Report No. 21858-BR Brazil Eradicating Child Labor in Brazil December 6, 2001 Human Development Department Brazil Country Management Unit Latin America and Caribbean Regional Office u Document of the World Bank CURRENCY EQUIVALENTS Currency Unit - Real (R$) December 1999: R$ 1.79/US$ December 2000: R$ 1.95/US$ July 2001: R$ 2.50/US$ FISCAL YEAR January 1 - December 31 ACRONYMS AND ABBREVIATIONS BA State of Bahia CUT Central Unica dos Trabalhadores DF Distrito Federal ECA Statute of Children and Adolescents FGRM Fundo Garantia de Renda Minima ILO International Labor Organization IPEA Instituto de Pesquisa e Economia Aplicada IPEC International Program for the Elimination of Child Labor LDB Lei de Diretrizes Basicas MEC Ministerio de Educacao MOC Movimento de Organiza,co Comunitaria MPAS Ministerio da Previdencia e Assistencia Social NGO Non-governmental organization PE Pernambuco PETI Programa de Erradica,ao do Trabalho Infantil PME Pesquisa Mensal de Emprego PNAD Pesquisa Nacional Amostra de Domicilios PROGRESA Programa de Educaci6n, Salud y Alimentaci6n RAIS Rela,co Anual de Informa,oes Sociais SE Sergipe SEAS Secretaria de Assistencia Social SESAB Secretaria da Sauide do Estado da Bahia SEC Secretaria de Educavao da Bahia SEI Superintendencia de Estudos Econ6micos e Sociais da Bahia SEPLANDES Secretaria Estadual de Planejamento e Desenvolvimento Social SETRAS Secretaria do Trabalho e A,co Social da Bahia STR Union of Rural Workers UFBA Universidade Federal da Bahia UNICEF United Nations International Children's Fund Vice President, LCR: David de Ferranti Country Director, LCC5C: Vinod Thomas Sector Director, LCSHD: Xavier Coll Team Leader, LCSHD: Guilherme Sedlacek TABLE OF CONTENTS ACRONYMS AND ABBREVIATIONS PREFACE FISCAL YEAR IV ACRONYMS AND ABBREVIATIONS IV 2. INTRODUCTION 1 CHILD LABOR - THE POLITICAL AND LEGAL DEBATE 1 DEFINING CHILD LABOR 2 OVERVIEW OF THE INCIDENCE OF CHILD LABOR - WORLD, LAC, AND BRAZIL 2 PRIORITIZING THE WORST FORMS OF CHILD LABOR 2 3. CHILD LABOR INDICATORS AND TRENDS IN BRAZIL 5 INCIDENCE AND EVOLUTION OF CHILD LABOR IN BRAZIL 5 WORK INTENSITY: THE TIME TRADEOFF 7 PAID AND UNPAID CHILD LABOR 9 OCCUPATIONS OF THE WORST FORMS 9 4. THE DETERMINANTS AND OUTCOMES OF CHILD LABOR 11 DETERMINANTS 11 THE LINK BETWEEN CHILD LABOR AND POVERTY 12 THE STRUCTURAL DETERMINANTS OF CHILD LABOR 12 How ECONOMIC CRISES AFFECT CHILD LABOR 1 3 OUTCOMES 14 CHILD LABOR LEADS TO Low EDUCATIONAL ATTAINMENT AND Low EARNINGS 14 THE TRADEOFF BETWEEN WORK AND SCHOOL 15 THE IMPLICATIONS OF CHILD LABOR FOR ADULT WAGES, INCOME, AND POVERTY 1 6 5. BRAZILIAN EFFORTS TO ADDRESS CHILD LABOR 18 THE CHILD LABOR ERADICATION PROGRAM 18 PROGRAM OBJECTIVES 19 EVALUATING THE PETI 1 9 BEHAVIOR AND ACCESS TO SERVICES 25 AN ILO-IPEC STUDY "BODE-ESCOLA" IN THE SISAL REGION OF BAHIA 26 ii A QUANTITATIVE ASSESSMENT OF THE PETI IN PERNAMBUCO, BAHIA, AND SERGIPE 27 THE BOLSA EscoL4 PROGRAMS 29 BOLSA ESCOLA IMPROVES LEARNING OUTCOMES 30 CHILD LABOR IN URBAN AREAS 33 STREET CHILDREN 33 SEXUAL EXPLOITATION 35 DOMESTIC SERVICE 36 6. SUMMARY OF LESSONS LEARNED AND FUTURE CHALLENGES 40 OVERALL DESIGN ASPECTS 40 TARGETING AND PROGRAM EXPANSION 41 THE CASH TRANSFER LEVEL 42 THE AFTER-SCHOOL PROGRAM - JORNADA AMPLIADA 43 THE DURATION OF THE PROGRAM AND POST-PROGRAM OPTIONS 43 IMPLEMENTATION ISSUES 44 OUTSTANDING CHALLENGES AND RESEARCH AGENDA 50 CHALLENGES TO OVERALL ERADICATION OF THE WORST FORMS OF CHILD LABOR 50 CHALLENGES OF EXPANSION TO URBAN AREAS 50 BIBLIOGRAPHY 52 I ables TABLE 1. ESTIMATES OF CHILDREN WORKING IN THE STREETS IN SELECTED CITIES 34 TABLE 2. STREET CHILDREN PROGRAMS IN FIVE METROPOLITAN AREAS OF BRAZIL 38 Box3s Box 1: WORST FoRMS OF LABOR IN BAHIA AND PERNAMBUCO ........................................ 2 1 Box 2: BOLSA ESCOLA VERSUS PETI ................................................ 29 iii PREFACE This report was prepared by Guilherme Sedlacek (Task Manager, LCSHD) and Emily Gustafsson-Wright (Consultant, LCSHD). The major contributing studies are the following: "Trabalho Infantil no Brasil: Entre a Cana e o Sisal," (by A. S. Moura, A. Butto, P. B. Fontenelle, and V. L. Jatoba); "Evaluating the Impact of PETI on Child Labor Supply and Schooling Demand in Rural Northeastern Brazil: The Case of Pernambuco, Bahia and Sergipe," (by Yoon-Tien Yap, consultant LCSHD); "A Descriptive Look at Child Labor in 1999 and its Trends during the 1990s: The Case of Brazil;" (by Yoon- Tien Yap); "The Implications of Child Labor for Adult Wages, Income and Poverty: Retrospective Evidence from Brazil," (by Nadeem Ilahi, Peter Orazem, and Guilherme Sedlacek, LCSHD), "Microeconomic Instability and the Children's Human Capital Accumulation: The Effects of Idiosyncratic Shocks to Father's Income on Child Labor, School Drop-Outs and Repetition Rates in Brazil," (by Marcelo Cortes Neri and Emily Gustafsson-Wright, consultants LCSHD), Guilherme Sedlacek (LCSHD), Daniela Ribeiro da Costa, and Alexandre Pinto; and "Gender Note on Child Labor and Street Children in Brazil," (by Emily Gustafsson-Wright, (consultant LCSHD), and Hnin Hnin Pyne (LCSPG). This effort is a result of the World Bank's collaboration with UNICEF, ILO, and the Inter- American Development Bank. The main intended audiences for this report are the federal, state, and municipal governments of Brazil, the World Bank staff, donors and partner organizations. The authors would like to specially thank Peter Orazem for the methodological guidance throughout the work. We would also like to thank Alexandrina Sobreira, Sonia Rocha and Ricardo Neri for the support and excellent background papers. The authors would also like to thank Cesar Granados and Fabiana Imperatriz (LCSHD) for assistance. Last but not least, we would like to thank the government of Brazil for their collaboration and the countless individuals in Brazil who contributed time, information, and insight into the production of this report. While this report has been discussed with institutions and individuals of the Brazilian Government, the views expressed in this report are exclusively those of the World Bank. iv CURRENCY EQUIVALENTS Currency Unit - Real (R$) December 1999: R$ 1.79/US$ December 2000: R$ 1.95/US$ July 2001: R$ 2.50/US$ FISCAL YEAR January 1 - December 31 ACRONYMS AND ABBREVIATIONS BA State of Bahia CUT Central Unica dos Trabalhadores DF Distrito Federal ECA Statute of Children and Adolescents FGRM Fundo Garantia de Renda Minima ILO International Labor Organization IPEA Instituto de Pesquisa e Economia Aplicada IPEC International Program for the Elimination of Child Labor LDB Lei de Diretrizes Basicas MEC Ministerio de Educacao MOC Movimento de Organizagao Comunitaria MPAS Ministerio da Previdencia e Assistencia Social NGO Non-governmental-organization PE Pernambuco PETI Programa de Erradicac,ao do Trabalho Infantil PME Pesquisa Mensal de Emprego PNAD Pesquisa Nacional Amostra de Domicilios PROGRESA Programa de Educaci6n, Salud y Alimentaci6n RAIS Rela,ao Anual de Informa,oes Sociais SE Sergipe SEAS Secretaria de Assistencia Social SESAB Secretaria da Sauide do Estado da Bahia SEC Secretaria de Educacao da Bahia SEI Superintendencia de Estudos Econ6micos e Sociais da Bahia SEPLANDES Secretaria Estadual de Planejamento e Desenvolvimento Social SETRAS Secretaria do Trabalho e Acao Social da Bahia STR Union of Rural Workers UFBA Universidade Federal da Bahia UNICEF United Nations International Children's Fund Vice President, LCR: David de Ferranti Country Director, LCC5C: Vinod Thomas Sector Director, LCSHD: Xavier Coll Team Leader, LCSHD: Guilherme Sedlacek v EXECUTIVE SUMMARY 1. Brazil has made eradicating child labor a policy priority. In light of Brazilian efforts and commitment to achieve that goal, this report discusses the history, determinants, and implications of early entry into the labor market on a child's well being and poverty. The report examines background assessments commissioned to inform this analysis, of the ongoing programs to prevent and eradicate the worst forms of child labor in Brazil, and assesses ongoing program expansion efforts and challenges to be faced. The report also seeks to identify promising strategies and, as part of efforts to expand coverage, seeks to address the needs of highly vulnerable children in urban areas who are engaged in such activities as the commerce of drugs, in prostitution, who are street vendors, or who are working in other dangerous activities. The main audiences for this report are the federal, state, and municipal governments of Brazil, the World Bank, and other donors and partner organizations. This report is a result of the World Bank's collaboration with UNICEF, the ILO, and the Inter-American Development Bank. 2. This report reviews the evidence of child labor in Brazil and the government's outstanding efforts to eradicate its worst forms. The Child Labor Eradication Program (PETI) is unique in the world in that it provides an implementable strategy to address by 2002 the incidence of the worst forms of child labor, both in urban and rural areas. The full 6-year program is being implemented in two phases. From 2000 to 2002, there will be a gradual scaling up of PETI to reach 866,000 children. From 2003 onwards, it is the consolidation and phasing out phase. Brazil is also outstanding for providing the social and political consensus for the execution of such an effort. The evidence presented on the risks and outcomes of early entry into the labor market, and the link between poverty and child labor, justifies the need for a program such as PETI. We have summarized the successes and lessons learned after five years of program implementation, and we have made suggestions based on these findings in an effort to contribute to the internal technical debate. This is clearly a success story. In an effort to expand, to include more urban areas, and to attempt to eradicate all of the worst forms of child labor, the resolution of some challenges is still part of an ongoing learning process (the issue of prostitution, for example, is being handled by a specific program complementary to PETI, the "Sentinela" Program). 3. Rationale. Is a program such as the PETI justified in Brazil? The evidence presented in this report highlights two central points for why the program is warranted: (a) the declining but large incidence of child labor that remains in Brazil, and (b) the significant impact that early entry into the labor market has on a child's future earnings vi and educational outcomes, not to mention the potential immediate and long-term developmental impacts. Nonetheless, resolving the problem of child labor does not simply entail forcing the removal of children from their place of work. Many poor families often depend on the earnings of their children, since this income may represent a significant portion of family income. Hence, simply cutting off this income would do little to help either the children or the rest of the family. A compensatory cash transfer is an appropriate mechanism to facilitate the transition. 4. While the original objective of PETI was to remove children in rural areas from the worst forms of child labor, the program uses education as the central tool to achieve its goal of addressing a much larger problem-poverty and its constant and persistent intergenerational transfer. As the former professor and first president of Tanzania Julius Nyerere once said, "Education is not a manner in which to escape poverty, it is a way to fight it." Cash transfer programs contribute to the removal of mechanisms, traditionally in existence in school systems, that exclude the poorest students. "Normally" functioning schools tend to weed out students from poor economic and social backgrounds, based on performance and promotion criteria. In the absence of the program, PETI beneficiaries would have trajectories that would lead them to early dropout. By ensuring that schools retain students with high probability of dropping out, the program promotes the rupture of one of the strongest mechanisms of reproduction and legitimacy of inequalities: early school exclusion. 5. Child labor incidence remains high in rural areas of the Northeast, and most working children combine work and school. Chapter 2 discusses several characteristics of child labor in Brazil, which supports the argument for the need of a child labor eradication program. First, while overall incidence of child labor in the country declined from 20 percent in 1992 to 15 percent in 1999, it is crucial to note that this decline has occurred mostly among children who only work. The incidence of child labor among children who combine work and school, which represents the largest portion of child workers, has remained virtually constant over the last decade. The incidence remains high in rural areas, with a rate of approximately 37 percent compared to almost 8 percent in urban areas. Agriculture is by far the largest employer of children, even in urban areas. Overall, the evidence demonstrates, fairly consistently, that female incidence of child labor is about half that of boys. Despite this evidence, the report recognizes that absolute incidence alone is a weak indicator of the reality facing the country and the children involved in child labor today. In fact, several other factors, such as work intensity and type of activity in which children are involved, must be examined. Additionally, data relating to children working in the streets, and girls involved in prostitution or domestic labor, may be inaccurate or lacking. 6. Early entry into the labor market has considerable consequences for the future earnings and educational outcomes of working children. Evidence presented in Chapter 3 demonstrates that child labor reduces lifetime earnings by 13 percent to 17 percent overall, and raises the probability of being poor as an adult by almost 8 percent (Ilahi, Orazem, and Sedlacek 2000). Findings suggest that working children tend to enter school four months later than children who do not work, and on average they lag behind by 0.5 years more than children who do not work. vii 7. Addressing child labor through cash transfer programs is promising in light of the link between poverty and child labor and the demonstrable impact of income shocks on poor households. Chapter 3 also notes that the largest proportion of child workers come from families who are in the lowest income quintiles in Brazil. Within the lowest income quintile, 27 percent of 10-to-14-year-old children work.' Also, parents' educational attainment, which may act as a proxy for income, is highly negatively correlated with child labor, suggesting a link between child labor and poverty. Additional findings suggest that children from poor households are more susceptible to income shocks and are more likely to work as a coping strategy in time of crisis. Children in the lowest income quintile are more likely to begin working as a result of a shock to their father's income, and are also more likely to drop out of school and to repeat a grade than the poorest children who experience no shock.2 8. Education is a tool to fight child labor through fighting poverty. Chapter 3 also presents retrospective evidence of age of entry into the labor market, which demonstrates that working during the first 12 years of life has an indirect effect on future income, by reducing completed years of education by almost three years, and on adult hourly wages. Policies that delay age of entry into the labor market may significantly reduce adult poverty incidence through the indirect effect of increasing years of non-labor and years of schooling. These policies are especially appropriate given the evidence of the intergenerational transfer of child labor, low schooling attainment, and poverty (see - de Souza and Emerson 2000). 9. Chapter 4 evaluates the effectiveness and sustainability of the PETI and assesses the Brazilian government's efforts to expand the programn to additional municipalities and urban areas. The analysis is based on an impact evaluation of the program in states of Bahia, Pernambuco, and Sergipe; on a qualitative and institutional assessment in the first two states; and on earlier assessments. The assessments examined five main areas with respect to impacts of the PETI: (a) whether the program was successful in targeting, (b) whether the program was successful in decreasing rates of child labor, (c) whether the program managed to increase school attendance, (d) whether there were noticeable changes in the behavior of program participants with respect to child labor and education, and if there were other positive outcomes of the program, and (e) whether the cash transfer amount is perceived as appropriate, given the expectations of intended impacts. 10. The program was successful in reaching its intended beneficiaries. A study of the states of Pemambuco, Bahia, and Sergipe reveals that about 90 percent of the children who participated in the program were eligible based on the selection criteria of (a) at least one child aged 7 through 14 in the household, and (b) per capita household income equal to less than half of minimum wage. A qualitative analysis of Pemambuco and Bahia, which describes characteristics of participating children and families, suggests that program participants are on average aged 7 to 11, and come from large families where parents have little or no education and are working in sisal or sugarcane, earning low 1Based on PNAD, 1999. Based on results from Neri et al., using PME data. viii wages. It also demonstrates that the program is reaching the poor in areas where some of the worst forms of labor occur. 11. The assessments demonstrate that the program has been successful in achieving its objectives of reducing rates of child labor. As a result of the program, the probability of child labor in the states of Bahia, Pernambuco, and Sergipe declined by 25%, 5%, and 10%, respectively. The impact on the hours worked was also the highest for Bahia. In a qualitative questionnaire, families were asked if children were working after the implementation of the program. The responses revealed that in Bahia there remained a small number of children still working who were benefiting from the program. In Pernambuco there were claims of no child labor. 12. School attendance has increased as a result of the program. Household levels of school attendance increased by 16 hours per week on average among children participating in the program since 1997. Individual estimates of increases in school attendance were 18 hours per week for those participating in the program in Pemambuco since 1997. A separate assessment of Pernambuco and Bahia demonstrated a decrease in nonpassing rates in Pernambuco, and even greater impacts in Bahia. Rates of nonattendance declined dramatically in Bahia. 13. Changes in attitudes and other changes have occurred as a result of the program. A survey of impacts with respect to attitudes about child labor demonstrated that a large number of families continue to perceive child labor as beneficial to the children and the family. Few described education as an important part of a child's development. Children, however, were found to be tidier and more positive about the future compared to children who had not participated in the program. Schools with children participating in the program tended to have better infrastructure and school materials than schools without the program. Families also noted the jornada ampliada (after-school program) to be of great importance because it provides the children with increased educational and extracurricular activities. Many families noted that the school lunch provided was an important contribution to the well being and health of the children. Mothers commented that the cash transfer (which is supposed to be given to the mother and was in most cases) had increased their independence and family responsibility. 14. Most families found the cash transfer amount to be appropriate. The transfer amount was approximately R$25 per child per month in Bahia, and R$50 (for one or two children) or R$100 (for three or four children) in each household in Pemambuco. (A common benefit criterion was set for all states from 2001 onwards; each family would receive a "Bolsa" for each eligible child.). Interviews with families in participating municipalities in these states revealed that the majority of families found the level to be fair, though a small percentage in Bahia felt it was too low. In both states, families claimed to have spent the majority of the stipend on clothing and food, followed by health, medicine, and school supplies. 15. Chapter 4 also discusses the problem of child labor in urban areas. Traditionally, in Brazil these efforts have been tackled by NGOs; however, in 1999 the PETI began a major initiative to address the need of vulnerable children in urban areas, including (by ix the end of 2002) 266,000 children and adolescents aged 10 to 14 working in such activities as street vending, garbage collection, and drug trafficking. The urban PETI program is similar to that in rural areas, though it provides a slightly higher cash transfer per child (R$40 per month) and R$10 per child per month to the jornada ampliada. Though this effort has yet to be evaluated, an evaluation of promising NGOs, addressing children working in the streets, was commissioned (jointly with the United Nations International Children's Fund, UNICEF) as part of this report.3 16. The situation of urban child laborers and street children is quite different from that of rural child laborers. Due to the public nature of work in the streets, urban child laborers are subject to a lack of rights, and are much more vulnerable to risks and hazards, such as drug addiction and violence. Links to families are often weak, even if children return home to families at the end of the day (80 percent are reported to return home weekly). Prostitution, which affects mostly girls, is increasing and becoming an area of employment for boys as well. This form of sexual exploitation, which has a heightened social stigma attached to it, leads to increased violence and the probability of HIV infection. 17. Programs that address child labor in urban areas must work together with families, schools, and the community, with particular emphasis on the rights and well being of the child. The lessons learned from the experiences of the NGO programs, which have a long history of combating child labor in the streets of urban areas, demonstrate that the PETI is well on its way to being on target for best practices in addressing the worst forms of child labor. The link that PETI establishes between the child, his or her family, and the school, and the conditional cash transfer given, provides an anchor in improving the rights and well being of the child. The methodologies typically implemented by NGOs can contribute with meaningful lessons for PETI strategies in terms of their emphasis on focusing on the self-esteem and social skills of children and the development of psychosocial activities and inclusion of parents. Their methodologies for coordinating and strengthening social institutions working in these communities, and preparing children for a working environment, are also crucial best practices to consider. 18. The Brazilian government has made a remarkable commitment (unique among countries with similar level of development) to address the issue of the highly vulnerable urban children. The current program arrangements assign to the municipalities the responsibility for program implementation; their efforts could be further strengthened by programs specifically designed to finance congruent NGOs initiatives. The report recognizes the on-going efforts of the Brazilian Government to develop a financial and monitoring system (as well as the program evaluations already contracted), but emphasize the critical need for better and more systematic data for targeting, monitoring, and evaluation, and the setting of the cash transfer level. These are some of the remaining challenges which are further discussed in the conclusions and policy recommendations. 