ........................ 21 Ill. EXIT STRATEGIES FROM POVERTY IN CEARA ............................................................. 36 111.1 Introduction ........................................................... 36 111.2 From Poverty Profiles to Polices and Programs ......................... .................... 36 111.3 Priority programs for poverty reduction in Ceara ........................ .................... 44 IV. THE CURRENT POVERTY REDUCTION STRATEGY IN CEARA - STATE POLICIES AND PROGRAMS ............................................................. 46 IV.1 Introduction ........................................................... 46 IV.2 Overview of Public Expenditures, 1990-2000 ................................................. 46 IV.3 Breakdown of State Spending: 2000 ........................................................... 49 IV.4 Ceara's Programs for Poverty Reduction ........................................................ 52 IV.5 Basic Indicators of Program Effectiveness for Poverty Reduction ......... ......... 56 V. POLICIES FOR ENHANCED POVERTY REDUCTION PERFORMANCE IN CEARA . . 61 V.1 Introduction ........................................................... 61 V.2 The Scope for Poverty Reduction via Growth In Ceara ............... ................... 61 V.3 Poverty Reduction via Education in Ceara .............................................. 63 V.4 Poverty Reduction through Redistributive Transfers .80 V.5 How can the expansion of economic opportunities be made more equitable? 89 V.6 Industry, services and spatial development .................................................... 94 V.7 Governance, participation and equity ........................................................ 103 V.8 Implications for spatial policy ......................... ............................... 109 TabEes Table 2.1 CEARA-- Decomposic,o da Taxa de Crescimento Acumulada do PIB, 1985-2000 (%) ........................................................... 12 Table 2.2 CEARA-- Produtividade Media da Mao-de-Obra (*)--R$ 1 de 1971, 1985 e 1999 ........ 12 Table 2.3 Poverty Incidence and Depth in Metropolitan, Urban and Rural Ceara, 1999 .......... 15 Table 2.4 Poverty Rates by Age Group, Urban and Rural Areas, 1999 .................................... 16 Table 2.5 The Relationship between Education and per capita Household Income ................. 18 Table 2.6 The Effect of Sector of Employment on per capita Household Incomes ................... 19 Table 2.7 Average per capita Household Income for Individuals of 10 or more, by Education and Sector, Ceara ........................................................... 19 Table 2.8 The Effect of Labor Force Status and per capita Household Incomes ...................... 19 Table 2.9 Premium for Men relative to Women in per capita Household Income ..................... 20 Table 3.1 Rural Poverty--Exit Strategies and Associated Policies ......................... ................... 41 Table 3.2 Urban Poverty--Exit Strategies and Associated Public Policies ................................ 43 Table 4.1 Ceara--Real Public Expenditures, 1990-2000: Total and Share in GDP .................. 46 Table 4.2 Ceara--Real Public Expenditures by Category, 1990,1995 and 2000 ........ ......... 47 Table 4.3 Ceara--Public Expenditures by Category, 1990,1995, and 2000-Share in total ..... 47 Table 4.4 Estado Do CearA--Despesas totais do Estado por Categorfas- Ano 2000 ............. 48 Table 4.5 Ceara--Programs for Poverty Reduction ....................................................... 53 Table 4.6 Program Spending in Ceara: Summary of Results ......................................... 55 Table 4.7 Estado do Ceara--Despesa por Categorfa de Gasto-Acao Finalistica .................. 59 Table 5.1 Poverty Reduction Impact of Economic Growths- Projected Poverty Rates for Ceara in 2004 & 2020 .......................................................... 62 Table 5.2 % of Urban Working-Age Population with 8 or More Years of Schooling, 1999 ........ 64 Table 5.3 Education in Ceara: Main Target Groups and Policy Recommendations ................ 74 Table 5.4 The Previdencia Rural in Ceara: Coverage and Targeting Indicators ...................... 83 Table 5.5 Implementing the Bolsa Escola Transfer Program in Ceara: Projected Poverty Impact and Costs .86 Table 5.6 Projected Poverty Reduction Impact of Growth combined with Income Transfers (Bolsa Escola) Over Time .................................................. 86 Table 5.7 Summary of Policy Recommendations-Income Transfers ....................................... 88 Table 5.8 Share of Employment in Agriculture, Spain & Portugal, 1960-1996 .......................... 90 Table 5.9 Summary of Policy Recommendations for Equitable Growth .................................. 102 Charts and Flgures Chart 2.1 Desigualdade e Analfabetismo ...................................................5 Chart 2.2 Ceara--Evolution of Poverty Rate 1981 to 1999 (%) ...................................................6 Chart 2.3 Ceara--Evolution of Inequality: Relative Income Shares, Wealthiest 1 % and Poorest 20% of Population, 1981 -1999 .6 Chart 2.4 Ceara--Evolution of Poverty: Metropolitan Fortaleza, other Urban and Rural, 1981-1999 (%) ........................................7 Chart 2.5 Initial Income and Economic Growth in Four Periods ........................................8 Chart 2.6 Income Growth among the Poor in Four Periods ......................................................... 9 Chart 2.7 Growth and Poverty Reduction in Four Periods ......................................................... 10 Chart 2.8 Cearc--llliteracy Rates by Geographic Region, 1999 (%) .......................................... 17 Chart 4.1 Economic Development Expenditures, 2000 .......................................................... 50 Figure Al Poverty Headcount since 1985 (IBGE/PNAD) .......................................................... 21 Figure A2 Illiteracy (percent of over 15s, (IBGE/PNAD) .......................................................... 22 Figure A3 Initial Income and Income Growth, 1993-98 (IBGE/PNAD) ....................... ................... 22 Figure A4 Income Growth for the Population and for the Poor, 1993-98 (IBGE/PNAD) ......... ...... 23 Figure A5 Income Growth and Poverty Reduction, 1993-98 (IBGE/PNAD) ................ .................. 24 Figure A6 Income Growth among the Poor and Poverty Reduction, 1993-98 (IBGE/PNAD) ....... 24 Figure A7 The Poverty Headcount and the Poverty Gap, 1991 -95 (IBGE/PNAD) ........................ 25 Figure A8 Ceara Enrollment by Age Group, 1993-99 (IBGE/PNAD) ............................................. 25 Figure A9 Enrollment 6-16 in Ceara, the NE and Brazil, 1993-99 (IBGE/PNAD) .......................... 26 Figure Al 0 Changes in Headcount Poverty from Including Education Transfers, 1999 (IBGE/PNAD). 28 Figure Al 1 Headcount Poverty Including Lower Education Transfers, 1999 (IBGE/PNAD) .............. 28 Figure A12 Baseline Poverty Path (no education) ........................................................ 29 Figure A13 Preliminary Simulation (poverty lines set including one primary pupil) ...................... 30 Figure A14 Preliminary Simulation (poverty lines set for same initial 1993 condition) ................. 30 Figure Al5 Ceara Simulations ...................................................... 31 Figure Al 6 NE Simulations ...................................................... 31 Figure Al 7 Brazil Simulations ...................................................... 31 Figure A18 Enrollments among the Poor and Non-Poor (Aged 6-16) ........................... ............... 32 Figure A19 lnfraestructure Services, Poor and Non-Poor 1995-99 (IBGE/PNAD) ........... ........... 34 Figure 5.1 CDFs of Years of Schooling in Ceara: Actual and Simulated ................................... 66 Figure 5.2 Simulation of earnings, mean of education 7 years (1 concave) .................................. 69 Figure 5.3 Simulation of earnings, mean of education 7 years (13 convex) ................................... 70 Figure 5.4 Simulation of household per capita incomes, mean of 7 years (1 concave) ...............: ... 71 Figure 5.5 Simulation of household per capita incomes, mean of 7 years (1 convex) .................. 72 Figure 5.6 Number of People living in Rural Areas, 1960-2000 .................................................. 91 Figure 5.7 PIB Agropecuario e Pobreza Rural ...................................................... 94 Boxes Box 3.1 Criteria for Assessing Targeting and Effectiveness of Social Programs .................. 38 Box 5.1 Note on Simulations ...................................................... 68 Box 5.2 Information and Communication Technologies and Education ................................ 80 Box 5.3 Bolsa-Escola type Programs and Poverty Reduction ................................................ 81 Acknowledgments This is a product of joint work between IPLANCE/SEPLAN of the Government of Ceara and the World Bank. The joint work was led by Michael Walton for the World Bank and Alex Araujo, then of IPLANCE. Principal contributors included Gisella Colares, Annuzia Pontes and Am6lia Costa (IPLANCE) Fatima Falcao (SEPLAN), and Gillette Hall (World Bank). Madalena dos Santos and Joachim von Amsberg (World Bank) provided key inputs and advice. Monica Amorim and Jair Amaral (SEPLAN/CED) provided inputs on options for equitable growth. Jos6 Ricardo Bezerra Nogueira and Rozane Bezerra de Siqueira (Federal University of Pernambuco) undertook the simulation work on transfers. Francisco Ferreira and Phillippe George Leite (then of PUC-Rio, currently World Bank) prepared the background analysis on the effects of education expansion on distribution. Mark Thomas (World Bank) prepared the Annex to chapter 2 in the background material on how multiple dimensions affect poverty and poverty trends. Paul Siegel (World Bank) provided the main inputs for the section on drought management. Ant6nio Magalhaes and Dorte Verner (World Bank) provided ongoing inputs and advice. Judith Tendler (Massachussetts Institute of Technology) provided invaluable general advice. Charles Kenny and Juan Navas-Sabater (World Bank) prepared inputs on information and communications technology. Christina Alquinta (World Bank) provided research support. It draws on extensive meetings in Ceara and on work by Francisco Ferreira Alves (IPLANCE), Naercio Menezes (USP) and Mark Thomas (World Bank). The work was undertaken under the overall supervision of Gobind Nankani (Country Director for Brazil, World Bank during the bulk of report's preparation) Vinod Thomas (current Country Director for Brazil, World Bank) and Monica Clark (Secretary of Planning, Government of Ceara, during the report's preparation). Particular thanks to Senator Tasso Jereissati (Governor of Ceara until 2002) for posing the questions that motivated this report and for his active engagement in the process of preparation. Drafts of this report have been discussed with the Government of Ceara and have been shared with the Federal Government of Brazil. The analysis of the report was undertaken in the period to August 2001, and includes information that was available up to that time. While the work was a product of a collaborative process, views expressed in this document are those of World Bank and do not necessarily reflect the views of the Government of Ceara, the Federal Government of Brazil or the Executive Directors of World Bank's Board. 1. Introduction 1.1 This background material provides more detail on the approach and results presented in the document Strategies for Reducing Poverty in Ceara. This first section provides the motivation and overall approach for the work. As noted in the main the companion volume, the work was undertaken at the request of, and jointly with, the government of Ceara. The motivation was the government's own question on social progress in the state of Ceara, on how to explain the contrasting evidence on non- monetary vs. monetary measures of poverty, and in particular the persistence of income inequality and slow progress in poverty reduction despite major development efforts. 1.2 Approach. The questions posed by the government of Ceara reflect the central development challenges faced by many societies in Latin America and elsewhere in the world. Inequality is notoriously slow to change. When societies grow, the poor typically gain in proportion to their initial share in incomes-and when societies are initially highly unequal, this is a correspondingly small income share. Ceara's experience of the past decade is thus more the rule than the exception. But that only underlines the importance of trying to face the question posed by the government-how to shift the design of policies and programs to maintain past successes while accelerating income gains for the poor. 1.3 In methodological terms, there are two problems in developing precise answers to the questions posed. First, there is no single technique for providing clear policy guidance on these issues: models of growth and distribution have so far proved incapable of handling all the complex relationships between structural conditions, external developments and policy choices. Indeed, some observers would argue that the world, and the development process, is intrinsically complex in a way that makes local assessment and judgment essential to the assessment and design of development programs.' Second, while there are techniques for evaluating the specific impact of individual programs on poverty outcomes (controlling for other influences), Ceara has not yet developed the information base for undertaking such an analysis. This lack of information and of detailed program impact analysis is again not specific to Ceara-the absence of program evaluation is a feature of most programs in Brazil, and of many industrializing countries around the world - but it does limit the scope and level of detail of policy guidance that can be given at this stage, particularly with regard to the ability to form judgments on the effectiveness of the state's existing poverty reduction programs. 1.4 The approach taken in the work underlying this study thus had to be pragmatic- examining trends in poverty indicators in the context of the state's structural heritage, drawing out the causal links between economic and social patterns and outcomes among different groups, and then forming judgments regarding how these relate to both external developments (including Brazil-wide conditions) and existing policies and programs within the state. This assessment forms the basis for developing possible 'See James Scott, Seeing Like a State. options for future policy enhancement. Precisely because of the impossibility of forming definitive conclusions at this stage, it is important that future shifts in policy are based on continued assessment of ongoing conditions coupled with efforts to improve state capacity for program monitoring and evaluation. 1.5 Coverage of bIckground mnter@iL This background material consists of five substantive sections: o analysis of the trends and structure of poverty and inequality in Ceara, and links between these outcomes and state and national patterns of growth (Section 11); o analysis of whether the multiple dimensions of poverty affect the interpretation of poverty trends o identification of the main processes or 'exit strategies' by which different groups of people can move out of poverty in Ceara, set within the state's historical context and the goal of more inclusive development (Section 111); o analysis of Ceara's current poverty reduction policies and programs in light of how well they support these exit strategies from poverty, including a review of state expenditures for poverty reduction and an analysis of relevant state programs in terms of objectives, beneficiaries, coverage, and costs (Section IV); o discussion of policy options in four main areas that together are likely to be necessary to accelerate poverty reduction and reduce inequality: (i) education strategies for redistribution; (ii) re-distributive transfers; (iii) employment- creating patterns of growth; and (iv) institutions; and finally a review of implications for spatial policy (Section V). 2 II. Poverty, Inequality and the Pattern of Growth in Ceara 11. 1 Introduction 2.1 In this section we look at the dimensions, history and nature of poverty and inequality in Ceara. In line with the concerns of the state government at the time of preparation of the report, this report places a primary focus on income poverty, while also outlining overall indicators of progress in social well-being and other dimensions with causal influence on incomes, especially education and institutions.2 This section first reviews trends in social conditions, incomes and inequality in Ceara. Then it focuses on income growth among the poor in Ceara, in light of average income growth in Ceara and Brazil-wide. Third, it assesses why growth in Ceara has been accompanied by persistent income inequality and hence only moderate income poverty reduction. Finally, it examines the current geographic and demographic structure of poverty within Ceara. 11.2 Overview of Poverty and Inequality Trends 11.2.A Measuring Poverty 2.2 The dimensions of Poverty. In assessing the structure and trends in poverty, two measurement issues have to be confronted: what dimensions are relevant? And what is the difference between being poor and non-poor? Both of these questions are in the end a function of the norms prevailing within a society. We adopt the following common approaches in poverty analysis:3 * use of measures of education and health status as direct proxies for the quality of life in terms of these core capabilities, and access to selected services as proxies for a range of influences on well-being; * income as a proxy for the influence of material conditions on well-being. 2.3 Poverty lines. The next step concerns the choice of poverty line. Here, it is of importance to note that the precise placement of a poverty line is arbitrary. In Ceara, as in Brazil and other countries, there are many poverty lines in use. Partial objectivity can be obtained by calculating the requirements for meeting a certain level of calorie inputs for food, and adding an allowance for non-food spending based on spending 2 Incomes are a very incomplete measure of well-being. Poverty-and freedom from poverty-encompasses every aspect of a person's life, including the capacity to live a healthy life, to have the skills for work and other human pursuits, to have a reasonable level of security, to enjoy self-esteem and to be able to influence institutions in that affect an individual. Poor people emphasize ill health, vulnerability and the disempowering consequences of lack of influence over formal and informal societal institutions as central dimensions of poverty, in addition to lack of material resources. See Sen, 1999 Development as Freedom, World Bank, 2000, World Development Report 2000/01, and Narayan et al. Voices of the Poor. 3 While the intrinsic importance of vulnerability and self-esteem is fully recognized, these concepts are difficult to measure. This remains a weakness of many poverty studies that can be partly compensated by participatory and qualitative work. 3 characteristics of the poor. An example of this approach is illustrated in the work undertaken for one of the poverty lines used in the previous World Bank report on Ceara.4 However, this approach still has a high degree of arbitrariness, in addition to a whole set of potential errors and options in measurement. Thus the total number or proportion of poor at a point of time is always dependent on the particular assumptions used in developing a poverty line. 2.4 Regardless of the line selected, poverty measurement is useful is in assessing the depth of poverty, trends over time and the structure of poverty in relation to characteristics of households. Assessing these aspects of poverty in Ceara is the focus of this report. No independent calculations of poverty lines were undertaken for this work, which instead draws on the following existing sources: o for comparisons with Brazil and other states, including trends over time, the work draws on poverty estimates by IPEA, which employs a somewhat higher line but is the only source for poverty estimates across geographic regions in Brazil. o for the analysis of the level, depth and structure of poverty in Cear&, the work uses estimates calculated by the government of CearA (IPLANCE), based on a poverty line of half a minimum wage, and an indigence line of a quarter of a minimum wage in 1999. .022l Tren1d in Onc1nme Pve13ty Over' Time 2.5 According to poverty estimates from IPEA that are comparable at the national and state level, Brazil's national poverty rate in 1999 was 34%, while 58% of Ceara's population was poor.5 The state's relatively high level of poverty is due to the fact that Ceara's average income is much lower than the Brazil average, and that it shares with virtually all states within Brazil one of the highest levels of inequality in the world. In terms of the share of the bottom 20 % of the population, Ceara is actually slightly more unequal than Brazil-itself competing with South Africa to be the most unequal country of significant size in the world. One of the starkest dimensions of both deprivation and inequality is that over a quarter of the population of about 15 years of age is illiterate. 4See in particular SGovernment of CearA III of the World Bank, 2000. 5 These poverty estimates are calculated from PNAD survey data by IPEA, and are based on a poverty line of R$ 68 in 1999 prices (based on the value of a minimum food basket, and roughly half a minimum salary). This poverty line is also close to the line calculated by the World Bank as providing the minimum spending to meet food requirements for healthy living for 1996. This was estimated at R$ 65, equivalent to R$ 76 in 1999. See World Bank 2000. IPLANCE, Ceara's state research organization, uses an alternative (slightly lower) poverty line, equivalent to exactly ½2 the value of the minimum wage; thus IPLANCE poverty estimates for the poverty rate in Ceara are somewhat lower. The IPEA series, which can be directly compared to national poverty estimates, is used for time-series analysis presented in Chart 2.2, while the current poverty profile estimates are drawn from IPLANCE analyses, and hence uses the alternative (slightly lower) poverty rate. All poverty estimates are not adjusted for differences in cost-of-living, urban-rural (which could result in an overestimation of rural poverty), nor adjusted for household composition using equivalence scales or economies of scale (the latter will lead to an upward bias in calculated poverty rates for households with children, and for larger households.) 4 This link between inequality and low education levels, vividly illustrated in Chart 2.1, will be a theme to which we return. 2.6 Over the past two decades, the poverty rate in Ceara has demonstrated a slow downward trend, as illustrated in Chart 2.2, from a peak of 80% in the early 1 980s to just under 60% in the second half of the 1 990s. Over the entire period, Ceara's poverty rate has closely followed yearly movements in the national (and regional) poverty rate, indicating significant correlations with national conditions.6 However, Ceara began the period with a poverty rate slightly above the regional Northeast average, yet ended the period at or slightly below that average. While this trend points to relatively greater success in poverty reduction in Ceara compared to the rest of the Northeastern states, this result is overshadowed by the virtual stagnation of the poverty rate in Ceara in the latter half of the 1990s, which hovered at or just under 60%, contrasting sharply with the state's higher than average rates of economic growth over the same period. Chart 2.1: Inequality and lliliteracy Inequality and Education International Comparison 9.0 - South a 8.0 K l o 7.0 - (I) | Vietna n Indonesia 7.0 U) @ 6.0 ° 5.0- al 4 0 ~ Uruguay _ Chl a___ .~4.0 0C3._-_ Chile _ -Brazil '0 3.0 E 2.0 Afica earae 0 ° 1.0 IL., 0 In 0 L Illiteracy Rate +15 6 The sharp decline in poverty observed in the data for 1986 represents largely an 'illusory effect' of the Cruzado Plan (survey fieldwork coincided with a period of rapid, and temporary, economic expansion). Data prior to 1995 are also considered less reliable due to sharp movements in the inflation rate that predominated over that period. 5 Chart 2.2: Cearbi - Evolutlon of povorty rate 19S1 to 1999% 90 , I , , , , , , . o , . . 80 - _L _ _ L9 1__ __J__ __ __ ___I_ ___L__*__L__ __ __ __ 70- ~ - -r----- so I 1I 40 -4- - e--F -4 t4 -\ -4-H- ----- - -4 -- F4 -4 - -4 1- -4 -4i -- 20 BasI--Nodse---------- Char _ 2.3_| : _ _r - _lvlto _o_ XeutiV _ : t_ _ F 14 m It I I F I I F N I I I I F I I I F I I _0 ' -r- - -4 4 - 4e--e--e H -4 -1 -4 1- -4 1- r - 4 e - -e- - 4 Wf-- 210.0 I I F I I I I I I F I I I I I F I I F I , I, F I . F F F F . F , F. 20.F0F FI L 0 . N (n ' In u' r CO Cs 0 N n Lf 0 N 0 a's 0c co co o0 cc xc co c aO O 0' 0' Os 0" 0" 0' 0' ' a , a, c a a, o C s o s a" a, a, o' a, a, c's as -4 -4 -4 - I - -4 - -4 - -4 -4 -4 -4 - 4~ -0--20% -I 10/ Source: IBGEIPNAD data, tabulated by IPLANCE 2.7 Indicators of inequality (Chart 2.3), provide a partial explanation for the state's slow progress in poverty reduction. Income distribution in Ceard has remained highly unequal over the past two decades, with over 15% of total income consistently accruing to the wealthiest 1 % of the population. The income share of the bottom 20% of the income distribution, on the other hand, has fallen fairly steadily over the past two decades, from 3.6% in the mid-I980s to 2.3% in 1999. 2.8 Poverty rates for Metropolitan Fortaleza, other urban areas, and rural areas (Chart 2.4) again demonstrate a slow downward trend. However, regional trends differ somewhat after the mid-1 990s, particularly since 1997. Rural poverty has continued to fall -- from 80.8% in 1997 to 76% in 1999 -- as has poverty in 'other urban' areas 6 (excluding Metropolitan Fortaleza) -- from 62.8% to 61.6%. Nevertheless, poverty in Metropolitan Fortaleza has risen from 38.1% to 42% over this same period.7 Hence the stagnation in poverty reduction arises in part from the combined effect of two diverging factors: continued decline in rural and 'other urban' poverty rates, offset by rising poverty in Metropolitan Fortaleza.8 Chart 2.4: Ceara - Evolution of poverty: metropolitan Fortaleza (RMF), other urban, and rural, 1981-1999 - % 100 90 O 1 _ _ J _ 0 _ _ _ L__ i_ _ _ , 1A_ ,-4 , - x,- ,, ,1 ,4 .4 , -I a -' 1 1 1 - 1 4 1 80 -0 X -T,ta L - ,-RL __ J_ -rI_n -0-r Sore IBG-PA daa tauae by IPLANIICEI___ 20 s r- - T - - -t - rar t h a - - -sig -dete - - r - - T - - s - - t i- - - the 10s:firs th Rea Plan bor,,ought ( asgniicnt imedat redctoninovrt between O O O 0 1993 nd 195 (a wel asNa imrovmen ininqalty. Seod h f 1 1sa 1th 1Bz-wide s d in 1 99 -99, n h w East Asia-Ru--n--Totalra--n--l RMF c Urises,and, in---1999a ma reTovr fhromru father99 dperouh.Tefre per to have bensgiiatdtriaffeted Fofpvrtayeza mt, con0s: isnt ith the metropolta aren beian imost ful integate dn witohin between 1993 and 199 (as wel as an impovement in ineuliy. Seod the end ~ of the 1 sa the Bail-wide slowdow in 1998-99 in the wae fh 2.9eTovr frortru fathe19orsgthefre appears to have bensgiiatdtriaffeted Fofpvrtaleza mothe con9s:irsten wthe thea Plantropolita are beniicng immedite feullyo intgae povethi national financial, labor and product markets. At the same time, the agricultural recovery more than offset any negative effects of the slowdown in rural areas.10 11.3 Patterns of growth in incomes, Ceara in the context of Brazil 2.10 Patterns of average household income growth in Brazilian states confirm that Ceara's economic performance has been above average under the post-1987 state 7A possible explanation for this divergence in urban/rural poverty trends is the 1999 recession which may have affected the urban formal sector (monetized economy) disproportionately. 8 Note: IPLANCE suggests drawing only on IPEA data for discussion in this paragraph. Will be changed. 9 However it is important to bear in mind that much of this observed drop in poverty may be driven by measurement effects of the major change in the price environment through disinflation dunng the Real Plan. la What is really noteworthy in the agricultural series is the apparent asymmetric effect on between droughts (for example the major 1998 one) and recoveries: as we discuss below, this is plausibly due to the effectiveness of public action in drought years. 7 governments." Chart 2.5 divides the period in line with Brazil-wide macroeconomic conditions (note that Ceara is abbreviated as CE in these graphs; other states are abbreviated similarly). While household incomes in CearA grew at roughly average (slow) rates in the first half of the 1980s, incomes declined relatively quickly in 1985-89 and then weathered the 1989-93 period relatively well. But the real difference in Ceara's performance becomes evident in 1993-98, when the state's average income growth of almost 5 % per annum is significantly higher than that of most other states. While on average the poorer northeastern states grew faster in this last period, there was a large divergence in growth rates even for this period, with some poor northeastern states growing faster than Ceara (Paraiba and Pemambuco) and some much slower, notably Bahia. Chart 2.5 Onitial Mncome and Economnic GrowRh in Four P3eriods 1981-85 1985-89 GO 2 S RS RN Ms GO SC MA MT RJ pi MG pi ES ES SE PR 6 P BA cO stA ~~~~~~ALA MGM A RSP 9 - 2 ~~~PE S, _ 3PP SE~~A Ms CE ES PB 3 4 5 6 3 4 5 B intal mean log Income 1981 initial mean log Income 1985 1989-93 1993-98 2 pi 6B PB~~~~~~~~~~~~~~~P 8 -.. CE A~ ~ ~~~~~~~~S C' NPE C' C' ~~~~~E ECE o- PA SC Pt->, RN ES Ad MA N Source MA MGnze An BAml20 AL ~~~~~~8 4 -A E ~~~~~~~~~~S SPC BA~~~A . 