3We follow the convention of defining children who work in the streets of urban areas as "street children." We view them as a subset of child laborers, meaning that while children working in the street are engaged in child labor, child laborers are not necessarily street children. By this definition, street children may either live and work in the streets or only work in the streets. x There is a wealth of institutional capacity ready to be energized in the NGO and civil society community in order to meet these challenges. 19. Expansion to urban areas presents challenges: * Child labor is more difficult to target in urban areas because occupations are much more heterogeneous than those in certain rural areas, and because many of the forms of child labor deemed harmful take place in the streets. Data for children vending items, collecting trash, selling drugs, and engaging in prostitution are difficult to obtain. * Establishing a cash transfer level in urban areas presents a challenge because the types of activities in which children are involved are much more heterogeneous. In the proposal for the urban component of the program, the government has recognized that the level must differ from that in rural areas, given that wages are higher in urban areas. * The types of labor in urban areas are more complex in terms of developmental and physical harms. However, the risks and exposures that street children confront daily, and the public nature of their work and their hardships, make it a priority concern. * For children living and working in the streets, exposure to drugs and violence adds to other harmful conditions, such as traffic accidents, extreme weather, poor sanitation, and psychological distress induced by ostracization and discrimination. These factors may require additional methodologies and support, such as counseling and rehabilitation. * Strategies to involve families in the program may be more complicated given the potential weak link between children and their families. 20. The report concludes with Chapter 5, which summarizes the institutional findings from the assessments of the PETI, expands upon the policy implications of these findings, and suggests future strategies for the governrnent and other organizations involved in an effort to eradicate child labor. The assessments note that the PETI is undoubtedly unique in the world and has indeed demonstrated impressive abilities in obtaining its goals of decreasing the incidence of the worst forms of child labor, increasing school attendance, and improving income and quality of life for its beneficiaries. In only five years, the program has already managed to reach almost 400,000 children, and it plans to eradicate all of the worst forms of child labor by 2002. As the program expands to other areas, including urban areas, there are a number of lessons that can be learned from their experiences. 21. The lessons learned and future challenges discussed highlight design aspects, including whether the program should have a preventive or a remedial focus, targeting and program expansion, the cash transfer level, topics regarding the jornada ampliada, and program duration/post program options. Chapter 5 also gives recommendations regarding implementation aspects, such as the selection of beneficiary families, program information dissemination, coordination between Social Assistance and Education, school and jornada ampliada capacity issues, the process of transferring the cash, monitoring and evaluation of the program, participation of civil society, and the necessity xi for complementary social assistance programs. The chapter closes with outstanding challenges and the future research agenda. Design Aspects Preventive versus Remedial Program Focus 22. The PETI was originally designed to remove and protect children in rural areas of Brazil from the worst forms of child labor, and thereby targeted children already working or those at risk of working. This targeting method has proved to be successful. The second phase of PETI, in which only working children were targeted, has the potential risk of attracting children to work as a means of entering the program. . The government has already made proposals to address this aspect. Projeto Alvorada (proposed in September 2000) solves this issue by only including in the PETI children who had worked for two years, and by including the rest of vulnerable children in a separate, but integrated, cash-grant initiative financed by the federal government program-Bolsa Escola (result of other federal education policies and programs such as Fundo de Garantia da Renda Minima). This national program will seek, over the next few years, to implement a Bolsa Escola (a cash transfer tied to school attendance) program in the whole country, including the poorest municipalities of the country. Targeting and Program Expansion 23. Prior to 2000, the target population of the PETI includes children aged 7 to 14 from poor families working, or who have the potential to be working, in activities deemed the worst forms of child labor, and households which have a per capita income of less than half of a minimum wage. (From 2000 onwards, the program lost its preventive character and became only remedial, i.e. eligible children are those working in activities deemed the worst forns of child labor.) The assessments found that the targeting strategy was appropriate, given that the largest number of children working in Brazil live in the areas selected by the geographical targeting, and are found among the poorest income quintile groups. For the future, it will become increasingly important to change the focus of the program from school enrollment to completion of basic education (grade 8) and progression to secondary education, once universal school participation of all children aged 7 to 14 is achieved. The two main concerns related to the program expansion are (i) how to adapt the targeting mechanisms, given that geographical targeting may be inappropriate for urban areas; and (ii) identifying other priority activities among the worst forms. * The assessments suggest that benefits be expanded to cover children aged 15 and 16, who have a higher risk of dropping out and of child labor. This recommendation is consistent with the Government strategy (based on Emenda Constitucional No. 20, which prohibits child labor before age 16) to address the needs of the graduates from the PETI program through the Programs Agente Jovem (children aged 15-17) and Agentes da Familia. xii * It is recommended that workers unions, universities, and other members of civil society, in cooperation with the Municipal Commissions for the Prevention and Eradication of child labor, should all take part in the task of detailed census and mapping of localities and types of child labor occurring in specific areas and municipalities. The report acknowledges that broad civil society participation is already present in the State Commissions for the Prevention and Eradication of Child Labor, but would emphasize the need to strengthen their participation and ownership on decisions related to the census and mapping tasks. The Cash Transfer Level 24. Originally, the amount and mechanism of the cash transfer level differed from state to state. In Pernambuco, a family received R$50 or R$100 per I to 2, or 3 to 4 children, respectively. In late 1999, when the programs were consolidated, a common benefit level of R$ 25 per child was established across all states, with the change to take place over a period of one year. The questions which arose in the assessments were whether the transfer should be given per participating child or by family; how much the cash transfer should be; and should it differ between rural and urban areas, across states, and across municipalities. 9 Our findings suggest that the PETI has demonstrated that the current rural benefit level is appropriate vis-a-vis the opportunity cost of a child working, and that the current benefit level may need to be differentiated arnong urban areas, to reflect differences in costs of leaving. The optimal design should consider flexible amounts depending upon area (urban versus rural, main activity, earnings, etc.). * Overall, the evidence gives support to family benefit designs on a per-child basis; however, in accordance with the PETI design, families should be expected to register all eligible children in the program. The After-School Program - Jornada Ampliada 25. A critical innovation of the PETI is the jornada ampliada, which is intended to provide additional educational and recreational stimulus for the children, and hinder them from combining work and school. It also provides the children with a school lunch, which improves nutrition and health and aids families financially. The social assessments revealed that families were satisfied with the jornada ampliada's ability to provide stimulus for their children, and to provide a nutritional supplement. Two aspects regarding the jornada ampliada that were noted by the assessments are the hiring and training of monitors, and the lack of a set curriculum with a pedagogical strategy. * The social assessments recommend that the Municipal Secretariats of Education formulate a more systematic and uniform mechanism to hire monitors, for establishing grounds for transparency, and ensuring the selection of the best- qualified candidates for the post. Training of monitors for the program should include aspects of PETI program design and implementation, the harms of the worst forms of child labor, and the benefits of education and gender sensitivity. xiii * In coordination with the state-level Secretariats of Education, the municipality should undertake the responsibility for establishing curriculum and activity guides for the after-school program. * As per PETI Operation Manual, Municipal Education Secretariats should seek a better coordination between the regular school and the after-school programs, which is essential to the successful continuation and effectiveness of the program. The Duration of the Program and Post program Options 26. The PETI attends families for a maximum of four years, with an interim evaluation after two years to ensure that the family continues to comply with the agreements for program participation, or until a child has reached 14 years of age. Two issues have been raised. One is that the child may be forced to exit the program before age 14, which may happen before completion of primary school; and the other is that there is a risk that program graduates will return to do the same job they did previously or were at risk of doing (which may be one of the worst forms of child labor). * The overall recommendation is that children should be allowed to remain in the program or should be covered by another type of education program until age 14, which is consistent with recent government initiatives under the Projeto Alvorada financial umbrella. Children exiting the PETI would be eligible until age 14 for a cash transfer under the Alvorada's Bolsa Escola program. The effectiveness and sustainability of the PETI would be improved by making all PETI graduates eligible for the Alvorada's Bolsa Escola Program. * It is recommended that innovative initiatives, such as some in Pemambuco and Bahia, which assist PETI graduates in continuing their schooling and in acquiring training for facilitating their insertion into the labor market, be replicated where appropriate. It should be ensured that there is a link between the skills and competencies needed by the poor youth to enter the labor market and the training provided. * One promising solution has been to commence the child's benefits (for the purpose of the four years requirement) at the moment his/her family is inducted into an income generating and social promotion program. This strategy is being developed in collaboration with PRONAGER (Programa Nacional de Geraado de Emprego e Renda), and will seek to enhance the income and productive capacity of the PETI families. Implementation Aspects 27. An analysis of the implementation of the PETI in the states of Pernambuco, Bahia, and Sergipe revealed the following points and recommendations: (a) Family selection process. Beneficiary selection was conducted in two stages. The State Commission for the Prevention and Eradication of Child Labor, with the help of the Municipal Councils of Social Assistance, identifies municipalities with large incidence of xiv the worst forms of child labor. The council then performs a socioeconomic diagnosis of the prioritized areas. It is the responsibility of the Municipal Councils for the Eradication of Child Labor to carry out the registration and selection of families, based on the established criteria for the target population. The assessments verified the overall targeting success of the program. Leakage was minimal in the three states examined. The evaluations of the targeting effectiveness revealed that in all three states many eligible families were not participating. * It is recommended that Pemambuco and Sergipe perform a complete census of all eligible families in program municipalities, which could prove invaluable to guide program expansion. The states also should seek the integration of children reaching age 7 into the program, and to monitor program effectiveness. * The basis upon which the beneficiaries are chosen in Pernambuco, and in the other areas to which the program will be expanded, should be improved by following a method similar to the one being used in Bahia, which contracted the Federal University to collect and maintain a database of families which fit the selection criteria. (b) Program information dissemination. The assessments suggest that each state government consider its role in organizing the dissemination of the program in cooperation with the National Forum for the Prevention and Eradication of Child Labor, and the State and Municipal Commissions. Greater involvement of civil society is paramount. * Regular seminars and workshops with families and school officials, as well as radio announcements and pamphlets, are needed to increase the knowledge base of the program in areas where the program has been implemented, and to disseminate innovations. (c) Coordination between departments of Social Assistance and Education. The assessment noted that there often is a lack of integration and coordination at the state and municipal levels between the secretariats and municipal-level departments of Social Assistance and Education. Since the program is administered primarily by the Ministry and the Secretariats of Social Assistance, Secretariats of Education do not have a strong role and often are not supportive. The strength of the PETI, vis-a-vis other cash transfer programs, is its specific focus on ensuring the provision of education services of acceptable quality for the poorest and most vulnerable children. The program mechanism is to finance the municipal costs of an after-school program, which doubles the typical schooling time of a child, which is consistent with the Lei de Diretrizes e Bases (LDB) requirements on the provision of full-time schooling (jornada integral) for all children, by 2006. • Better coordination between regular teacher and monitor's activities in the after- program should be considered. * The long-term sustainability of the PETI education initiatives clearly depends upon the sense of full ownership and participation by local education authorities. xv The municipal-level education departments should have an important role in the curriculum of schools and of the jornada ampliada, and also in the criteria for selection of monitors for the jornada ampliada. The role of the education departments in the administration of the PETI should be coordinated with the Social Assistance secretariats and Education secretariats. (d) School and jornada ampliada capacity issues. Participants and program teachers/monitors noted that there was lack of space or venues to hold the jornada ampliada, or that they were far from the school. The assessments also revealed cases in which monitors lacked knowledge of the PETI and revealed a general lack of training on the harms of certain forms of child labor and the benefits of education. The curriculum of the jornada ampliada, and its lack of integration with the regular school curriculum, is also of crucial concern. Another issue regarding the studies of the children is the lack of support they have, since their parents have little to no formal education. . There is a need to ensure the availability of enough classrooms. This is an important responsibility of the State and Municipal Governments as per agreements of participation in the PETI program. When the venues for the jornada ampliada are distant from the households, then it is also the responsibility of the Municipalities to provide transportation for the children/youth in the program. The costs incurred by the provision of these services are part of the agreed local counterpart financed costs for the operation of the program. * The jornada ampliada is a venue for children to receive integrated support for schoolwork. Monitors should be able to provide children with the needed help, and should, therefore, have the knowledge capacity so as to coordinate their activities with the regular school curriculum. (e) Process of transferring the cash. The cash transfer is sent from the federal government (MPAS/SEAS) to the State Treasury, and the mothers of participating children collect the money from bank accounts, at official banks or mobile bank units, upon confirmation that their child have fulfilled the requirements of the program of 90 percent school attendance and attendance of the extended school day program. The assessment found no evidence of teachers misrepresenting the attendance of the children in the program, but did find cases of delays in payment. No cases were found of families pulling children out of school as a result, though many threatened that they would if there were recurrent delays in payment. Delays in the transfer of funds from the Federal Government also occur due to delays in Municipal Governments providing accounting verification of amounts previously transferred, lack of certification of INSS (Instituto Nacional de Previdencia Social) payments, as well as weak institutional capacity in some states and municipalities. * The transfer of funds from the federal treasury should be periodic and according to the agreed schedule, in order to minimize the hardships on the families and to break the implicit contract between government and parents. * State and Municipal Governments should provide prompt auditing verification of funds previously spent, and INSS certification. xvi * A promising solution is the forthcoming strategy to pay the benefits directly to families through electronic banking transfers. * State Governments should provide technical assistance to municipalities to enhance their institutional implementation capacity. 69 Monitoring and evaluation. The assessments noted that the official monitoring system established by the program was not being used in any of the states evaluated. Additionally, the monitoring of school attendance was shown to be erratic and the importance of this step was not well understood. Management of program-related data for the purpose of evaluation was found to be poor, though in Bahia, the link with the University has ensured a better system. * The experience of Bahia demonstrated promising methodologies. * To guarantee proper monitoring and evaluation of program impacts, it is recommended (i) the creation of maps of localities and types of child labor and of poverty, and (ii) the creation of a database with PETI information including demographic characteristics of participating families and schools. * PETI program information needs to be recorded in a census. It is suggested that this process be reviewed and implemented, and that this be an area highlighted to be regularly reviewed by the states. * Teachers should be trained on the importance of accurate attendance records, and the implementation of spot checks by third party is also suggested. The report fully supports the Government's proposal to institute a computerized management, monitoring and evaluation information system, linking all states - with the objective to improve overall program monitoring. (g) Participation of civil society. Some areas demonstrated much better abilities to include civil society in the program. Promising examples were the Associations of Mothers and the Associations of Women in the Municipalities of Goiana and Vicencia, respectively. In Bahia, the Agentes de Familia (implemented by the NGO MOC with financial assistance by UNICEF) serve as a link between the PETI families and the schools in the educational attainment of the children. The component emphasize teacher training and competencies, curriculum adequacy, income generating activities. * It is recommended that these promising practices be analyzed, and the experiences disseminated and used as models of best practices of involving civil society and local businesses, employers, and employer associations in the effort to eradicate child labor. The municipalities should provide local town hall meetings and information workshops which include incentives for the participation of members of the civil society. (h) Complementary programs. The objective of the PETI is to eradicate the worst forms of child labor. The program attempts to reach the root of the problem by providing both cash transfers and requiring school attendance. The assessment revealed that currently xvii there is little link between the PETI and other social assistance programs, which is necessary for addressing the larger problem-poverty-to which child labor is linked. * The implementation of the PETI should not be conducted in isolation from other safety net policy programs. Only through the integration of complementary programs can maximum impact be achieved. * The findings in this report support the proposal of the government to complement the PETI with a preventive program intended to encourage the completion of primary education. In Programa Alvorada, families would be provided with a stipend or other assistance, as deemed appropriate by the selection criteria. * Social assistance programs, both governmental and non-governmental, which provide literacy program for parents, especially mothers, are strongly recommended as part of this effort. In addition, programs which provide professional or vocational training to children exiting the program would be highly beneficial. The concession of credit for poor families from the national credit program would be highly complementary to the PETI. Future Research Agenda 28. In this report, we provide suggestions for expanding the efforts to include more urban areas, and for attempting to meet the challenge of eradicating all of the worst forms of child labor by 2002. We have summarized the successes and lessons leamed after five years of program implementation, and we have provided suggestions based on these findings, in an effort to contribute to the internal technical debate. This is clearly a story of success. In an effort to include more urban areas and to meet the challenge of eradicating all of the worst forms of child labor by 2002, the resolution of some challenges is still part of an ongoing learning process. Since this is a learning process, we reviewed remaining issues to be addressed. A formal impact analysis of the existing program and the related Bolsa Escola program is critical to providing more insight into these challenges. 