4 2S E__ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 3 4-B6 245 initial mean log Income 1989 InItIal mean log Income 1993 Source: Menezes arnd Thomas, 2001 11 This section is based on Menezes and Thomas, 2001. National accounts estimates of income differ from household survey estimates both in concept (national accounts include capital income and value added in non- government organizations) and due to measurement differences 8 2.11 Did the poor participate in this income growth? In both Ceara and Brazil, incomes of poorer groups tended to move in line with average income, but significant differences do emerge between states and time periods (Chart 2.6). The relatively poor, measured as the bottom quarter of the Brazil-wide distribution (that includes, for CearA, roughly the bottom half of the income distribution) on average experienced income growth below mean income growth for the past two decades, implying some worsening in inequality for the bottom of the distribution. However, both Brazil-wide and in Ceara in particular, inequality trends show some improvement over time. While income growth in the bottom of the distribution was significantly lower than average for the 1981-85, 1985-89 and 1989-93 periods, for 1993-98 income growth was broadly in line with the state average - in Ceara growth in incomes of the relatively poor was about 5 % per annum in this period, slightly above average income growth in the state. However, this relatively faster growth may in part reflect the once-off distributional gains in the wake of the disinflation of the Real plan, and there is some evidence that growth in incomes was faster for those closer to the poverty line, and slower for the indigent.12 Chart 2.6 Income Growth among the Poor in Four Periods 1981-85 1985-89 2 Pi MA R%_ - ~ ~~~~~~~~ ~ ~~~~~ RJ ° B/sRN S 1 - ~~~~~~~~~~~~~~~~~~~PE E AL SC _-gf1 Ms E Ms E /8MG - SE PR GO PR o EPB 'PC PRS MT o - ,3ArPN rc CE SP -2 - '1 0 1 2 -1 0 1 2 mean log income growth 1981-85 mean log Income growth 1985-89 1989-93 1993-98 1 - PR RS 6 GO MT ' a a ~~~~~~~~~_PPE 0 0 aB 5 CE , a - 2 - BA ALMASP MG M MAS RJ RJ SSE PR CE - 4 - 2 C -- ~~~~~~ES E BA MA E R o AL B c -2 3 RS M a ~ ~ ~ ~ ~ ~ ' X~~~~~ MT SP PE -~~~~~~~~~~~~~~~~ -4 -2 0 2 2 3 4 5 6 mean log Income growth 1989-93 mean log Income growth 1993-98 Source: Menezes and Thomas, 2001 12 The mean income gap for the poor increased at least till 1996; see World Bank SEM, SGovernment of Ceard 3. 9 2.12 Ceara's performance in a Brazilian context is shown in Chart 2.7. Modest reduction in poverty incidence in 1989-93 was followed by more rapid gains in 1993-98, though with less progress than in a number of other Northeastern states, including some that grew.faster than Ceara, such as Paraiba and (marginally) Pernambuco, and some that grew more slowly, including Bahia, Maranhao, Alagoas and Rio Grande do Norte. Chart 2.7 Gvrowth and Poverty Reducton in Four Perfods 1981-85 1985-89 PB GO RN MA - E 05 05 o5 c c AL R AS PR MG B P1~~S SP 1 0 AS --R AJ PBP BA 05 -06s EE.L I1 0 12 -1 0. mnean log Inoome growth 1981-85 mean log incomie growth 1985-9 1989-93 1993°98 pi2- PB - - ~~~~~~~~~~~AL R Os - ~~~~~SE7 9 BA MAAN B o GO MA / > ES; P5 t~~~~~ - OS- AUEo J|1 S«9N(R BA S 05 -4 -2 32 2 3 .4 S 6 mean logl Intcome growth 18898-93 mean log Income growtl 1 993-98 Source: Menezes and Thomas, 2001 2.13 These results provide a quantitative picture of relative income changes for Ceara. They confirmn that, for Ceara, as for the rest of Brazil, the poor gain from overall growth. Ceara's relative good economic performance in the past decade or so has indeed yielded income gains for the poor. However, these gains have been modest both because of very high initial inequality and, in some periods, a slight worsening in inequality. 10 11.4 Growth Performance and Patterns in Ceara 2;14 Why was Ceara's overall growth performance relatively good, and yet, why did this growth fail to make inroads into inequality-and, consequently, why has poverty reduction been moderate? Lacking a complete statistical analysis of determinants of growth in Ceara, this report provides a preliminary assessment based on existing descriptive information."3 2.15 There are two common interpretations of Ceara's relatively good aggregate performance. The first falls squarely within the vast economics literature on cross- country growth. Ceara's vigorous stabilization and public sector reforms after 1987 helped support a healthy overall investment environment for the private sector, and provided the fiscal headroom for significant expansion of productivity-enhancing public investments detailed further in Section IV). The second aspect emphasizes Ceara's sectoral investment policy, including the vigorous policy of attraction of private investment-especially from Brazilian firms from the Southeast, through a mixture of specific fiscal incentives and a proactive strategy of projecting a modernizing image for national investors.'4 2.16 These interpretations are partially supported by statistical evidence. The fiscal adjustment and reforms have been widely discussed elsewhere.15 By contrast, the aggregate investment ratio actually fell from 20.9% in 1985-90 to 16.7% in 1991 to 1996, though the share of private investment within this rose from 50% to 62% between 1991 and 1996.16 Both case study work and the total numbers attest to the significant levels of industrial investment attracted to Ceara -- totalling R$ 5.7 billion since 1994. though there is some controversy as to how much of this investment can be attributed directly to the fiscal incentives program (tax breaks), and how cost-efficient these incentives have been. However, in the most recent period, industry has become the state's fastest growing sector, followed by services, with agriculture showing no average growth. 17 13 Statistical results from comparative state performance are available from the World Bank study on Brazil's prowth performance. 4See Ferreira, 2001, Vasconcelos 1999, Vasconcelos et al, 1999, World Bank 2000a and Tendler, 2000, 77Te Economic Wars Between the States. The World Bank's State CEM emphasizes the first set of factors; the others also recognize these but also argue that industrial policy had a positive role in attracting industrial investment. 15 See references in previous footnote. 16 See Ferreira 2001. 17 Note that agricultural growth is heavily affected by the weather cycle. 11 Table 2.1: CEARA-Decomposi98o da Taxa de Crescimonto Acumulads do PIB,1985-2000 (%/o) Setores Crescimento Acumulado Contrdbullo Setorial no Perlodo P/Tsxa de Cresc.(') Ag ropecuaria *1,61 -0,40 Industria 104,67 54,20 Servi0os 59,91 46,20 Total da Economia 69,97 .0ooa, Fonte: IPLANCE, C4lula de Contas Regionais, see Ferreira, 2001. (') M6dia ponderada, obtida pela multiplica9ao das taxas de crescimento Setorias pela participag5o de cada setor no PIB estadual. 2.17 Of particular interest to poverty and inequality is the productivity of each sector, measured as the value added per worker. As Table 2.2 shows, there are major differences between the three sectors, with industry showing large increases in productivity (rising to a level 20 times that of agriculture in 1999), while services experienced stagnant productivity change. These results are consistent with relatively capital intensive industrial growth. The pattern of production has shifted more markedly to industry than has the pattern of employment. Table 2.2: CEARA-Produtiividade iddia de Moo-de-Obro (*)-R$ 1 de 1999 1971,1985 e 1999 Anos Total d Agropacu rainda Economic. 1971 3.321 1.506 6.097 7.100 1985 5.006 1.665 10.386 6.750 1999 5.938 802 16.814 6.989 Cresc.Acumulado 1971 -99 (%/.) 78,8 - 46,7 175,8 * 1,6 Fontes: SUDENE/CPE; IBGE/PNAD e IPLANCE, see Ferreira 2001. (*) POB/Populaq&o Ocupada. 2.18 Movement of labor between sectors, as well as migration, can be a major source of poverty reduction - indeed, both of these factors were an important part of the poverty reduction process in East Asia. In Ceara, movement out of agriculture - in part associated with geographic migration -- would be expected to bring large income gains, with productivity in services held down by the continued urban influx. However, 12 the pace of reduction in agricultural employment has been insufficient to offset the major adverse shocks to the sector, notably the rapid decline in the cotton sector.18 Migration to the Southeast is another major source of income gains. However, for reasons that have yet to be analyzed, one of the striking features of recent years has been the reversal in net migration flows in the Northeast, with the region becoming a net importer of labor. 2.19 The state government's public spending strategy has placed major emphasis on investment in human capital (especially education), as well as on infrastructure and industrial incentives (see Section IV). General educational and infrastructure expansion is consistent with the overall growth performance, while fiscal incentives-sometimes linked to specific infrastructure spending-are consistent with the acceleration in industrial growth in the second half of the decade. 2.20 A preliminary comparison of with the determinants of growth in mean income and of the poor across all Brazilian states, as undertaken by Menezes and Thomas, yields the following results of interest to Ceard.19 * Less political volatility and policy continuity is associated with more growth on average for the poor, whereas good public administration and overall state- wide policy is associated with higher average growth, but not for the poor; * More education is associated with higher subsequent growth for households on average, but not for the poor-indeed in the 1981-98 and 1989-98 periods higher initial education is slightly disequalizing. However, within education there is some evidence that lower secondary education is associated with relatively rapid growth of poorer groups, while more high school and college education favors non-poor groups; * Higher infrastructure spending is associated with higher average growth and, to a lesser extent, with higher growth for the poor. However, this finding combined with evidence that (other things equal) higher initial income is associated with higher subsequent growth, supports the view that geographic location does matter-that location in a poorer state, with lower levels of state-wide infrastructure, is associated with slower income growth * Higher shares of employment in industry and (for the 1990s) agriculture are associated with higher average growth, but not with different levels of growth of the poor. However, both higher agriculture and service employment shares are associated with disequalizing growth. Industrial growth appears to be relatively favorable to the poor. 2.21 While these results should be treated as only suggestive of patterns, the consistency with the qualitative assessment of Ceara is striking. Good state-wide policy helps growth, but past general education expansion has had limited effects on the 18 See World Bank (2001e) Tendler (1997) on cotton strategy. 19 See Menezes and Thomas (2001). This is based on an econometric analysis of the correlates of income growth for age cohorts pooled by year (1981-1998), state and ten age cohorts. Since it follows the same cohorts, but not the same households, it is a quasi-panel. 13 incomes of poor. This makes intuitive sense: in the initial phase of education expansion, focusing on primary education, the effects on household incomes will be negligible -- and even negative, to the extent that children cease working in order to attend school. The result that infrastructure spending has tended to be disequalizing is consistent with the continued skewed pattern of growth. The negative impact of higher agriculture employment on inequality is consistent with the fact that slower agricultural growth has an immediate dampening effect on the incomes of those engaged in agriculture, who are disproportionately poor. Movement of labor to other sectors has been insufficient to offset this result. The principal sector attracting employment has been the services sector, which has experienced stagnant productivity growth in Ceara. 1JL5 Tihe VruWure Of lPoV@y in Coard 2.22 The discussion thus far has explored the pattern of poverty and inequality over time, and possible correlates with trends in overall income growth. This sub-section focuses on the structure of poverty in Ceara for the most recent year for which survey data are available -1999. While the analysis is based on a snapshot of one year, it helps fill in factors behind the long-term correlates of poverty, by examining the structure of poverty according to location, age, education, household head sector of activity, and labor force status. In addition to descriptive statistics, the results draw on statistical analysis of the correlates of income. HJ,5.A Location 2.23 Table 2.3 provides summary poverty statistics for 1999 using a poverty line of half the minimum wage-or R65-which is close to a 'food only' measure, calculated for past work on Brazil and CearA and an indigence line of a quarter of the minimum wage-R37.5.20 Almost half of Ceara's population are poor by this measure, and over one fifth are indigent. In terms of absolute numbers, more of the poor live in urban areas, but in terms of incidence and depth, poverty is substantially worse in rural areas But almost three quarters of the rural population live below the poverty line and over 40 % below the indigence line. Rural areas account for over one-third of the total poor and almost half of the indigent, even though they account for less than a quarter of the population. More disaggregated earlier work on the 1996 PNAD found that the peripheral areas of Fortaleza also had relatively high levels of poverty incidence- higher than other urban areas.2' 20 See World Bank (2000). 21 Ibid. 14 Table 2.3: Poverty incidence and depth in metropolitan, urban and rural Ceara, 1999 (thousands of people and %ages) Metropolitan Other urban Rural CearA Fortaleza Population Numbers 2746 4729 2349 9824 Share of total 27.9% 48.1% 23.9% 100.0% Poverty Numbers 1027 2029 1717 4772 Incidence 37.4% 42.9% 73.1% 48.6% Share of total 21.5% 42.5% 36.0% 100.0% Gap 10.2% 16.9% 26.4% Depth 27.3% 39.4% 36.1% Indigence Numbers 367 801 982 2151 Incidence 13.4% 16.9% 41.8% 21.9% Share of total 17.1% 37.3% 45.7% 100.0% Gap 1.8% 2.2% 6.7% Depth 13.5% 13.0% 16.0% Notes: incidence is P0 (the proportion below the relevant line), gap is Pl (the incidence times the average distance below the poverty line) and depth is the distance below the line just for the poor, or P/PO. Source: calculations by IPLANCE from PNAD, 1999. 2.24 Measures that capture the average level of incomes below the poverty line (the 'poverty gap'), and that place greater weight on an individual's degree of poverty the further income falls below the poverty line ('depth of poverty') are indicate of Ceard's severe income inequalities, particularly among urban populations. Average incomes of the poor vary between 27 % below the poverty line in metropolitan Fortaleza to 36 % in other urban areas; mean incomes of the indigent vary from 13% below the indigence line in urban areas to 16 % in rural areas.22 1.5.B Poverty by Age Group 2.25 Ceara's population is predominantly young; one-third of the population is under age 15 -- some 2.4 million children and young adolescents - and poverty is also heavily 22 The size of the poverty gap and the severity of poverty have important implications for policy reaching the poorest will likely achieve the greatest welfare gains, as opposed to simply shifting individuals over the poverty line: the further an individual's income is below the poverty line, the worse off they are, and the greater the welfare gain that will arise from each incremental increase in income. 15 concentrated among this population.23 Rural poverty rates for those under age 15 are 57% and 87% respectively, far higher than the average urban and rural poverty rates 43% and 73% (Table 2.4). Thus 44 % of the poor - over 1.6 million people - are under age 15, and about two-thirds of young people live in poor households24. The fact that poverty rates are high among the young is of central importance to future success - or failure - of poverty reduction in Ceara, given the central role that inter-generational transmission plays in the perpetuation of poverty over time. Table 2.4: Povety Rates by Age Group, Urban nd RuraD Areas, 1999 URBAN Age Group lnd L nt % poor _ Total population 0-6 years 190,119 26.8% 404,943 57.1% 709_413 7-14 years 204,115 24.2% 483,821 57.5% 842,044 15-18 years 77,558 18.7% 191,877 46.3% 414,297 19-25 years 74,292 12.5% 222,063 37.4% 593,964 26-40 years 139,193 13.8% 377,384 37.3% 1,012,122 41-64 years 106,595 12.2% 291,542 33.4% 873,807 65+ 9,604 3.4% 57,338 20.2% 283779 All 801,476 16.9% 2,028,968 42.9% 4,729,426 RURAL AaeGroua MndIent - eoo°r L - Total population 0-6 years 230,296 59.3% 336,306 86.6% 388,355 - 7-14 years 265,438 52.9% 419,696 83.6% 502,242 15-1 8 years 91,179 40.6% 169,236 75.4% 224,443 19-25 years 83,293 32.6% 172,409 67.4% 255,678 26-40 years 168,321 42.2% 303,421 76.1% 398,463 41-64 years 138,310 32.0% 269,837 62.5% 431,683 65+ 5,636 3.8% 45,684 30.8% 148,533 All _ 982,473 41.8% 1,716,589 73.1% 2,349,397 Total 1,783,949 25.2% 3,745,557 52.9% 7,078,823 Source: IBGE/PNAD data, as calculated by IPLANCE. 2.26 By contrast, the incidence of poverty among the elderly (over age 65) is relatively low. Only 20% of the urban elderly and 31% of the rural elderly are poor. More remarkably, urban and rural indigence rates among the elderly are both very low and almost the same - at 3.4% and 3.8% respectively. This is the only category for which rural poverty and indigence rates approach urban rates, and the low rate of indigence in 23 While results have to be treated with caution in this area, since they are affected by assumptions over equivalence scales and economues of scale in households, it is likely that the basic result will still hold if such adjustments were made, even if the degree of difference would fall. Equivalence scales adjust per capita income values for household composition, based on the assumption that children consume less (and therefore require less monetary income to achieve a certain standard of living) than adults; however this assumption is often called into question, and there are no agreed-upon norms for calculating the equivalence scales); one can also adjust per-capita income measures upwards for larger households, on the assumption that some consumption goods can be shared by all and thus Mrovide 'economies of scale' for larger households. These are almost certainly upper bound estimates-if adjusted for equivalence scales which assume lower per capita consumption requirements of children and for economies of scale in larger households, and therefore can lead to lower estimates of poverty rates for these two groups. 16 rural areas likely reflects the effects of the federal rural old-age pensions program (previdencia rural) -- demonstrating the power of well-targeted transfers in lifting very poor groups out of destitution (see section V). 11.5.C Education 2.27 Ceara has made tremendous gains in increasing access to basic education, achieving near-universal initial primary enrollment at entry in the 1990s (enrollment rates among 7-14 year olds have risen from 44% in 1970 to over 95% in 1999).25 However, the state started with a large historical deficit in educational attainment -- 53% of the adult population (ages 15-64) and 79% of rural adults are "functionally illiterate", defined as having less than four years of basic schooling (Chart 2.8). More importantly, this deficit is not confined to older populations, but is significant even among young adults who are just entering the labor force; 46% of rural 15-24 year-olds, and 20 % of urban youth in this same age group fall into this category. The problem is also concentrated among the poor: among the poorest income quintile, functional illiteracy among 15-24 year olds rises to 39% in urban areas, and to 45% in rural areas. These figures reflect widespread problems in the level of educational attainment as well as quality of education services, which lie at the root of Ceard's weak indicators in educational outcomes, which in turn are a key factor in the state's potential for successful poverty reduction26. Chart 2.8: Ceari - Illiteracy Rates, by Geographic Region, 1999 (%) 100.0 82.2 80.0 60.1 60.9 60.0 42.1 44.0 40.0 7177ii 29.8 40.0 [ 't ,n278 20.0 ., 14.3 Total RMF Urbano Rural OTx. Analfabetismo (15+) OTx. Analf. Funcional (10+) 25 Source: IBGE/Census (1970) and PNAD (1999). Further, Ceara's primary school enrollment rates were below average for the northeast in 1970, but by 1999 they had surpassed that average; this result suggests proactive and successful policy in this area by the state government. 26 See Section V. Rates of child labor are also significant, especially amongst the poor: in the bottom two rural quintiles, over 10 % of 5-9 year olds and half of 10-14 year-olds are working; in the bottom two urban quintiles only 1 % of 5-9 year olds and 12-14 % of 10-14 year-olds work. 17 2.28 Table 2.5 examines the impact of increased education on per capita household incomes, . based on a multivariate analysis that includes employment variables and other characteristics as determinants of income.27 . The analysis yields two results of interest: First, after controlling for other factors, having up to three years of education has virtually no impact on per household income; and second, there are sharply rising household incomes for people with 7-10 and especially 10-14 years of education, with a more modest additional gain for 14-15 (which is probably an anomaly in the data). Tab[le 2.5: The IRelationship bRw@en Educaion an d Per Capfta DHouisehold lncomre Years of schooling Increase in income associated with incremental education (in %) 0-1 0,03 1 - 3 0,89 3-7 11,42 7 -10 25,21 10 -14 81,67 14 -15 24,11 Source: IPLANCE estimates. HD.5.0 Sector oD Economic Activity 2.29 Table 2.6 examines the effects of sector of employment and of labor force status on incomes. The results show that after controlling for education and other individual characteristics, working in agriculture is associated with substantially lower incomes relative to industry: 37% less for all of Ceara, but notably 14% less in rural areas. Mean incomes are even lower for people in these sectors, since educational status is on average also worse. Employment in the service sector is associated with higher household incomes than industry, again in urban and rural areas. This result suggests that industrial employment is on average relatively unskilled compared with services, and that the higher labor productivity measures reported in Table 2.6 below, predominantly affect much higher levels of capital-intensity (and capital incomes). 27 This is not a rate of return to education, which would have to use individual income data; it is rather a measure of the welfare in terms of household per capita incomes of those with education (and later, other characteristics), thus reflecting the demographic characteristics and earnings of others in their household as well. 18 Table 2.6: The Effect of Sector of Employment on Per Capita Household Incomes (relative to industry) Ceara RMF Urb. Nao RMF Rural Agricultura -36,51 % -31,63% -35,26% -14,33% Servi,os 13,85% 11,83% 14,58% 42,75% Outras atividades 16,63% 26,22% -14,58% -58,02% Source: IPLANCE estimates. 2.30 The relationship between education and sector of employment is examined in a different way in Table 2.7 (which gives simple averages of per capita income). The results illustrate that level of income is influenced by both by sector of employment and by educational status (though notably, higher education of individuals in farming is not associated with higher per capita incomes up to the completion of basic education). Table 2.7 Average Per Capita Household Income for Individuals of 10 or more, by Education and Sector, Ceara Anos de estudo Setor de Atividade Agricultura Industria Servi,o Sem instru,co ou com at6 3 anos de 85.63 175.93 177.76 estudo Fundamental incompleto 69.14 217.94 219.72 Fundamental completo 66.90 246.76 322.60 Medio 157.29 514.91 509.38 Superior 1,787.51 2,475.75 2,024.29 Source: PNAD, 1999 2.31 Surprisingly, individuals in the informal sector live in households with higher per capita incomes than those with a carteira (formal sector job) on average, after controlling for their othdr c aracteristics, including higher levels of skills (Table 2.8). In Metropolitan Fortaleza the same result applies to the self-employed. As expected, employers live in significantly richer households, and domestic workers in substantially poorer ones. Finally, other things equal, men live in households that are 12 % richer than do women, with this being especially marked in rural areas (Table 2.9). This is of course only a partial measure of gender differentials in poverty and welfare. Table 2.8: The Effect of Labor Force Status and Per Capita Household Incomes Ceara RMF Urb. Nao RMF Rural Sem carteira 7,38% 0,54% 6,85% 13,41% Conta pr6pria -2,83% 4,50% -1,34% -12,48% Empregador 112,66% 113,52% 149,62% 62,25% Empregado domestico -17,73% -15,78% -18,15% -35,46% Source: IPLANCE estimates. Note: Results indicate income differentials relative to formal sector employees, or "com carteira". 19 Tale 2.9: Premimg Man Reg7ti9 e tD9 o Womien in Pair Capita eiD Ohid oe I Ceara RMF Urb. Nao RMF Rural Men 12,44% 2,17% 12,79% 17,02% ID(S CaoncusDon 2.32 Income poverty is deep and inequality very high in Ceara. Both of these factors are due to historically major differences in productivity between types of economic activity, along with large inequities in the distribution of education as well as in ownership of land and other productive assets. Recent growth has had only a modest impact on poverty, and none on inequality. The expansion of basic education services and infrastructure appear to have only a modest effect on incomes of the poor to date. 20 Annex to Chapter 2. How multiple dimensions of poverty affect poverty trends-preliminary results,1993-1 999.28 Background and Purpose This Annex explores further the relationship between income poverty and other dirriensions of well-being. As discussed in the main report, despite political stability, investment-led growth, improved social policy, and sound public governance throughout the 1990s, Ceara has found that levels of income-poverty have remained stubbornly flat, after falling significantly in the early to mid 1990s. Figure Al shows headcount poverty for Ceara, the Northeast (NE) region, and Brazil. It can be seen that despite catch-up in Cearai in the early 1990s, lately its trend on this measure (relative to a poverty line worth R$65 in 1996 in Sao Paulo) has been flat, as elsewhere in Brazil. At the same time, Ceara has made progress on other measures of social welfare, in particular related to health and education (illiteracy is shown in Figure A2). The subject of this note is therefore to what extent income poverty is a fair measure of progress, or, conversely, to what extent building in money-metric measures of other social indicators will change the picture. Figure Al: Poverty Headcount since 1985 (IBGE/PNAD) 70 0- - = . 50 - .1~~~~0 V.~~~~~1 10, - . I-[ Brzi 2 20-.;. ,. R <>,X W §1 s 41.9. .'58.7 ~ ~ ~ 41. 1° 8 - q X 1988 4- -1990 St| 1993 .l 19O15i NS997z ~Bl _6 4. __1 33~7~1 i _ _ _~v4 '- ..~Caran 58.6 A86. 7 7.2.~ 46 47 Income Growth versus Poverty Reduction As discussed in the main report and chapter 2, Ceara has been a well-managed state in the 1990s. The state has followed an economic regime that could be described as one of public investment- and industry-led growth, and this has manifested itself in a fairly high growth rate compared with other Brazilian states throughout the 1990s. Using 28 This chapter was prepared by Mark Thomas, World Bank. 21 IBGE regional GDP data from 1995-99 (deflating using the IPCA consumer price index), Ceara performed slightly above average among NE states with a real growth rate of 5.4 percent (fifth among nine NE states). Nationally (excluding the Northem region which has grown faster through immigration and extraction), Ceara ranked seventh out of 19 states for GDP growth measured this way over 1995-99. Figure A2: lliReNracy (percent of ov@r 15s, R NAO) 20 Perhaps a more reliable indicator of welfare is household income. Figure A3 maps household log-per-capita income growth against initial income for the period 1993-98 and shows that Ceara is a high performer both in absolute growth (ranking third of 19) and controlling for initial income (it lies slightly above the line of best fit). Figure A3: DnitUMO Dncom3e and Ilncom3e Groh, 1@93-98 (BOEMPNAD) .6 - PB CE P RN ES RJ 20 MA X .4 -Nos E SP B SE SGF' .9 PR ~ c 3- MS SC ' RS 3 4 5 6 initial nman log incomne 1993 To link this growth with poverty reduction, Figure A4 charts income growth among the poor against average income growth using the same data as Figure A3 (PNAD), and shows that in 1993-98 Ceara performed well in channeling income growth to the poor, 22 again both in absolute terms (ranked third overall) and controlling for its average income growth (it lies well above the line of best fit for the data). This is an important finding, since it contradicts the often-heard thesis that the somewhat industrial pattern of growth in Ceara has not been pro-poor when compared with other states in Brazil. Figure A4: Income Growth for the Population and for the Poor, 1993-98 (IBGE/PNAD) 6 CO ~~~~~~~~~PE P 2 19S3 : ,...1995 i 1997 : .;1999,f, 29 Figure A13: Preliminary SimufiDon (poverty MDae $ incudng one primary pupD) '' 3 Source- IBGE/PNAD and 30 as st,r- .ILA Fiur N4 Peawrlr $uaSRun pw@eyu@$$@ $9UXBrEl830)1) Figure Al15: Cearei Simulations 6090 '7 ,4 Figure A17: NEzi Simulations 40% - 30%~ ~~~~3 LroFimEiny Coy c§uZiona It should be emphasized that this work is preliminary, and that the presentation glosses over a number of technical subtleties to which the results may be sensitive. Nonetheless, it seems that introducing educational transfers into household income measures does significantly change the time path of poverty headcount measures. The order of magnitude of the difference between poverty simulations incorporating educational access and baselines ignoring it seems to be of about three percentage points, although this quantitative conclusion may change in subsequent refinements of the work. Figure Al8: IEnroGGments among Rhoe PQoir and M!on-Por (Aged 6-1 6) n~~~~~~~ 11 ,'' _ _ _ _ _ _ _ _ _ _ ''- p i i Finally, these results may be clarified by returning to education statistics in CearcA and Brazil over the period considered. Figure Al18 shows school enrollment rates over 1993-99 in Ceard and Brazil, but distinguishes between the poor and the non-poor. For increased access to have a net effect on poverty after readjusting the poverty line to reflect the additional implied household income, it should be the case that enrollment has expanded faster among the poorthan among the non-poor. Figure Al8shows that, according to the PNAD, this has indeed been the case in CearcA and in Brazil, with particularly strong catch-up taking place between 1995 and 1997. Two extensions are conceptually easy to discuss. The first concerns increases in educational spending per student. A 2001 joint study for Ceard by IPLANCE and the World Bank reports real spending growth in "human capital development" (mainly health and education) of 36 percent for 1990-95 and 21 percent for 1995-2000 (for a total of 65 percent for 1990-2000), while in 1990-1999 actual enrollment increased by 42 32 percent. Calculating annualized growth rates and assuming that education spending moved in tandem with this wider spending category, these numbers imply a real spending increase per student of 2.4 percent for 1990-95 and of zero for 1995-99. It is therefore unnecessary to build these spending changes into the simulation for Ceara to state that they will not change the general pattern of welfare changes since 1995. The second extension concerns other social spending categories, in particular health, which makes up a larger share of State spending than does education. Here, the main problem is a lack of information on beneficiaries. Anecdotal evidence for NE Brazil generally suggests that the incidence of public health spending is fairy similar across income brackets, that is, spending is not strongly biased towards the rich or the poor. However, unlike children, who are clearly identified in the PNAD as benefiting from education spending (i.e., attending school), health beneficiaries are not identified over time in the PNAD. Data from IPLANCE suggest that the state of Ceara spent approximately R$106m in 2000 on its main health programs. The population of Ceara is about 7.4m, so equal incidence across the income distribution would imply a transfer of about R$14 per person per year, or just over R$1 per month. As above for education, we may also assume that health spending grew with human capital development spending at an annualized rate of about 4 percent in 1995-99, or about 2.5 percent per capita. At present we do not have information on changes in the incidence of this spending (i.e., (2) or (3) from our list on page 5), and it becomes clear that using changes in the State's real health spending per capita for 1995-99 is not going to change by much the story told by Figures A12-A14. This approach understates health spending considerably, however, by leaving out federal and municipal transfers. Brazil's total health spending is of the order of R$25b or more than R$100 per inhabitant. It therefore remains an open question to what extent changing patterns in health spending may affect welfare and poverty measures. It may well be that measuring and valuing outcomes rather than spending (and attaching money values to those outcomes) will best capture the welfare effects of health services in a sensible way. Basic infrastructure provision such as electricity, water and sanitation, and garbage collection are also measured in PNAD, can be priced fairly easily, and tend to concentrate their marginal effects among poorer groups. Incorporating these measures into welfare should probably take precedence over health care effects, for this reason. Figure Al 9 gives some penetration rates of different services within and outside the poor population for Ceara. Figure A19 suggests that there are significant welfare effects to be modeled here (particularly for drainage/sewerage). 33 A health module was included in the 1998 PNAD, which is presently being analyzed for the case of Bahia. Results are not yet available, but since changes in the incidence of spending across time periods are not recorded, all measured welfare changes must still come from real spending changes. 33 1FIure Al 9: InDfrasftructure ServDces, PVoor and Man-Poor 1995-99 (DBIEIPAD) - _ - _==- Among the Water 78 82 (4.6) 82 (0.2) MonrDPoor Dr Dainaj9 L 4 7 Y 48 L_ff Garbage 72 77 (6.1) 77 (1.1) Among the Poor Water 36 42 (18.0) 44 (4.9) j' rain Je * 13 3 C L4 L 17 ) l Garbage 30 37 (22.4) 38 (2.8) I ledeOnt I'r L I 72 ''-L '( > I so (245 Our discussion so far should be taken as a progress report on exploratory research. Clearly more can and should be done on the topic. But the results here suggest paths to take and limitations. First, education, since changes in incidence are well measured and concentrate resources on relatively few individuals, will probably dominate most other social spending in welfare calculations. This remark should be qualified in two ways. Measuring and valuing improvements in health outcomes may give a wide range of estimated effects depending on the methods used. And improved service delivery has been focused in poorer segments of the population, implying that there should also be additional poverty reduction effects from this source. Second, the education effects themselves do seem to be significant. Of course their magnitude depends on the monetary value one imputes to them, but for reasonable ranges (and reasonable compensating alterations to the poverty line) the results seem noteworthy. Third, in making choices about which welfare components to analyze, we will logically try to select those where the improvements have been most marked: to do otherwise would be to expend unnecessary effort for little likely insight. However, the same logic should apply to any facet of social welfare that has deteriorated over the same time period: to do otherwise could invite accusations of bias. What might such facets be? Crime or violence suggest themselves by their high profile in public debate in Brazil. These phenomena are hard to value in money terms, but a fuller analysis than the one here might try to include worsening as well as improving indicators, however crudely. Fourth, the work, involving multiple simulations across a series of distinct household surveys, is technically a little complicated. This sometimes makes the results hard to interpret, and necessitates investigation of cause and effect that have by no means been exhausted here, even on the limited subject matter we have admitted. For instance, what is driving the changing patterns of education-related welfare effects between the periods 1995-97 and 1997-99? Poverty reduction occurs faster 1997-99, 34 whereas educational catch-up among the poor is faster in 1995-97. The answer lies in a more detailed understanding of the simulations, beyond our scope here. Finally, the early discussion of pure income measures suggests that in states such as Ceara, where poverty is deeper than in other states on average, measuring changes in the poverty gap (Pi) rather than the poverty headcount (Po) may do more justice to the effects of poverty reduction policies. 35 HUODL l2ronucDon 3.1 One of the major challenges for understanding poverty reduction is documenting the dynamic process of how individuals and families move out of poverty, and how public policies influence this process. In Ceara, as in most places, the information base for analyzing this process is weak. This section develops a schematic analysis of the relationship between poor groups, their exit strategies, and the policies and programs that best support their exit from poverty. We briefly describe the principles of an ideal approach, and then lay out the pragmatic approach that was actually followed given the limits of existing information. A major message of this report is that the process of ongoing diagnosis and continuous evaluation and redesign of policies is more important than a one-time assessment. Thus the approach developed here can be steadily fleshed out as more diagnostic information becomes available. 01102 From [Povery ProfiDee to 13of00e and Progrmz 3.2 The underlying idea is simple: the poverty profile provides a snapshot of the structure of poverty (who is poor), yet all households are in the midst of a dynamic process, which for the poor, focuses on efforts to become non-poor. Policies and programs can potentially influence this process, accelerating the transition out of poverty.34 The poverty profile relates income levels to various characteristics of the poor, including geographic location as well as individual characteristics such as sector of employment, demographic and educational status of household members, etc., as presented Section II. Movements out of poverty are then driven by the choices made by individuals and households (regarding, for example, employment activities, migration, and the acquisition of education), by changes in the economic and social environment, and by public policies that influence those choices and that environment.35 3.3 The pathways of causation are complex, both because of the intricacies of the decision-making process of individuals and households, and because the effects of policies and programs are embedded within a whole range of influences on these decisions. To begin disentangling these factors, it is useful to distinguish between two aspects of the diagnosis: 34 The focus here is on income poverty, but this approach could equally be applied to the structure of deprivation along other dimensions of well-being-health status or skills, for example. And while the focus here is on the poor, the approach applies equally to all population groups. 35 A particular dimension of this inter-relationship concerns the importance of adverse shocks: of weather, economic decline, ill-health, and/or loss of a breadwinner. Individuals and households actively respond to such shocks, often drawing on relations with other households in a community. 36 o Identification of the processes of change in income in relation to different household characteristics.36 o Diagnosis of the relative influence of policies and programs on these processes and decision, in the context of other external influences. 3.4 In implementing these steps there is a large gap between the ideal and the feasible, given the state of existing information. For the first step, the ideal situation would be to have detailed information on income changes for different groups in the population that could be related to their individual characteristics and to their environment. Information that tracks the same individuals or households (panel data) supplemented by participatory techniques to assist the interpretation of patterns of change and choices is ideal. In practice, the readily available information is on general trends-and does not track the same people over time. For the second step, the ideal information specifies the impact of policies and programs on different population groups. There are statistical techniques separating the influence of an intervention from the range of other factors, such as the characteristics of individuals or households and other external influences. This is the standard approach to impact evaluation.37 However, there are virtually no evaluations of this kind available in Ceara, and in general the data that would be required to conduct such an analysis are not routinely collected, nor built in to the process of project design' and implementation. Turning to policies that affect the whole economy, the challenge is even greater, since there is no simple way of determining what would have happened in the absence of the policy change. Before- and-after analysis, comparison with other economies and models that can simulate the effects of different factors provide some answers, but there is no clear-cut solution. 3.5 The pragmatic approach developed here is based on the following available information: e data on the structure of poverty from survey evidence (as outlined in Section 11), using IPLANCE's poverty line of half a minimum wage in 1999; along with judgments over dynamic changes and processes from what is known about past patterns in Ceara and international experience; * Evidence on the incidence of programs from available analysis from Brazil's Living Standards Measurement Survey-the PPV (see box 3.1), coupled with information on the design, and intended impact of existing programs, and international experience with other like programs, to form judgments over the probable impact of policies and programs. 36 We are particularly interested in transitions out of poverty for different groups in the poverty profile. For many purposes the individual is making decisions, but the household is the primary unit of analysis for poverty changes (and indeed the household is to some extent a decision-making unit). 37 See Judy Baker, 2000, Evaluating the Impact of Development Projects on Poverty, for an overview. 37 Bao 3.1: Assessing TeRin9g iand Effec&Dens$ a godagU IPr ogr2el in 3V&M538 This analysis illustrates the incidence of different categories of public spending on the poor in Brazil in terms of the proportion of the poor reached and the extent to which programs are targeted exclusively to the poor. Each bubble in the figure represents an area of social spending. The size of each bubble represents the estimated spending per beneficiary household and, thus, the relative importance of the program for beneficiaries. The horizontal position of the bubble represents a program's targeting to the poor (share of spending to the bottom 20% of the population) while the vertical position represents reach among the poor (share of the bottom 20% of the popualtion who are program beneficiaries). For example, public kindergarten education is roughly at 0.4 along both the horizontal and vertical axis, indicating that the program is approximately 40% targeted to the poor and accessible to approximately 40% of the poor 1 (DGeneral Investment Growth of 4% p.a. Public Health Primary Education 0.8- 0 0~ 0~~~~~ a 0.6 Water 0 Merenda Escolar E 0.4 C Kindergarten Soci Assistance JSewerace X 0.2 X Xensions Unemplo Secondary Educatin LOAS/BPC Insura -Adult Education Creche 0~ ~ ~ ~0.2 0.4 0.6 0.8 1 Targeting to PCoo Source WB staff estmates 3.6 Tables 3.1 and 3.2 summarize the results on exit strategies for major groups among the poor, for both urban and rural areas.39. The tables classify the different poverty groups-the poverty profile-in rows, organize the principal exit processes along the columns, and then identify major programs or interventions in each cells. 38 Source: World Bank. 2001. Attacking Brazil's Poverty: A Poverty Report with a Focus on Urban Poverty Reduction Policies. Washington, DC. 39 Note that these tables are designed to illustrate the two steps in one summary format-in order to present the results in an economical fashion. This would really require three dimensions to show the processes against the different poverty groups and then the potential impact of programs and policies on these processes. 38 3.7 Programs and policies are then classified into categories as "primary" and "secondary" based on judgments on their expected impact on poverty reduction. Building on the criteria set out in box 3.1, programs are categorized as being of primary importance to poverty reduction if they meet the following three criteria: (a) high impact on poverty reduction (significant positive effect on the permanent incomes of the poor); (b) high coverage among the poor; and (c) cost-effectiveness (large income effect for the poor per $ spent). Programs that have a high impact on poverty reduction (criteria a) through non-income means, or by providing short-term income protection, are also categorized as primary but indicated in italics. Programs of secondary importance are those that meet only two of the three above-mentioned criteria.40 111.2.A Rural Poverty: Exit Routes and Associated Policies 3.8 Breaking the rural poor into four main groups (by age, and, for the working population, by main sector of economic activity), Table 3.1 identifies the set of priority programs for poverty reduction for each population in that group. These programs are categorized across five main exit strategies from poverty for the rural population. These include: raising small farm productivity (incomes), raising commercial farming employment, raising rural non-farm employment (incomes), increasing human capital, direct income transfers, and migration. 3.9 Rural Children and Youth (ages 0-18): There are 900,000 poor rural children and youth - representing 24% of all Ceara's total poor41. This group thus represents one of the single largest groups among the poor in Ceara, second only to urban children. Further, this population group has an extremely high poverty rate of 81%, significantly higher than that for urban children (54.8%), and in fact evidencing the highest poverty rate among all population groups in the state. Programs which are of primary importance to facilitating this group's exit from poverty would be ranked of highest priority for poverty reduction, given the pre-eminence of this group among the poor in Ceara. 3.10 The main exit strategies which promote movements out of poverty for this group - and associated programs of primary importance - include: (a) increased human capitatl pre-school and basic education (including secondary schooling); (b) raising small farm or rural non-farm incomes, (i.e the income of these children's parents, in order to move their dependents out of poverty); (c) income transfers- Bolsa Escola- type programs which provide income support to families with poor children in exchange for educational attainment; and (d) migration. Programs of secondary importance 40 The results in Box 3.1 provide a guide to who is receiving the benefits from different categories of policies and programs in Brazil. This analysis captures the incidence of public spending, and not the dynamic impact on processes of change, and so constitutes a point of departure for forming a fuller assessment of effects of these programs. However, it is important to note that incidence analysis is no substitute for ex post evaluation of effects of interventions. 41 As noted in Section III, this figure does not make an allowance for differences in consumption across individuals or economues of scale in household consumption. 39 include other income transfers directed to the rural poor, such as old-age pensions (i.e. Previdincia Rurao, of which a minority of people in this population category may become indirect beneficiaries. 3.11 Rural amrmers, Ages 164: There are 500,000 poor small farmers of working age in Ceara, representing 13.5% of the total poor population42. The poverty rate among this group is 78.7%. The primary exit strategies from poverty open to this group are likely to include: (a) raising small farm productivity, which can be facilitated through programs providing basic training, input and community development (e.g. Projeto Sao Jose), land reform, irrigation for small farmers, etc., and (b) income transfers such as Bolsa Escola type programs, which can provide permanent income support, and risk management programs such as drought relief, which can provide temporary protection from income fluctuations. Strategies of secondary importance to adults in this category include increasing human capital (adult literacy programs can be effective in teaching basic skills, but are of limited impact in actually raising incomes among this age group, thus resulting in low cost-effectiveness in terms of poverty reduction); other income transfers directed to rural areas (of which they may only be indirect beneficiaries); and migration-information is not available on the migration propensity of this group, so the classification of this process as secondary is somewhat arbitrary. 3.12 Rural workers (farm arnd non=farm), ages 1@964: this group numbers 250,000 people, representing 6.7% of the Ceard's total poor43. The poverty rate among this group is also high, at 78.7%. Exit strategies of primary importance for this group are likely to include: (a) raising rural non-farm employment and incomes, including programs directed towards increasing small and medium farm surpluses, and investments in rural infrastructure; and (b) income transfers (Bolsa Escola type). Secondary exit strategies, similar to poor farmers, include raising human capital and other income transfer programs, including old-age pensions and drought relief, of which this population group may be indirect beneficiaries though coverage will be low. 3.13 RAural e1idely, over age 65: This group numbers only 46,000 people, and represents only 1.2% of Ceara's total poor. This population group also exhibits a relatively low poverty rate of 30.8%. Income transfers (such as existing Previdencia Rural) represent the only primary means of exit from poverty among this group. 3.14 AtI rural poor: the rural poor total 1.7 million, and represent 46% of Ceara's total poor. Increasing well being and health through provision of basic water and sanitation services is a priority policy intervention for poverty reduction, facilitating the exit from poverty through non-income means. 42 Rural farmers and dependents total 1 million, and represent 27% of all poor. u Rural workers and dependents number 520,000, and total 13.9% of all poor. 40 Table 3.1: Rural Poverty - Exit Strategies and Associated Policies Exit Strategies: Population Raising Raising Raising rural Increase Income Outmigration Group small farm commercial non-farm human Transfers productivity farming employment capital Number of (incomes) employment and Incomes poor, (% of total poor in Ceard) _______ Rural Primary Primary Primary Primary Primary children & (a significant (a significant Pre-school, Bolsa Escola (education youth (0-18) share of rural share of rural basic and facilitates) children will children will secondary Secondary 900,000 become find rural off- education Rural (24% of all farmers) farm work) Child- health pensions poor) programs (indirect beneficiary) Secondary _____________ Tertiary educ. Rural farmers Primary Secondary Secondary Primary Secondary (ages 19-64) Programs for Adult literacy Bolsa escola small farm Drought relief 500,000 productivity (13.5% of all Land reform Secondary poor) Rural Indirect Infrastructure beneficiary of Irrigation for rural pensions small farmers Rural Secondary Primary Secondary Primary workers Irrigation for All policies See box Bolsa escola large farms affecting above 250,000 Crop-specific small & Secondary (6.7% of all programs medium farm Drought relief poor) Labor market surpluses Indirect reform Rural beneficiary of infrastructure rural pensions Rural elderly Primary (over 65) Rural pensions 46,000 (1.2% of all poor) Total Rural Primary Poor Water and sanitation 1.7 million Health poor services (46% of all poor) 41 119112.I3 Urban Poverty: Exit loutes and Associated Policiez 3.15 Table 3.2 summarizes the judgments made over urban exit strategies and the associated programs which are of primary importance for poverty reduction among major urban population groups. The five main exit strategies include: increasing formal sector jobs for unskilled and semi-skilled workers, raising employment and earnings for small business owners, increasing human capital, bringing excluded groups into the mainstream, and income transfers. 3.16 Urban children and youth, ag@es 018: There are 1.1 million poor urban children and youth, making this the single largest group (29%) among Ceara's poor. The poverty rate among this age group is 54.8%. Policies directed towards facilitating the exit from poverty for this group are thus of highest priority for the state, given the size of the group. The main exit strategies and supporting public policies for this group are similar to those for rural children, and include: (a) increasing human capital pre-school and basic education (including primary and secondary schooling); and (b) income transfers--Bolsa Escola-type programs that provide income support to families with poor children in exchange for educational attainment. Programs of secondary importance include tertiary education, as these services at this point largely accrue to the non-poor". 3.17 Urban formal sector workers, ages 19=64: Poor formal sector workers number just 58,000, and represent only 1.6% of Ceard's total poor population'5. Further, the poverty rate for formal sector workers is relatively low, at 32%. The existing number of poor in the formal sector suggest that this group is a low priority - in terms of targeted groups for poverty reduction policies, but also indicates that access to formal sector employment is a primary means for exiting from poverty. Primary exit strategies - or means of broadening access to the formal sector among today's poor -- of primary importance include: (a) increasing formal jobs for unskilled and semi-skilled workers (hrough support for labor-intensive industry); and complementary policies include (b) income transfers (such as Bolsa Escola). 3.18 A broad range of programs are classified as being of secondary importance to poverty reduction for this group, and represent attempts to increase formal sector job opportunities and/or incomes which have limitations in terms of cost-effectiveness or coverage. These include programs such as fiscal incentives (which appear to be of low-cost-effectiveness to poverty reduction), minimum wages (which while raising wages among the semi-skilled, can exclude the unskilled from the formal sector - thus producing low coverage among the poorest), and training programs, for which international evidence indicates low-cost poverty reduction impact.46 44 While subsidies to tertiary education are not cost-effective from a poverty reduction standpoint, targeted demand-side interventions, such as scholarships for the poor to attend tertiary education, would meet all three criteria for primary importance. See section V.3 45 Poor formal sector workers and their dependents total 240,000, and represent 6.5% of all poor. 46 Dar and Tzannatos, 1999. 42 Table 3.2: Urban Poverty -Exit Strategies and Associated Public Policies Exit Strategies Population Increasing formal jobs Raising small firm Increasing Bringing Income Group for unskilled and employment and human excluded Transfers semiskilled earnings capital groups Into Number of the poor, (% of mainstream total poor) Urban Primary Primary children Pre-school, Bolsa (ages 0-18) basic Escola education 1. 1 millmon Child- health (29% of all programs poor) Secondary terciary educ. Formal sector Primary Secondary Primary workers (and Support for labor- See box below Bolsa dependants) intensive industry Escola Secondary 240,000 Fiscal incentives Secondar (6.5% of all Labor market policy y poor) (minimum wage) Unemploy Urban Infrastructure ment Training programs insurance Informal Secondary Primary Secondary Primary sector See box above Support for small Adult literacy Urban workers (and firms: investment workfare dependants) environment, Bolsa regulationAicensing Escola 1,2 million Land titling in (30% of all favelas poor) Slum upgrading Microcredit Urban elderly Primary (over 65) Urban pensions 57,000 (1.5% of all poor). All urban Primary poor Water and Sanitation 2 million Health (54% of all services poor) Special urban Secondary Secondary Primary "excluded" (will benefit from greater (will benefit from Programs to groups job opportunities) greater job bring these Child opportunities) groups into prostitutes work and community 43 3.19 UJrban inlormaIl sector workers, ages 19-=4: the number of poor in the urban informal sector totals 785,000, representing (21%) of all poor in Ceara -- the third largest category among the poor as categorized here. The poverty rate among informal sector workers, at 50.2%, is also substantially higher than for formal sector workers. Exit strategies of primary importance for this group include (a) raising small firm employment and earnings (through programs to facilitate the investment environment, such as microcredit), as well as targeted infrastructural investments (urban slum up-grading) and land-titling; and (b) income transfers programs for both permanent income support (Bolsa Escola), and risk management programs such as urban workfare which can provide short-term income support as temporary protection from unemployment spells. 3.20 UOrban elderGy, over age 65: The urban elderly poor number 57,000, representing just 1.5% of all poor. This population group also exhibits the lowest poverty rate among those categorized here, of 20.2%. As in the case of the rural elderly, Income transfers represent the only primary means of exiting from poverty for this group. 3.21 AIJ uirban poor: The urban poor total 2 million, and represent 54% of Ceara's total poor. As with the rural poor, increasing well-being and health through provision of basic water and sanitation services is a priority policy intervention for poverty reduction for all of the urban poor. IJD.3 Priority Programs for Poverty RAeduction O Ceant 3.22 The challenge for government is to identify where it can most effectively intervene to hasten the pace of poverty reduction because of failures in markets, or by supporting positive behavior among households, communities, firms and other agents. We draw on the above analysis to generate an initial list of categories of priority programs for poverty reduction in Ceara where there is a case for additional government intervention, ranking these categories in terms of importance according to the size of the group of poor to whom they are directed; in other words, priority programs identified below are likely to have a strong, positive impact on permanent incomes among the poor, can achieve high coverage among the poor, and are cost- effective: o Income transfers (rural and urban children/youth, and working-age poor) o Pre-school, primary and secondary education, and child health (rural and urban children/youth) o Programs to raise urban small firm productivity and incomes: microcredit, other investment environment (urban informal sector and dependents), and risk-reduction (urban workfare) o Programs to raise small farm productivity and incomes: e.g. Projeto Sao Jos6, land reform, and risk reduction (drought relief) (rural farmers and dependents) 44 e Support for labor-intensive industry (urban formal sector) * Programs to raise rural non-farm employment and incomes (rural workers) * Old-age pensions (rural and urban elderly poor) 3.23 It should again be repeated that this analysis is based on limited available information, without the benefit of historical analysis of past shifts in the structure of incomes and well-being and of impact evaluations of specific programs. Work in both of these areas would strengthen the basis for future policy formation within the overall approach taken for this report. With this caveat, the above results can be used as a general template against which to assess Ceara's current allocation of state resources and composition of programs against the ranking of priority programs that emerge here (Section IV). 