1 2. INTRODUCTION CHILD LABOR- THE POLITICAL AND LEGAL DEBATE 2.1 The reality of child labor throughout the world precedes recorded history. At the international level in recent years, a rise in the consensus that children's early entry into the labor force may be harmful to their health and development is demonstrated by the adoption of the United Nations Convention on the Rights of the Child in 1989, the establishment in 1992 of the Intemational Program for the Elimination of Child Labor (IPEC) as part of the Intemational Labor Organization (ILO), and the adoption of ILO Conventions 138 and 182.4 2.2 In Brazil, the federal constitution, in 1988, pronounced children and adolescents to be an "absolute priority"-as stated in article 227, "it is the duty of the society and the state to ensure that children and adolescents are an absolute priority-that they have the right to life, health, food, education, leisure, training, culture, dignity, respect, familial and community relations, and to ensure that they are not subject to any form of negligence, discrimination, exploitation, violence, cruelty or oppression."5 The government also supported the prohibition of dangerous, unhealthy, or nocturnal labor for children less than 18 years of age and any type of labor for children less than 14 years of age, except in a situation of learning or professional training. In 1990, Law 8.069-the Statute of Children and Adolescents (ECA)-article 60, ratified the prohibition of labor for children less than 14 years of age.6 The government has set as a goal eradication of the worst forms of child labor by 2002. 2.3 In its policy on the protection of children, the government also vowed to support programs and projects that stimulate equality in conditions of access and return to schooling, and in the supply of socioeducational activities, sports, and leisure before and after the normal school day. 4The minimum age convention of 1973 (ILO Convention 138) states that the minimum age for work shall not be less than the age of completion of compulsory schooling, and in any case shall not be less than 15 years. However, in the case where a member's economy and educational facilities are insufficiently developed, this member may, after consulting with the organizations of employers and workers concerned, initially specify a minimum age of 14 years. The minimum age of working at any type of work, which by its nature is likely to jeopardize the health, safety, or morals of young people is 18 years. This convention does not apply to work done by children in schools for vocational or for training purposes. 5Programa de Erradicacao do Trabatho Infantil. Serie Programas de Assistencia Social. 6 In 1990, Law 8.069-the Statute of Children and Adolescents (ECA}-article 60 established 14 as the cutoff age for working. In December 1998 the age of illegal labor was raised to 16, but was lowered again in March 1999 based on a ruling by a Minas Gerais judge that the change was unconstitutional. 2 2.4 Efforts in Brazil to eradicate harmful forms of child labor have been demonstrated through the establishment of the federal Program to Eradicate Child Labor (PETI) and other programs at the state and municipal levels. The objective of the PETI is to increase educational attainment and reduce poverty, though its main goal is to simultaneously eradicate the "worst forms" of child labor. The program provides stipends of approximately R$25 to poor families which have children aged 7 to 14, which is given to the mother of the beneficiaries. The stipend is conditional on the promise that the children will attend school, participate in after-school activities, and agree to not work. The after-school activities-the so-called jornada ampliada-ensure that children do not both attend school and work. Additional efforts of civil society organizations, such as worker unions, churches, and NGOs, have demonstrated an increase in the concern for the issue of child labor in Brazil. DEFINING CHILD LABOR 2.5 When speaking of child labor, the World Bank defines neither child nor labor, because these are highly dependent on the locality of these elements.7 However, when speaking of child labor in the context of Brazil, it is important to note that child labor refers to any work, paid or unpaid, for at least one hour per week, and that any type of labor is illegal for children less than 14 years of age. 8 This report primarily examines children aged 10 to 14 due to data restrictions and the fact that the majority of working children are among this age group in Brazil. OVERVIEW OF THE INCIDENCE OF CHILD LABOR - WORLD, LAC, AND BRAZIL 2.6 Today, there are approximately 250 million children between the ages of 5 and 14 working in the developing world, of which at least 120 million are working full time. The highest concentrations are in Africa (26.2 percent), Asia and Oceania (12.8 percent), and Latin America and the Caribbean (9.8 percent).9 Brazil is the third largest employer of children in Latin America (following Guatemala and Ecuador). The highest incidence of child labor in Brazil occurs in rural areas (comprising 60 percent of all Brazilian children aged 10 to 14), where it reaches 37 percent of children in this age bracket compared to almost 8 percent of children in this age bracket in urban areas (PNAD 1999). Between 1992 and 1999 the incidence of child labor in Brazil declined by more than 5 percentage points, to 15 percent of all children aged 10 to 14. PRIORITIZING THE WORST FORMS OF CHILD LABOR 2.7 The growing effort in Brazil to tackle the issue of child labor stems from the social and economic consequences of children working in harmful activities. It is important to note that often the situation is one in which the entire family is highly vulnerable, in a context of poor enforcement of labor codes and standards, and workers' regulatory protection. Children in these situations, however, are likely to face larger risks than adults due to a 7See World Bank Child Labor Initiative. 8In Brazil, before 1991, work was characterized as 15 hours per week. 9From ILO (1996), and Basu (1999). 3 combination of prior malnutrition, poor health outcomes, and low body mass. Working conditions (long hours worked, exposure to physical or mental strain or repetitive actions, or exposure to hazardous substances) and other physical or psychological risks or abuse can impact negatively on children's welfare. Child labor is also associated with low human capital investment-poor education and health outcomes, and future poverty incidence. Low rates of productivity growth further contribute to the intergenerational cycle of poverty and restrict economic growth. 2.8 It is argued that not all types of labor are harmful to children and that child labor may even be beneficial to a child. Child labor can, in fact, be associated with a nurturing working environment in the home with family members. Informal education and training are some of the benefits of this type of labor. In many parts of the world, child labor is viewed as an integral and positive part of a child's development. A working child is seen as receiving discipline and training for the labor market in adulthood; besides that, working occupies time which might otherwise be spent "getting into mischief." In addition, because children can be important contributors to family income, their working can substantially improve living conditions for the family where the alternative is even more extreme poverty. Working may even enable poor families to send their children to school because the children's income can help finance school fees, supplies, or uniforms. This report will demonstrate that nevertheless, for many children, the type of labor in which they are involved can impose substantial harm to their physical and mental health. 2.9 In addition to the adverse impacts of child labor on education and future earnings, some forms of child labor pose immediate and long-term health threats to children. Although there is little data on the impact of child labor on children's health in Brazil, it is generally accepted by public health researchers that some general risks and hazards are associated with activities that commonly engage children, such as agriculture, construction, and street work. Hazard is defined as exposure to objects, substances, and/or conditions that have the potential to have adverse effects on humans. 2.10 In Brazil, the agricultural sector is the main source of labor for both male and female children, particularly in the rural areas. It is also consistently ranked as one of the most hazardous industries in terms of morbidity and mortality (Fassa et al. 2000). Hazards include farm machinery; strenuous labor (lifting and carrying heavy loads, working in uncomfortable positions, etc.); chemicals, such as pesticides; and adverse weather (for example, heat). 2.11 Injuries abound among children working on the farrns in Brazil, such as on sisal (jute) or sugarcane plantations, caused by long knives and machetes used for cutting, piling, and hauling the crops (Buckley 2000). In addition, processing machines and heavy machinery are frequent sources of occupational injuries. The heavy manual labor involved in agricultural activities places both physical and emotional strain on the child workers. In addition, children working in agricultural activities face an elevated risk of being exposed to chemicals and pollutants. Poor field sanitation and lack of potable water for drinking and washing facilitates the transmission of communicable diseases, and exacerbates the effects of pesticides and heat. 4 2.12 The construction sector in Brazil employs only about 5 percent of working children in the urban areas (6 percent among boys, and 1 percent among girls), and less than 1 percent in the rural areas. It is, however, a harmful occupation for children because of the elevated risk of accidents, and exposure to, among other things, noise, silica, asbestos, dusts, and heavy loads (Fassa et at. 2000). 2.13 In urban areas children may be engaged in such activities as street vending, garbage collection, and illegal occupations such as selling drugs or prostitution. These activities have other risks and harms associated with them related not only to the physical harns but also to the emotional and developmental risks associated with the stigmas and social exclusion which they face on a daily basis. In addition, because of the public nature of their work, these children lack rights to which they are entitled and often suffer from poor family ties or exposure to violence both in and out of the home. 2.14 In Brazil, key activities identified thus far as the priority forms of child labor to address are grouped by geographical area, urban or rural.'0 In the urban areas, they include illicit activities such as drug trafficking, prostitution, or activities such as street vending (newspapers or other products). In rural areas, they include activities which involve the collection or production of charcoal, agave, cotton, vegetable products, sugar cane, tobacco, horticultural products, citrus, salt, flour, fish, wood, textiles, tiles or ceramics, and activities related to the extraction of stones and gems (mining). 2.15 The definition of priority forms of child labor to eradicate in Brazil is narrow; its focus is on the impact of activities on physical, rather than psychological, well-being, and on manual labor conducted in public spaces (work that is more visible). This is due in large part to the fact that this overwhelming problem must be addressed in steps, considering the absolute worst forms first. 10The Worst Forms of Child Labor Convention of 1999 (Convention 182) states that all members must take immediate and effective measures to secure the prohibition and elimination of the worst forms of child labor as a matter of urgency. For the purposes of the convention, "child" refers to all persons under age 18 and "the worst forms of child labor" comprises: (a) All forms of slavery or practices similar to slavery, such as the sale and trafficking of children, debt bondage and serfdom, and forced or compulsory labor, including forced or compulsory recruitment of children for use in armed conflict; (b) The use, procuring, or offering of a child for prostitution, for the production of pornography, or for pornographic performances; (c) The use, procuring, or offering of a child for illicit activities, in particular for the production and trafficking of drugs as defined in the relevant international treaties; (d) Work which, by its nature or the circumstances in which it is carried out, is likely to harm the health, safety, or morals of children. 5 3. CHILD LABOR INDICATORS AND TRENDS IN BRAZIL INCIDENCE AND EVOLUTION OF CHILD LABOR IN BRAZIL 3.1 Household-level data of both rural and urban Brazil demonstrates a declining trend in the incidence of child labor for children aged 10 to 14 over the last decade from 20.4 percent in 1992 to 14.9 percent in 1999. Incidence remains high, however, particularly in rural areas, in the Northeast, in agriculture, and among boys, as has been the trend throughout the last decade. I l 3.2 In 1999, in rural Brazil, 37.3 percent of children aged 10 to 1412 were working relative to 7.8 percent in urban areas. This represents a decline of almost 9 percentage points in rural areas since 1992. The Northeast, in which 23.4 percent of children are working, has the largest incidence. The lowest proportion can be found in the Southeastern states, where the regional child labor rate was 7.5 percent. Examining state- level prevalence of child labor, we find that six out of the nine Northeastern states make up the top six states with respect to highest incidence. These states, in descending order, are Maranhao, Piaui, Ceara, Sergipe, Bahia, and Paraiba. The state ranked seventh is Mato Grosso, located in the Central-West section of Brazil, with a 20.9 percent incidence of child labor. Pernambuco falls into eighth place, with a 20.04 percent incidence of child labor. Nevertheless, children living in the Southern rural areas have the largest probability of working, 45.5 percent, followed by those in the rural Northeast, where 41.1 percent of the children work. The lowest concentration of child labor is found in urban sections of the Southeastern part of Brazil, where 4.8 percent of children work. 3.3 Examining the data by industry, agriculture stands out as the primary employer of children, even in urban sections of Brazil (see Appendix 1, Table 1). For the nation as a whole, agriculture employs 86.6 percent of rural working children and 31.7 percent of urban working children. In urban areas, nonagricultural industries play a larger role in employing child labor. In particular, the service industry employs 27.4 percent of urban '" Consistent with the findings of Saboia and Saboia in "Trabalho de Criancas e Adolescentes no Brasil nos Anos 90: Uma Analise Desagregada por Estado e Microregiao," a study done in collaboration with the World Bank and UNICEF, November 1999. [2 In this analysis we focus on children aged 10 to 14 partly due to data constraints and partly due to the fact that the legal work age is 15. In addition, the majority of working children fall into this age cohort. 6 working children, followed by the commerce/retail industry (15.6 percent) and the manufacturing industry (10.3 percent). 3.4 Agriculture's share of the child labor pool has remained fairly stable over the years. In 1992, agriculture employed 59 percent of working children in Brazil, and this proportion has increased slightly to 63.2 percent in 1999 (see Appendix 1, Table 2). The other two significant employers of child labor, the retail and service industries, have also had fairly stable shares of the child labor workforce during the 1990s. The service industry employed 15.5 percent of working children in 1992 and 13.1 percent in 1999. The retail industry employed 11.7 percent in 1992 and 12.7 percent in 1999. Neither of these has declined greatly. Proportions in manufacturing have declined from 7.4 percent in 1992 to 4.9 percent in 1999. Child labor in all other industries has consistently been under 3 percent. 3.5 Throughout the decade boys have been about twice as likely to work as girls (see Appendix 1, Table 3). Nevertheless, the gender gap in the probability of work has been shrinking slightly. In 1992 males were 13.2 percentage points more likely to work than females. In 1999 this gap shrank to 9.7 percentage points. The male child labor rate in particular dropped from 27.1 percent in 1992 to 19.8 percent in 1999. The female child labor rate fell from 13.8 percent in 1992 to 10.3 percent in 1999. 3.6 An analysis of children aged 10 to 14 and 15 to 17 using the PNAD (1997), the Brazilian demographic census, and the dataset Relaqao Anual de Informaqoes Sociais (RAIS) confirms the findings that child labor remains a problem, particularly in the Northeast, among boys working in agriculture (Saboia and Saboia 1999). The use of these other datasets in comparison with PNAD is useful because they provide data at a micro regional level and provide insight into the formal labor market, which PNAD does not do as well. In addition, they define child labor as 15 hours per week rather than 1 hour per week, which is important in that it confirms that, despite this definitional difference, the incidence and trends remain virtually the same. The report notes that among adolescents (aged 15 to 17), child labor predominates in the South and the Southeast and to a lesser extent in the Center-West. Child labor in the capitals of each state is relatively small compared to the interior areas of these states. The study also highlights that child labor exists primarily in the poorest states and micro regions. For adolescents, the findings are even more stark. 3.7 While the statistics described above provide a great deal of useful information, they may hide a spectrum of possible outcomes associated with different types of activities and intensities of labor. First, intensity-as measured by number of hours worked-may differ substantially among children and by gender, area, and region. As measured in this survey, child labor constitutes at least 1 hour of work per week. Most children are working more than 20 hours per week, however, and at the same time, the majority of children are combining both work and school. This heterogeneity makes it difficult to interpret incidence alone in terms of outcomes. Second, the statistics are difficult to interpret because there is heterogeneity in the types of work among the children. Some forms of child labor, such as certain agricultural or factory work, may be associated with substantial physical and developmental risks, while others may be less so. 7 Additionally, child labor may be understated for certain activities such as domestic work for girls, whose work often goes unrecorded due to its more isolated and private nature, and for children working in the streets whose labor is also difficult to measure. The following section attempts to break down the statistics into more detailed descriptions of intensity and types of activities as much as the data allow. Work Intensity: The Time Tradeoff 3.8 As mentioned, the statistics above measure any type of work independent of number of hours worked per week above 1. Breaking down the classification into categories of "work only," "work and study," "study," and "neither" into actual number of hours worked allows for more appropriate interpretation of the trends in the statistics by gender, area, and region. This is important because it is likely that the more hours that a child works, the less time he or she will put into school. Figure 1.1. Child Labor and Schooling Trends, 1992-99 Disttibution of Children Aged 10 throtugh 14 in Work and Schooling Js- 5I 7 I 23 ;3w 141 162 143 117l 1?$! .3 3 l I ,!: 40 1992 ~~1992 1995) 199 19913 1I.3 Year 9113I U ywork y sLigolOm*& studyeihe Source: PNAD 1992-1999. 8 3.9 Over the last decade, there has been a monotonic decrease in the percentage of children who "work only," while the category of "work and study" has remained relatively unchanged, declining only slightly from 14.1 percent in 1992 to 13.4 percent in 1999 (see Figure 1.1). The proportion of children who only study has increased by about 10 percentage points, from approximately 72 percent in 1992 to 82 percent in 1999. The proportion of idle children appears to have experienced a roughly monotonic decline during the 1990s, from 7.5 percent to 3 percent. 3.10 The state which contributes the largest number of children who only work is Bahia, followed by Minas Geris, Pernambuco, and Maranhao. Together these four states make up 38 percent of the nation's children who are engaged in work alone. The largest decline in number of children only working occurred in the state of Sao Paulo, where child labor (working only) declined by 86 percentage points in the last four years alone. Bahia and Minas Gerais remain the two states with the highest counts of children working only. 3.11 While we may expect the number of hours worked to be greater among those in the "work only" category, it may be of interest to examine mean hours worked by area, region, and state to see whether children working in the localities with the highest proportions of child labor work longer or shorter hours. The first obvious pattern is that among children who "work only," urban children have significantly longer work weeks than rural children-39 and 31 hours, respectively. As we will observe later, this may be due to the higher percentage of paid labor in urban areas, because children who are paid generally work longer hours. For children who both work and study, the hour differential between urban areas is still present but is less pronounced, at 23 and 21 hours, respectively. 3.12 Children in the Southeastern states and urban section of the North appear to be working longer hours compared to children from the rest of the country. Children in the Northeast, who only work, spend on average 32 hours per week working compared to 39 hours per week in the Southeast and 36 hours per week in the South. Regional work hour differentials are less pronounced among children who engage in both work and study. 3.13 The states with the longest mean hours worked per week are first Acre, with a mean of 48.3 hours, followed by Paraiba, Rond6nia, Goias, and the Distrito Federal, with an average of 45 hours work per week. Comparing proportions of children working only by region, industry, and urban/rural classification suggests that children employed in manufacturing, construction, service, and transportation industries work the most-on average 40 hours per week. Children working in agriculture spend on average 31 hours per week on the job, and children in social services work the fewest hours, at about 25.8 hours per week. This may again be due to the fact that children who are paid work more hours on average than those who are unpaid. 3.14 There are a greater number of boys working, whether solely or in combination with school (see Appendix 1, Table 3). While about 2 percent of boys were only 9 working, less than 1 percent of girls were in that category. A larger percentage of boys, almost 18 percent, and 9 percent of girls, were combining work and school. About three out of every four boys, and almost 87 percent of girls, were in the "study only" category. Examining weekly hours worked by gender demonstrates that despite the fact that a larger number of boy are found in both the "work only" and the "work and study" categories, girls work on average 1 hour more per week than do boys. In urban areas, however, they tend to work on average 3 hours more per week than their male counterparts, and in rural areas they work about 3 hours less. Given that girls spend more time in school as well, this indicates a greater tradeoff between work and leisure for working girls than for boys. Paid and Unpaid Child Labor 3.15 The majority of working children in Brazil in 1999 were unpaid-only 27 percent of children working were given remuneration for their labor. In urban areas the distribution is more even, though still more children were unpaid. In rural areas the difference is stark, where 87 percent of children were unpaid and 13 percent were paid. Figure 1 in Appendix 1 shows that the proportion of urban children in unpaid jobs has generally declined from 1992 to 1996, after which it has increased to levels surpassing those in 1992. Moreover, most years have seen lower proportions of unpaid females relative to males, with the exception of 1998 where the proportion of unpaid urban females has surpassed the proportion of unpaid males by 2.6 percent. 