45 W.0 The CuvvsM Powey Rsd ucUn &T082gy 0 n Caud - MMU DW1 Onoduudcon 4.1 The essential elements of Ceara's current poverty reduction strategy - economic growth, investment in public services, and protection from income volatility and basic support (transfers) for the hard-core poor - are described in Consolidando o Novo Cearl: Plano de Desenvolvimento Sustentavel 1999-2002 (Estado do Ceara, 2000), and broadly reviewed in the World Bank State Economic Memorandum for CearS (World Bank, 2000). This section builds on that analysis with, first, an analysis of state expenditures in light of the goal of poverty reduction (state spending over time and across major policy categories); second, a mapping of existing programs against the priority set of programs that emerge from the analysis of exit strategies from poverty in Ceara (Section 111); and third, a basic assessment of the existing set of programs based on key indicators such as targeting criteria, coverage rates, and costs per beneficiary47. Thus this section provides a summary evaluation of Ceara's current public resource allocation and program mix from the perspective of poverty reduction. W.2 O veMeQw of IPubifO Empa9n(dRuire@s, 199)02000 4.2 Public expenditures in Ceara have been roughly stable over the 1990s as a share of GDP (beginning and ending the period at about 18% of GDP), and have risen substantially in real terms (Table 4.1), from R$ 2.9 billion in 1990 to 4.4. billion in 2000- an increase of just over 50%. In per capita terms, this represents an increase from R$472 to R$612, or a spending increase of about 30% per person in Ceara over the decade of the 1990s (Table 4.2). Table 4.1: Ceara - Real Public Expenditures, 1990-2000: Total and Share In GOP 1990 1995 1996 1997 1998 1999 2000 Total Expenditures 2,925.7 3,298.3 3,508.4 3,858.7 4,654.3 5,899.1 4,410.0 Total Exp. (1990=10 100 113 120 132 159 202 151 Total Exp./GDP 18.3 16 15.1 16 18.7 n/a n/a Values in million Reais, constant pnces (Dec. 2000) Data for total expJGDP from IPLANCE calculations in current prices - no deflated GDP data yet. Source. State Govemment of Ceara. 47 A review of federal and NGO programs for poverty reduction operating in Ceara was attempted to accompany this analysis of state programs and spending. However, information currently available provides an insufficient degree of detail, particularly on amounts spent by each program in Ceara and on coverage data, to conduct a comparable analysis. 46 Table 4.2: Ceara - Real Public Expenditures by Category, 1990,1995 and 2000 (index 1990=100) Category 1990 1 1995 2000 Reals (mill.) index Reais (mill.) index Reais (mill.) index Legislative, Judiciary and General Admin. 498 2 100 658.1 132 612.8 123 Economic Development 411.9 100 340.9 83 643 6 156 Human Capital Development 705 5 100 960.4 136 1,162.1 165 Social Insurance and Assistance 349.6 100 396 3 113 603.4 173 Housing, Public Services and Safety 178.1 100 201.1 113 388.2 218 Interest, Amortization & Municipal Transfers 782.4 100 741.5 95 999.9 128 Total 2925.7 100 3298.3 113 4410 151 Total per capita (Reais) 472 100 492 104 612 130 Values in constant prices (Dec. 2000) Source: State Govemment of Ceara. 4.3 Spending increased across all categories over the decade, but in terms of shares, was reallocated away from some categories and into others. The two largest categories of state spending are Human Capital Development and Interest, Debt Amortization and Municipal Transfers, which together absorb roughly half of all state spending (Table 4.2). In terms of reallocations, spending on Legislative, Judiciary and General Administration and Interest, Amortization and Municipal Transfers increased the least (by 23% and 28% respectively over the decade), while spending in the categories more prominently directed towards poverty reduction efforts - Economic Development, Human Capital, Social Insurance and Assistance, and Housing, Public Services and Safety increased by 50% or more. As a result of this reallocation, by 1995 Human Capital expenditures represented the largest share of state spending (29%), and remained the largest spending category in 2000 (26.7%), while spending on public officials and administration (Legislature, Judiciary and General Admin.) fell from 17% to 13.9% over the decade (Table 4.3). Thus the pattern of spending suggests, on the face of it, an increasing drive to direct resources towards poverty reduction efforts by targeting economic and human capital development for spending increases, as well as through better state management, both fiscal and administrative. Table 4.3: Ceara - Public Expenditures by Category, 1990, 1995, and 2000 - share in total Category 1990 1995 2000 Legislative, Judiciary and General Admin. 17.0% 20.0% 13.9% Economic Development 14.1% 10.3% 14.6% Human Capital Development 24.1% 29.1% 26.4% Social Insurance and Assistance 11.9% 12.0% 13.7% Housing, Public Services and Safety 6.1% 6.1% 8.8% Interest, Amortization & Municipal Transfers 26.7% 22.5% 22.7% Total 100.0% 100 0% 100.0% Source, State Govemment of CearA 47 TME M4: EST= W CESPSAi OTrMs noETMF AEM- - A!iOf 2=~ I_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _( 1t M I) DESPESA PESSOAL aismo- IFimisnCA VALOR _____________________ ______ _ __V_lJ V,% VALCI _ YALCR TOT!AL I - Dsapmms GVTJ 351,517.28 5s. 23A7M52 9.9 1,=83 0.24 587,68.3 Lsi§atho 49,736 60.1 33,074.03 39.9 - 8286.39 JuLdiirio 157,405.37 79.1 41,4e0.66 2D.8 140.54 0.1 199,00858 AbM dsiasro 144,319.55 472 160,225.84 52.4 1,246.29 0.4 305,791.68 2 - lZswclv. Ebmitin-DOO 51,719.68 4 18,313.37 .0 547,198 88.7 617,MC5 Agrticultra 30,640.60 34.7 4,8835S 5.5 52,657.34 59.7 8B,191.50 EiraeRe P c Nos Mrisas - - 124.39 124.39 Irdtria, COnm o e Seigos 2,671.86 0.9 a049.73 1.0 29257892 98.1 298,3M0.51 Trspte, CaTu*xa e Obras 17,CB0.25 8 2 8,589 4.3 182,354.50 87.5 208,410.64 Tuisno 1,324.97 6.0 1,41619 64 19,454.85 87.7 22,196.01 3- lBavw. CstWl Iwno 521,413AS M.8 ,170 7.2 5112,19 4.0 1,114A867 E&caqo 400,761.34 52.9 53,477.76 7.1 303194.34 40.0 757,433.44 Sa e Sareanto 120,652.14 352 26,70124 80 185,723.05 55.8 333,076.44 TrabE1ho 23,97&79 100 23,97.79 4- Sagwideda e AsasL Sadal 476&5M3 5.4 33,98.7 5.9 6,155.20 11.8 57347.5 Prevd&-da Sade, ( Inatiw e Sa6- 455,988.15 94.9 24,239.97 5.1 - - 48058.12 de do Ser. PcOoo) Assis&idaSocial 20,535.68 20.9 9,698.50 9.9 68,155.20 69.3 98,389.38 5 - Qmldea da fide 181,U51 M9 5%47M3 136 138,12.5 37.6 37 Habbto e Urbsrmn - 11,274.22 135 72,390.19 865 83,664.41 Go Aiental 7,771.85 16.7 5,480.00 11.8 33,187.23 71.5 46,439.09 O.Wtura, Dasporto eazer 29,Z.46 i12 1,972.94 12.1 11,387.82 69.7 16,343.2 JLaitigae Segwuarwa Pica 171,145.19 75.8 31,743.18 14.1 22,947.31 10.2 225,835.68 S - E1}3gm Ebztals 23,4L7 2.4 M395,485.9 97.6 - -% m52 Jiun - 217,221.44 100 217,221.44 Anotiza es 205,357.95 100 - 208,357.95 Trartsferdn. a lbrdidpio - - 509,886.55 100 509,886.55 2ros hagos 23,490.57 100 23,490.57 7- I daQms TOTAL 1,63 4.35 MO 11,33,157.65 _ 0 1,2. 3A5 77 n4.0 48 IV.3 Breakdown of State Spending: 2000 4.4 Table 4.4 provides a more detailed breakdown of total state spending for the 20004. Expenditures are broken down into sub-categories (columns) and across three expenditure types (rows): personnel, other administrative, and 'finalistica'49. With a view of the effectiveness of this pattern of spending for poverty reduction outcomes, the following general results emerge: * Personnel and administrative costs remain high (70% of total expenditures). * Economic development spending is centered on industrial incentives and is urban-focused. * Education spending is sizable (18% of total expenditures) but analysis in Section V indicates that it may be insufficient to rapidly breach the existing deficit in state education services; e The state employee pension program is costly to operate (11% of total expenditures, and 28% of total state personnel expenditures) and requires a heavy state subsidy (85% of outlays). Raising the fiscal sustainability of this program would release significant funds for reallocation to activities with a direct impact on poverty reduction. . The volatility of the budget cycle has been reduced, however mid-year allocations persist, some of which seem to penalize disproportionately those programs targeted to very vulnerable groups. A fair and transparent process of mid-year allocations should be implemented. 4.5 Personnel and Administrative Costs. The breakdoWn across three expenditure types reveals the predominance of personnel in the state budget (representing almost 40% of total spending). Personnel plus administrative costs together absorb 70% of the state's budget. Budget data by sub-category also suggest high personnel costs in some areas delivering direct benefits to the poor (for example, personnel costs in the sub-category of Social Assistance total 20.9% of the total budget for these programs). It should be noted, however, that in some categories, such as education and health, a large part of service delivery is in fact done by state personnel (teachers, nurses), such that high personnel expenditures in this budget categories should not be taken as indicators of inefficiency in service delivery. Further investigating should be done to determine the weight of spending on salaries for administrative staff vs. teachers and health practitioners, and perhaps using comparisons with other states, such that the scope for altering expenditure patterns across these categories can be determined with greater precision. 48 Total spending in table 4.4 for year 2000 is slightly lower (4,229 million) than that shown in table 4.2 (4,410 million), given that the former figure is expressed in current Reais for year 2000, while the latter is expressed constant Reais (Dec. 2000). 49Gasto Finalistica refers to expenditures on individual state programs, as well as expenditures on direct provision of goods and services within an individual budget category. Note that Table 4.7 provides detailed breakdown of this expenditure 'finalistica' by program, and this is what is then used in the mapping exercise. 49 4.6 EconomGc IeveDoprmenl. The cornerstone of Ceara's economic development strategy is the industrial incentives program (R$278 million, 48% of economic development spending), focused on large firms and urban areas, including urban areas in the interior of the state (World Bank, 2000). Together with investments in general infrastructure (R$ 182 million, (28%), these two categories make up about three- quarters of all economic development spending (Chart 4.1). The bulk of the remaining budget is spent on urban infrastructure (12%), and other urban development opportunities (tourism). Investment in agro-industrial development and rural development programs together totals less than 9% of all economic development spending (R$52 million). Thus one reason why the growth achieved during the 1990s has not yielded more rapid results in poverty reduction may be the apparent discrepancy between the current pattern of investment in economic development - which is largely directed towards urban industry and related urban infrastructural investments - and the current distribution of poor - roughly half of whom continue to live in rural areas. (Char M.-0 EC[OMMDC DBUMOPMeMn EKP8nDK0UP3$9 2()0100 Share of Expendltures In Economic Development Category z>su04816~~~081% 03 14% 9 '\ 304% \' \ / Development of Poor, Rural Agriculture \ Agro-lndustrial Development, Special 05o %terest 0o02% 0 Urban Industrial Development O General Infrastructure O Urban Infrastructure Oili 0 Rural Infrastructure 02 25% 4.7 Human IOeveOopment. The state of Ceara allocates 26% of its total spending to human development activities - R$ 1.1 billion in 2000, equivalent to R$ 237 per capita. These funds are divided between three sets of activities: education, which absorbs the bulk of spending (68%), followed by health (30%), and employment (2%). Education 50 expenditures total R$ 757 million, or R$ 329 per school age child, of which 80% is allocated to teacher and administrator salaries (Table 4.4)5°. In health, spending is far lower - R$ 333 million, or R$ 46 per capita5l. This strong emphasis on education spending is likely a large factor behind Ceara's rapid success at expanding access to basic education,- but a tremendous deficit in public education service provision persists, both in access to services above the primary level, and in quality at all education levels. Section V examines these issues in detail. 4.8 Social Insurance: Public Employee Pensions. Total spending on social insurance and social assistance programs is R$ 578 million (14% of public sector spending), however the bulk of these resources is allocated to the public employee retirement pension and health insurance, Previdencia Saude (R$ 480 million). As with other Brazilian states, Ceara's pension program is not financially sustainable as currently designed, with benefits outpacing contribution rates52. Costs of the system have been rising over the 1990s, and in 2000 the state financed about 85% of benefits paid out by drawing from general tax revenues. Benefits paid out through the pension program totaled R$ 456 million in 2000, representing 28% of all state expenditures on personnel in that year (Table 4.4). The portion of these costs subsidized from general tax revenues (85%) represeints 9.2% of total state expenditures in 2000. The system thus implies costly subsidies financed from general tax revenues which are by and large not directed to the poor. 4.9 Reallocations between budgeted and actual spending. Ceara's budget cycle illustrates substantial deviations between budgeted and actual expenditures,, particularly during the early part of the decade (the ratio of actual to planned spending reached 150% in 1990, for example, and plummeted to 30% by 1995). This reallocation process raises the question of how funds are reallocated across programs, particularly to accommodate budget shortfalls - which have been frequent in the latter half of the 1990s (actual spending ranged from 30% to 107% of planned spending between 1995- 2000, but was in shortfall for 5 out of the 6 years). Focusing in particular on how the de facto reallocation.process affects programs targeted to the poor, it is apparent that spending on certain programs tends to be reduced in the budgetary cycle. Some programs that direct resources to the poorest -- notably for rural basic infrastructure - had actual budgets cut to zero for year 2000. 50 Total spending on personnel in education equals the $400 million spent by the state on salaries, as noted in table 4.4, plus R$ 205 million spent on the FUNDEF program (Table 4.7) which is a transfer to municipalities for education largely used to finance teacher salaries at the municipal level. 51 are these spending levels appropriate? How do they compare to other Brazilian states and/or international averages ? 52 See World Bank. 2000. Brazil: Critical Issues in Social Security and World Bank. 2001. Brazil: Broadening the Base for Growth: A Report on the State of Bahia. 51 4.10 How appropriate is the current set of poverty reduction programs operating in Ceara, vis-&-vis existing groups of poor and the exit strategies associated with their movement out of poverty? This sub-section draws on information for over 80 programs currently operated by the state of Ceara to begin answering this question, including target group, number of actual program beneficiaries in 2000, and year 2000 budget (these data are compiled in Table 4.7).53 4.11 We draw on this information to examine basic indicators of program effectiveness that emerge from Table 4.7 -- what proportion of the target population is actually reached by a given program in a given year? what is the cost per beneficiary of the program? -- in order to draw broad conclusions about program coverage among the poor, and efficiency in service delivery. 4.12 Table 4.5 illustrates the results of a program mapping exercise where the existing set of programs and interventions for poverty reduction currently operated by the state of Ceara are matched against the set of priority programs for poverty reduction that emerges from section III (Tables 3.1 and 3.2). This exercise compares the 'ideal set' of priority programs that would support Ceara's main groups of poor in their exit from poverty with the set of programs that actually exist in the state. Program budgets are also mapped into the analysis, by program type and group to which the program is directed, in order to generate broad comparisons of amounts spent by type of program and target group. Programs currently in operation by the state government of Ceara to each target group are then mapped into the table, along with amounts spent on these programs. 4.13 These results allow some general conclusions to be drawn regarding: (a) the composition of the state's current poverty reduction strategy (which programs exist and to whom are they directed); (b) the allocation of expenditures on these programs across geographic regions and target groups; and (c) priority programs for poverty reduction, given the state's geographic and age composition of poor, that are nevertheless not currently included in the state's portfolio of programs (highlighted in the table). 53 Budget records for individual programs do not include general operating costs (personnel and administration) for the state ministry that runs a particular program; thus budgets here correspond to 'gasto finatistica' only. Thus the total budget of program spending in Table 4.7 (1,269 million) corresponds to the total for 'gasto finalistica' in the state budget for 2000, table 4.4) 52 Tlba e 45: Ceare -Programs for Poverty Reduction (Continue) I. Programs of Primary Importance for Poverty Reduction Programs that are both urban Total State Key Group Urban Programs/Policies Rural Programs/policies and rural Expenditure A. Youth a Programs targeted to * Education, primary and excluded groups - vulnerable . Education: preschool secondary (II.A.1 and II.A.3: R$ urban children & youth (III.A: 279 million) R$30 million) * Health: child-oriented * Education: preschool (R$ 8 programs (II.C.R$ 22 million) million) . Child-targeted income transfer (Bolsa Escola) R$ 38 million R$ 301 million R$ 339 million B. Working-age * Support for labor-intensive . Programs to raise small farm population industry productivity (I.A.1: R$12 million * Small and medium enterprise * Land reform(l.A.1: R$1,2 development, i.e. titling, million) deregulation, market * Risk management - drought development relief (III.B: R$ 17 million) * Credit availability . Irrigation for small farmers * Land-titling (favelas) (Caminhos de Israel - I.A.4: R$ 0.8 million) 0 R$ 31 million 0 R$ 31 million C. Elderly (raising * Old age income support * Old-age income support incomes) (pensions) (Previdencia Rural - not financed by state) D. General a Water (IV.A.1, IV.B.1: R$ 50 a Water (II.A.5 - Projeto Sao Jos6 * Water (IV.2: R$ 8 million) Population million) - Infra-estrutura - R$ 17 million) . Sanitation (II.C.3: R$ 20 million) * Sanitation (II.C.3: R$ 8 Sanitation * Basic health services (II.C.1: million) * Other rural infrastructure: R$109 million), + special health * Other urban infrastructure electrification (R$ 100,000) programs (II.C.2: 25 million) (including housing: IV.B.2: (II.A.5 - Projeto Sao Jos6 - R$ 46 million) Infra-estrutura - R$ 4,4 million) * Special programs targeted to excluded groups (III.A.1 - Assistencia a Grupos Vulneraveis - R$ 12 million) R$ 21.5 million R$ 162 million R$ 299.5 million R$ 116 million nTotd1,pendin R$ m6n t ; aiRS 5225 m * +. ; _ . , , 69, milio 53 Table 4.5, Continued 11. Programs of Secondary Importance for Poverty Reduction Programs that are both Total Key Group Urban Programs/Policies Rural Programs/policies urban and rural Expenditure A. Youth B. Working-age o Fiscal incentives for large o Irrigation for large farms, and o General infrastructure population firms (I.C.1: $292 million) crop specific programs (roads, ports, etc. o General urban infrastructure: (I.A.2: R$ 15 million) I.C.3: R$169 million) transport, communications o Adult literacy (II.A.2: R$ 6 (I.C.3: R$ 12 million) million) o Tourism Development (I.C.2: R$19 million) o Labor force skills development - training (II.A.2: R$8 million e II.B: R$ 18 million) o Employment Assistance (II.B: R$ 6 million) R$ 355 million R$ 21 million R$ 169 million R$ 545 million C. Elderly D. General Population . Source: Table 4.7. Notes: (1) Codes following each program type indicate location of program budget information on table 4.7. (2) Highlighting indicates the programs and policies identified as essential (of primary or secondary importance) to poverty reduction in Ceara according to the exit strategy exercise, that are nevertheless not currently in operation. 54 4.14 The current strategy is weighted towards policies that support exit strategies for the urban poor. Of all program spending on poverty reduction by the state of Ceara, 40.1% is currently directed specifically towards policies that support exit strategies for the urban poor. The corresponding value for the rural poor is 5.8% (Table 4.6) though they potentially benefit from expanded urban opportunities through migration. Even assuming an equitable distribution of the remaining general expenditures (55% of total spending) - which may or may not be the case - given that close to one half of the poor (47%) live in rural areas, an equitable distribution of targeted spending would suggest that close to one-half of that spending be allocated towards supporting rural exit strategies from poverty. The case for a greater emphasis on programs of primary importance to rural poverty reduction is - in particular the implementation of a Bolsa Escola-type income transfer -- is discussed in Section V. Table 4.6: Program S ending in Ceara: Summary of Results Programs that are Total Program Urban Rural both urban and Expenditure Category Programs Programs rural Programs of primary R$ 154 million R$ 52.5 million R$ 463 million R$ 669.5 million importance for (52.7%) poverty reduction (_.7%) Programs of secondary R$ 355 million R$ 21 million R$ 169 million R$ 545.0 million importance for (42.9%) poverty reduction ( ._ _ Other programs54 R$ 54.5 million R$ 54.5 million _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 4 .3 % ) Total R$ 509 million R$ 73.5 million R$ 686.5 million 1,269.0 million (40.1 %) (5.8%) (54.1%) (100%) Source: Table 4.5 4.15 The current strategy is only slightly weighted towards policies of primary importance to poverty reduction. Program spending is currently only slightly weighted towards programs of primary importance for poverty reduction (52.7%), vs. programs of secondary importance (42.9%)55. Thus the opportunity may exist for the state to redirect spending into priority poverty reduction programs. Alternatively, some secondary programs might be redesigned to place greater weight on activities of primary importance for poverty reduction. The current program of fiscal incentives - which absorbs over 50% of all spending on secondary programs - presents just such an opportunity, in that support might be stepped-up for labor-intensive industry such that 54 Other programs implemented by the state but not on the list of primary or secondary programs for poverty reduction: Cultural Development (IV.C: R$ 11 million), Public Safety: police and prisons (IV.D: R$23 million), Higher Education (IH.A.3: R$8 million), Other (various: R$22 million) 55The remaining 5% is spent on programs that fall outside of these parameters (i.e. police and public safety). 55 the program would be positioned as a priority intervention for poverty reduction among the urban, working age-population.56 4.16 Ceara's current portfolio of programs and interventions includes many priority programs (all programs not highlighted in table 4.5), but also lacks some programs of primary importance to poverty reduction among key population groups (highlighted in table 4.5). These programs are identified below: Cearg: Four priority progranm types currentRy missing or underrepresented in state poverty reduction strategy Program type Target Population o Child-targeted income transfer Poor families with children o Rural pre-schools Rural children ages 4-6 o Policies in support of labor- Poor, urban working age intensive industry population o Rural basic infrastructure (water, Rural poor sanitation) O#.5 B$asic DidDcatoar (D IPro'gTv 1Effctivenes Vov Povauy ReduUon 4.17 The program data collected also provides some general indicators of program effectiveness -- what proportion of the target population is actually reached by a given program in a given year? what is the cost-per-beneficiary of the program? -- in order to draw broad conclusions about program coverage among the poor, and efficiency in service delivery. In general, results indicate that Ceara's portfolio of poverty reduction spending includes a few very costly programs from the perspective of poverty reduction - with low coverage and therefore extremely high costs per beneficiary, combined with a wide range of smaller programs - many of which are priority programs for the poorest - but suffer from low budgets and therefore low coverage rates. 4.18 lndustrial Gncenttvas - high cost and low covargs. The goal of the program is to generate greater and/or higher paying jobs among the urban economically active population. In 2000, new firms signing agreements under the fiscal incentives program were forecasted to generate 17,150 jobs. Total jobs in firms currently receiving incentives (as a result of prior agreements with the government) is estimated at 47,560 in 2000 (2.2 % the urban economically active population),at a cost of R$ 5,866 per job created. 4.19 Rural Poverty Reduction Programs - low Davel off enpenditures vios-vis size of target population. Funding for rural small scale agriculture programs totals 56 The restructuring fiscal incentives is the subject of detailed work by the Israeli team, as a companion to this report. 56 R$35 million (6.1% of total spending on economic development), equaling an annual expenditure of R$20 per poor rural inhabitant. Of the 6 programs in this category, Combate a Pobreza Rural (Projeto Sao Jose) is by far the largest, with a budget of R$30 million. This program's coverage is relatively extensive, reaching 9.2% of the rural poor in 2000 (though cumulative coverage since the program's inception is estimated to be up to 75% of the rural population). Its impact is modest however - estimated to provide beneficiaries with a 10% increase in income - and total spending per beneficiary is low, at R$ 192 per year, such that the program is unlikely to provide an impact sufficient to raise beneficiaries and their families above the poverty line. Five other smaller programs offer virtually insignificant coverage; none of them reaching more than 2% of the rural poor in 200057 (Table 4.7). Cedula da Terra, one of these programs that has a relatively high level of spending per beneficiary (R$1,024), represents state funding for technical assistance and basic infrastructure for beneficiaries of land reform programs (land acquisition costs are covered by federal funds). A recent evaluation finds that this project has significant impact among beneficiaries (raising their quality of life as measured by the IQV (Indice da Qualidade da Vida) by 70% on average)58. However, extremely low coverage suggests that this program, as with the majority of other rural programs, has minimal impact on lowering overall rural poverty. 4.20 Expenditures in Agro-Industrial Development are concentrated in the Desenvolvimento de Agropolos program, which finances irrigation projects for large farms. The number of beneficiaries of the program in 2000 was 2,851 - but the total number of future jobs estimated to be generated via this investment totals 13,548, at a relatively high cost of over R$4,300 per job. 4.21 Basic Infrastructure Services - large investments, primarily urban. State expenditures on infrastructure are sizeable, totaling R$468.5 million, or 37% of all programmatic expenditures. The bulk of infrastructure investment is specifically urban (50.2%), with an additional 42% spent on general infrastructure programs directed to both urban and rural areas (where it is not possible to deduce the allocation between rural and urban investments from current budget figures). Infrastructure investment directed specifically to rural areas totals only 7.8% of total infrastructure spending. Water and sanitation programs are largely directed towards urban areas, and hence provide very low coverage of the rural population vis-&-vis size of the population currently without access to these basic services. In water, there are five programs which focus on water provision specifically. However, only two of these programs, PROGERIRH and PROAGUA, offer services for rural households without water - but cover urban areas, too - and coverage appears to be low (PROAGUA, for example, 57 SEPLAN has attempted to supplement this information with data on program coverage for a over the past 5 years. However in most cases these data do not provide a reliable estimate of the total number of people served by a given program, due to double-counting (many programs serve beneficiaries for more than one year); thus a simple addition of the number of beneficiaries in each year leads to overestimation of total beneficiaries over time. 58 The ICV index includes the following indicators: durable goods owned by the household, education level, health, housing conditions, and access to water and sanitation services. (Khan, Ahmad. 2000. Reforma Agraria Solidaria no CearA e Assistencia Tecnica: Novo Modelo de Desenvolvimento Rural no Estado do Ceara. Governo do Estado do Ceara. Fortaleza. 57 reaches just 2.4% of the total state population without access to water per year). Rural electrification expenditures by the state are extremely low (R$106,873); however, actual expenditures are significantly higher (no exact figure currently available), financed under agreement with the state by the privatized electricity company. Further infrastructural investment is financed with federal funds. The largest of these include the Barragem do Castanao dam construction which received an additional R$43 million in federal funding in 2000, and MetroFor (urban transportation) with an additional R$118 million. 4.22 Soca Assistance - low budget, urban-ocu's and high administratve costs. Social assistance programs currently garner a very low proportion (5.7%) of state spending (R$ 68 million, or 5.7% of programmatic expenditures, and 1.6% of total state expenditures), and all are directed to urban areas. Assistencia a Grupos Vulnerables is the largest program in this category, but covers only 5% of the target population (poor urban youth ages 0-18, and elderly poor (over age 65). Program expenditures total R$441 per beneficiary. The most costly program per beneficiary is Protecao Social e Medidas Socio-Educativas, a program for youth in conflict with the law, which costs over R$700 per beneficiary to operate. No further effectiveness indicators or evaluation results are available for these two programs. Spending per beneficiary is very low for the three additional urban social assistance programs: Apoio as Reformas Sociais para o Desenvolvimento de Criancas e Adolescentes (PROARES) (R$55), Vivendo e aprendendo / Crian9a fora da rua dentro da escola (R$46), and Crianca Feliz (R$75) - while evaluations are also not available for these programs, their poverty reduction impact may be questionable given such low expenditure levels. The one rural program -- rural risk management (drought relief) Defesa Civil Permanente operated with a budget of R$17 million in 2000, benefiting 1 19,000 people - or 12% of the rural indigent population at a cost of R$ 142 per beneficiary. 58 Table 4.7: Z9TAB67UAT1A- ESPESA POR CATEGORIA DE GASTO - Beeiiry Population ORCAMENTO ANUALT Covere& CATEGORIA/Treed 20 Estimakted 2000 DSE Spending In Targeted K~ofTar8et Beneniciarl ESE Beneiclar Target Group, _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ~~~~~~~~~~~~~~~2000 1. DESENVOLVIMENTO ECONOMICO ________547,190,016. 43. ___ I A AGRICtILTURA 52,667,342.3 9.1 Agneultora L'.''~~~~; SrmIIrt~~~isJagrs. , 171~~6....9 r..d '55'1.49.15695 0 274'0 19 .A~Fn,n(Refbrmm Rossi W.-an, Wkrs wIout 1,716.589ritral 14.265 I "28 0 ~ I 099 * ~ ~~ .,i,, , ,~~~~~ l716.SS9rural 4 *~~~' 1.50.896 0 . 227: , 3001 ,49 .7 .009 6950 04 Combat Poh~~zaRursdPi*~~V~ ~ ra1a~rtcJWkrs.W/out' I'1716,59SnuraI 15.50 .8002 192 p. A.go.-Ind.srr4aldeveIopnwU. special interest Desenvohvunento da Coton 21,400 2,43.4,659 I 0 113 7 Desenvolvunento de Agroind & pop in 6 2,851 1358,253, 2 4.334 7 Outius 2.178.045, 0 I B ENERGIA E RECURSOS MINERAIS 124,391 O3 U-veralizaces do Atendimeuto mm Energut Rural popul Wnout 1,106,35 91,480 106,873 3 0 117 839 Energia noSo 17,518 0 0 IC 1 NDtSTRJA COM&9RCIO B SER VJ(OS 292,578,922. 53.! -. 'Almmo'de Ii 'deMedioc ' ~ ~ ~ 44cjuein 6,6751 ..1 .9.4 1 , 2 1.992 4 Finane,xesitodoDesevo1vsento~,,, . ~ J tan econ.l Acttwvev,, i< 2,132.43 ' 710oprgs 1 -624(9472 S . _166.23 .08g ID WRIJSUO 19,44,56- 3.4 DesroWurlveInto do Tuns=n no Nordeste Meto Fortaleza & west 14.869.610 2 FotnentoecPrsoa,cado PEiA m 2,567 4.371,746 al 1.703 0 Desenvolyunento das Mwcrregim3 Population mn 6 80,0547 0 tummtias O,0103 '133,445 C 0 IlB TP4NSPORYE COSti,.I1C4(.4 E08&OeRS 182.364,804. 33 Rodovtanodo Estanlo dDCna',I .., State- j ;.~; .'0 634776' 19 METiROFO Metro 8,854,36 I 0utros Ptogrnnu (Telefoma. _____________ 2.903,074 0. ___ IL DESENVOLVIMENTO DO CAPITALH8UMANO _ _______512,896,181. 40.,___ II.A EDUC4ACO 303.194,333. 59.~ Primary and Swcoodary Education Desenvol,nonento do Ensin Fundamnwtal Poor clukdren ages 1.036,92 685,295 16.156.64 3 23 5 66 1 Deaenvolvtmento da Gestno Educational Cooperanvra e, oo chuikren ages 1.036,92. 22,829,798 4 FornmcaD a Valorawsco do Malpstem ~ wUrnufictt SCID c 4.202 - 3.456.711 0 822 6 Expansao e Mehsr om Infra-esmtruun, e Gesano do Poor yvutIh ages 190.30 12 600 9.749.20 I 773 7 6.61 Luertc, and trnuu,,g I ~~Cap.,t .oAitdal)ap..L.4a' Ce..ra.i "7.. ,, 2,364 38 * ~ .i% ,6.54'291 Kilt 0 CltlTeClIOgIn (CVT a Co,.v Ia !.r.s.n ~ '. 004(1 ,, ,, ,,81746 I .,19 J Ees,no~~~~~~~~~~~~~~~~~~~~~~~~~~~~I I 0 8.56590 EnLoSpnrcOutros .*.1 -5 5,401 Emnploiment.4iswsizra . t-I ., tI C SACODE ESANEA MENTO 185,723,054. 36. PubUic Heail Services 59 P NEXd1 1 ''' x,I L 5?3'' 219gA66L3 4.34 Uo oS= bdo6emo5E) Es~~~~~~~~~~~D&mxiD b W012.2613 4C0 S5=MilUb=dwMsapm bEdo U dx ppd wat I D 16S 48C pi3lW= &If S514 _. IH ASS1572ASa4L oS,155,99S S. 111A As515srimsa4 S 67,7YA5 99.4 PpVx-ffW=b7WhsesbxabgMp ' AphH.t=s - p(oRlli\dC- Rnwb=y W1W7-14 136.72D pa 759.738E6 IIo6. Vrlo de Nn1 I Cwfmdam nn Lcsda P=esma Rwrnw 7-I7 E6S 3983 is E46C2 RasiloSd e i SBa th vs C 17 s o EWBOhllm SM 6,212S9 17 9 1 m 77C7 C< Fi' bs ihbmya.06 logW)f BA6621 11. 7466 A s aGkpwuh.nod P u yuWh (Si3 kiddy 563S6px 24"S.463.65 365 441.29 iCiil' Pi r., RmDd m_fc -tn *uW 9,47 119" i5p== 17=23,9121 25i 1a424 12291 Om= g.. y__ 404,93IL95 (1 _ [VDilCI M$3943O311A P-Clmb 4523.37521 3 Damlinf Uib cG=wi Rim tidos(PRLifid) iopldioowbd5t 805.474 5Z,6W abis(emn4 14,425fil9 82 10, 27423 659i GwrmnMoe iobW-dw i~ M6d.- PROE, Umpdnoth io wto 8D5,474 6.102 43 44 imwidmofwd 9e R=ne WAz ipxs o Suam4& V= M iib I k ii i wta iiko q2wa ID V== 29CIL05 75,261 i~ 6E91.931.98 4,9 91.57 2.6% IV,B R4BffA(iiOE B t1 72,VA177.24 5i.7 UR* I 1,914 16 , '0 - i t%xr. ] ! * t '>;. , ,i; i'r, ', , - - t baUswkirwrc ; .,, : , 4621Oi , 0 15,aR46,21072949lt*.7 1131,3 33376 U~~~~~~~~~m & I v, ! e,^+, . w ;3,4i747r*Z Dfimsc V r rdcAh SDio4vseA t m.iW &q=ymqo p5016.OD 0.l OA= t . . 941.46L3( CL7 AVD JV,7AE SBg, ,NCA POUC ZZ r, , O3, I._ _I S t 2N7i14 3~~~~~~~~~~~~~~~~~~~~~~~l S!aiS c~ SdM ri Vft Fbi I ig ik'ZMai72 93 ,. NE-~~~~~~~~~~'- a The distribution of retums to work. The distribution of wages and returns to self-employment is driven by complex supply and demand factors within Ceara, within the national labor market, and from outside Brazil. There has been a tendency to widening of the wage distribution in relation to skills in the past two decades in Brazil-with a flattening of returns to low levels of education and rising returns to higher secondary and tertiary education.68 However, it is difficult to determine whether these past trends will continue. Thus the simulation assumes three possible alternatives for the medium to long-term pattern of returns to skills: (a) a reversal in the trend, with a relative reduction in differences between relatively unskilled and highly skilled workers (this happened in Korea in the 1970s, for example); (b) no change in the pattern of returns from the 1999 level; and (c) a further significant increase in differences. No probability is assigned between these three alternatives. 67 The distribution of schooling across individuals Is a function of individual and household characteristics, based on patterns observed in the data to date -in practice this involves some flattening of the education distribution. See Ferreira, F.H.G. and R. Paes de Barros, 1999, "The Slippery Slope: Explaining the Increase in Extreme Poverty in Urban Brazil, 1976-96", Revista de Econometrica, 19 (2) 66 5.14 All other results in terms of household characteristics flow from the relationships within the model, based on the statistical analysis of the past: the shifts in education influence occupational choice, including, importantly, the decision as to whether to work or not, and fertility decisions. Any simulation has to be treated as essentially illustrating what might happen if past relationships were to persist, rather than a prediction for the future. One factor that was not changed is location, since it was not possible to estimate determinants of internal migration. This is a restriction that is important to keep in mind. 5.15 Finally, it was assumed that there would be underlying income growth independent of effects of education within the model: a cumulative total of 30 percent per capita was taken (this could be 2.5 percent per annum over 11 years, or a slower rate over a longer.period-this is again intended to illustrate changes in the medium to long term). The total effects on incomes then depends on this exogenous factor plus effects of rising education, participation and fertility decisions across the distribution. The results are driven by three main effects: * Changes in the pattern of earnings driven by the combination of different education levels and assumptions on the distribution in the returns to education. * A substantial increase in labor force participation that flows from the model's results on the relationship between participation and education, especially among poor women-this generally increases the proportion of low income earners, but has positive impacts on per capita household incomes, owing to the reduction in dependency ratios. * A reduction in fertility, also associate with greater education, leading to lower household sizes, and an additional reduction in dependency ratios. 5.16 The results show that of the above effects have different consequences for the pattern of earnings and incomes. For earnings, all scenarios find an increase in average earnings, but a very small decline in overall inequality. The decline in larger, but still modest with the reduction in wage dispersion. The net effect of all the above changes on the Gini is a reduction from 0.542 to 0.512 in this case-in part because of the rise in low income earners. Nevertheless the impact on changes from 1999 are quite different across the distribution: in particular, if earnings differences do not narrow there could be a significant problem of large numbers of low paid workers in the state, even after the education expansion. These are illustrated in Figures 2 and 3. 5.17 In terms of per capita household incomes, which is what determines poverty levels, the results are rather different. The overall impact on poverty is significant, with reductions in poverty incidence between a quarter and a third. However, it is noteworthy that there is a substantial positive impact from the effect of increased participation in work and reduced fertility in addition to the effects of earnings. Indeed under all scenarios the larger part of the impact on poverty levels comes from these 67 effects: an important illustration of the potential impact of education on behavior, especially for women. As a consequence, the differences in outcomes in the labor market are relatively muted: while the effect of additional participation on earnings tends to reduce mean earnings at the bottom of the education distribution (see figures 2 and 3), it naturally increases mean per capita income (see figures 3 and 4). By contrast to the effects on poverty, there are again only modest impacts on inequality. The largest decline in inequality is from a Gini of 0.619 in 1999 to 0.56 in the scenario of reduced dispersion of returns to education. This is a non-trivial change, but would still leave Ceara very unequal. IBox 5.1: Note on MMUGonz The graphs show a subset of the simulations to illustrate the most interesting results. Each graph shows the difference (in logs) in incomes between the scenario and the original 1999 position for all households, ordered in terms of their education levels. This is done first for individual incomes and then for household per capita incomes. > Pconwave refers to the scenarios in which it is assumed that the distribution of returns to education in the labor force will become less unequal, with, in particular, a narrowing of differentials between un- and low-skilled workers and those with college education. '> pconvex then refers to the scenarios in which the opposite is assumed, with a widening in the dispersion of the returns to education in the labor force. For each case, the contribution to differences in incomes is shown in three, cumulative stages: > the first line ([JU) shows only the contribution of differences in returns to education in the labor market, where Qis a constant termpirefers to the parameters for Mincerian earnings equations, and fOcto the residuals; > the second line (y, .oo°) then adds to this the contribution of the effects of changes in occupational choice associated with education expansion, with the choice to participate in some work being the most effect here; y refers to the parameters of the occupational choice model; > the third line (r, y, QiOO) finally adds the impact of educational expansion on fertility; iV refers to the parameters of the demographic choice model. Note in particular that the effects of adding the second and third effects are quite different for individual and household incomes. For individual incomes the effect of occupational choice is to reduce gains at the bottom of the distribution. This is because the scenario is only reporting incomes of those in the workforce, and those who are joining work are primarily women, with low levels of education (but higher than in 1999) earning low wages. The addition of the fertility effects has, of course, no impact on the pattern of individual earnings. Then, for household per capita incomes, both these effects work, strongly, to increase per capita incomes, the first because in effect monetary work is being substituted for home activities, and the second because the number of children per household has declined. See Ferreira and Leite, 2001, for details of assumptions. 68 Figure 5X2: Simulation of earnings, mean of education 7 years r3concave 0,6- 0 ,5 - - - - - - - - - _ _ - - -- - -- - -- - - - - - - - - - _ __ _ _ _ _ log diff ncre3 -5!- a, 03, ca2 e O-- y, ax, P, ca2 WI y, , ,x , a2 0,1 0- 10 20 30 40 50 60 70 80 90 1 0 -0,1 percentiles 69 13 ura 5.3: muDaIn of sairnhg$, mean of e3ducSaItion 7 years 7' conve 0, 6. .. 0.5 0,4 ----- lo 9 0,3 . . . ... . . . 8 a, , a2 dif 0,a,3,cr fer Y,a, , 2 en - , y, a, ,B, a2 co 0,2- 0,1 -0,1 percentiles 70 Figure 5.4: Simulation of household per capita incomes, '3concav Mean of education 7 years 0,6 0,5 10 0,3 t, g -'6a, 0, a2 dif fer , ,Do 0,1 - - 0 10 20 30 40 50 60 70 80 - 90 01 -0,1 percentiles 71 1F3guSe 595: !Sim e n el hu dhJd pper cuphst hcomeasq convex Mean of educaltion 7 yesirs 0,6 Do 0,3- g -B- a, 03, a2 0,- Y, a, 3 X, ''2 0- 5 ) 10 20 30 40 50 60 70 80 90 )0 -0, 1 -0,2 72 5.18 In addition to these overall results the simulations suggest a rising concentration of poverty in rural areas-these areas had lower levels of education to start with, and, if past patterns are repeated, will experience more modest gains in the future. This underscores the importance of changing past patterns of educational attainment, and pursuing a vigorous proactive program of education in rural areas (an issue taken up in the next sub-section). Moreover, the model did not account for impacts on migration, which would certainly continue, and could actually increase if educational expansion in rural areas were to spur rural-urban flows. Finally, the effects of a targeted program to reduce illiteracy are also positive, but are generally smaller than the overall education expansion: this only underscores the importance of seeing this as a complementary policy. 5.19 Any simulation from past statistical results needs to be treated with caution: there are always errors, not everything is captured, and the world changes. Yet these results are suggestive of the potential impact of expanded education: it can have a significant impact on reducing income poverty, in large part due to the significant impacts on the work and fertility decisions of women, but is likely to have only a the small effect of inequality. Thus education is absolutely necessary, but is by no means sufficient, to shift inequality. And special efforts will be needed to provide quality education in rural areas, and to the poor in particular. The next section develops specific policy options for the state in fostering such a broad-based expansion of education services. V.3.B Education Policy in Ceara 5.20 Raising education levels is likely to have a large medium-term impact on poverty, as seen above. Thus orienting new policy initiatives around tackling the existing deficits in education services, particularly for the poor, should be is a priority focus for the state government. This section identifies 5 main bottlenecks in the provision of education in Ceara, and outlines specific policy options for addressing these problems. These include: low enrollment in and poor quality of early child development programs; low quality basic education services; * severe low quality in municipal schools, grades 5-8; * a large proportion of under-educated youth leaving school and entering the labor force; and * low enrollments and poor quality in secondary and higher education. 73 Tabe 5.3: Education in Ceav&: MinTarg3et Groups & Pocy Ri ecommendations (Continue) Target Group Problem Diagnosis Policy Recommendations Poor children ages 1. Low enrollment Low quality services: 39% Medium-term: 0-6 rates and poor of pre-school teachers 1. Use PROARES to train quality of Early have not completed pre-school teachers Child secondary school. 2. Train community heaith Development Municipalities are agents as ECD service (ECD) and responsible for providing providers preschool these services, so service 3. Use NGOs as service programs coverage and quality is providers worst in the poorest 4. Increase state funding for (11% of children municipalities. municipal ECD and pre- ages 0-3 atfend school services ECD programs; 72% of children ages 4-6 attend _preschool) Poor children and 2. Low quality of Low level of teacher Medium term youth ages 6-18 basic education qualification and skills. 5. Expand PRODEM to all services Lack of operational municipalities (funding standards that apply contingent on improved (46% of students uniformly across schools. recruitment and educational complete grade 8 Lack of educational standards) on time, state test materials in schools 6. Increase state provision of scores below education materials national average) Poor rural school- 3. Pocket of Municipal teacher Vext 12 months: age children severe low quality recruitment is not merit- 7. Implement system to in municipal based. monitor use of FUNDEF schools, (esp. Municipal schools lack (federal funds for municipal grades 5-8); rural access to teacher training education service provision). children fall to and monitoring systems Wedium-Rerm: advance beyond that currently operate in 8. Supplement PRODEM lower secondary state schools with technical assistance to Poor management municipalities on educational (indicator?) capacity, limited access to financial management. financial management systems Poor new entrants 4. Large Low quality basic Next 12 months: to labor force (ages population of education services. 9. Evaluate adult education 15-24) under-educated Adult education and programs for impact on young adults training have limited employment and incomes of entering gabor coverage and beneficiaries force effectiveness is unknown. 9iedium-term: 10. Diversify adult education (60% of urban curriculum; offer 'cerfificacdo population ages de competencias" 15-24 haz less than 8 years of schooling) 74 Table 5.3 (Continue) Secondary and 5. Low Distribution of secondary Medium-term: higher education- enrollments and schools urban-focused;. 11. Provide scholarships for aged students poor quality in disparities in access and low-income students in secondary quality between income secondary school (Bolsa education groups. Escola) Curricula unmatched to 12. Implement program for (30% of population labor market demands. college students as teachers ages 15-19 17% of secondary school in remote areas. enrolled in teachers failed to meet 13. Renovate state TV secondary school) state standards. channel (TVC) facilities; widen TV curriculum offerings V.3.C Low Enrollment and Poor Quality of Initial and Pre-School Education (0 to 6 years old). 5.21 There is increasing awareness of the crucial role played by early childhood education in poverty reduction, in reducing repetition and evasion, and in increasing student achievements. 69 Yet the services provided in Ceara fall far short of desired levels of coverage, quality and management. As of 1999, overall enrollment of the 4 to 6 age brackets was 355,161 or 72.4% of this population group; the corresponding figures for the 0 to 3 group were 56,371 and 11.4% respectively. At both levels, access is lowest among the poor. Low quality is pervasive, mainly due to insufficient-training of teachers and the lack of standards and materials. Of a total of 18,858 pre-school teachers, 7,298 (38.7%) have not completed secondary education. 5.22 In ECD (programs for children 0-3) the newly-proposed non-formal programs implemented by community health agents have the potential to function well, but will take time to increase coverage and quality - coverage is projected to increase from 11% to 22% in 5 years. Further, when PRODEM is implemented, its funding (conditional on ECD and pre-school provision) will create a further incentive to provide these services in the 54 poorest municipalities. 5.23 However, these initiatives are not sufficient because of the size of the existing deficit in service provision, which is particularly severe for those in the 0-3 age range. While the ECD and pre-school programs are the major responsibility of the municipalities, as opposed to the state, these programs are of such high priority for poverty reduction among today's children that it is critical that the state play as active a role as possible in promoting better service provision. Options for the government to consider include: (a) continue training Community Health Agents to provide ECD 69 International evidence indicates there are high returns for public investment in ECD and pre-school services. Children that attend early childhood activities complete primary school in a shorter time and attain higher levels of education. In Brazil, the lifelong increase in income of those who attend early childhood activities is about 10% per year, and for pre-school it is estimated to be around 17% per year. The complementarity between these two levels of education is reinforcing, particularly for poor children (Barros and Mendonca, 1999b). 75 services; (b) increase state funding for municipal-level provision of ECD and pre-school through an incentive program similar to the UNICEF "Sello de Qualidade'; (c) evaluate the effectiveness of PROARES (Programa Reforma de Apoyo Sociais) to determine whether it may be used as an instrument for improving quality of ECD and pre-school services; and (d) consider greater involvement of NGO service providers (some useful models include the Educaci6n Inicial program developed by CONAFE in Mexico, and the proposed municipal program in Rio de Janeiro. V.3.D The Qualy of Basic Education 5.24 Although access to basic education has increased substantially in Ceara (from 57% to 98% between 1990 and 1999), the major problem that remains to be addressed is the low quality of education, especially for schools in poorer areas. Repetition and dropout rates in primary education are still very high (a combined total of 25%) and only 46% of the students complete 8th grade within the expected eight years of schooling. In 2000, about 56% of students were two or more years older than they should be for their respective grades. Age-grade distortion is the cumulative result of late entry to school, chronic grade repetition syndrome, and low basic standards in most schools, leading to low student achievement. 5.25 One of the major underlying causes of these problems on the supply side is the low level of teacher qualification and skills. A detailed 1998 Teachers' Census undertaken by the State Basic Education Secretariat's (SEDUC) Teacher Education Unit indicated that, of a total of 39,225 active fundamental education teachers in Ceara, 11,406 (29.1%) were considered "lay teachers" who did not meet the National Education Law (LDB) requirements. Approximately 10% of them had not completed the eight-year fundamental education level. The Unified Teacher Competition (Concurso Unico), jointly designed and held in December 1997 and June 1998 by the state and municipal public systems to fill existing vacancies with higher-level professionals, helped improve the situation. Yet it is still a serious problem, especially at the municipal level. Furthermore, low content knowledge and poor teaching skills are common among teachers and contribute to low student achievement and high repetition and dropout rates. The lack of attractive career plans and incentives affects the entire teacher system and has a negative impact on education quality.70 5.26 A second factor restricting current education quality is the lack of basic operational standards that would apply to all schools, and the educational materials 70 Results of the 1997 SAEB's evaluation showed that in Ceara the average achievement level (measuring the probability of a correct answer to curricular items on a scale of zero to one hundred) in Portuguese was 46.5% for 4th graders and 59.4% for 8th graders, compared with 49.4% and 65.8%, respectively, at the national level. In mathematics, the average achievement level for 4th graders was 27.1%, and 32.2% for 8th graders, while national levels were 29.5% and 35.8% respectively. These below-average results indicate that student achievements in Ceara could negatively affect the possibilities of economic development of the State, especially for initiatives in areas outside the state capital. 76 necessary for providing an adequate classroom environment for teaching and learning. Thus, there is a tremendous variation in the provision of textbook and other didactic materials, libraries, labs, computers, such that many schools, particularly municipal schools in the poorest areas, lack these basic inputs to education (World Bank, 2000) 1. 5.27 The state is currently taking steps to tackle these problems: the Magiste program provides university-level training to teachers (with only secondary school qualifications) through 4 local universities; the Pro-Formacao (federal) program provides secondary- level education and certification to currently non-qualified teachers in rural areas; the state also operates an intensive in-service training program to familiarize teachers with updated teaching methodologies. 5.28 What is lacking from the state's existing education strategy, however, is an accreditation system and a permanent system of incentives for teachers based on performance and student results. This is an area where the state can play an active role in designing their own strategy under existing general federal guidelines. Specifically, the state could create meaningful mechanisms for school-level accountability through census-based (universal) student assessment systems, "quality assurance" teams for participatory school reviews, teacher networks and other forms of professional support. The role of participation of parents and communities is likely to be of particular importance in this process. While there is not a strong tradition of parent participation in Brazil, Ceara has recently actively promoted it by electing members of the parent-community in school councils; the process of making this effective takes time. 5.29 While the problem of quality is widespread, a particular problem exists in municipal schools grades 5-8, which merits specific policy attention by the state. Based on the Sistema Nacional de Avafiacao da Educacao B.sica (SAEB) 1999 evaluation results, there is evidence that students in grades 5-8 in state public schools perform significantly better than those in municipal schools at the same grade level. The problem is again rooted in poor teacher qualifications, stemming from the fact that recruitment in municipal schools is not generally based on merit; nor do municipal schools currently have access to the SAP teacher training and monitoring system that operates in state schools. While municipal schools are not under the direct control of the state, mechanisms exist, and can be created, through which the state may bring about a solution to this problem. Specifically, one mechanism already in place is PRODEM, a system which leverages desirable policy changes in municipalities where the state cannot exert direct control through funding, which is contingent on improved education and recruitment standards, student assessment, and basic operational standards in line with the current state system. However, out of 184 municipalities in the state, PRODEM will only operate in the 54 poorest municipalities. In order to bring about a complete solution, however, this mechanism should be extended to all municipalities. FUNDEF funds are also a source for addressing quality deficiencies in 71 Ceara Basic Education Quality Improvement Project, Project Appraisal Document. 72 Further details on these proposals and policy options are clearly laid out in "Brazil: Teacher Development and Incentives: A Strategic Framework". February 2001. World Bank. 77 municipal schools - however, since there is currently very little control over how these funds are actually used at the municipal level, the state should establish a monitoring system tying effective use of FUNDEF funds to additional (PRODEM) funding73. Such monitoring schemes can work well in association with mechanisms of social accountability-using both participatory structures at the school level, and public information across schools that allow communities and the government to compare and assess progress. V.3.E Young and Under-Educated Aduhl s already ouft l Scthool 5.30 As seen above, Ceara's young working-age population suffers from extremely low education levels. These are the cumulative result of a historical lack of access to basic-schooling for low-income populations, especially in rural areas.. Poverty reduction among this group hinges both on the state of the labor market for unskilled labor and on their lack of earning potential because of low skills. Policies linked to skills upgrading concern: (a) training and education opportunities targeted to this group; and (b) better employment opportunities for those who receive training. (Issues of direct income support for the elderly and other very poor groups are covered in the next section). It is important to note that world-wide experience suggests that upgrading unskilled adults is often a tough proposition, because of the major complementarities between core learning and subsequent training. These may be exacerbated in the difficulties of reaching socially excluded groups. 5.31 The state currently operates a set of programs designed to provide training and adult-education services: Tempo de Avan9ar (Telecurso 2000) is an accelerated televised program for students to complete grades 5-8 and secondary schooling; Comunidad Solidaria provides literacy programs for adults in conjunction with the Universities. Centro Vocacional Tecnol6gica (CVT) provide technological training in urban areas, and Projeto Sao Jos6 can be included in this category as a program that provides basic training for rural communities. These existing programs are apparently well-implemented, however they have in general limited coverage vis-a-vis the large size of the at-risk population (see Section IV). Furthermore, the quality of the education and training obtained has not yet been adequately evaluated. Finally, a newly proposed program, EducagAo Contextuafizada74, will provide general basic education content (primary school level) and job-skills for students in the 15 to 39 year-old group who recently became literate. The program plans to reach 90,000 people over the course of 5 years - an important contribution, but again small vis-a-vis the size of the target population. 