3.16 On the other hand, the proportion of unpaid rural child workers increased dramatically from 1992 to 1998, with an equally dramatic decline from 1998 to 1999. The gender differential in the proportion of unpaid jobs is more pronounced in rural areas of Brazil. Children who are paid tend to work more hours than those who are not both in urban and rural areas. In urban areas, girls who are engaged in paid labor work almost 5 hours more per week than boys. Occupations of the Worst Forms 3.17 The "worst forms" of child labor may be defined in terms of work conditions, occupational hazard, whether children in those occupations are more likely to work longer hours, and whether added time pressure introduced by work induces the child to leave school. The Program to Eradicate Child Labor (PETI) designates occupational hazard as the main identifier of worst forms of child labor. In rural areas, occupations labeled as such include the collection and production of coal/charcoal, sisal, cotton, sugarcane, vegetables and fruits, tobacco, salt, flour, timber, tiles or ceramics, fishing activities, and quarry. 3.18 Through detailed coding of occupational category, the PNAD data allow us to identify children in these occupations. There appears to be an overall decline in the number of children working in sugarcane plantations (see Appendix 1, Table 4 ). In particular, the largest annual decline occurred between 1992 and 1993. Child work in cotton fields has also displayed an overall decline during the 1990s. On the other hand, 10 the number of children working in the fishing industry seems to be on the rise. Its current level is comparable to its level in 1992, involving about 40,000 children. 13 3.19 In terms of absolute counts, most children who work seem to be concentrated in horticulture, which includes the cultivation, planting, and harvesting of vegetables and fruits. There were about 457,000 children performing horticultural tasks during 1992, and this number has dropped only slightly (in a percentage sense) to 449,000 in 1999. The number of children in rice fields has dropped slightly from 158,757 in 1992 to 125,370 in 1999. In terms of absolute changes, the largest decline occurred in corn/maize plantations, which saw a decline of over 170,000 in the size of its child labor workforce. The number of children in coalmines appears to fluctuate around 5,000 to 7,000. During 1993 there was a surge in the number of children working on sisal plantations to slightly over 10,000 children, but this number has dropped since then to about 4,000 children. 3.20 Two states in particular stand out with the concentration of child labor in the worst form occupations (see Appendix 1, Table 5). Bahia is the only state in which child labor on sisal plantations occurs. It is also the state with the largest number of children working in horticulture and on sugarcane plantations. Children in fishing industries appear to be concentrated in Maranhao, which also employs the most children in rice fields and manioc plantations. 3 The numbers of children in rice, manioc, corn, and horticulture work are not reported because they are disproportionately large. 11 4. THE DETERMINANTS AND OUTCOMES OF CHILD LABOR 4.1 Having examined the incidence and trends of child labor in Brazil, we now analyze the determinants and outcomes of child labor. Our purpose is to explore and evaluate the government's strategy in prioritizing the worst forms for eradication, and to identify program designs which effectively target and achieve the goals of not only eliminating child labor, but also improving school outcomes and eventually reversing the process which perpetuates the intergenerational transfer of poverty. DETERMINANTS 4.2 Children work for a variety of reasons. Poverty is a large contributing factor since poor families can be forced to send their children to work to contribute to family income. This does not mean that they are necessarily indifferent to their children's welfare. Many families may not see an alternative to this choice because the children working may be necessary for the family's own survival, it may be the cultural norm, or the alternative of education may not exist. Additionally, the constraint to work is placed on children where the opportunity cost of receiving an education (not working) is high. Baland and Robinson (2000) demonstrate that credit-constrained households will choose inefficiently high levels of child labor. Numerous studies have found that as household income increases, the incidence of child labor falls. 14 Children whose parents worked as children may also have a higher probability of working, indicating an intergenerational transfer of child labor. Increased incidence of child labor and lower child enrollment rates may also be induced by stronger market demand for child labor.'5 Hence, if society perceives work experience to be more instructive than formal education, significant proportions of working children may be found even among households at the top of the income distribution. 4.3 A low-quality education system and high opportunity costs of school attendance constitute demand- and supply-side factors that may result in a low valuation of the child's return to formal education. Policies and program designs should strengthen the educational component of child labor programs and therefore also address the effectiveness of the education system to provide services of adequate quality. In recent 4 See for example Grootaert and Patrinos (1999), Jensen and Nielsen (1997), and Tzannatos (1998). 1 Rosenzweig and Evenson (1977) and Levy (1985) found that higher child wages led to increased child labor participation and decreased enrollments. King, Orazem, and Paterno (1999) found that stronger local child labor markets increased the incidence of school drop-out. 12 years the education system has considerably increased the number of children at school - from 83% in 1993 to 97% in 2000. The Link Between Child Labor and Poverty 4.4 Contemporaneous per capita household income data suggest a strong link between income quintiles, poverty, and child labor. Within the poorest income quintile, 23 percent of children are working (in the "only work" and "work and study" categories combined); in contrast, almost 14 percent of children in the second income quintile work (see Appendix 1, Table 6). We also find that the incidence of child labor is highest among the poorest households in rural areas. Child labor in urban areas is more equally distributed among the bottom three quintiles. This may be due to the high relative size of the agriculture industry in rural areas, and the close relation between agriculture and poverty. In urban areas, the relative size of agriculture is smaller, and at higher income quintiles the service industry becomes the main employer (see Appendix 1,Table 7). The Structural Determinants of Child Labor 4.5 Chapter 2 and the section above analyze the determinants of child labor from a descriptive perspective but do not attempt to identify the causal determinants of the incidence of child labor. A recent study on the link between poverty and child labor provides evidence from estimates of four different models that child labor persists intergenerationally (Emerson and de Souza 2000). Based on the assumption that the head of the household makes the child labor decision, they find that children whose parents worked as children have a higher probability of working. Additionally, they find evidence that childrens' education has a positive correlation with parents' education. 4.6 Another analysis of children age 10 to 15 using data from the PME dataset of six metropolitan areas of Brazil, by Neri et al. (2000), provides an empirical analysis of a correlation between structural household and market characteristics and the probability of working and beginning to work (as the main activity). 4.7 This analysis demonstrates that boys have a 13 percent probability of working and are twice as likely to be working as girls. Boys are more likely to begin working at a younger age, and for both boys and girls the probability of working increases with age. This evidence is consistent with the PNAD data examined previously. Boys at age 15 have a 29 percent probability of working. A child's being behind in school for their age increases the probability of their working by 17 percent. A father's and mother's education both have a negative relationship, with the probability of a child working with father's education having a very slight stronger effect. The difference between the effect of parents' education on boys and girls is negligible. Living in the metropolitan area of Porto Alegre increases most the probability of a child working, followed by Sao Paulo and Belo Horizonte, in descending order. For all locations the probability is proportionally two times greater for boys than for girls, except Salvador, where the proportion is about 3 to 1 for boys to girls. Children whose fathers are in the middle 13 income quintile have the highest probability of working, followed by the second lowest and the lowest income quintiles, according to this analysis. 4.8 A dynamic analysis of a child's entry into the labor force demonstrates similar patterns in that boys, older children, children whose parents have little to no schooling, and children in the middle income quintile all have the highest probability of entering the labor market as their main activity (see Appendix 1, Table 8). The same holds true for children living in Porto Alegre, Sao Paulo, and Belo Horizonte. This analysis also examined seasonal patterns and sub periods (year intervals) as possible determinants of the probability of a child beginning to work. Children were most likely to begin working during the months of January through March. They were least likely to begin working during April and May. The year interval in which children were most likely to begin working was in 1986 through 1987, followed by the previous year interval and then the subsequent one. The probability of beginning to work decreased to 3.6 percent in the interval 1998 to 1999, from 10.4 in 1986-87. How Economic Crises Affect Child Labor 4.9 Neri et al. (2000) evaluate the effects of economic instability on human capital accumulation. They study the effects of shocks measured at individual and family levels on child time allocation in Brazil, in particular, whether adverse shocks to father's occupational status measured through income losses lead to early entry of a child into the labor force, low attainment rates, and poor grade advancement outcomes observed in Brazil. The paper decomposes the effects of shocks on grade advancement into drop-out rates and repetition rates. The large rotating panel dataset used allows for the estimation of the impacts of changes in occupational and income circumstances of adults on changes in child work and education-related decisions. Throughout the paper Neri et al. also assess other possible determinants of child time allocation, such as child characteristics (gender, age, if he or she is behind in school for age), parent characteristics (grade attainment and income), and time and location variables. 4.10 The main variables analyzed are dynamic proxies of impulses and responses, namely: shocks to household head's income and unemployment status, on the one hand, and child's probability of dropping out of school, repeating a grade, and starting to work, on the other. The findings suggest that father's income has a significant negative correlation with child's probability of beginning to work, dropping out of school, and repeating a grade. Adverse shocks to father's income for those children whose fathers are in the lowest income quintile increase the probability of working both outside and inside the home by about 23 percent. A child whose father experiences a shock and is in the lowest income quintile has an approximately 42 percent greater probability of dropping out of school than a child in the same quintile whose father does not experience a shock. Children in the second and third income quintiles have about a 30 percent greater probability of dropping out than those whose father experiences no idiosyncratic shock to his income. Grade repetition increases by about 23 percent for children in the lowest income quintile when their father experiences an income shock. Both gender and age were highly significant, with boys and older children being more likely to work, drop out and repeat grades; however, girls were 90 percent more likely to begin working in 14 household labor than boys. Parents' education was an important determinant of all of the variables, with children whose mothers have little to no education having a stronger effect than fathers with the same amount of, or lack of, education. Children who were already behind in school were more than three times more likely to drop out of school and about 43 percent more likely to repeat a grade. These findings strongly indicate that household income shocks affecting the current poor also likely impact their children's welfare and have permanent long-run consequences to their human capital investment and probability of moving out of poverty. 4.11 It is indisputable that the determinants of child labor are difficult to ascertain since the decision for a child to work is often determined simultaneously with the decision to go to school and engage in leisure. The decision, however determined, clearly demonstrates patterns in both the contemporaneous evidence and the structural static and dynamic analyses examined above. Children from poor households, particularly those in rural areas, supply a large proportion of Brazil's child labor. This may be related, as mentioned, to supply-side factors such as lack of adequate school facilities. However, the results of the static and dynamic regression analyses which control for, among other things, region, still indicate that the largest proportions of child labor occur among the poorest income quintiles. Parents' education, which may act as a proxy for household income, also has a negative relationship with child labor. The significance of seasonal indicators suggests support for the hypothesis of stronger markets for child labor increasing the probability of child labor. The evidence from Neri et al. (2000) supports the argument that credit-constrained households will cope with crises by changing their children's time allocation by substituting away from school toward work both in and outside of the household. The longer-term outcomes of these decisions will be discussed below. OUTCOMES Child Labor Leads to Low Educational Attainment and Low Earnings 4.12 The empirical evidence shows that child labor leads directly to lower years of schooling (Psacharopolous 1997), or at least to lower learning per year of schooling (Akabayashi and Psacharopolous 1999). Child labor and child schooling may not be mutually exclusive, however. Most working children are also in school. Ravallion and Wodon (2000) found only weak substitutability between child labor and child time in school, and Patrinos and Psacharopolous (1997) found evidence that child labor and enrollment were complementary activities. The argument is that child labor need not reduce the probability of a child being in school. One explanation could be that child labor may increase household income sufficiently to make school affordable, either for the working children themselves or for their siblings.'16 Child labor may also buffer the household from shocks to parental labor income which might otherwise force children to 16 Note that in many countries even public schools have associated fixed costs for materials, uniforms, transportation, and other fees that can lower enrollments of poor children (Alderman, Orazem, and Paterno 1996). 15 drop out (Jacoby and Skoufias 1997). Alternatively, household income and school attributes themselves may be complementary inputs into human capital production, so that child labor income actually makes time in school more productive. 4.13 It is even likely that child labor can directly enhance the lifetime earnings potential of children. Work experience raises wages most likely because human capital is generated by learning by doing (Mincer 1974). It is feasible that returns to work experience dominate the returns to a year of schooling, particularly if the schools available to poor households are of poor quality. It is also probable that by increasing current household income, child labor allows the parents to build an endowment of physical assets that can be transferred to the child at maturity. These physical assets may have a greater return in credit-constrained developing countries or even middle-income countries such as Brazil than do the foregone human capital assets.17 The Tradeoff Between Work and School 4.14 As discussed in Chapter 2, the evidence for Brazil seems to suggest that child labor tends to lead children to substitute away from both school and leisure. It was noted, however, that the majority of children who do work in Brazil are also enrolled in school. In fact, school enrollment among children aged 10 through 14 remained high in 1999, at 95.4 percent. However, among those who are working, 86.7 percent are enrolled, and among those who are not working, 96.4 percent are enrolled. Thus, there appears to be some reduction in quantity of schooling in the presence of child work (see Appendix 1, Figure 1). Enrollment for nonworking children remains strictly above those of working children across all ages. Moreover, the plots diverge across increasing age cohorts. This may be due to the fact that older children who do work are more likely to have left school completely to work full time. Younger children who work are more likely to be attending school at the same time. 4.15 Using data on age and educational attainment, it is possible to calculate the number of grades at which the child is lagging. The customary age at which children begin first grade is 7. The number of lagged grades is calculated as GRADELAG = (AGE- 7) - YEARS OFATTAINED EDUCATION. 4.16 For example, by society's norms a child should start first grade at age 7 and complete the grade within a year; thus, he or she should have 1 year of attained education by the time he or she is 8 years old. In this case, GRADELAG = 0. This sample deals with children aged 10 through 14; the means are therefore truncated upward. Note also that children born in later months of the year are more likely to have slightly higher grade lags due to the incompatibility in school year and aging intervals. 17 However, Parsons and Goldin (1989) found that in U.S. households in 1890, child labor income primarily went toward current household consumption and little if any physical assets were transferred to working children when they reached adulthood. 16 4.17 The average child aged 10 through 14 is 2.2 years behind the expected grade attainment. Among children currently working, the average lag is larger (2.3 years) relative to those not working; those only in school are 1.8 years behind in school attainment. The latter may be slightly biased upward by previous work experience.'8 4.18 Working children typically enter school 4 months later than nonworking children. Those children who work lag behind in grade by 0.05 years for every year they remain working. Given the short age range due to data constraints, the main part of the adverse effect of child labor on school achievement appears to be late entry into school. Examination of children aged 7 through 14 should indicate higher slope differentials. 4.19 Child labor negative impact on educational achievement is larger in rural areas when compared to children in urban areas. Rural children have greater grade lags relative to urban children regardless of work status. In fact, rural children who do not work actually have higher grade lags than urban children who do work, regardless of age. This may reflect supply-side differences: rural schools may be providing substandard services relative to urban schools. Most likely they also include demand-side effects: rural households are more likely to be poorer and less likely to be able to afford keeping the child in school or to be able to afford sending them to better schools. 4.20 In rural areas, children who work enroll in school 3 months later than those who do not work. The differential is 4 months in urban areas. For every year that children work, they are likely to lag behind by 0.04 years in both rural and urban areas. On average, rural children lag behind by 2.4 months for every year in school relative to urban children. Males who work typically delay enrollment by about 4 months, females by about 3 months. Rate of grade lag accumulation appears to be comparable between males and females. 4.21 The correlation between early work and school attainment at the level of micro regions is negative, as expected. The highest incidence of child labor can be found in the regions with the worst schooling indicators. In many states, however, this association is not statistically different from zero, which confirms that factors other than early entry into the labor force influence school retention and learning outcomes. The Implications of Child Labor for Adult Wages, Income, and Poverty 4.22 Early entry into the workforce reduces lifetime earnings by 13 to 17 percent. It raises the probability of being poor later in life (that is, falling into the bottom 40 percent 18 Figure 3 plots the average grade lag by age and current work status. It should be noted that the definition of a grade lag imposes an upper bound of + 1 and a lower bound of 0 on the slopes of these plots. To give a clearer picture, we fit linear trends over each of the plots and record the equations of the fitted trends. Note that theoretically the fitted trends should be piecewise functions consisting of three discontinuous sections over the preschool, schooling, and post-schooling age ranges, with horizontal sections over the pre- and postschooling age ranges. Moreover, if limage<7' GRADELAG >0 the plots imply late entry into school. 17 of the income distribution) by 7 to 8 percent. Both direct effects (independent of the effect on education) and indirect effects (through the effects on education) are statistically significant. The direct effect of child labor is about 3 percent for adult wages and 4 to 5 percent for poverty. The indirect effects through education are also quite large in magnitude-1I to 14 percent for wages and about 3 percent for poverty. Ilahi, Orazem, and Sedlacek (2000) analyze the lifetime earnings implications of child labor, using a unique dataset on adult earnings in Brazil. In their survey, respondents were asked retrospectively if they worked when they were children. By treating child labor symmetrically with child schooling in regressions explaining adult wages, they were able to isolate the long-term impact of child labor on human capital accumulation. 18 5. BRAZILIAN EFFORTS TO ADDRESS CHILD LABOR 5.1 Programs such as the Child Labor Eradication Program (PETI) and the federal government program - Bolsa Escola are an important part of a larger social assistance strategy in Brazil to promote the demand for social programs in a sociopolitical climate of improved access and transparency. The rationale for these types of programs relates to the level of poverty, low educational attainment, and high incidence of child labor in the country. It is indisputable that besides poverty and the high incidence of child labor, the poor quality of education in Brazil is also a contributing factor to the observed low levels of educational attainment. Programs that seek solely to improve educational quality would complement the necessary demand-side intervention of a program such as PETI, and the federal government program - Bolsa Escola, but would not on their own adequately address the needs of child labor or the long-term cycle of poverty. Because PETI provides monetary incentives to attend school, the demand for education is expected to rise, leading to an increase in long-term income. For many poor families the constraint of the need to work may outweigh the perceived benefits of education. The after-school programs that are part of the PETI design are a key a mechanism to ensure that the children do not both work and attend school, and have a chance to access schooling of acceptable quality. The Child Labor Eradication Program Program Implementation History and Coverage 5.2 The Child Labor Eradication Program (PETI) (see Annexes 1 and 2 for a detailed program description) was instituted by the federal government in 1996 with the implementation of a pilot program in the state of Mato Grosso do Sul, an area of high incidence of children working in the production of charcoal. In January of the following year, the program was implemented in the state of Pemambuco in the Mata Sul zone, which has high concentrations of children working in the cultivation of sugarcane. In July 1997, the program was implemented in the sisal region of the state of Bahia. By 1999 the program had managed to reach 166 municipalities in eight states and provide assistance to more than 131,000 working children. 19 Program Objectives 5.3 The program was initiated by the government with the objective of eradicating the worse forms of child labor in urban and rural areas.19 By providing cash grants to families with children of school-going age (7 to 14), the program requires that children attend school at least 80 percent of the required number of hours, and attend the jornada ampliada, which is in essence an extended school day or after-school activities. Because PETI provides monetary incentives to attend school, the demand for education is expected to rise, leading to an increase in long-term income. For many poor families the constraint of the need to work outweighs the perceived benefits of education. The program attempts to mitigate this constraint directly while also providing education and knowledge which may change attitudes and deep-rooted cultural perceptions of education and child labor (see the Operational Manual in Annex 3). Evaluating the PETI 5.4 In this report we will synthesize the findings from four qualitative assessments that were conducted to evaluate the program in Pemambuco and Bahia, and one quantitative assessment of the same states and including the state of Sergipe. Using in- depth interviews, a preliminary study assessed the effectiveness of the program from the perspective of local stakeholders in the first 13 municipalities involved-state and municipal government officials, educators, and school administrators, civil society organizations, and parents and children (Sobreira de Moura 1999). Another study employed a case study approach to review the child labor situation in Brazil and to evaluate the program based on field observation in four municipalities (Rocha 1999). The most recent study, which was a qualitative and institutional assessment commissioned by the World Bank, used a formal field approach using questionnaires distributed in six municipalities in two PETI beneficiary states-Bahia and Pemambuco. Within these two states, three beneficiary municipalities and three nonbeneficiary municipalities were chosen based on distinct criteria (Sobreira de Moura et al. 2000). An additional study, commissioned by the International Labor Organization (ILO) and the Movimento de Organizaqdo Comunitaria (MOC) focused on examining the PETI within the socioeconomic and political context of Bahia, where the program is taking place. This study also used a field interview approach in 17 municipalities in Bahia. 5.5 All of the assessments found that the PETI has been quite successful in its implementation in many aspects. First, the program has actually made it possible for poor children to attend school, contributing to the eradication of child labor, by 19 The Key Activities identified as "Worst Forms of Child Labor" are, In urban areas: Illicit activities such as drug trafficking, prostitution, or activities such as street vending (newspapers or other products). Taking into account the need for a specific approach, the risks of prostitution are addressed by a specialized program outside PETI- Program "Sentinela". In rural areas: Activities which involve the collection or production of charcoal, agave, cotton, vegetable products, sugarcane, tobacco, horticultural products, citrus, salt, flour, fish, wood, textiles, tiles or ceramics, and activities related to the extraction of stones and gems (mining). 20 guaranteeing a minimum income for a large number of poor families. Also, the program demonstrates a considerable positive impact on community life by mobilizing parents, school administrators and teachers, the municipal and state governments, local workers unions, and NGOs. The workers unions were identified as a key successful component of the program in terms of selecting families and monitoring children's compliance with the program requirements. The assessments also asserted that the program helped stimulate the relatively weak economies of these municipalities by inducing a reconstruction of their institutional capacity, and helped induce the development of health and other social assistance programs. In addition, they noted some move toward changes in cultural patterns with respect to family behavior and work and producing new perspectives for the future of the children, though this was one area noted in which the program could be strengthened. A Qualitative and Institutional Assessment of PETI in Bahia and Pernambuco 5.6 This section summarizes the findings from a qualitative analysis of the PETI in Bahia and Pernambuco.20 The aim of the assessment was to examine the development of the PETI in the two states considering the perspectives of the institutional segment, the educational segment, the PETI beneficiaries, and civil society (see Annex 4, Box 1). In each segment, the study examines the level of participation and the relationships established for the conception and implementation of the PETI from its coverage to its evaluation and monitoring instruments. Specifically, the analysis attempts to capture the opinions of those entities involved regarding institutional aspects such as dissemination of program information and objectives, targeting, appropriateness and timeliness of the program's stipend payment, and relationships between the government and those involved, and the communities' perspectives regarding the issues of child labor, education, and perceptions of changes in these elements as a result of the program. Working Conditions and Perceptions of Child Labor 5.7 During various periods in recent decades, especially during the redemocratization of the country, claims were made against the working conditions in the areas of sugarcane and sisal production in Bahia and Pernambuco. Civil society, in particular workers unions, conducted public awareness campaigns about the consequences of the use of the manual labor of children in these activities considered dangerous for all workers. The assessment highlights the importance of civil society, which continues demanding government action for the resolution of what they consider a breach of children's rights. Despite the broad consensus on the negative impact of child labor, until recently little was found in the agendas of local organizations or municipalities to address the issue. The exceptions have been the church and the workers unions, and in some cases women's organizations. 20 Gustafsson-Wright (2000), based on Sobreira de Moura (2000). 21 5.8 The policy agenda of preventing and eradicating child labor was present at the state level in both Bahia and Pernambuco; however, the assessment reveals that the states differ slightly in the objectives of their programs. In Pernambuco, the program is envisioned as a contributing factor in the attempt to prevent and eradicate child labor, which itself is part of a larger social problem. In Bahia, the programn is seen as the central tool in addressing the problem. Both states support an expansion of the PETI to urban areas and including other activities such as quarry, brick making, and flour and textile factories. As a result, in Bahia they have begun to include other activities as part of the selection criteria, and in Pernambuco the government mandate includes the expansion of the program to areas beyond the sugarcane region. Box 1. Worst Forms of Labor in Bahia and Pernambuco Bahia Bahia produces 80 percent of Brazil's sisal, which is concentrated in 27 municipalities in the northeast of the state. Sisal is a plant that is used as a substitute for oil in the production of some synthetic products. Sisal production boomed in the 1970s due to the oil crisis, but in the last two decades production has declined-due partially to drought. The culture is based on small enterprises and family-run farms, where women and children represent 55 percent of the labor force. The technology is rudimentary-the processing entails manual working of the land; planting and harvesting, which is done mostly by women and children; gathering and transporting the leaves, breaking down the fiber using a dangerous machine, cleaning the husks from the machine and spreading out the fiber for drying; and mixing the fibers, classifying them, and packaging them for sale. On average, sisal workers work 7 to 17 hours per day in temperatures up to 38-degrees Celsius. Children working generally earn less than half of what adults earn (R$5/week), which itself is less than minimum wage. The health and safety of children is compromised by working in these conditions; accidents are frequent, and they are forced to carry loads heavier than their weight and to walk long distances. In addition, they are exposed to asphyxiating powder, which causes asthma, and to acid from the plants, which causes sores and burns on arms and legs which often become infected due to lack of medical attention. Pernambuco Sugarcane is concentrated in the Northern Mata Zone, which is made up of 51 municipalities. In the past it represented over 90 percent of the land of the region, though, as with sisal, since the 1980s, production has declined. The production process is rudimentary and uses intensive manual labor which usually involves the use of entire families, including children. The harvesting period is from September to March, during which time children are called upon to help their families. The conditions are poor and accidents are frequent-many with irreversible consequences. 22 Participation in and Organization and Implementation of the Program 5.9 At the state level, Bahia and Pemambuco differ substantially in their organizational structure with respect to the implementation of the program. In Bahia, the state government (SETRAS) runs the program with the assistance of the Federal University of Bahia (UFBA), which ensures the technical base for the selection of families, the community organization MOC (Movimento de Organizafdo Comunitaria), and UNICEF, which integrate the decision process in various phases of the program. In Pemambuco, the state secretariat of social assistance and planning (SEPLANDES) undertakes the role of administrative support for the municipalities and has much less organized interaction with NGOs or civil society. 5.10 The assessment reveals that the selection of teachers/monitors is more rigorous in Bahia, where the state has made selection competitive. The selection in Pernambuco is based on an analysis of the curriculum, interviews, and level of teaching experience. With respect to monitoring and evaluation, the two states demonstrate dissimilarities as well. Bahia has shown to have more frequent and regular evaluations, in large part due to the efforts of MOC to involve beneficiaries in defining the directions for change. In Pemambuco, the evaluation is performed on a case-by-case basis, with the ability to facilitate evaluation at distinct stages of program use. 5.11 While the municipalities, both with and without the PETI, recognize that the benefits of social assistance reduce the social demands that they face, they do not necessarily recognize the extended school day and its educational bearing to be of great importance in terms of social benefits. Instead, what they value and what awakens greatest local attention is the stipend payment, which represents a supplement to family income. The municipal councils play a large role in maintaining partnerships within municipalities and in ensuring transparency of local actions where the program has been implemented. Despite these efforts, some cases of corruption have been recorded, and the state government has responded to these problems. 5.12 In the municipalities observed, local NGOs have a good basic knowledge of the PETI. The participation efforts of the NGOs have emphasized the importance of education and its value in terms of improving quality of life and the socioeconomic dynamics of the beneficiary municipality. In Bahia, the establishment of a relationship between schools and families is mandated by the state, and the entity responsible for this is MOC. In Pernambuco, the organization differs by municipality. The selection of beneficiaries was in large part undertaken by the Rural Worker's unions, as was the monitoring and evaluation, though it was noted in the study that this does not necessarily indicate that monitoring is systematic. However, the unions have been recognized as one of the entities that continue to be the voice of the local population. 5.13 Other members of civil society include the church and Agentes Voluntarios da Familia-voluntary family agents who act as links between UNICEF, the families, and the schools. The Associations of Mothers and Women in Bahia were good examples of civil society participation, in particular supporting the mothers of the beneficiary 23 children. In one municipality they provided a course in gender and development with the goal of integrating both women and men in a sustainable development process. Dissemination of Program Information 5.14 Promotion of the program takes place through different channels in Pernambuco and Bahia. In Bahia, SETRAS and UNICEF produce pamphlets and books about the PETI, and these are distributed based on the criteria database established by UFBA. Radio announcements and community meetings are methods used in both states. In Pernambuco, the state has little to do with this process. Interviews with beneficiaries, teachers, and monitors revealed, however, that often many are not aware of the selection criteria of the program. An understanding of the selection criteria and objectives of the program is a key element in achieving long-term perceptions and behavioral changes with regard to child labor and education. The assessment reveals a weakness in the program's ability to change these deep-rooted values (see Annex 4, Table 1). Targeting 5.15 Family Composition. Interviews with beneficiary families demonstrated that the families which have benefited from the PETI tend to be relatively large-up to seven members. While the majority of families have three to five children, in Pernambuco families are slightly larger. Children benefiting from the program fall in the age 7 to 11 cohort, with children in Bahia being slightly older. This evidence demonstrates that the program is reaching the intended target group. In addition, by capturing this younger cohort, the program is able to reach children before they begin to fall behind in school. The proportion of girls and boys is about equal in Bahia, while the number of girls in Pernambuco is slightly larger. Parents are relatively young-most ranging from 26 to 45 years of age. 5.16 Average Monthly Family Wages and Distribution of Labor. Interview questions relating to income and labor revealed that fathers and grandfathers contribute most to family income-mothers follow, and finally children contribute the least to family income. It appears that in Bahia, grandparents play a larger role in family income than in Pernambuco. In Bahia fathers work mostly in sisal, and in Pernambuco they are concentrated in sugarcane production. This reveals that the program appropriately targets families that engage in the agricultural activities deemed "worst forms" of labor. Fathers tend to work on average between 37 and 48 hours per week. Wages in Pernambuco are higher, with the majority of fathers earning R$31 to R$50, while in Bahia, the majority earn only R$10 to R$20. The activities in which women are involved are generally quite low paying. In both states, mothers are found to work in agriculture, although, like the fathers, in Bahia they earn less than their counterparts in Pernambuco. Those not working in agriculture are involved in domestic labor. 5.17 Parents' Educational Attainment. To examine whether the program is targeting a group where the perpetuation of the cycle of poverty and low educational attainment is a high risk, mothers were interviewed about their own and their spouse's level of education. The majority of mothers completed first grade in both states, and a little more 24 than one third never went to school. Fathers in Bahia demonstrate slightly more education than those in Pernambuco. This reveals the risk of perpetuating the cycle of poverty among this population. The Cash Transfer 5.18 The family stipend (bolsa) is transferred from MPAS to the state social assistance fund to the municipality social assistance fund and then to the families. In Bahia the transfer is equal to approximately US$12 (R$25) per child per month, but in Pemambuco the amount varies by the number of children in the household, and in some cases the amount is equal to US$25 (R$50). 5.19 The assessment reveals that number of bolsas per family-between one and two-is approximately the same in both states. The majority of families in both states responded that they felt the level was fair, though slightly more in Bahia felt it was too low. Claims of delays in payments were made in both states and most perceived this as a problem at the federal level. Interviews with beneficiaries revealed that in the majority of the cases, the transfer is being made to the mother, which is an important criterion of the program (see Annex 4, Tables 2 and 3). 5.20 In both states, the beneficiaries responded that they spend the majority of the bolsa first on clothing and second on food, though families in Bahia spend more of the bolsa on health and medicine while beneficiaries in Pemambuco spend more on school supplies. The allocation of the bolsa on basic necessity goods such as food and clothing demonstrates that the program is well targeted. 5.21 There is an additional transfer payment made for the extended school day- jornada ampliada. This transfer is made directly from MPAS to the municipalities and is equivalent to R$25 per child in the program, and is intended for the purchase of such things as school lunches, school materials, and transportation. In Bahia, SETRAS finances through an entity related to the University (UFBA), the contracting of jornada ampliada monitors for 20 hours per week. Impacts Attitudes about Children's Education vs. Child Labor 5.22 The survey questions regarding children and work versus school reveal that less than half of those heads of households interviewed feel child labor is detrimental to the development of children. In Bahia, however, they are more adamant that child labor is dangerous and puts children at risk of accidents. In Pernambuco, 43 percent value the need for children to work, while only 23 percent in Bahia held that opinion. The opinion that children need to study was held by 11 percent in Pemambuco and 18 percent in Bahia. This is relatively low and again accentuates the lack of value placed on education (Annex 4, Table 1). 25 Behavior and Access to Services 5.23 A small number of beneficiaries in Bahia admitted that there were children working, while in Pemambuco the families denied any child labor. Interviews with nonbeneficiary families demonstrated that the number of children working was considerably higher than in families involved in the PETI. In terms of housework, girls generally dominate, and their work increases with age. Girls in Bahia are more likely to be involved in housework than their counterparts in Pemambuco. As a result of the program, the children are spending less time in housework than before. 5.24 The study revealed notable differences between the behavior of the children in the PETI and the nonbeneficiary children. The PETI children were found to be happier and more tidy and, in general, demonstrated greater capacity to express themselves and different, more positive visions about their future. Tabulations of attendance rates and rates of nonpassing (Tables 4 and 5) demonstrate an increase in attendance rates and a decrease in nonpassing rates in each of the municipalities from Bahia and Pemambuco. The impacts are considerably less significant in Pernambuco than in Bahia, where the nonpassing rates went from 16.7 to 7.36 between 1995 and 1998. In Pemambuco, 1997 represented a year when nonattendance rates increased, particularly in rural areas. In Bahia, nonpassing increased in both 1997 and 1998, but then declined considerably in 1999. 5.25 The beneficiary mothers' perceptions of the impact of the PETI on their children revealed that overall they found the program to have measurable benefits for the children, family, and community (see Annex 4 for Tables 6 through 16 for Bahia and Pernambuco). While they mention school as both an important aspect in improving their children's future and a tool that could help prevent the cycle of poverty, again it is the stipend that they see as the primary benefit of the program. When the payment is delayed, many threaten to remove their children from school, evidencing the minimal value placed on education and the little change in attitude about child labor. 5.26 Because the mothers receive the payment, it gives them a sense of independence and responsibility because they make the purchases for the family. The majority of the mothers responded that their lives had changed because they now had more housework and more work in the fields as a result of their children going to school and the extended school day. 5.27 This assessment found that the majority of beneficiary families appreciate the program and think it operates well. Some of the challenges the program faces are related to aspects that are part of the learning process in the implementation of a unique program like this, and at such a large scale. The final chapter of this report discusses the challenges, policy recommendations, and future research agenda with respect to the PETI and strategies of child labor eradication in Brazil. 26 An ILO-IPEC Study "Bode-Escola" in the Sisal Region of Bahia21 5.28 This study evaluated the program "Bode-Escola", a variation of the PETI concept, which is being implemented in 17 municipalities in Bahia with the assistance of the ILO/IPEC program (International Programme on the Elimination of Child Labour) and MOC. This study examined the municipality of Retirola.ndia experience where the Union of Rural Workers of Retirolandia (STRR) set up a pilot project which provided goats to families who agreed to send their children to school instead of work. The families received the goats on the condition that they would also use the milk to feed their children and repay the program with as many goats as were given to them after having bred the livestock. More than 60 goats were distributed to about 30 families initially which resulted in a preliminary removal of about 100 children from hazardous work. The program also sought to improve the beneficiaries' property (about 100) and to introduce anti-drought crops (palms), and, one-hundred and fifty families received new water cisterns for public areas such as schools. The evaluation focused on the income generating and welfare increasing outcomes 5.29 The municipality of Retirolandia, was chosen for the evaluation as a representative locality for the purpose of examining the social and political context of the sisal region of Bahia. This semi-arid area suffers from poverty, which is made particularly worse during times of drought. Without a safety net to minimize the effects of these droughts, small producers and their families run the risk of falling further into extreme poverty and may attempt to cope by putting their children to work. This municipality falls into the 294th place out of 415 in the ranking of social development in the state (SEI - Superintendencia de Estudos Econ6micos e Sociais da Bahia. Classifica9ao dos Municipios Baianos, 1998). The families with no income represent approximately 11% of the municipality while 34% earn less than one half of minimum wage. Almost half of the families do not own land. Of the families that do own land, three-quarters do not have access to credit. 