73 Further operational details are laid out in Ceara Basic Education Quality Improvement Project, Project Appraisal Document. World Bank. 2001. 74 The Educacao Contextualizada program, combines general basic education content with activities that help students to develop skills and abilities useful for entenng the labor market. The 18-month program will be developed with a total of 180 television "modules". Approximately 90,000 students will be attending the program during the life of the project, with intense evening and weekend use of school facilities. The program will be jointly implemented by an eligible university and the Comunidade Soliddria Program, in close coordination with SEDUC. 78 5.32 Given the uncertainties over effectiveness from work elsewhere, the priority action on the state's part for improving current policy on adult training and education is to evaluate the quality of services being provided under these existing programs (focusing on relevance for generating income-opportunities among those trained). The expansion of these programs should wait until such evaluations are complete. In the meantime, greater coordination between the agencies providing services to reduce overlap between population served and services provided will likely enhance current program impact. V.3.F Low Enrollment Rates and Poor Quality in Secondary Education; little to no access to university level education among the poor. 5.33 Only 30% of the population aged 15-19 is enrolled in secondary education. The distribution of school places and enrollment is highly unequal, resulting in striking disparities among and between income groups, the capital city of Fortaleza and the rest of the state, and urban and rural areas. There is also a well-known mismatch between secondary school curricula and labor market demands. Poorly qualified teachers are an issue on its own. The Unified Teacher Competition radically changed the profile of secondary teachers, but 1,730 (or 17%) of the total 10,224 still did not meet the 1996 LDB requirements. 5.34 The current policy of increasing the number of secondary schools and lyceos (IDB and Projeto Alvorada) is likely to have a strong and positive impact on increasing access to this level of schooling. However, access among the poor is likely to remain constrained unless this policy is complemented by scholarship programs for rural students, or low-income students in urban marginal areas. Models to consider include income-transfer programs that cover secondary schooling (PROGRESA, see next section), and the scholarship program included in the secondary school education project for El Salvador75. 5.35 At the university level, main and lower campuses now exist in at least 12 cities throughout the state. In order to better promote access to this level of education among secondary school graduates from poor families, the state should consider the following two alternatives: (a) providing a preparatory course for low-income students (including a small amount of financial support while attending the program) in order to better compete in university placement exams (the state of Pernambuco is currently piloting this type of program, providing a useful model to follow); (b) on a selective basis (merit and financial need) providing full university scholarships for those students coming from the public schools who obtain better scores on the secondary school national examination. Alternatively, secondary and higher education financing can be provided contingent on a public service contribution (teaching) in the rural and remote areas of the state (the non-formal education model implemented by CONAFE (Consejo Nacional de Fomento Educativo) in the rural areas of Mexico provides an excellent model for this type of program). 75World Bank. 1997. Project Appraisal Document: Secondary School Education Project - El Salvador. 79 5.36 Positioning Cearc's education system to make best use of Information and Communication Technologies (ICT) is a policy issue that cuts across each of the above areas. Box 5.1 synthesizes current views on the use of ICT in education service provision in Ceara. Sox 5.2 lnfornmalon and Communicaton TechnoOoglez (OCT) and Education Throughout the world there is excitement over the potential role of ICT-especially the internet-in revolutionizing education. Clearly today's children will grow up in a world in which use of the internet will be increasingly part of their working and personal lives. What are the implications for education strategy for a society such as Ceara's? Evidence to date suggests a relatively cautious stance on the internet-relative to other potential uses of ICT, for example in e-government and small-scale business. Even in rich countries there are not yet studies to support the view that computer-based learning and internet connections are cost-effective means of improving skills. Where resources for education are limited, and problems or teaching and leaming so extensive, it is of great importance that resource are not diverted to this-and away from the more fundamental capacity, institutional and social issues. By contrast, there is a longer history of use of more traditional information and communications technologies-especially radio and television. Radio programs can reach those without access to telephone or Intemet, they can be produced locally at low cost, and they have a strong development record. An FAO study on rural-radio in the Philippines found that the cost of delivering radio content is approximately US$ 1.60 to 3.10 per thousand inhabitants. Conversely, reaching people through television costs US$ 32.80, through fliers it costs US$ 38.50, and local print costs US$ 86.70 ( FAO, 1999). There is now some forty years of experience in the use of radio in activities ranging from literacy training (from Mexico to Mali) to primary education, public health and civic education (See Nwaerondu and Thompson (1987). Ceara has been a pioneer in the use of television for educational purposes--recently 25,000 students completed 5-8 grades in less than 11 months with courses using the television. What does this imply for CearA? Continued use of radio and television clearly make sense. With respect to the internet, the state level decision is made easier by the Federal decision to have a major program on the infrastructure side for the intemet. The focus should be on positioning itself to rely on infrastructure expansion from the Federal program, focusing on the institutions to make this work. On the specific issue of computers and internet in schools, it will be important to ensure that community institutions have access during non-school hours, since benefits may be even more important at the community level in the short run. VA PoverRy IReduMon0D0 2hrough0 1Red6aMRb9utiv Ym1naftw 5.37 Redistributive income transfers are an effective poverty reduction policy, in particular as a complement to growth and education initiatives. They form part of a longer-term vision of development, playing a significant role in poverty reduction in OECD countries. However the role for such programs may be particularly important in countries such as Brazil, where inequality slows the poverty reduction results of growth, 80 and is reinforced by public expenditures which have not traditionally been directed to the poorest (Camargo and Ferreira, 2001)76. 5.38 Recent innovations in Brazil and Mexico illustrate the promise of income transfers in achieving sizable and rapid poverty reduction results, and, where benefits are conditioned on human capital acquisition (education and health), in addressing the longer-run determinants of poverty. In Brazil, the Previdencia Rural successfully reduces poverty among a localized population (the rural elderly poor) by targeting them for direct income support. Brazil's Bolsa Escola and PETI programs and Mexico's Progresa have demonstrated success at linking income transfers with various positive behaviors, such as keeping children in school, and in the case of Progresa, obtaining regular medical services.77 5.39 Both Bolsa Escola and Progresa have Box 5.3: Bolsa-Escola type have succeeded in raising family incomes Programs and Poverty Reduction while improving schooling attainment among program beneficiaries. In Brazil, there is Income transfers conditioned on evidence of immediate poverty reduction education have four main objectives. among Bolsa Escola beneficiaries, particularly First, they hope to increase in the federal district, where the program is educational attainment among today's estimated to have reached 42% of the children and thus reduce future population below the poverty line, and to have poverty. Second, by restricting the reduced their income gap from 41% to 4% of grants to the current poor, the programs the poverty line by providing a per-family aim to reduce current poverty. Third, benefit of one minimum wage - two times the by requiring children in beneficiary value of the poverty line (World Bank, 2001)7 . households to have minimum The education effect has also been positive. attendance in school, the programs aim Though enrollment rates are generally high in to reduce child labor. Last, an implicit Brazil, these figures mask the fact that many objective is that by providing income children - especially the poorest - enter late, support to poor families, they act as a attend infrequently, and have higher rates of partial safety net-i.e., they prevent grade repetition and drop out. And it is these these families from falling further into problems with education attainment - over poverty in the event of an adverse and above school attendance per se -- that shock. Bolsa Escola-type income transfer programs 76 While social expenditures represent two-thirds of Brazil's federal expenditures - and 20% of GDP -only a small fraction of this is on redistributive transfers; the bulk goes to the social security system for formal sector workers and to the education and health sectors. n Brazil's Bolsa Escola program, initiated in the municipalities of Campinas and the Federal District of Brasilia in 1995 and now operates in 52 municipalities in Brazil, alongside two major Federal Programs (FGRM- Fundo de Garantia de Renda Minima and PETI-Programa de Eradicacao do Trabalho Infantial). Mexico's Progresa is a simular program which has pioneered expanding eligibility conditions more broadly to include both education and healthcare components, and has yielded positive results (for further information on these programs, see sGovernment of Ceara). 78 Note however that coverage among the poor does not appear to be as high in other municipalities - in some cases as low as 10% -- due in part to eligibility requirements that vary by municipality (World Bank, 2001). 81 have shown positive evidence of addressing79. Similarly, in more rigorous evaluation techniques Mexico's Progresa has demonstrated success in raising education attainment among children whose families participate in the program, vis-a-vis a control group of non-participants.! VM4.A Mncome Tlransfers end Poverty ReducUaOon in Cer 5.40 In Ceara, the federal pension program illustrates the scope for achieving substantial, rapid poverty reduction via income transfers. Household survey data indicate that indigence among the rural elderly has been virtually eliminated -- their indigence rate now stands at 3.8%, compared to 42% for the rural population as a whole. Further, simulation results indicate that rural poverty rate would be 14 percentage points higher than it is today without the income receipts from this program (table 5.3)*8 In other words, rural poverty has been reduced from 88% to 72% as a result of this program. 5.41 Rural pensions are also well targeted. Almost 90% of beneficiaries have incomes that would be below the poverty level without pension income (and over 60% would be below the indigence level). Given that the Previdencia Rural is the largest pension program operating in rural Ceari (other pension receipts would likely come for people previously in public sector employment), these results reflect positively on the poverty targeting and poverty reduction impact of direct beneficiaries of the Previdencia Rural (the elderly rural poor). The total cost of the program is difficult to estimate precisely from the available data. Total rural pension receipts as measured by the PNAD household survey sum to R$488.9 million. Since this amount includes all pension receipts, however, it is not the relevant figure in estimating the cost of the poverty-targeted Previdencia Rural program alone that is probably over a R$1 00 million lower. 12 79 Evidence from the Bolsa Escola program in Brasilia indicates that dropout rates are far lower for program beneficiaries (0.4%) than for non-beneficiaries (5.6%), and that promotion rates to the next grade are higher -- 80% vs. 72%. While these results are not based on statistically robust evaluation techniques (that is, they are based on simple comparisons of beneficiaries with the non-beneficiary population, as opposed to results obtained from a 'matched' control group) they do provide a general indication that the program is succeeding in improving basic educational outcome indicators among program beneficiaries. ( World Bank, 2001, Brazil: Eradicating Child Labor). The companion program PETI has formed part of a highly impressive program to tackle child labor in Brazil, that first targeted the worst forms of rural labor, and then has moved into the more complicated environment of urban labor79. It has reached some 400,000 children, and evidence from Bahia, Pernambuco and Sergipe found significant reductions in child labor. One of its instruments has been the introduction of a jornada ampliada or after- school care system. 80 For example, in the first two years of the program alone, experiment and control group evaluation techniques showed that participation in Progresa raised primary enrollment rates by 2% higher, and secondary enrollment rates by 5% (Schultz, 1999; World Bank, 2001). 81 For all rural households reporting a pension receipt in the 1999 PNAD, the simulation creates a simple estimate of total household income with and without the pension receipt, and then estimates which households would be poor without their pension income 82 This total cost includes all pension receipts reported by rural inhabitants in the PNAD survey. Of these recipients, 11.5% were among the non-poor. It is likely that many of these pensions come from sources other than the Previdencia Rural (the average reported pension amount among the non-poor was R$184, well over the amount of the Previdencia Rural pension (R$136). The total amount of these reported 82 Table 5.4: The Previdencia Rural in Ceara: Coverage and Targeting Indicators Total number of beneficiaries 278,847 (Share of rural population) (12%) Share of beneficiaries with pre-pension income below poverty line 88% Total number of people (beneficiaries + dependants) moved out of 332,891 poverty as a result of pension income (Share of rural population) (14%) Share of beneficiaries below indigence line (pre-pension income) 64% Total number of people (beneficiaries + dependants) moved out of 422,468 indigence as a result of pension income (Share of rural population) (18%) Beneficiaries with less than 2 dependents 87.5% Source: simulations based on PNAD survey data, 1999. Notes: 1. Poverty line is set at 112 minimum wage (R$ 68) per capita. 2. The simulation models the effect of current rural pension receipts on poverty - the 'before' poverty rate is the simulated poverty rate that would hold in rural Ceara without pension receipts, and the 'after' poverty rate is the current (1999) rural poverty rate. 5.42 While the Previdencia Rural demonstrates success at poverty reduction among Ceara's rural poor, given that this is the main income transfer program targeted to the poor in Ceara, the vast majority of poor remain uncovered by the current system of benefits. The rural pension beneficiaries represent only 12% of the rural population, and most recipients have fewer than 2 dependents (87%), so the 'spillover' effects of these income benefits to the overall rural population are limited. The program provides no coverage among the elderly poor in urban areas (a gap in the state's current poverty reduction strategy noted section IV) - however, the urban poor represent a very small share of the total population in Ceara (0.8%) - such that expanding the current old-age pension system to the currently uncovered elderly poor (to both urban and rural areas) would apparently have only a very small additional impact on poverty reduction -- the state poverty rate is estimated to fall by less than 1 percentage point, from 51.2% to 50. 9%.83 5.43 The lack of a child-targeted income transfer program emerges as a far more significant gap in the current income transfer system. Children ages 0-18 have higher income poverty rates than any other age group, and make up over half of the total poor in Ceara (section 11). pension receipts is R$74.6 million, or 16% of all pension receipts. The total spent on pensions received by the rural poor, which could be used a gross estimate of the total cost of the Previdencia Rural pensions, is R$ 347.9 million. 3 However, the program would be relatively cost effective, moving over 1,000 people out of poverty for every R$ million spent, and would be low-cost in terms of the overall state budget (0.3%), given the small number of intended beneficiaries. 83 V.43 Implemenenting a ChGnd-Targeted Oncome TransVfr In Cear,: the Baolsa IEscOD and Options Vor a Supplemental State Program 5.44 At the time of preparation of this report, there were various proposals to extend Brazil's Bolsa Escola to all states under federally-agreed guidelines, in exchange for federal co-financing of the program. In this section we explore one of these that would involve targeting to families below a poverty line of 1/2 minimum wage, and would provide a benefit of R$15 per month for each child ages 7-14 in school, to a maximum of R$45 per family (three children). The goal of the program is to reduce poverty by providing additional monthly income to poor families, and provide incentives for families to keep their children in school (the criteria for receiving the benefit - covering up to three children). 5.45 Simulations indicate that under these assumptions, a Bolsa Escola program will not have a strong poverty reduction impact in Ceara (table 5.4). Even assuming the program successfully reaches all eligible families (an assumption that is optimistic in that it entails no targeting errors of exclusion), the indigence rate is projected to fall by just two percentage points (from 25.5% to 23.4%). The poverty rate would fall by an even narrower margin (from 51.2% to 49.7%)84. Under the federal program, the value of the transfer is insufficient to move most beneficiary households out of poverty, and excluding families with children below the age of 7 prohibits some of the most vulnerable families (with very young children) from participating.85 5.46 In order to achieve greater income-poverty reduction impact, the state could mandate (and fund) a supplement to the federal Bolsa Escola program. 84The table actually presents results for a program targeted just to families below the indigence line, and for an alternative program targeted to all poor families, in order to illustrate cost differentials of different targeting criteria. A program targeted to all poor will have the same impact on the indigence rate as a program targeted to just the indigent, with an added impact on the poverty rate. 85 All simulations estimate the impact of a given transfer program on poverty and indigence rates assuming all households who qualify for the benefit actually receive the benefit. No adjustments are made for changes in behavior that the household might undertake as a result of receiving the income benefit - for example, should households cut back on labor and thus reduce their earnings from employment, the poverty impact would be less than that estimated here. Only costs of the total value transferred are calculated - total costs to implement the program would include administrative costs, plus any errors of inclusion in targeting not accounted for in the above calculations (households who do not officially qualify for the program, but nevertheless receive benefits). 84 Simulations were undertaken to provide some quantitative basis for assessing this option. By way of illustration, the following two scenarios estimate the added poverty reduction impact and cost of a supplemental state program. In both cases, estimates are calculated for a program targeted just to indigent families, and to all poor families (table 5.4): o Supplement A: the state increases the Federally-mandated Bolsa Escola benefit to R$25 per child (maximum benefit R$75); o Supplement B: the state increases the benefit as above, and extends coverage to all families with children ages 0-14; 85 TabMe S.S. Wpleme@ntng ih@e Moma Esco0 Transffer Program Gn C@ar: Pvojsecedd Poerty mpacSt & CoWs INDIGENCE RATE POVERTY RATE Amount transferred Net cost, Program Description After program After program Cost JR. mill) as % of Before (1999) implementation Before (1999) implementation Total Net state exp. 1. Federal Bolsa Escola: R$15 per child ages 7-14, max. benefit R$45: o Targeted to indigent families 25.5 23.4 70 0 0 o Targeted to all poor families 51.2 49.7 147 0 0 2. State supplement A - raises benefit to R$25 per child, max. R$75: O Targeted to indigent families 25.5 21.0 117 47 1.1% o targeted to all poor families 51.2 48.5 246 99 2.3% 3. State supplement B - raises benefit as above, and covers all children 0-14: o targeted to indigent families 25.5 15.7 257 187 4.4% O targeted to all poor families 51.2 46.0 443 296 7.0% Note: The net cost to the government of CearA of each state supplement to the Federal Bolsa Escola is calculated by subtracting the cost of the Federal Bolsa Escola program from the total cost of the supplement. YabOe 5.6.' PvoX@fd Poverty Reduction Hmpact tfl giromh Cobgnod w2h Mcame Tran$ffie$ (Bo0$e EecDeD Over Scenario INDIGENCE RATE POVERTY RATE After 5 years After 21 years After 5 years After 21 years 1999 (2004) (2020) 1999 2004) (2020) 1. 4% Growth 25.5 23.7 15.9 51.2 47.6 34.7 2. 4% growth + Federal 20.3 13.2 45.8 31.8 Bolsa Escola 3. 4%growth+state 17.9 11.4 44.3 30.0 supplement A 4. 4% growth+ state 12.9 6.9 40.6 25.9 supplement B 86 5.47 Simulation results indicate that substantially more rapid progress can be made in reducing poverty in Ceara by expanding on the Bolsa Escola system, at reasonable cost to the state. The more generous of the two supplements modeled (raising the per-child benefit to R$25 and extending coverage to all families with children 0-14) can be targeted to only indigent families at a cost of approximately 4.4% of state expenditures, and achieve a reduction in the indigence rate from 25.5% to 15.7% (table 5.4) -- slightly better than that achieved after 21 years of economic growth at a rate of 4% per year (table 5.5). If targeted to all poor families, the state poverty rate is estimated to fall from 51.2% to 46%, at a net cost of 7% of total state expenditures. 5.48 The results are even more positive over time, when a well-targeted income transfer program is combined with steady economic growth, as illustrated in Table 5.5. Row 1 illustrates the results that can be expected from economic growth in reducing poverty over time - after 5 years, sustaining current growth rates of 4% GDP will move Ceara's indigence rates from 25.5% to 23.7%, and poverty from 51.2% to 47.6%. If the federal Bolsa Escola program is implemented, results will be only slightly better: after 5 years, the indigence rate will stand at 20.3%, and the poverty rate at 45.8%. However, if the state of Ceara were to supplement the Bolsa Escola with the more generous benefit plan (option B), after 5 years the indigence rate would stand at 12.9%, while poverty will have fallen to 44.3%. And after 21 years, indigence will have been almost eliminated (at 6.9%), and poverty reduced to roughly one-quarter of the population (25.9%). V.4.C Policy Recommendations 5.49 The evidence from existing income transfer programs operating in CearA (i.e. rural pensions for the elderly poor) indicates that such programs have substantial capacity for having a strong, immediate, positive impact on poverty reduction in the state. The PrevidOncia Rural is estimated to reduce poverty in rural Ceara by 14% -- the rural poverty rate, which currently stands at 71.7%, would be over 85% in the absence of this program. While this program fails to cover urban areas, expanding the existing program to cover the urban elderly does not appear to be a policy priority, given the relatively low indigence rate among this population, and the fact that that most elderly have few dependents who would be moved out of poverty as indirect beneficiaries of the program. 5.50 However, there is ample scope for implementing an income transfer program to poor families with children, building on the experience of Brazil's existing programs, notably the Bolsa Escola and PETI, and Mexico's Progresa. These programs achieve the dual benefits of immediate poverty reduction via a direct income transfer to poor families, and to human capital development, by making benefits conditional on school attendance (and other conditions, such as basic health care visits). Basic simulation results suggest, however, that the poverty reduction impact of the program will depend on the design of the program - in particular the benefit level and eligibility criteria, as well as the institutional context that will determine the effectiveness of implementation. Interestingly, simulations indicate that the current federal Bolsa Escola program, if 87 implemented in Ceara under federal guidelines, will have a very modest impact on poverty reduction in the state. Obviously greater poverty reduction is achieved with higher transfer amounts and broader coverage, and Cearca has the option of making a larger impact on poverty by financing such supplements to the Federal program. An active role of the state in design will also allow it to build on existing areas of local institutional strength86, while supplementing not only benefit amounts but eligibility with additional groups, such as pre-school children, and in the medium term, poor secondary secondary-school children. 5.51 In addition, it is recommended that the state continue its program of rural drought relief, and consider administering the bolsa escola via the same established targeting and delivery mechanisms in order to reduce administrative costs of the new Bolsa Escola benefit. Table 5.7: SummarV of Policy Recommendations - Mncome Transfers Population Problem/Inddicator Diagnosis Policy Recommendations Group/Exit Strategy Poor families with 1. Income poverty rates Current rates of growth Next 12 months: children ages 0-14 falling slowly despite projected to reduce poverty 1. Supplement Federal Bolsa economic growth; poverty rate by less than 3 Escola program (increase concentrated among the percentage points in 5 benefit amount from R$15 to young years). R$25 per month, with a Existing income transfer maximum benefit of R$75, and (Poverty rates stagnant in programs exclude poor expand coverage to all children Ceard since 1995; 44% of families with children. ages 0-14) the poor are under age 15; Children from poor families Impact: poverty rate reduced two-thirds of young have low educational by 5 percentage points, people live in poor attainment, high grade indigence rate by 10 households.) repetition and dropout rates, percentage points. and enter labor force early at Cost: R$296 million (benefits below-poverty wages. only), 6.9% of state 2. Proposed Federal expenditures. Bolsa Escola Program will have little Impact on Medium term: income poverty The program is projected to 2. Expand coverage of Bolsa reduce poverty by less than Escola to secondary schools (the program is projected one percentage point, and (i.e. Mexico's Progresa model) to reduce poverty by less indigence by 2 percentage Introduce smart cards and other than 2 percentage points) points (benefit levels of R$15 ICT options to increase per child are too low, and efficiency and transparency of coverage restricted to benefit delivery systems. children age 7-14.) Poor rural farmers 1. Vulnerability to Continue existing program extreme income loss of drought relief during drought years 86 For example through the involvement of the agentes de saude in inter-sectoral teams, as is already being explored within the state 88 V.5 How can the Expansion of Economic Opportunities be Made More Equitable? 5.52 Education and transfers have the potential to play a major role in a poverty reduction strategy. But these need to be complemented by policies that not only support relatively rapid growth, but seek to shape the pattern of expansion to be create more equitable opportunities, especially for work. The state of Ceara is in the middle of a transition from a predominantly rural to a predominantly urban society. The dynamics of this process are central to the state's overall development strategy, and to the pattern of work opportunities. Less than a third of Ceara's population currently live in rural areas -compared with two-thirds in 1960 and half in 1980. If current trends continue, the rural population share will fall to a quarter by 2010, and to 10 percent by 2030. 5.53 What does international experience have to say on the rural-urban transition, employment and poverty reduction? Migration is clearly part-possibly a large part-of the answer to rural poverty, but migration tends to be concentrated amongst those better equipped to move, in skills, savings and contacts. And, on the other side, there is a concern that excessive migration will only swell urban poverty. The most compelling successes in poverty reduction have typically involved a dual strategy: effective creation of employment opportunity and services in urban areas on the one hand, in both industry and services, with continued support for rural productivity increase and rural economic and social services on the other. In the last half of the 20th Century this dual strategy was a central aspect of the East Asian successes in poverty reduction and economic development, in countries now richer than Brazil, including Korea, Taiwan (China), Malaysia and Thailand, as well as Japan over a longer term, and also for countries such as China and Indonesia that are now richer than Ceara. Vietnam, still much poorer than Ceara, has been rapidly following a similar path in the past decade or so, and has enjoyed rapid declines in poverty.87 5.54 Of particular importance-for both poverty reduction and overall development-in all of these cases was the development of a dynamic of rising rural productivity, with strong linkages between agriculture, rural non-farm activities and small town economic and social development. As seen in Chapter 2, Ceara, by contrast, has been characterized by falling rural productivity, especially in wake of the collapse of the cotton sector. In successful cases, rural non-farm activities was driven be a mixture of rising agricultural surpluses (especially of small and medium farms) and local economic infrastructure. Family farms lay at the center of this process, whether from longstanding structures of small farm structures-as in Indonesia and Thailand-redistributive land reforms, as in Korea and Taiwan, China-or the recreation of family farming in the land reforms of China and Vietnam. However, labor intensive estate farming also played a role in some cases, notably in Indonesia and Malaysia. 