5.30 The local elites just tolerate the participation of civil society in the program, in particular the workers unions 'Sindicato de Trabalhadores Rurais de Retirolandia' (STRR). The relationship may be strained and possess an element of mutual mistrust which has affected the progress of an establishment of a partnership between civil society and the government. In addition, the community radio plays a large role in the dissemination of infornation in conjunction with other organizations in particular religious organizations. Other regional entities and NGO's including several women's organizations are involved in an efficient effort to raise awareness among families regarding education and women's health. 5.31 The evaluation demonstrated that the general and immediate objectives of the Bode-Escola program, particularly income generating, were achieved. They estimate that families who received 4 goats would be able to produce 8 kids per year total. These kids could be sold for about US$120 and their milk for US$240 making a total family income 21 Evaluation Report, ILO-IPEC Brasilia, August 2000. 27 of approximately US$360 (a monthly income of US$30). This would represent a 40% increase in income for those earning less than 74% of minimum wage and an 80% increase for those earning less than half of minimum wage. These estimates were not confirmed with a formal evaluation using a database of systematic data and this study recommended this for future evaluation. Families with more land received three goats in 1996, now have 12 and those who received 9 in 1997, now have 27. In most cases, it was found that the milk was used to feed their children and that goats were sold to buy necessity goods such as medicine. The program actually exceeded expectations in terms of overall goals. 5.32 A financial assessment of the program reveals that given the annual investment per family of US$223 and the generated family income of US$400, the return on investment was 79% without considering other costs and benefits of the program such as the land (property) improvement, the positive impacts on education and the improved quality of life. 5.33 The basic initial activity of the project demonstrates itself to be sustainable if the resources to obtain the initial goats are provided. The expansion of goat husbandry and the improvement of property could continue by forms of credit from other means. Nevertheless, further technical assistance and agricultural support are necessary. The program demonstrates sustainability to the extent that it improves the productivity of small-scale farmers, their income and that child labor is not being utilized and that quality of life is improved for their children. A Quantitative Assessment of the PETI in Pernambuco, Bahia, and Sergipe 5.34 A summary of an impact evaluation of the PETI based on a household survey administered in the states of Pemnambuco, Bahia, and Sergipe demonstrates the impact of the program (Yap 2000). In each of the states the household survey covered six municipalities; three that were participating in the program and three nonparticipating (control) municipalities. This survey provided quantitative data for evaluating changes in a child's hours in school, the child's probability of working, attained years of education, grade-for-age, and program impact on the distribution of children in hazardous occupations. 5.35 Due to different program implementation techniques across states and the possibility of general equilibrium adjustments within clusters of participating children, the study estimates program impacts separately at the individual, household, and municipality levels. 5.36 The findings suggest that, overall, the program increases household demand for schooling by roughly 15 hours per week, and individual demand by about 17 hours. This corresponds roughly to a doubling of hours spent in school. Estimates are somewhat smaller for Sergipe. 28 5.37 Estimates indicate a massive 25 percent reduction in child labor probability attributed to the program in Bahia. Also in Bahia, control subjects in programn municipalities are predicted to behave like program subjects with regard to child labor. Redefining child labor to those working at least 10 hours a week resulted in smaller estimates all around. In particular, estimates for Bahia and Sergipe are not as robust with respect to this more strict definition of child labor. Program impact in Bahia falls from 25 percent to about 9 percent. With this definition it appears more likely that there have been some general equilibrium adjustments in the regional child labor supply market. Nevertheless, magnitudes of contamination bias are small, and in most cases the inclusion of control subjects living in program municipalities in the comparison group does not introduce any significant changes in the estimated impacts. 5.38 The program reduces weekly work hours (in the full sample of children, not just for those who remain working) by about one to two hours in Pernambuco and Sergipe, and by about four to five hours in Bahia. There is no evidence that the program reduces only hours worked without eliminating the incidence of work altogether. The program in Bahia succeeds in concentrating on children who work longer hours. The program in Sergipe concentrates on children in the middle of the hours worked distribution. The program in Pemambuco produced roughly equivalent reductions in child labor incidence regardless of hours worked. 5.39 In a simple classification of most hazardous occupations based on those found to have caused most accidents among interviewed children, it was found that the program in Pernambuco is concentrated on the less hazardous occupations out of the possible types of work. In Bahia the most hazardous occupations experienced the greatest reductions in working children, and in Sergipe the prograrn targets the mid-ranked occupations. It is noteworthy that this classification is extremely simplified and that, in fact, the magnitude of hazard for an occupation should be based on a wide range of factors, including duration of work and repetitive action, exposure to dangerous or toxic elements, and emotional stress. Some of these elements may be more difficult to measure. 5.40 On the whole, the study finds some treatment contamination of control group subjects in Bahia with regard to program impact on child labor incidence and attained education. Even so, the direction of contamination is sometimes not as predicted by general equilibrium adjustments. 5.41 The magnitude of contamination is generally small enough such that the use of control subjects living in program localities in the comparison group will not significantly change the estimated program effects. Nevertheless, two-sample tests for the equality of means and probit estimation of participation provide overwhelming evidence that treatment placement at both municipality and household levels is not random. It definitely constitutes good practice to have covariate controls. 5.42 The study concludes that estimated magnitudes of program impact are sufficiently robust to legitimize a simple means-comparison technique as a method of producing unbiased estimates of program impacts. Estimates are generally robust with respect to 29 the presence of covariate controls and the presence of ineligible subjects in the analysis sample. THE BOLSA ESCOLA PROGRAMS 5.43 The Bolsa Escola programs are poverty-targeted social assistance programs similar to the PETI that give cash grants to poor families with school-age (7-to-14-year- old) children. These grants are given on the condition that the children attend school a minimum number of days each month. These programs began in 1995 at the municipal level in Campinas and the Federal District of Brasilia. By 1999 there were 60 programs in operation in various urban municipalities. These programs have four objectives. First, they hope to increase educational attainment among today's children and thus reduce future poverty. Second, by restricting the grants to the current poor, the programs aim to reduce current poverty. Third, by requiring children in beneficiary households to have minimum attendance in school, the programs implicitly aim to reduce child labor. Fourth, by providing income support to poor families, these programs act as a partial safety net; that is, they prevent these families from falling further into poverty in the event of an adverse shock. Box 2. Bolsa Escola versus PETI * Bolsa Escola is preventative while PETI has remedial components. * PETI is in rural areas and Bolsa Escola is located in metropolitan areas. * PETI is a Social Assistance initiative while Bolsa Escola is an Education Department initiative. * Bolsa Escola, mainly the federal government program, selects families based on an income-means test and a scoring system, while PETI targets the worst forms of child labor. * The state and municipal Bolsa Escola does not have the supply-side component of the after-school program, while PETI does. The federal government program - Bolsa Escola, has the supply-side component of the after-school program as a counterpart of the municipalities. 5.44 Based on existing evidence, the preliminary assessments find that the Bolsa Escola programs are generally appropriately designed and well administered. They have a role to play in the larger social assistance strategy of Brazil. They are likely to remain successful because they enjoy broad support in the government and in civil society, as does the PETI. However, the concerns raised in the preliminary assessments should be kept in mind when replication or expansion of these programs is being considered. First, the assessments note that there are limits to adjusting the size of the cash transfer or the selection criteria in order to make the programs affordable. Second, the administering municipalities and the federal and state governments must agree on cost sharing for the 30 programs. A related issue is that the federal government needs to prioritize geographical areas where interventions are needed most so as to remain consistent with the objective of reaching the most needy in Brazil. Third, there is a need, perceived in the current federal government program - Bolsa Escola, to either expand the existing Bolsa Escola programs to include the "noncovered" populations (such as those younger or older than the program criteria) or to serve them with other complementary programs. Fourth, the Bolsa Escola programs should complement rather than substitute for investments in school quality, aspect being emphasized on the current federal govermment program - Bolsa Escola. Fifth, since the programs act only as a partial safety net, there is a need to examine how to reach the transitory poor through complementary programs. Finally, there is a need to carry out a broad-based and systematic evaluation of these programs. A formal evaluation, such as the one for PROGRESA in Mexico and those for the PETI discussed in this report, should include control and experiment groups that are tracked over time so that it is possible to make a scientific judgment on the programs' effectiveness (PROGRESA 1999). Bolsa Escola Improves Learning Outcomes22 5.45 While there is strong evidence that cash transfer programs increase enrollment and school attendance, and reduce both repetition and drop-outs, there is much weaker evidence of improved learning outcomes, which is an explicit objective of all these programs. To address this concern, and in collaboration with ILO, IPEA, and the government of the city of Recife (Pernambuco), an in-depth impact evaluation was conducted of the Recife Bolsa Escola program, created in 1997, and today covering almost 1,600 families. This study analyzes which aspects associated with the Bolsa Escola program have the strongest impact on children's learning outcomes and the vulnerability and welfare of families and beneficiaries. 5.46 While strong evidence supports the view that cash-transfer programs increase school enrollment and attendance, and reduce grade repetition and drop out rates, evidence that they also improve learning outcomes (the explicit objective of such programs) is much weaker. With a view to investigating this concern, an in-depth evaluation was conducted by the Institute for Applied Economic Research - IPEA, in collaboration with the municipal government of Recife (Pernambuco), the International Labor Organization -ILO and the World Bank, to assess the impact of its Bolsa Escola Program which, created in 1997, currently pays benefits to some 1.600 families. The study consisted of an analysis of factors which affect learning outcomes among children attending school under the Program and which impact upon the social vulnerability and economic welfare of the families benefited. 5.47 The study sought to: (a) to identify the impact of the Program on the economic activities of adult family members; (b) to assess its contribution to increasing family 22 The study was coordinated by Lena Lavias (IFP-SES), and in it participated as a team: Lena Lavinas, Maria Ligia Barbosa, Eduardo Garcia, and Octavio Tourinho as ad-hoc consultant. 31 income (other than the cash benefit itself); (c) to determine how long a family must remain in the Program in order to escape poverty; (d) to ascertain whether child labor had definitively been eradicated among families participating in the Program; and (e) to appraise Bolsa Escola's real contribution, in terms of the children's school performance, aside from ensuring high rates of attendance. 5.48 The study was based upon a sample of 1.218 beneficiary families and a control group consisting of 268 families (with characteristics similar to those of the beneficiaries but which were not participating in the Program). Two interviews were conducted with each participating family: the first at the time when they entered the Program, and the second, one year later. 5.49 The study showed that: * The Program was well targeted, both in terms of the criteria for selecting children to participate, and in its focus upon the neediest of families. It was found, however, that guidelines were lacking for extending the Program to other schools, i.e., for the progressive expansion of the Program. * The Bolsa Escola Program was well received: both teachers and school principals consider that the Program reinforces their work as educators. Teachers also look to the Program as a source of methodological change, seeing in it an opportunity to enhance their own work, and thereby improve learning performance on the part of the children. - School resources could be better utilized: ideally, investments should concentrate on improving teachers' skills, since teacher performance is the greatest differentiating factor when it comes to achieving higher student performance (grades). - Assessment of individual features - such as age, sex and participation in the Program - provided no indication of factors to explain differences in school performance. However, it appears, according to the data, that girls are at higher risk of dropping out of school than are boys. 5.50 Main conclusions: Schools, as institutions, are the central element in the execution of a policy of income transfer. Some schools succeed in enabling poorer students to achieve similar performance to that of their slightly more affluent colleagues. Furthermore, these schools manage to obtain from the Bolsa Escola students slightly better results than they would achieve at other schools. The school and the teachers, are the preeminent factors accounting for improved school performance. . The Bolsa Escola Program has contributed to the breakdown of mechanisms traditionally used by schools to exclude the poorer students. Bolsa Escola commits families to ensuring that their children attend school and, at the same time, obliges the schools to keep on students who would otherwise be at high risk of dropping out. 32 Under "normal" circumstances, schools - on the pretext of "purely academic" criteria - tend to gradually edge out students from deprived socioeconomic backgrounds. The Bolsa Escola students tend have lower school achievement (as compared to students in the control group) and this, were it not for the Program, would tend to lead to their dropping out of school. * The Bolsa Escola Program has proved an effective means of breaking one of the most pervasive mechanisms for reproducing and legitimizing inequalities: namely, early exclusion from school. 5.51 Challenges: * Its very low coverage is the weakest feature of the Bolsa Escola Program in Recife: currently, only 2% of the target public is served. A clear time horizon and objective goals are needed if the Program is to become consolidated. Despite budgetary constraints and low revenue-raising capacity on the part of the Municipal Government it would, nonetheless, be feasible to extend the Program to no less than 3.200 families (twice the number currently benefited) if a mere 1% of current municipal revenues were allocated to the project. Were this allocation to be increased to 2%, a cash benefit, worth one minimum wage, could be paid out each month to 6.200 families, i.e., 80% of those families that meet the criteria of the Program, and 9% of the potential target public. * The Law that instituted the Bolsa Escola Program in Recife - like most similar programs elsewhere - does not define deadlines or objectives, thereby tending to undermine the effectiveness of the Program as a mechanism for combating poverty and reducing social inequalities. Its positive spillovers and strategic potential are thus under-exploited. A Program of this nature could provide the mainspring needed to induce a redefinition of the social protection system in Brazil, since it strengthens universal principles by taking a selective and focused approach. * With respect to the value of the benefit, it is our belief that the Municipal authorities were right to establish two measures for the benefit - of one, or of half of the minimum wage - according to the number of dependants. Simulations conducted within the scope of this study found that the principal trade-off - and greatest impact - in terms of the increased costs of the Program, was not so much the value of the benefit, but rather, the scope of program coverage. * The Program does not constitute a disincentive to work, but rather, the contrary. Family income (not including the benefit) increased significantly in the first year that families were in enrolled in the Program, despite the difficulty of finding jobs on an weak labor market. More than 50% of the adults in the Program and/or their spouses are illiterate, or barely literate. Despite such shortcomings, the level of occupation (employment) among the benefited families rose to a level that fulfilled their basic economic needs, notwithstanding an unfavorable economic environment characterized by recession. Thanks to the monthly cash-benefit received over the 33 period of one year, over 2/3 of the families in the Bolsa Escola Program were enabled to rise above the poverty line and reduce their degree of social deprivation. The impact of the Bolsa Escola Program on eliminating child labor has been much smaller than had been expected, indicating that schools must concentrate on providing a better response to this challenge. Since the condition for receiving the Bolsa Escola benefit is that the child must attend school regularly, its impact is limited to school hours and it can not be expected to influence the time spent working, or doing household chores, while the child is at home. One positive point worth mentioning, however, is the Bolsa Escola students tend not to be engaged in paid work. CHILD LABOR IN URBAN AREAS 5.52 In Brazil the problem of children working in the streets of urban areas has been traditionally addressed by NGOs. While PETI activities toward the eradication of rural child labor are well documented and tested, starting in 1999 the PETI began a major initiative to address its worst forms, including 266,000 children and adolescents aged 10 to 14 working at illicit activities such as drug trafficking, prostitution, and other activities such as street vending (newspapers or other products) and trash collection. In the urban program, each participating child receives US$20 per month (compared to US$12 in rural areas) and US$12 per child is given per month to the jornada ampliada (the after-school program). These efforts are yet to be evaluated. This section reviews promising approaches of NGOs working in urban areas in an effort to inform the PETI's urban expansion. Additionally, the issue of domestic labor and its implications for the girls normally employed in these jobs is also discussed. Street Children 5.53 In this report we follow the convention of defining children who work in the streets of urban areas as "street children." We view them as a subset of child laborers, meaning that while children working in the street are engaged in child labor, child laborers are not necessarily street children. By this definition, street children may either live and work in the streets or only work in the streets. 5.54 The urban situation of working children presents a different set of issues and concerns for interventions. Child labor in urban areas leads to more vulnerabilities for the child him- or herself, rather than the entire family, as in rural areas. Children's rights, the vulnerability, risks, and exposures of working on the streets, and the very public nature of the work have caused a heightening of concern about addressing the issue of child labor in urban areas. However, not only for these reasons does expansion of the PETI to include urban areas present challenges. The selection of beneficiaries in urban areas presents a much more difficult task than the selection of beneficiaries in rural areas where the child labor is more homogeneous. 5.55 The 1995 national household survey (PNAD 1995) indicates that about 161,000 children (6.2 percent of all working children) work in the roadway or in public areas. It is 34 questionable, however, as to how accurate these statistics are for the purpose of identifying program beneficiaries. Presumably the numbers are greater. Table 2 shows the number of children working in the streets in three Brazilian cities. In the city of Salvador, it is estimated that there are about 16,000 children who are working in the streets; however, only 3 percent live and sleep in the street. In fact, the percentages of children living/sleeping on the streets in the three cities are relatively low (about 20 percent), compared to an estimated 25 percent of street children in Central America (Takahashi and Cederlof 2000). Table 1. Estimates of Children Working in the Streets in Selected Cities Sao Paulo 4, 520 895 (19.8 percent) Fortaleza 5, 962 184 (3.1 percent) Salvador 15, 743 468 (3.0 percent) Source: Rizzini, Barker, and Cassaniga (1998). 5.56 Though generalizing the findings to other cities or regions of Brazil is difficult, data from a study of street children and adolescents (under 18 years of age) in Brasilia provides an insight into the population's gender composition. The survey conducted in 1996 and 1997 found that although there was a dramatic drop in the absolute number of children in the second year, the gender composition remained the same: 89 percent of the street children were boys, and 11 percent were girls. Overrepresentation of boys among the children working in the streets is a pattern that is also witnessed in other Latin American countries. Boys comprised 72 percent of the working youth population in the streets of Mexico (UNICEF 1993), and 70 percent in Honduras (Takahashi and Cederlof 2000). Surprisingly, however, the Brasilia study found no gender differences in the number of hours worked, the time of day or night worked, or the likelihood of sleeping in the street (Araujo 1998). 5.57 Concerns about child labor are further amplified when considering the condition of children working in the streets, particularly those making a home in the streets. For these children, exposure to drugs and violence (within the home and on the street) adds to other harmful conditions, such as traffic accidents, extreme weather, poor sanitation, and psychological distress induced by ostracization and discrimination. 5.58 A study of children working in the streets of Brasilia revealed that in 1996 more than one half of the boys and one fifth of the girls experienced some form of violence in the streets (violence perpetrated against each other or by police and other adults) (Araujo 1998). In 1997, however, there was a significant decrease in violent encounters among boys, and an increase among girls. In 1996, 21 percent of the boys, compared to 12 percent of the girls, reported having used drugs. This pattern was reversed in 1997, when 35 one third of the girls, compared to 22 percent of the boys, experimented with drugs. This dramnatic increase in exposure to violence and drugs arnong the girls in 1997 could be the result of a sampling error, rather than an actual trend. The profile of the girls in the 1997 sample is reported to be dramatically different from that of 1998. Sexual Exploitation 5.59 One of the most harmful activities for children on the street is prostitution, posing an additional set of issues for program intervention strategy. Girls are most often the target of recruitment and coercion into the sex industry. The Brazilian Center for Childhood and Adolescence of the Ministry of Social Assistance estimates that the incidence of adolescents in prostitution is a growing and critical problem in Brazil, from the streets of large cities nationwide to migrant settlements in Amazonia (Dimenstein 1994). Though smaller in number, boys in prostitution exist and cannot be ignored; however, there is a lack of data to determine the scope of the problem. The invisibility and heightened stigma attached to male prostitution causes the problem to be hidden. The plight of children in prostitution and other work in the street-their vulnerability to violence, substance abuse, and HIV infection-is, however, well documented (Dimenstein 1994; Green 1998). 5.60 The PETI efforts to address child prostitution were meant to be both preventive and combative, targeting at-risk children aged 7 to 14 from families with per capita income less than half of minimum wage.23 By providing US$20 to each child and US$12 per child per month for the jornada ampliada, the government proposed to address 8,500 children working in the sex industry. The preventive measures included a national campaign to raise awareness, annual state-level seminars, and workshops for program coordinators regarding the gravity of the problem of prostitution. The combative measures included an integration of their plan with councils, facilitating access to social services, and a guarantee of interaction between the families, schools, and the community. They also proposed to implement a system of evaluation which allows for the possibility of measuring the efficiency and effectiveness of this program. In addition, these combative measures proposed to improve the social and cultural environment of the targeted children, to increase opportunities of literacy, and to stimulate training for the purpose of income generation. As of yet, this component of the PETI has not been evaluated, though the structure for evaluation and monitoring has been established by the government. 23 The federal constitution includes two laws which address sexual exploitation. Article 226, paragraph 8 states that, "the state will ensure the family mechanisms with which to hinder violence within their relations," and Article 227, paragraph 4 severely punishes abuse, violence, and sexual exploitation of children and adolescents. The Statute of the Child and Adolescent (ECA), which implements Article 227, had not in January 2000 been implemented. 36 Domestic Service 5.61 Domestic service is another form of labor which is primarily concentrated in urban areas (four out of five domestic workers are found in urban areas) but which does not take place on or in the streets. It is disputable whether domestic service should be considered among the worst forms of child labor, though given the large number of domestics in Brazil and the potential harm to which these girls may be exposed, we felt it important to discuss in this report some of the issues surrounding their work. 5.62 According to PNAD 1998, there were approximately 400,000 girls aged 10 to 16 working as domestics, representing 8 percent of all domestic workers. Of all working girls aged 10 to 16, almost 23 percent were domestics. Most domestics were found to be black or mulatto and 13 percent resided in their place of work. The private nature of the work locality causes children working as domestics to be invisible and isolated. While not all children working as domestics suffer neglect, abuse, and exploitation, and while the work itself is not always dangerous or inhumane (Innocenti Digest 1999; Green 1998), it is important to recognize that the invisibility of the work creates the need for attention and scrutiny, rather than neglect. 5.63 The ILO characterizes the domestic service as harmful under the following conditions: (a) the child has been sold; (b) the child is bonded or has to work without pay; (c) the child works excessive hours, in isolation, or at night; (d) the child is exposed to safety or health hazards; (e) the child is subjected to physical or sexual harassment and violence; and (f) the child is very young. Girls, who are more likely than boys to engage in this form of work, also experience heightened vulnerabilities, such as sexual coercion and abuse, because of their gender. 5.64 In addition, work as domestic servants, which is often associated with an inferior status fueled by stigma and discrimination, can induce psychological and emotional distress and erode self-esteem, particularly among those who start out at very early ages (Innocenti Digest 1999). Although there exists anecdotal evidence of exploitation and abuse of children working as domestic servants, the prevalence of harmful conditions is not known. Girls who work in domestic service generally are working without an employment card (carteira assinada). While they earn less than 60 percent of minimum wage, they are still earning almost 30 percent more than other workers. Unpaid workers number less than 5 percent. 5.65 Domestic service, like other forms of child labor, has effects on education and thereby on future earnings. One third of domestic workers do not attend school compared to 18 percent of other workers, and 7 percent of children neither attend school nor work. At age 14, only 11.5 percent of domestic workers have attained seven years of schooling compared to 24 percent of other workers and 31 percent of nonworking children. Girls who live in their employers' homes attain less education. While, as mentioned above, domestic workers earn more than the average child worker, the effects of decreased educational attainment may be greater in the future, though evidence is lacking to prove this. Half of domestic workers come from families with per capita 37 income less than half of minimum wage. The intergenerational transfer of poverty is clearly present among domestic workers as well. 5.66 Domestic labor, though not indicated as a worst form of child labor, nevertheless indicates a cause for concern, especially considering the gender implications. The psychological and developmental consequences and the effects on educational attainment of young girls working as domestics flag the need for intervention in this area. While a program such as the PETI, which specifically addresses the worst forms of child labor, may not be appropriate in this case, intervention which improves working conditions and rights of these girls and guarantees completion of at least primary school is crucial and timely. Promising Efforts to Eradicate Urban Child Labor 5.67 Addressing the problem of children working in the streets of urban areas is now included in the PETI and is an issue that has a long history of being addressed by NGOs in Brazil. For the purpose of this report, in collaboration with UNICEF, an assessment was commissioned of promising programs developed by 11 governmental and non- governmental organizations dealing with street children and adolescents in five Brazilian capitals (Rio de Janeiro, Sao Paulo, Salvador, Recife, and Belem) (OEI/UNICEF 2000). This assessment looks at the promising methodologies of programs to address children in situations of high vulnerability (working in prostitution, commerce of drugs, as street vendors, and in dangerous urban activities such as picking garbage). Through the analysis of methodologies and pedagogical practices, forms of organization and sustainability, and challenges and successes, this study shows all of the stages involved in the process, from leaving the streets to building a new life for these children. 5.68 As discussed above, combating child labor in urban areas presents a set of challenges different from rural areas. Children working in the streets are at increased risk of being without rights and being vulnerable to the hazards and hardships that street work presents. While the problem clearly stems from a vulnerable family structure or that the child comes from a poor household, when the child enters the street to work, the vulnerability of the child exceeds that of the family. Hence, the methodologies with which to tackle child labor in urban areas cannot be replicas of those used in rural areas. 5.69 The assessment revealed several points regarding methodologies with which to eradicate children working in the streets. There are two channels which are commonly used to combat the problem of street children: (a) the prevention methodology, and (b) the intervention methodology. 5.70 An effective prevention methodology focuses on the self-esteem of the children and adolescents and develops actions that guarantee that children stay with their families and in school. In addition, this type of methodology emphasizes improving children's and adolescents' social skills; developing psychosocial activities; educating children about health issues and practices, rights as citizens, and culture; and promotes these types 38 of activities with families. In addition, this methodology focuses on coordinating and strengthening social institutions working in these communities. It also attempts to prepare the children for a working environment. 5.71 The basic structure of an intervention methodology focuses on returning street children to their families through street education. The institutions working with street children attempt to offer children and adolescents the opportunity to aspire to a life as a person with rights. These methods include integrating children not only in family life, but also in the education system, and they strive to work with both children's and families' self-esteem and income. Increasing social participation of children and families leads to increased social inclusion. Table 2. Street Children Programs in Five Metropolitan Areas of Brazil CITY INSTITTION Fundacao Joao XXIII/FUNPAPA Belim (Governamental) Par6quia da Confissao Luterana (ONG) ____________________________________ Programa Murere II Recife Grupo Ruas e Pragas (ONG) Coletivo Mulher Vida(ONG) Salvador Fundacao Cidade Mae (Governarental) Projeto Axe (ONG) Rio de Janeiro Fundacao Sao Martinho (ONG) Secretaria Municipal de Promocao Social (Governamental) - P. Vem para a Casa Sio Paulo Funda,ao Projeto Travessia (ONG) Meninos e Meninas de rua de Sao Bemardo (ONG) MOVIMENTO NACIONAL DE Comissao Local de Recife MENINOS E MENINAS DE RUAS Source: Soler (2000). Conclusions on How to Address Urban Child Labor 5.72 As noted, the urban situation of working children presents a much more diverse set of issues and concerns for interventions than the interventions in rural areas with which the PETI has much experience. First, data for street children are often lacking and inaccurate, and urban child labor is more heterogeneous, making it more complex to target. Second, given the heterogeneity of work, setting an appropriate cash transfer amount is more difficult than in rural areas. Third, child labor in urban areas is more linked to problems for the child him- or herself and his or her exclusion from the family. This indicates higher vulnerability and a lack of rights of the child and calls for 39 interventions beyond compensatory income and conditional school attendance. Fourth, the public nature of working in the streets heightens the probability of exposure to violence, drugs, and health risks, such as disease. These risks and exposures require interventions with both psychological and health components. Finally, due to the ambulatory nature of much urban child labor and the poor link to the family, monitoring and evaluation of urban efforts to eradicate child labor may present hurdles and additional costs. 5.73 Given these challenges, how can the PETI best work in urban areas and what lessons can the program learn from the experience of the NGO street children interventions discussed above? The central element of the findings is that programs must act together with the family, the school, the community, and civil society to guarantee an integrated scheme for addressing this problem. Compensatory and benevolent policies are not enough to fully combat the complexity of the problem of urban child labor. The PETI model of linking working children with school fits well into the intervention methodology, as does the improvement of family income through a cash transfer. The PETI components of involvement and participation with families also fit into this methodology. 5.74 What PETI can learn from the NGOs are the best practices in how to establish links with families and methods with which to encourage their participation in the lives of their children and their children's education. Taking from the prevention methodology, the PETI should consider components which focus on the improvement of self-esteem and social skills of the children and adolescents, the development of psychosocial activities, and the education of children about health issues and practices, rights as citizens, and culture. In addition, the PETI should consider coordinating and strengthening social institutions working in these communities. Another component of this methodology is to prepare children for a working environment. This is another area in which the PETI can learn from the best practices of NGOs with experience in these methodologies. The remaining issues are those of lack of accurate data for targeting, monitoring, and evaluation, and the setting of the appropriate cash transfer level. These challenges will be further discussed in the conclusions and policy recommendations. However, finalizing the most important realization is that there is a wealth of institutional capacity ready to be energized in the NGO and civil society community. The final recommendation can only be that the Brazilian government's remarkable commitment (unique among countries with similar level of development) to address the issue of the highly vulnerable urban children could be further strengthened by programs specifically designed to finance congruent NGOs initiatives. 40 6. SUMMARY OF LESSONS LEARNED AND FUTURE CHALLENGES 6.01 The evidence presented in this report of the high incidence of child labor in Brazil (in spite of recent gains) and the negative effects of early entry into the labor market on a child's education and future earnings, reaffirms the absolute necessity of a program to eradicate worst forms of child labor in Brazil. Nonetheless, resolving the problem of child labor does not simply involve the removal of children from their place of work. Where there is a vulnerable child, it is likely that there is also a vulnerable family. The labor of children is often necessary for the survival of a poor family, in particular during an economic crisis. There is a clear need to compensate credit-constrained families for the lost labor and earnings of a child if that child is removed from the labor market. Tying this compensation to school attendance, as the Brazilian Child Labor Eradication Program (PETI) has done, is a promising strategy for breaking the intergenerational poverty trap. The assessments discussed in this report demonstrate that the PETI has been successful in targeting and achieving its goals of reducing rates of child labor and increasing school attendance. 6.02 This chapter summarizes the successes and remaining challenges of the program. These are organized into three main categories: (a) overall design aspects, (b) program implementation aspects, and (c) outstanding challenges and the research agenda. Among design aspects there are five main subtopics: (i) having a preventive versus a remedial program focus, (ii) targeting and program expansion, (iii) the cash transfer level, (iv) the after-school program (Yornada ampliada), and (v) the duration of the program and post program options. The assessments also reveal crucial opinions and experiences regarding the implementation of the program in the states of Pernambuco, Bahia, and Sergipe. They discuss the process and organization of the different entities involved in implementing the program, such as (a) family selection process, (b) program information dissemination, (c ) coordination between Social Assistance and Education, (d) schools and jornada ampliada capacity issues, (e) process of transferring the cash, (f) monitoring and evaluation, (g) participation of civil society, and (h) complementary programs. OVERALL DESIGN ASPECTS A Preventive versus Remedial Program Focus 6.03 The PETI was originally designed to protect and remove children in rural areas of Brazil from the worst forms of child labor, and therefore targeted children already working or those at risk of working in worst forms of child labor. To reach the target of 41 eradicating the worst forms of child labor by 2002, the federal government at the end of 1999 revised the targeting mechanism to prioritize those children demonstrably working. This transition toward a greater focus on remedial actions vis-a-vis prevention was an effort to improve program targeting and effectiveness to help achieve the goal of reaching 866,000 children by 2002. However, in the states with ongoing programs, a gradual transition was proposed over a one-year period. The outstanding weakness was the incentive problems associated with lack of preventive actions of a Bolsa Escola nature on the same areas covered by the PETI. Targeting with a preventive strategy could avoid the potential risk of attracting children to work for the purpose of entering the program and receiving the cash stipend (only children previously working for two years would be eligible). The existing preventive program, Fundo Garantia de Renda Minima (FGRM), had problems with very low cash transfer levels, complex benefit design, and target exclusion of many poor municipalities in the Northeast and North. 6.04 Projeto Alvorada. In September 2000 the Brazilian government took steps to scale up and integrate separate cash-grant initiatives financed by the federal FGRM and the PETI under the Alvorada program. Implementation would continue to be the primary responsibility of states and municipalities. Over the next few years, this national program will gradually implement a Bolsa Escola program in poor municipalities with the worst indexes of social development in the country. This policy initiative represents a major step in program design, fiscal commitment, and management. It integrates within one coherent program both preventive initiatives (Bolsa EscolalFGRM with 2 million beneficiaries in 2000) and the remedial PETI (with 392,000 beneficiaries in 2000). Alvorada will permit the Ministry of Education (MEC) to address design and implementation issues hindering the effectiveness of the FGRM program. By building on the complementarity of existing conditional cash transfers to the poor, and improving coordination with educational initiatives aimed at universalizing basic education, it has the potential to significantly increase targeting and effectiveness. Targeting and Program Expansion 6.05 The target population of the PETI includes children aged 7 to 14 from poor families working in, or who have the potential to be working in, activities deemed worst forms of child labor. The initial implementation of the program in the charcoal, sugarcane, and sisal areas was based on a geographical mapping of areas with high incidence of these types of labor, and the selection of households which have a per capital income of less than half of minimum wage. The assessments reveal that the targeting strategy was appropriate given that the largest number of children working in Brazil fall into the 10 to 14 age group, live in the areas selected by the geographical targeting, and are found among the poorest income quintile groups. 6.06 There are two main concerns related to the program expansion: (a) how to adapt targeting mechanisms as the program expands (the original method of geographical targeting may be inappropriate since other areas have more heterogeneous formns of child labor and the boundaries are less clear as to where it occurs), and (b) how to take the next step in determining the worst forms. Our findings show that highest incidence of child 42 labor does not necessarily indicate either longest number of hours worked or actual working conditions. 6.07 Possible solutions require: * More detailed census and mapping of localities and types of child labor occurring in specific areas and municipalities. It is important that workers unions, universities, and other members of civil society be fully involved in working with the Municipal Commissions for the Prevention and Eradication of Child Labor in this task. * It is also recommended that state and municipal governments take into account the length of the typical workday of different activities as an input to prioritize activities and worst forrns of child labor for intervention by the PETI. The Cash Transfer Level 6.08 Originally, the cash transfer amount and mechanism differed by state. In Pernambuco a family received R$50 or R$100 per I to 2 or 3 to 4 children. There was variation across other states. The consolidation of the programs in late 1999 created a common benefit level across all states at R$25 per child. There was also the decision to adjust all ongoing programs to the common benefit level over a period of one year. 6.09 Two questions exist. First, should the transfer be given per participating child or by famnily? If the transfer is given by famnily, there is the risk that families will choose which children to send to school and defeat the purpose of attempting to remove all children from child labor of the worst forms. Second, how much should the cash transfer be and should it differ between rural and urban areas, across states, and across municipalities? 6.10 The various evaluations of the Bolsa Escola programs and other cash transfer programs in Latin America provide little guidance to set the benefit level. . The assessment of the PETI has demonstrated that the current benefit level is appropriate vis-a-vis the opportunity cost of a child in a local labor market. Since social assessments also revealed an overall satisfaction among beneficiary families with the cash transfer level, it is suggested that the issues of the transfer level continue to be monitored (for evidence of beneficiaries quitting the program) and be subject to a formal impact evaluation. These finding should be disseminated among the federal and state governments, the Forum, Commissions, and Councils for the Prevention and Eradication of Child Labor. * Our findings also suggest that as the program is expanded to urban areas, the current benefit level may need to be adjusted. The optimal design should consider flexible amounts depending on area (urban versus rural, main activity, earnings, etc.) and local financial sustainability in the context of their overall strategy to expand the program to address as many children as possible. This solution is not inconsistent with the current design, where local urban governments have the flexibility of setting large benefit levels (according to local conditions); however, they would be responsible for financing the additional expenditure (above the R$40 per child). 