87 See World Bank (1993, 1995) 89 5.55 The other side of the process was the pull of labor into urban employment, especially into labor-intensive industries and services, that nevertheless had higher levels of productivity and wages than agriculture. Here there was an important interplay with dynamic changes in education and age: in Korea, for example, newly educated youth were the primary entrants to industrial firms; adults tended to migrate into service jobs.88 The labor policy and institutional framework for this job creation was quite varied: some repressed unions and substituted paternalistic labor codes (notably Korea, especially pre-democracy) but issues of labor conditions and rights has become of increasing importance, whether from domestic or international pressures. 5.56 The spatial mix varied: urban production was dispersed in Taiwan, China, and "township and village enterprises" were central to employment growth in China in the 1980s. Production in Thailand was highly concentrated in the urban metropolis of Bangkok. So did the firm size mix: in many cases, small and medium sized firms have played an important role, though in Korea domestic conglomerates played the leading role. Foreign investment played a central role in many of the cases, and is of increasing importance, with relocation of capital from richer to poorer areas within East Asia at the core of international and national process of employment creation in the 1990s. This has of course been part of the strategy for Ceara and the rest of the Northeast, in attracting capital from the South East to lower wage locations. 5.57 The process of rural-urban transition takes decades. This can be illustrated by two now-developed countries. Spain and Portugal both shifted from a 30 percent agricultural employment share to around 10 percent in about 25 years (see Table 5.1) And lagging rural regions, often in low agricultural productivity areas, show continued relative backwardness. This is also a feature of many of the East Asian successes: for example, there are many parts of inner China that are resource-poor, rain-fed, and have much higher levels of deprivation than the dynamic coastal regions. Now-developed countries all have their lagging regions-from the Appalachians in the United States to Andalucia or Extremadura in Spain. And while outmigration from such "backward" regions is always significant, it typically takes time before there are large declines in the absolute numbers of the population fall. In Cear6 the absolute numbers of the rural population have only fallen by a 20 percent between 1970 and 2000-back to about the same numbers as in 1960 (Figure 5.2). Table 5.8: Share of Employment in Agrilculture, Spain and Prtgal, 1960-96 1960 1970 1980 1985 1996 Spain 44 30 19 16 9 Portugal 44 30 27 22 12 Source: background paper on Spain and Portugal, ECA paper on poverty and inequality. 88 Se Kim and Topel. 90 Figure 5.6: Number of people living in rural areas, 1960-2000 Ceara: # people living in rural areas 3,000,000 2,500,000 - - - - - - l' - - - 2,000,000 T - . 1,500,000 - 1,000,000 -.I--- 500,00 - 1960 1970 1980 1991 2000 V.5.A Policy Options and Scenarios 5.58 The past spatial and structural pattern of growth has sustained, not reduced inequality. Is there an alternative that will help reduce inequality-and accelerate reduction in poverty? This implies heightened efforts to rebalance the spatial and structural mix, in two areas: * Rural productivity * Industrial and service employment and the pattern of urban development These will be linked by migration processes, that will themselves be supported by investment in education and improved communication systems, including both transport to ease travel and telephony and other forms of personal communication to help reduce the costs of maintaining linkages across dispersed family and community structures. 5.59 Rural productivity and the future of agriculture. One view is that agriculture has no real future in Ceara, that the drought-prone sertao has not only meager long-run prospects for productivity increase, but will only continue to be a periodic drain on government resources in the provision of drought relief. This would imply that the East Asian path of rising small farm productivity spurring poverty reduction on the farm and underpinning rural non-farm development is not available to Ceara. 5.60 There do not appear to be any magic bullets for transformation of agriculture. There will indeed be parts of the sertAo that will experience, over the long term, large scale outmigration. However, agriculture and rural development will continue to play a role of great importance to poverty reduction and economic development. It can play a significant part in major increases productivity to provide decent living for a fraction of the farming population, and a broader role for many more in the multi-decade transition 91 to an highly urbanized society. While there is no magic bullet, there are a diverse range of possibilities, that are reflected in the account of exit strategies of poverty for the current rural poor (see chapter 3). Supporting such multiple routes will require molding institutional support and public action to the particular needs of different rural groups. 5.61 A fuller account of. agriculture and rural development strategy is available in World Bank report, that includes a particular focus on the Northeast. 89 Here we outline some of the key points for Ceara's poverty reduction strategy, that can be summarized as: o Continuation of broad-based programs, including Projeto Sdo Jos6 o Systematic exploration of opportunities to support the (minority?) of small- scale farmers who can shift to higher productivity agriculture o Support for commercial and agro-industrial activities that are clearly economically viable where there are bottlenecks where public intervention makes sense, including on links to small-scale production o A complementary policy on (social and economic) infrastructure, with basic services throughout the state, including the poorest municipios, but more targeted efforts on areas of potential; this will also facilitate rural non-farm development o Exploration of options for strengthening risk management 5.62 Broad-based programs--Projeo $ao Joz. While potential for major productivity increases will be small for much of the sertSo, outmigration will be a solution that will take decades and in the transition tend to leave the poorer behind. The breadth of coverage of Projeto Sao Jos6, as well as a design structure that seeks to adapt investments to local conditions with community participation is highly appropriate. The evidence indicates that this brings modest but important gains in incomes (as well as significant other benefits in better quality of life and social capital) from relatively modest investments. In the longer-term this may be a transitional mechanism for many farmers, but it has a sound design as one element of an agricultural strategy. The interface between participation and municipalities is discussed in the fourth section on institutions. 5.63 Water management, landl reform and higher impact progirams. Activities with potential to transform income possibilities will depend on the particular ecological and farm conditions in different parts of rural Cear6. The principal themes are the inter- relationships between water management, high-productivity farming, land reform, the public-private interface and risk management. New detailed work on this issue was not undertaken for this report, but the preliminary position is that there is significant potential for substantial income generation amongst a subset of poor farmers through more efficient exploitation of existing water resources-including existing schemes. Benefits will be reaped by the poor especially where this is combined with small-farm production or tenant production, and in crops with high levels of employment intensity (for example grapes). Land reform can play an important role in this, in allowing a subset of landless 89 World Bank 2001c. 92 farmers to acquire sufficient land to be economically viable: this is being implemented within the C6dula da Terra program. Since this is quite new results are not available, but the program is notable for the inclusion of a careful evaluation structure that will allow the government and others to assess its effectiveness in the future. 5.64 However, the real challenge is institutional-on how to manage the interface between government action, the private sector and associations of farmers or workers. Cases of success have managed to do this that falls between either a simple "market- friendly" approach or a strongly state-directed approach, as in the past success of the agroindustrial poles of Petrolina/Juazeiro and Baixo Vale do Rio A,u. Petrolina/Juazeiro. In these the Sao Francisco River Vally (CODEVASF) effectively recruiting commercial firms from elsewhere in Brazil, with a competitive process of firm selection and subsequent performance-based management of their operations. 90 Based in part on this success, there is a growing consensus that any future for irrigated agriculture must be based on full commercial orientation based on production of high value crops91 5.65 In addition to irrigated production, there is likely to be a class of activities in which a subgroup of farmers are able to develop high productivity alternatives. This is a common theme of work on farming in areas that are generally resource-poor. A minority of small and medium farms are often able to manage resources, innovate on production and find market niches that are productive. Recent work on Andean agriculture-also a resource-poor region in long-term productivity decline-found this phenomenon92. An example from Ceara (documented in Tendler, 1997) is the success of the Santana farmers cooperative, in which a group of small farmers who had recently received land from a state-wide land distribution, decided to shift into intensive livestock production. The lesson from this is one of the role of customized and client-driven services between an effective local organization and outside government agents, in this case in the extension service. 5.66 Risk management. Cutting across all considerations of rural development is the issue of risk management. Dealing with drought risk will continue to be a perennial concern at least for those outside irrigated areas, and of course part of the rationale for both irrigation and other forms of diversification flow from the gains from risk reduction. The most important instrument for managing this in the past has been the major drought relief programs. There is indirect evidence that these have been effective: or at least the combination of public action and private coping measures have prevented major reductions in agricultural income having a significant impact on rural poverty in the 1993 and 1998 drought years. (See Figure 5.7). 90 See Daimani, 1999, as quoted in Tendler 2000. 91See Bando do Nordeste, 2000, Projeto Novo Modelo do Irrigaqdo, Ceara State Government, Secretary for Irrigated Agriculture Irrigando para a Competitivadade. 92 See Government of Bolivia/World Bank 1999 93 Filgure 5.7 Agricultural ProducN6on and RuralU Pov@ry in Ce&iM PIS Agropecuhrlo e Pobreza Rural Sdrie hist6rica 1985/99 - 1990-100 140.0 , , , , , . . . . . 120.0 8'0.0 ' - -t a' a' a' a' -0.-PObreza Rural -- PI Agropecuro Source: Calculations by IPLANCE, CearcA national accounts and PNAD. 5.67 However, the drought relief programs are financially costly and-with the exception of the 1987 case-are frequently accused of being embedded in clientelistic behavior and with limited long-run benefit. Lesson from international experience suggests designing these schemes such that they can expand very quickly-and as far as possible automatically-when risks increase, and then decline when conditions improve. This can imply linking these long-term structures of project planning and institutions-so that a shelf of projects are available to be brought into action when the need arises, with below market wages to support effective targeting. Integrating drought management with the Projeto S o Joso is a promising possibility.93 In the longer term it will be valuable to embed such schemes in a overall assessment of the risk profile of different socio-economic groups, and efficient options for dealing with the risks. Where rainfall is the key risk, there is new thinking on options for market-based insurance, but these would have to be based on careful analysis of climate and the availability and financial advantages of such instruments. See the main report for further discussion of options. sugest Odusignin thesevschees such Sp tial theycarn epn erucl-adaa 5.68 Cearci's pattern of non-agricultural development has been characterized by rapid growth in service employment, and-especially recently-rapid growth in industrial output, but with quite minor employment impacts. The state has pursued an active policy of industrial policy, and has been quite successful in pro-actively attracting investments through a mixture of tax write-offs, infrastructure support and a general See Worild Bank 2000 State economic memorandum WDR 2000/01, World Bank 2000, Securing our Future and SGoverment of Ceare 1 for the case for linking drought relief to the Projeto Sdo Jos n 94 projection of a "modern" and reliable investment climate. In the recent past the policy of encouraging firms to move to towns in the interior has led to some shifting of industrial projection out of Fortaleza. This policy has been criticized in the past as largely supporting profits, being confined to relatively capital-intensive industry (compared with tourism), focused on non-state firms, being costly in foregone revenues (to the extent the firms would have relocated to Ceara in any case) and having had much lower employment creation than originally promised.94 5.69 Here we sketch some of the issues in achieving job-creating investment, and in particular on options for the state to spur a pattern of development will be conducive to more rapid employment growth. More detailed work on some of these issues is being undertaken in parallel.95 5.70 Policies for supporting private investment for jobs can be divided into three categories: . the overall investment environment, encompassing political and macroeconomic stability, financial sector conditions, low levels of corruption, labor market conditions and the regulatory environment * factors that affect the investment environment for micro and small-scale firms96 . sectoral strategies designed to shape the pattern of growth of industry and service production in line with Ceara's competitive possibilities. 5.71 The first set of issues are clearly necessary. We take as given that a sound overall investment environment, decent infrastructure services, efficient financial services, attempts to reduce corruption and the quality of public services are all necessary for high levels of investment. Some of this falls within a state's purview-for example local infrastructure service and quality of local public service delivery; some clearly is Brazil-wide, notably the level of interest rates. Some fall within the state government's role-indeed the state has been successful in developing a reputation for a reliable business environment.97 Overall financial sector and labor market conditions are mainly under federal influence. We have a few comments on the second and third areas. V.6.A The Investment Environment for Micro, Small and Medium Firms 5.72 Micro and small farms typically account for the bulk of employment-especially for poorer workers-though their dynamic role varies with the country. For example, in 94 See World Bank (2000a) for a specific version of this criticism, that falls within a long tradition of critique by economists. For a general discussion see Tendler (2000), who quotes Vasconcelos et al (1999) who found that less than a quarter of firms approved in the 1991-94 period were operatingi in the late 1990s, with a correspondingly lower level of employment 95 See Bar-El (2001) 96 Definitions differ, but typically micro-enterprises are categorized in terms of the self-employed sector and firms with family or very few employees (say less than 5 employees); small firms may be up to 20 or 50 employees depending on the definitional custom. 97 See World Bank (2000a). Tendler, Judith (2000) 95 the Northeast of Brazil, enterprises with less than 10 employees (a typical cut-off for micro firms) accounted for almost half of employment in manufacturing, 60 percent in construction, 80 percent in commerce and 90 percent in other services in the late 1990s. There is growing evidence on the role of the combination of obstacles in limiting the growth for both micro and small and medium firms, including bureaucratic and regulatory impediments to licensing, lack of property rights, lack of access to credit and uncertain product markets. It is likely that action in all of these areas will be necessary to support more rapid growth. Ongoing work is assessing the environment for the micro, small and medium firm sectors in the Northeast that should allow a filling in of these recommendations in the coming months. 5.73 Titling and regulation. Problems of regulation, bureaucracy and lack of titling is most vividly laid out in the work of De Soto.9 who documents both the complexity of processes required for small firms to go legal and the substantial levels of capital in housing stocks that could potentially be used to back business ventures. This is done for countries as diverse as Pern, the Philippines, Haiti and Egypt. The core message of this diagnosis is that the legal and regulatory structure needs to be moulded to the conditions prevailing within the informal sector with strengthening of property rights a key element. A specific implementation of this approach is underway in Peru, in a project in which legal titling is a key element. 5.74 There are not comparable diagnoses for Ceara. However, there is evidence that licensing can take months and bureaucratic hurdles matter.99 And anecdotal evidence for the urban slum upgrading project in Ceard, that relies on self-help schemes and leads to greater security of tenure as well as physical upgrading. Initial results suggest higher levels of incomes in these areas, with qualitative feedback suggesting that this flowed from both the multiple benefits of local construction activities and from the greater security. Extending this approach would involve assessment of the prevailing conditions for micro and small firms to identify constraints, burdens and gaps. If justified, this could lead to design of titling and de- or re-regulation. Fiscal implications would be modest. 5.75 Advisory and credit services. Micro credit is an important source of support for micro-enterprises.100 The international movement that emerged in the 1980s and 1990s was built on successful large scale schemes in Bangladesh, Indonesia and in many parts of Latin America, notably Bolivia. This was in marked contrast to the dominant earlier model of subsidized credit from state banks, that systematically failed in terms of both banking and enterprise development. The new successes typically grew from NGO or other non-bank activities, using innovative mechanisms, such as group lending, and close interactions between the institution and borrowers to get good repayment records. Financial viability has been a core aim, with market like interest rates, even though many institutions required subsidies for administration for many years. Some, 98 De Soto (1995,2000) 9 See the preliminary map for the micro, small and medium enterprises sector for the Northeast of Brazil, IFC/World Bank, May 2001. 100 This subsection draws on Schonberger (2000). 96 such a Bancosol in Bolivia 'have become financial institutions. While most are non- government or private, a few, including the large-scale Indonesia program were run by state banks. 5.76 Evidence on poverty reduction is relatively limited, but what is available from other countries indicates that micro-credit is generally most successful at reaching groups around or above the poverty line. The poorest are generally not included, for lack of complementary resources. Observers generally place most emphasis on gains in assets, with risk reduction an important benefit. 5.77 Despite a significant range of directed credit lines of the Banco do Nordeste, a 1997 survey of microenterprises found that a very small minority used formal financial services (3 percent of those surveyed in Fortaleza), a significant minority used moneylenders, suppliers credits or family and most didn't borrow at all. Interest rates were some 20 percent per month. 5.78 Two microcredit programs have already gone to significant scale in the Northeast: the CrediAmigo scheme supported by the Banco do Nordeste, and the Federacao Nacional de Apoio aos Pequenos Empreendimentos (FENAPE) network of non-government micro-credit institutions. These fall very much within the mould of international experience outlined here: charging market-like interest rates (4-5 percent per month) using solidarity groups, achieving high repayment rates and on a path to subsidy independence. While evaluation results are not yet in, the challenge is not to introduce new schemes, but to help emerging micro enterprises gain access to this scheme. Public subsidy programs for this sector have a dismal track record and make little sense. 5.79 Small and medium scale firms should become be eligible for financing from more traditional financial sector firms. Here there are two categories of issue. First, the general cost of capital is high in Brazil-owing to overall financial sector conditions. There is not much that can be done on that at the state level. Second, private banks have typically paid little attention to the small and medium scale sector; here there may be scope for encouraging private banks to open credit departments for this sector. 5.80 Market access. The bulk of policies to support the micro and small scale enterprise sector has focused on supply-side support and the overall business environment. There can also be scope to work on the demand side, and Ceara has experience of this. The Government of Ceara introduced an innovative program of reserving public procurement for the small scale sector in the late 1980s and early 1990s, that is documented in Tendler (1997). This did not provide guaranteed markets for firms, since production had to be acceptable to purchasing ministries. But it did create a market, conditional on resolution of quality and supply side issues. And it converted the support agencies-especially SEBRAE-from focusing primarily on generalized supply side support for the sector, to tailored demand-driven, problem solving activities. In some cases, there was effective organization of supply, delivery at higher quality and lower cost, after initial start-up issues, and major expansion of the 97 micro and small scale production in a form of cluster. (Tendler, 1997) The legality of the procurement procedures were subsequently, successfully challenged (by larger firms) on grounds of preferential treatment. However, there are options of designing tender procedures in a fashion that seeks to level the playing field for smaller firms, without using the proxy of reserved procurement. Details would have to be worked out by procurement specialists, but this could include the unbundling of tenders into small lots, active efforts to share information, and encouraging of procurement in local areas. 5.81 There may also be other means of fostering market linkages for smaller firms, notably through encouraging large firms (especially those receiving public support) to have as part of their business strategy link-ups with small scale firms. There is further likely to be substantial potential from use of information technologies to link up small firms, avoid middlemen, and solve difficulties of bundling orders. This is an area of anecdotal experience, but one of immense potential that warrants active experimentation and support. V.6.13 BEenchmarkinjg andi Sup;portor flEfrend ParR of the SuppGy ChaDn 5.82 Sector- or spatially specific policy may be able to complement action on the overall investment environment. Can the government and private sector work together in shaping the pattern of growth? While there are many areas where there is potential role for the government-in solving coordination problems, infrastructure planning, information-the history of "industrial policy" has been mixed, though also controversial on both sides. But this is an issue that cannot be avoided: whatever the positions on principle, or interpretations of history, the government of Ceara is clearly engaged in shaping the pattern of industrial and service growth, both directly through its industrial subsidy policy, and indirectly through a whole range of choices on infrastructure. Moreover, Ceara is part of a system of Northeastern states all offering sets of incentives to attract mobile capital-in a situation sometimes referred to as guerras fiscais.'°' Does this need to be modified to support competitive growth with more rapid employment growth? 5.83 Experience in the Northeast and elsewhere provides some support for the view that strategic policy can bring benefits. Benchmarking of different activities is key to guiding any action, with particular attention to areas of natural advantage or production traditions with potential for growth. Such benchmarking could be undertaken with institutions involving both public and private sectors, depending on the particular institutional context. Assessment of the supply chain (for industrial and service activities) can help identify areas where there is a rationale for public intervention, notably on infrastructure provision, information, possibly on standards and in attracting and negotiating with large firms in line with a well-crafted sectoral investment strategy. There are a number of cases of success within Brazil: for example Bahia's long-term strategy of promoting tourism, the Petrolina-Juazeiro scheme and fruit exports, ' Tendler, 2000. 98 garments in Ceara in the late 1970s and early 1980s (all cited in Tendler, 2000) and more recently in production processes in Pernambuco. 5.84 Attention to the relationships between large and small or micro enterprises within the supply chain make sense: it will be in the interests of large firms to develop quality and efficient production of the small. There has frequently been considerable focus on the role of large firms purchasing the products of small farmers or firms, since they will be interested in developing a broad market, rather than focusing on a limited number of supply. Of potentially greater value is large firms seeking markets: Ceara attracted a zipper manufacturer has played a significant role in supporting quality production of small-scale jeans producers (also cited in Tendler, 2000) 5.85 Clusters. It is well recognized that spatial allocation of economic activity matters. The state government has tended to use its subsidies to encourage footloose firms to disperse to different small towns-hopefully providing some employment, and associated multiplier effects, throughout the state. This tends to run against patterns of private development of industrial and service centers, in which groups of similar firms often move to similar locations to reap agglomeration economies, pool knowledge, develop suppliers and build pools of industry-specific labor. It is also the case that urban centers need to be a minimum size (say 50000 people) for sustained industrial development. These considerations support the ongoing revisiting of spatial policy in Ceara, in favor of greater concentration of support on relatively few secondary centers, either around natural advantages for agro-industry or tourist development, or around a sufficient scale to support footloose industrial production, where processes of concentration are already underway. (Note that this approach would be complementary to a possibly more spatially focused approach to rural development, flagged above, with the latter involving smaller population centers.) 5.86 The focus on benchmarking, the supply chain and clusters based on natural advantage or existing dynamic processes does provide support for active spatial and structural policies. However, this is likely to involve a reshaping of current support to private capital, with less use of tax write-offs, and a more careful assessment of the role for public intervention. Support to large firms may be justified where they are important externalities, but these are better in the form of once-off supports, such as initial infrastructure or land. Furthermore, the issue is not just one of financial support-a second message from successful cases concerns the great importance is that engagement with local institutions is important-in defining the issues, in linking with small firms, in negotiating outcomes. 102 5.87 The role of information and communication technology. ICT is clearly changing business throughout the world, in internal processes and in interactions between firms, suppliers and markets. While evidence is still scattered, an area of particular potential can be in help in making markets more efficient for small firms. An example of this include the "virtual souk", that links artisanal production in North Africa to markets in industrial countries, amongst a growing number of international cases. In 102 See Tendler 2000. 99 the Pernambuco cluster activity, one of the elements is to use the internet to link small- scale producers to markets.103 TabDe 5.9: Siummary of IPofloy RecommendaUana or Equi tNae Growth (Continue) Population Problem/Indicator Dlignosis Policy Recommendations Group/Exit Strategy AB income-earners Aggregate growth of Above average growth Short term Ceara above Brazil can be attributed to Maintaining good overall average, but still effective governance and governance and fiscal modest including fiscal performance, backed by sound management, policy infrastructure and sectoral continuity, infrastructure, policies (see below) and industrial policy Faster Brazil growth may Medium term help in the future, but- Management of the transition to combined with the the new government will be devaluation-South East important. firms may be less likely to Building the sklls-base and relocate. economic environment for new industries and services will necessary for long term Rural farmers and Low-productivity, Sertao has good soils but Short term workers stagnant and semi-arid conditions make Implement next phase of drought-prone broad-based expansion Projeto Sao Jos6 as means Exit strategies: rising agriculture difficult. This has been of getting modest farm productivity, exacerbated in the past by productivity increase (plus rural-non farm the cotton crisis and social benefits) to many. employment, unequal land. Out- Drought management migration to migration is part of long- should rely on established secondary cities and run solution but takes mechanisms of work Fortaleza decades. programs, linked to There are both modest participatory mechanisms opportunities for many, and (including in PSJ or other potential for large agricultural services) productivity increases for a few. Medium term Implement new 103 Case still to be documented. 100 Table 5.9 (Continue) Urban formal and Slow growth Industrial policy Short term informal workers industrial and successful in attracting Implement new strategy now service jobs, industry, focused on under formulation that: Exit strategies: concentration in newcomers, but with on > Assesses existing formal and informal Metropolitan average limited job activities in industry and job growth in Fortaleza creation. Incentives for services Fortaleza and location in interior have > Supports concentration secondary cities been too dispersed and in secondary cities discouraged possibly where this makes productive clusters. economic sense Relatively little attention to > Uses benchmarking existing production against comparable activities until recently. activities in other Tourism has been growing states/countries to fast; is employment- assure potential intensive. competitiveness > Identifies infrastructure and market bottlenecks > Uses public-private existing institutions or new partnerships for diagnosis support > Phases down of industrial subsidies; where used link to local activities with high job content Medium term Active monitoring, evaluation, participatory assessment and continued redesign of strategy 101 Tsble 5.9 Informal and self- Low productivity of High regulatory burden for Short term employed workers nicro, small and MSMEs including six Reintroduce mechanisms for medium enterprises weeks for formalization public purchases from Exit strategies: (MSME) and registration. MSMEs within legal ruling. Productivity growth Limited markets and Roll out of Crediamigo and of informal firms market information. other micro-credit (formalization of Expensive short-term Develop program for radical some) working capital; limited improvement of business long-term capital; in part environment for MSME linked to weak titling. sector. SEBRAE has broad coverage of services, but Medium term unclear how big an impact Emerging technology > Titling sector, but fixed line > Streamlining of licensing phone density low. > Demand-driven training Telephone uncertain and service provision- markets e.g. with vouchers > Encourage smalVmedium lending departments in banks. > Possible sectoral policies e.g. in tropical fruit processing and export > Industrial/service policies to encourage links with large firms and use of IT 102 V.7 Governance, Participation and Equity 5.88 The state has a central role to play in the modernization process. However, this requires an effective state, and one that is responsive to the needs and demands of the poor. 5.89 The theme of the centrality of formal institutions and their interface with society in the process of poverty reduction has moved to center stage in much development work. Poor people throughout the world, including Brazil, talk about the ineffectiveness of government institutions in responding to their needs. And, as emphasized in work on social exclusion, structures of social inequality can perpetuate poverty, including through state structures that respond to elite interests. This was discussed in the World Bank's 2000/01 World Development Report, that emphasized the necessity to undertake institutional reforms and tackle social barriers to give poor people voice and influence over the institutions that affect there lives. There is a longstanding literature on both Brazil and the rest of Latin America on the mechanisms whereby the poor are excluded, or captured by elite interests, whether in the traditional coronelismo of Brazil, or broader concerns over clientelist behavior by local and national governments. 5.90 Work on Ceara both shows the importance of these themes and the scope for successful government action. A common position is that the participatory process that was an important part of the Tucano governments was used precisely to tackle traditional problems of clientelismo. While this is disputed by some observers, the work of Tendler and co-researchers clearly documents how in some cases effective state action was capable of getting better service delivery, for example in the health agent movement in the late 1980s and early 1990s, the management of the 1987 drought, and aspects of the agricultural extension service.104 This work also underlines additional themes of great importance: the key role of action to motivate, make effective, indeed "empower' public sector workers; the role of strategic action by the state government to foster effective local servic-e delivery-often going against the short-run desires of mayors and municipal governments; and the role of proactive information campaigns, notably by radio, in spreading information, celebrating success and providing the knowledge of community groups outside government on what is being done in their and other munic(pios. 5.91 A further significant exemplar of the role of participatory engagement in the Nordeste has been the series of community based rural development projects-in Ceara the Projeto Sao Jose. An evaluation of this program found that participation was significant in strengthening-or making use of-local social capital. Under one of the delivery models, decisions are made by municipal councils-with majority representations from communities-either without (FUMAC model) or with (FUMAC-P model) pre-assigned municipal funding level and decision power. The same evaluation 104 Tend]er (1997) 103 found that these had better project results and also helped strengthen overall community influences on municipal decision-making. 5.92 While Ceara stands out for some of its past success in state action to improve service delivery and use of participatory measures, there is a major outstanding agenda. Overall Ceara's institutional structure continues to have many characteristics of inequality and dualism that are so marked a feature of the economic and social structure. Past successes clearly do not cover the full array of sectors or municfpios. There are major inequities in the quality of public services, for example in the education sector. Moreover, there is evidence of a reversion to more clientelistic practices precisely in some of the rural service delivery areas of success. An approach to drought relief in the 1987 drought, that effectively reduced the power of "drought industrialists", and put community influence and needs at the core of the work (inter alia through taking power away from mayors and making agricultural extension agents the leading player) was not followed in subsequent droughts, and concerns over clientelistic practices in the use drought relief re-emerged. Similarly there is at least some evidence of project choices in Projeto Sao Josd in the mid to late 1990s being primarily influenced by design firms or local politicians with communities pragmatically going with choices that increased their probability of at least getting some projects.105 One interpretation has been that the modernization agenda has increasingly focused on urban or formal structures, whilst more traditional clientelistic approaches have persisted in rural areas, with quite variable consequences often dependent on the quality and personality of the mayor.'06 5.93 Even where there have been successes in participatory engagement and processes these have often not been fully integrated into local government processes. For example, the municipal councils under Projeto Sao Jos6 are only beginning to look to possibilities of coordination with other services. And there would be scope for redesigning drought relief mechanisms so that they were integrated within institutionalized Projeto Sao Josd processes. 5.94 Meanwhile the demands on municipios are on the rise, with in particular an important expansion in resources going to poorer munic[pios under the federal Projeto Alvorada, especially with respect to sanitation and the federally funded Bolsa Escola (discussed above). Many of the areas for redistributive strategy outlined above also involved a heightened focus on local action, including the role of secondary towns, state-enhancements to transfers and the central role of municipalities in the provision of education services. 5.95 Beyond, but related to, inequities or problems in formal institutional systems are the informal structures of social exclusion, that are embedded in highly unequal societies. This applies in Ceara both to traditional areas of rural deprivation-where poor rural households have little influence of elite structures-and to products of the process of urbanization of change, especially in underemployed urban youth involved in 105 See Tendler, 1997 and 1999. 106 See Tendler 1999. 104 drugs and violence, child labor (notably child prostitutes), laid off adult men prone to alcoholism (for example in caju areas where women have greater access to work). Socio-cultural conditions have an important role to play here, both as a source of strength-in the traditional strengths of sertao life-and inequality or exclusion-as in cultures of violence and drugs, of gender inequity and in the relations across the classes. 5.96 The remainder of this section outlines some issues around the next phases of institutional change for equitable and effective service delivery. These are issues, rather than detailed proposals, since the latter would require in-depth institutional analysis. It should again be emphasized that this draws both on international lessons and specific experiences where Ceara has been in the lead. V.7.A Participation and State Activism 5.97 There is now considerable experience on the complexities of getting effective local service delivery. A rosy view was that decentralization would automatically be pro- poor, especially if combined with greater participation, since this would bring the services close to the people and local agents would be more responsive to "their" communities. This is now generally recognized to be too optimistic. Both local action and activism of higher levels of government can make the difference between responsive service delivery and mechanisms that inefficient, biased to local elite interests or clientelistic. The state has a role to play both in affecting allocations between municipalities or regions, and in designing mechanisms that affect allocations within areas.107 5.98 With respect to getting equitable resource use within municipalities (or regions) a mixture of action is desirable. Participation is a crucial element in assuring better design and implementation of programs. However, it needs to be designed in a way that recognizes local capabilities and social structure and is linked to explicit measures to tackle inequities. 5.99 The challenge for Ceara is in building on existing participatory mechanisms with strong traditions-for example in Projeto Sao Jose and the agentes de saude and linking these with core municipal behavior and other services. This is already happening to some degree with Projeto Sao Jos6, though as noted above, there is also evidence of dominant influence by local design firms or dominant influence by local mayors. The agentes de saude program is also being broadened to link up with other services, in order to provide more integrated access to social services by households. 5.100 Budgets are key to local process. In some attempts to foster equitable local action a portion of untied resources are transferred to local levels, linked to mechanisms for participation to make choices over their use. The participatory budgeting system developed in Porto Alegre is a major example of this, in which local communities, and 107 For a general discussion of decentralization of social programs in Brazil, see Arretche, 2000. 105 civil society are part of the process of making decisions of the use of resources.108 The fact that resources are at stake create strong incentives to participate, and the real devolution of decision-making power with budget constraints, helps break through the problem of lengthy lists of unfunded requests that are a common feature of social fund type mechanisms.'09 5.101 An example of a quite comprehensive participatory process in a poor region comes from Rio Grande do Norte, The region of Serid6 developed its own assessment of priorities for a semi-arid region, working with the government (and with resources from a World Bank supported rural development project). This involved a process of internal diagnosis, that brought together available quantitative information from IBGE with a qualitative assessment by the communities of the nature of the development problems and their vision of the future. It also led to development of a vision for the future, including both a strong emphasis on building on strengthening the sense of identity with the region, and quantitative targets for economic and social service expansion. This was worked through with the government to define budget-constrained priorities. Indicators have been developed that would be monitored, evaluated and debated. While particular leaders-including the Bishop of the region-were key actors in this case, this illustrates the features of a comprehensive approach, that the government of Rio Grande do Norte is seeking to replicate in other regions. V.7. l Ungormation, Transpearency and Social AonoufalaMlhy 5.102 A necessary complement to decentralized resource allocation and service delivery is transparency in information and social mechanisms to track and assess resource use-and debate effects. 5.103 An example from Uganda illustrates: a survey of resources for education in the early 1990s found that about a quarter of resources intended for schooling actually arrived in the classroom. A public campaign to raise this ratios involved actively sharing information at all levels, using radio and other mechanisms. By the end of the 1990s this proportion had risen to over 90 percent. This could only have been achieved because the information system to track resources existed, there was strong commitment from the national government and appropriate means of sharing information was used.110 5.104 Openness in information, and social accountability is also a central feature of the participatory budgeting in Porto Alegre, though there the system has not gone as far as tracking results. As already noted Ceara has in the past periodically, and with great effect, used public information campaigns to share information on new initiatives, to compare and celebrate success, and to let citizens know their rights. Such approaches could support an institutionalized system of participatory and results-based municipal management. 108 Abers ( "9 See Tendler, 1999 "° This example is reported in World Bank 2000, WDR. 106 5.105 The potential of e-government. Information and communications technology has immense potential for making government more open, efficient and less corrupt. It should, however, be seen as a complement, even a catalyst, but not a substitute, for institutionalized structures of transparency and social accountability. 5.106 ICTs are an excellent tool in improving the performance of the budget preparation and execution cycle. Integrated financial management systems are becoming the norm across the region, covering the main areas of budgeting, accounting and treasury management within an interconnected plafform, that allows for decentralization of operations and centralization of control and registration of operations. Such systems have not only allowed policy-makers to better align resources with needs and improve the effectiveness of social welfare programs, but have also greatly enhanced the transparency and accountability of the managers of public monies. Successful examples using the latest Web-enabled technologies are available in Argentina, Guatemala, Bolivia, and many others are following suit. 5.107 When Mexico implemented the first phase of its on-line procurement system, just the fact that the information on open contracts was made available to potential bidders, beyond the reduced number of usual contractors, caused prices for certain services to drop by almost 80%. In other areas the savings have not been as dramatic, but it is estimated that an average savings of at least 20% is being realized due to the increased competition and transparency of this system. The U.S. Department of Commerce also estimates that the use of e-procurement can amount to savings of about 20% of the public budgets. The experience in Chile has been similar. The new technologies can reduce opportunities for graft. In the US, 'collusion detection software' is being used to root out impropriety in public bids. In Singapore, even the courts are being hooked up to electronic filing and retrieval systems to boost productivity. 111 5.108 It is our understanding that the Federal Government and some states are already planning the implementation of an e-procurement system, and some states already are implementing such a system. If not already underway, this would make sense for Ceara, at the same time allowing local governments to conduct their procurement online. Again this will work differently in different contexts: where there are many potential firms, in urban and more densely populated areas, competitive pressures can reduce inefficiency and corruption. But it won't solve problems- where there are few firms in dispersed areas. 5.109 In a number of developing cities and regions, ICTs are being used to speed delivery and reduce corruption in the issuing of permits. In Andhra Pradesh, India, for example, ICTs have been used as an important tool of the Government in its effort to become "SMART" (simple, moral, accountable, responsive and transparent). One example is the registration of Deeds and Stamp Duties in AP. Using traditional methods this took 13 cumbersome steps in a highly opaque process that invited bureaucratic delay and corruption. It took from three to as many as 15 days-and the process Forbes January 11, 1999 107 involved the registration of over 120 million documents a year. Using a new, recently introduced computerized system, the same task can be accomplished in just over two hours, with far less opportunity for graft. 5.110 The State Government of Bahia has created the "Shopping Mall for Public Services" as a one-stop shop (accessible by multiple means: in person, through the Internet, by phone) including land registration services, information on government services, support in conducting transactions with the public sector, and sources of assistance, in a format that can be easily accessed by low-literacy groups. It gathers in a single place different public services delivered to the citizens by national, sub-national and local institutions, reducing the need for citizens to walk around for public services to get the officials documents. The SAC units covers 14 sectors and have the capacity to serve 600 citizens per day. It offers public services such as renewing ID cards, social security numbers, company registrations public contract information, unemployment insurance, tourist information and consumer support. A mobile SAC has also been developed, that allows geographically isolated groups access to government services not usually available. The system might have a particularly important role in improving access to land registration for the poor. Medical and police information could also be provided. 5.111 Finally, ICT could be used as an input to mechanisms that allow citizens to compare performance-for example across municipalities, schooling, clinics and other services. Use of radio has been an important mechanisms of sharing positive results of innovations in Ceara-and used actively by the state government in earlier campaigns (for example in the agentes de saude program'12). Both radio and internet based information mechanisms could play a deeper and more institutionalized role if linked to information systems that made tracked, and made public, resource allocations to muncipalities, plans and results. V.7.C Combating SocWG Inequalitfies andi OnvoMvlng the SocWiOy Exciudedi 5.112 A recurrent theme of work on decentralized delivery is that the extent that local governments will be more pro-poor depends fundamentally not only on their technical capabilities but on the social structures in which they are embedded. If anything, the risks of elite capture are more likely at the local level (as is well known in the Brazil case). One study of Bangladesh found that provision of a particular service intended for the poor (stipends for school attendance) was more or less targeted to the poor in implementation in a systematic way: service delivery was not pro-poor where local organizations were those mainly of the better off; it was pro-poor where the dominant local organizations were of the poor.113 5.113 A common result of studies of the nature of networks amongst poor groups is that ties within extended families or communities are frequently strong-and an 112 Tendler (1997) 113 Galasso and Ravallion 108 important source of coping, especially in the management of vulnerability. However, ties to other groups and especially to formal institutions are typically weak, diminishing their capacity to obtain economic and social services. 114 5.114 The irriplications for public action depend on local conditions and the profile of those excluded. Where there are organizations of the poor but their influence is weak, effective participatory process can be effective, that gives such organizations a voice and influence over plans, budgets and in evaluating progress. Where this is not the case, more proactive measures will be needed to bring in the excluded, or strengthen capabilities for collective action. 5.115 Of particular interest is groups that are outside the social mainstream, victims of past deprivation or survival strategies. Child workers are a powerful example, that has rightly received both national and international attention. Reaching such groups requires focused effort, frequently engaging with civil society groups, and often working to support or break cultural practices. Brazil is rich in cases of individual action. For example in Fortaleza, there are many programs to get streets off the kids--the Programa Crian,a Fora da Rua, works on building partnerships between the state, companies and civil society. It works directly with the families, using awareness campaigns in local media and education grants. 0 Projeto Auto-Estima das Criangas Negras in Maranhao aims to build the cultural identity and self-esteem of very young children-between 0 and 6-working in poor rural areas. The Projeto Casa de Zabe/i in Teresina works with adolescent girls at risk, using artistic, sporting and educational activities, but with financial incentives to stay in school. The Casa do Caminho in Fortaleza provides a place to stay, clothes, food, health and legal services for women victims of violence."15 And the Projeto Quatro Vargas uses a range of sporting, psychological, education, artistic activities and legal services to work with excluded groups. 5.116 These examples are illustrative of the vitality of specific local action. They typically involve activism of civil society groups, sometimes linking with state services. Some make use of incentives to bring in excluded groups, along principles similar to the Bolsa Escola. The state government is already interfacing with some of these schemes. It can provide support through systematically providing the mechanisms and space for civil society groups to interact with public service delivery, engage in budget processes and monitor and debate results. V.8 Implications for Spatial Policy 5.117 Most of the areas of sectoral or economy-wide policy have a spatial dimension Both the poverty and income profile and the likely dynamics of change have strong spatial characteristics. The implications are outlined here. 114 See Wold Bank, 2000, WDR 2000/01. Ties within communities are there termed bonding social capital, to other external groups and formal institutions "bridging and linking" social capital 115 All these examples are taken from the Programa Gestdo PNblhca e Cidadania of the Getulio Vargas Foundation and the Ford Foundation. 109 5.118 As described in Section II Ceara displays two marked features of the spatial distribution of the population and well-being. First, a large minority of the population, and almost half the poor, live in dispersed settlement patterns in rural areas, primarily in the semi-arid and drought-prone sertao. Many live in conditions of extreme income poverty, and suffer relatively high levels of deprivation along other dimensions of well- being, including health status, skills, access to basic economic and social services and insecurity. Large family sizes imply a relatively high fraction of children. 5.119 Second, there is an unusually high concentration of urban population in one single metropolitan (Fortaleza) region. This metropolitan region itself has high levels of inequality, ranging from modern sections with incomes and facilities of a developed country through to low income favelas and semi-rural areas, with high levels of income poverty, and in some parts sub-cultures of violence, drugs, and alcoholism. There are some ... [see child labor report ]street-children, according to the PNAD, with child prostitution the most widely known area of exploitative or dangerous child work. The next sized town is a small fraction of Fortaleza and there is considerable interest in fostering a more dynamic policy of secondary towns both to reduce the pressure on the metropolis and to foster strong linkages between such towns and their rural hinterland. 5.120 As discussed in the section on growth the difficult agro-ecological conditions of the sertgo greatly limits the scope for a broad-based rural productivity transformation that was so marked a feature of equitable growth in all cases of East Asian success, except Hong Kong and Singapore. However, international evidence shows that the dynamic transition to a predominantly urban location takes decades, and generally the deepest poverty remains rural. This applies even where there is a dynamic agriculture: if much of the rural areas are only going to enjoy modest productivity increase the risks of many getting left behind are high. 5.121 This leads to the following spatial "mapping" of the sectoral and institutional recommendations. Area of strategy SpaffaW dimension Education Strongly equalizing across space: additional affirmative action needed to provide quality basic education in all rural areas (to include most deprived groups), backed by high quality upper secondary and tertiary education in secondary cities Transfers Equalizing across space: priority to indigent groups will lead to disproportionate rural focus, with also relatively high focus on periurban areas in secondary cities and RMF Growth Basic support for some core infrastructure modestly equalizing; focus on productive opportunities, driven by local conditions likely to lead to concentration of production both in secondary towns and continued growth in RMF. This would be complemented by special efforts to reach poorer locations/favelas in RMF Institutions Institutional design and effort needs to balance access and responsiveness to all-especially dealing with inefficiencies and clientelist practices in rural areas-while paying attention to the efficient scope, size and reach of institutions (as already evidenced Iin some excessively small municfpios); this will be service-specific. 110 5.122 Alternatively, this spatial approach can be summarized in terms of three features: (a) Broad-based provision of core services to support human development, connectivity, basic economic infrastructure and reduce vulnerability for all, using the geographic structure of poverty as an important indicator for targeting, balanced by considerations of the minimum efficient scale of municipalities and services (e.g. secondary and tertiary social services). (b) Support for a concentration of resources in areas with productive potential outside the metropolitan region-especially in secondary cities, but also in those areas with productive potential in rural areas. This is intended to reap productive opportunities within secondary'cities and reduce the relative preponderance of the metropolitan area. (c) Support for equitable growth and inclusive services within the metropolis- that under plausible scenarios will continue to play a dominant role in the state in the next 20 years and beyond--especially with respect to peripheral metropolitan areas, informal production and vulnerable groups. 5.123 This would have to be underpinned by more effort on the overall data and diagnostic base for the spatial distribution of public effort, driven both by poverty mapping and assessment of dynamic changes and potential across space. While the geographic information base is very rich in some areas, there is only limited information on the spatial or rural-urban distribution of sectoral spending. 111 For references see the bibliography in the main report. 112