43 * Overall, the evidence gives support to family benefit designs on a per-child basis. However, in accordance with the original PETI design, families should be expected to register all eligible children in the program. The program is a commitment to pull the family away from child labor toward new values and greater investment in the human capital of their children. The After-School Program - Jornada Ampliada 6.11 The jornada ampliada is intended to provide additional educational and recreational stimulus for the children and hinder them from combining work and school. It also provides the children with a school lunch, which improves nutrition and health and aids families financially. The assessments reveal that families are pleased with the jornada ampliada's ability to provide both stimulus for their children and a nutritional supplement. The after-school program is a critical innovation of the PETI. It balances the demand intervention (cash grants-to ensure that all children want go to school) with the need to ensure the provision of minimum quality education services. However, having this additional component clearly increases the complexity of the intervention. The first issue regarding jornada ampliada is the manner in which monitors are hired and trained. Assessments revealed cases in which monitors were unqualified and inadequately trained for the position. Another aspect observed in the assessments was the lack of set curricula with a pedagogical strategy for thejornada ampliada. 6.12 The following recommendations were obtained from the social assessments: - To ensure transparency and the selection of the best-qualified candidates for the post of monitor, a more systematic and uniform mechanism should be established for hiring. The municipalities should execute this responsibility in conjunction with the State Secretariat of Education. * Training of monitors for the program should include courses in the PETI design and implementation aspects, the harms of worst forms of child labor, and the benefits of education and gender sensitivity. * In coordination with the state-level Secretariats of Education, the municipality should undertake the responsibility for establishing curriculum and activity guides for the after-school prograrn. . As per PETI Operation Manual, Municipal Education Secretariats should seek a better coordination between the regular school and the after-school programs is integral to the successful continuation and effectiveness of the program. The Duration of the Program and Post-program Options 6.13 Finally, we have identified issues related to the duration of the program and the post program options for program graduates. The PETI attends families for a period of a maximum of four years, with an interim evaluation after two years to ensure the family continues to comply with the agreements for program participation, or until a child has reached age 14. Two issues have been raised. First, the child may be forced to exit the program before age 14. The result is that the child may still have not completed primary 44 school. Second, when a child exits the program, there is a risk that he or she will return to do the same job they were doing or were at risk of doing previously (and which may involve the same activities classified under the worst forms of child labor). Overall, in different assessments, program stakeholders clearly and consistently identified these issues as the most important from their point of view. * The central recommendation was that children should be allowed to remain in the program or should be covered by another type of education program until age 14. This is consistent with recent government initiatives under the Projeto Alvorada financial umbrella. Children exiting the PETI would be eligible until age 14 for a cash transfer under the Bolsa Escola program. The outstanding issue that remains to be addressed is that the Projeto Alvorada does not cover all the municipalities where the PETI is being implemented. The effectiveness and sustainability of the PETI would be improved by making all PETI beneficiaries eligible for the Alvorada's Bolsa Escola program. . There were reports that many municipalities in Pernambuco and Bahia have developed promising initiatives to assist PETI graduates in continuing their schooling and in acquiring training which may facilitate their insertion into the labor market. On the negative side, the assessments also revealed a predominance of supply-driven craft training activities with little or no relationship to the skills and competencies needed by the poor youth to enter the labor market. These initiatives need to be evaluated and replicated where appropriate. - One promising solution has been to commence the child's benefits (for the purpose of the four years requirement) at the moment his/her family induction into income generating and social promotion programs. This collaborative strategy with PRONAGER (Programa Nacional de Geraqdo de Emprego e Renda) seeks to enhance the income and productive capacity of the PETI families. - While the targeting and benefit design of the PETI was found to be effective in relation to the program objective of eradicating the worst forms of child labor in the 7 to 14 age group, some great challenges remain. As universal school participation of all children aged 7 to 14 is achieved, it becomes increasingly important to change the focus from school enrollment to completion of basic education (grade 8) and progression to secondary education. Under these conditions, benefits should be expanded to children aged 15 and who have a higher risk of drop out and child labor. IMPLEMENTATION ISSUES 6.14 This section summarizes the social and institutional assessments of the PETI in the states of Pernambuco, Bahia, and Sergipe. It provides a discussion of the main issues identified by key stakeholders. (a) Family Selection Process 6.15 Beneficiary selection is conducted in two stages. First, the State Commissions for the Prevention and Eradication of Child Labor, with the help of the Municipal Councils of Social Assistance, identify municipalities with a large incidence of worst forms of 45 child labor. The council then performs a socioeconomic diagnosis of the prioritized areas. It is the responsibility of the Municipal Councils for the Eradication of Child Labor to carry out the registration and selection of families based on the established criteria for the target population. 6.16 The assessments verified the overall targeting success of the program. Leakage was minimal in the three states examined. The evaluations of the targeting effectiveness revealed that in all three states many eligible families were not participating, and some eligible participating families were not included in the program. However, comparing the effectiveness of the program in Pernambuco and Bahia, the family selection process in Pernambuco leads to a higher probability of eligible children not participating. Pernambuco also lacks a database of families in the program municipalities. There were recorded cases in which within a family some eligible children were participating in the program while others were not. 6.17 To improve targeting: - Pernambuco and Sergipe should perform a complete census of all eligible families in program municipalities. This database could prove invaluable to guide program expansion and the integration of children reaching age 7 into the program, and for monitoring program effectiveness. * The basis on which the beneficiaries are chosen in Pemnambuco, and in the other areas to which the program will be expanded, should be improved by following a method similar to the one used in Bahia, where the state government contracted the Universidade Federal to collect and maintain a database of families which fit the selection criteria. * In all three states, the prevalence of eligible families not participating in the program suggests the urgent need for a complete evaluation of the targeting mechanism with special emphasis on identifying the family exclusion mechanism, and the development of strategies to minimize these outcomes. (b) Program Information Dissemination 6.18 Promotion of the program normally occurs through different methods depending on the municipality and state. The assessments revealed a lack of knowledge of program details and requirements. They also revealed little change in values among the community and among families participating in the program regarding the dangers of certain forms of child labor and the benefits of education. 6.19 Dissemination activities are key for the sustainability of the program. It is critical that all stakeholders be informed about the rationality and design and implementation guidelines for the program. The assessments suggest that: * Each state government consider its role in organizing the dissemination of the program in cooperation with the National Forum for the Prevention and Eradication of Child Labor and the State and Municipal Commissions. Greater involvement of civil society is paramount. 46 * Regular seminars and workshops with families and school officials are needed to increase the knowledge base of the program in areas where the program has been implemented, and to disseminate innovations. Examples of areas of innovations are the design of the curriculum for the after-school program and training initiatives for the graduates of the program. * Radio announcements and pamphlets have proven to be effective mechanisms to disseminate program information. * All these methods should be considered in the new areas as the program expands so that these are clearly included and identified in the budgets of the social assistance strategy to eradicate child labor. (c ) Coordination Between Social Assistance and Education 6.20 The assessment noted that often there is a lack of integration and coordination at the state and municipal level between the secretariats and municipal-level departments of Social Assistance and Education. Because the program is administered primarily by the Ministry and the Secretariats of Social Assistance, Education Secretariats do not have a strong role and often are not supportive. 6.21 The strength of the PETI vis-a-vis other cash transfer programs is its specific focus on ensuring the provision of education services of acceptable quality for the poorest and most vulnerable children. The program mechanism is to finance the municipal costs of an after-school program, which doubles the typical schooling time of a child. This approach is consistent with the Lei de Diretrizes Basicas (LDB) requirements of the provision of full-time schooling to all by 2007 (jornada integral), and provides an opportunity for state and municipal authorities to pilot test the approach. Additionally, only with better coordination between the activities of the regular teacher and those of the monitor in the after-school program can the maximum benefits be achieved. 6.22 Only by ensuring that all school-age children are in school and achieve acceptable leaving outcomes will the continuity and effectiveness of the program be assured. Consequently, the long-term sustainability of the PETI education initiatives clearly depend on the sense of full ownership and participation by local education authorities: . The municipal-level education departments should have an important role in the curricula of schools and the jornada ampliada and in the criteria for selection of the monitors for the jornada ampliada. * Their role in the administration of the PETI should be participatory with the Social Assistance secretariats and Education Departments. * It is also recommended that the coordination with the State Education Secretaries and municipalities be improved, with the states taking clear responsibility and a supportive role for the municipalities in the process of implementing the PETI. (d) School and Jornada Ampliada Capacity Issues 6.23 Participants and program teachers/monitors claimed that there was lack of space or venues to hold the jornada ampliada or that they were far from the school. The assessments suggested evidence of some children missing school and the jornada ampliada due to transportation problems. The assessments also revealed cases in which 47 monitors lacked knowledge of the PETI, and generally lacked training on the sensitivity of the harms of certain forms of child labor and the benefits of education. The curriculum of the jornada ampliada and its lack of integration into the regular school curriculum is also of crucial concern. Another issue regarding the studies of the children and the curriculum is the lack of support and help in the home for many of the children whose parents have little to no formal education. 6.24 In all states there was strong support among stakeholders for the after-school program. Among the key recommendations we found are: * State- and municipal-level governments (Social Assistance and Education), with the participation of the commissions and councils, need to ensure that there are enough classrooms for the children to attend and a means to get to the jornada ampliada if it is not on the school grounds. Besides being subject to a formal hiring process, the jornada ampliada monitors should be provided with training for their positions. This training should include gender sensitivity and information regarding the PETI implementation details and monitoring. * It is the responsibility of the State and Municipal Governments, when the venues for the jornada ampliada are distant from the household place of residence, to provide transport for the children/youth in the program. The costs incurred by the provision of these services are part of the agreed local counterpart financed costs for the operation of the program. * Since the jornada ampliada is a place for children to receive integrated support for schoolwork, monitors should be able to provide children with the needed help and should therefore have knowledge of and coordinate their activities with the regular school curriculum. (e) Process of Transferring the Cash 6.25 The cash transfer is sent from the federal government (MPAS/SEAS) to the State Treasury, and the mothers of participating children collect the money from bank accounts of official banks or mobile bank units upon the confirmation that their child has fulfilled the requirements of the program of 90 percent school attendance and attendance at the jornada ampliada. The assessment found no evidence of teachers misrepresenting the attendance of the children in the program. However, the assessments did find cases of delays in payment. When asked why they thought that the payments were delayed, families responded that they believed that the cash had not arrived from the federal government. No cases were found of families pulling children out of school as a result, though many threatened that they would do so if there continued to be delays in payment. Delays in the transfer of funds from the Federal Government also occur due to delays in Municipal Governments providing accounting verification of monies previously transferred, lack of certification of INSS (Instituto Nacional de Previdencia Social) payments, as well as weak of institutional capacity in some states and municipalities. * There was consensus that delay in the payment of the transfers brings hardship to the families and breaks the implicit contract between government and parents. The transfer of funds from the Federal Treasury should be periodic and according to the agreed schedule. 48 * While there was no evidence that teachers misrepresented the students' attendance, this is one area of concern. The findings from the report such the recommendation of periodic spot checks by an independent organization to verify the accuracy of the attendance records prepared by the teachers. * State and Municipal Governments should provide prompt auditing verification of funds previously spent, and INSS certification. . A promising solution is the forthcoming strategy to pay the benefits directly to families through electronic banking transfers. * State Governments should provide technical assistance to municipalities to enhance their institutional implementation capacity. (D) Monitoring and Evaluation 6.26 The assessments revealed four main issues. First, the assessments noted that the system of monitoring PETI goals and impacts was weak in both states evaluated. In neither case was the official monitoring system established by the program being used. Second, the monitoring of school attendance was shown to be erratic and the importance of this step was not well understood. Third, the assessments revealed that, overall, there is poor management and lack of program-related data for the purpose of evaluation, although in Bahia the link with the University has ensured a better system. Finally, the monitoring of the cash transfer disbursements should be improved. 6.27 The overall assessment is that the monitoring and evaluation efforts under the project are weak and need to be strengthened. The experience by the state of Bahia demonstrated some promising avenues. . To guarantee proper monitoring and evaluation of program impacts, it is recommended that maps be created of localities and types of child labor, that poverty maps be created, and that a database of PETI information be created, including demographic characteristics of participating families and schools. * PETI information needs to be recorded in a census. It is suggested that this process be reviewed and implemented and that this be an area highlighted to be regularly reviewed by the states. * Teachers should be trained on the importance of accurate attendance records, and the implementation of spot checks by a third party is also suggested. * It is recommended that there be regular auditing of the cash transfer process by an entity outside the government. 6.28 The report fully supports the Government proposal to institute a computerized management, monitoring and evaluation information system, linking all states - with the objective to improve overall program monitoring. (g) Participation of Civil Society 6.29 The assessments revealed some promising examples of participation of civil society in many areas and noted the need for more involvement in others. Promising examples were the Associations of Mothers in Goiana and the Associations of Women in the Municipalities in Vicencia. Both of these provided an important informal relationship with families in which mothers assumed an important role in the determination of the 49 expenditures of the PETI cash transfer. Additionally, in the case of the latter program, 70 percent of the mothers of the PETI were provided with a literacy program. In Bahia the Agentes Voluntarios de Familia were created by UNICEF to serve as a link between the PETI families and the schools in the educational attainment of the children. 6.30 Involving families, civil society, and NGOs is a crucial element of ensuring ownership and improved results in terms of sustainable outcomes. Key stakeholder suggestions are: * That these promising practices be analyzed and the experiences disseminated and used as models of best practices of involving civil society. * Beyond this is recommended the greater involvement of local businesses, employers, and employer associations in the effort to eradicate child labor. Ultimately, the incidence of child labor critically depends on compliance among employers. Efforts to simply develop and implement minimum labor codes and standards (consistent with no child labor) bypasses efforts to engage employers and restructure the way labor contracts are designed and implemented. * The municipalities should provide local town hall meetings and information workshops which include incentives for members of civil society to participate. (h) Complementary Programs 6.31 The objective of the PETI is to eradicate worst forms of child labor and by design attempts to reach the root of the problem by providing both cash transfers and requiring school attendance. Clearly, the issue of child labor does not stand alone and a number of issues are associated with it, such as the level of education of parents, and the mental and physical health of both parents and children, not excluding the problem of alcoholism. The assessment revealed that there is little link between the PETI and other social assistance programs. 6.32 Implementation of the PETI should not be conducted in isolation from other safety net policy programs. Only through the integration of complementary programs can maximum impact be achieved. * The findings in this report support the proposal of the government to complement the PETI with a preventive program intended to encourage the completion of primary education. In Projeto Alvorada, families would be provided with a stipend or other assistance deemed appropriate by the selection criteria. . Social assistance programs, governnental and non-governmental, which provide literacy program for parents, especially mothers, are strongly recommended as part of this effort. . Programs that provide professional or vocational training to children exiting the program would be highly beneficial. However, we do not mean to replicate current programs that often offer supply-driven training programs that fail to provide poor youth with the skills and competencies required to enter the labor market. * Finally, lack of access to credit is one of the biggest constraints placed on poor families, and the concession of credit for poor families from the national credit program would be highly complementary to the PETI. 50 Outstanding Challenges and Research Agenda 6.33 This report has reviewed the evidence on child labor in Brazil and the government's outstanding efforts to eradicate the worst forrns. The PETI is unique in the world in that it provides an implementable strategy to address the incidence of the worst forms of child labor within a six-year horizon. Brazil is also an outstanding example of a government providing the social and political consensus for the execution of such an effort. The evidence presented on the risks and outcomes of early entry into the labor market and the link between poverty and child labor without question justifies the need for a program such as the Child Labor Eradication Program (PETI). We have summarized the successes and lessons learned after five years of implementation of the program and we have provided suggestions based on these findings in an effort to contribute to the internal technical debate. The evidence is impressive and warrants praise. In the efforts to expand to include more urban areas and to attempt to meet the challenge to eradicate all of the worst forms of child labor by 2002, the resolution of some challenges is still part of an ongoing learning process. We are convinced that part of this resolution should include a formal impact analysis of the whole program, adequate to guide an effort of such magnitude. Challenges to Overall Eradication of the Worst Forms of Child Labor 6.34 The greatest challenges to achieving the eradication of the worst forms of child labor in Brazil include: C How to establish an appropriate cash transfer level that achieves the goal of removing children from their place of work and inserting them into school, and which also allows the program to reach the maximum number of beneficiaries and is sustainable in the long run. * How to achieve the goal of changing not only behavior, but also values regarding the harm of child labor and the benefits of education. * How to address the issue of children graduating from the program (finding an appropriate exit strategy). * How to address idiosyncratic shocks to vulnerable households. Can child labor eradication programs act as a safety net in times of crises by smoothing consumption over time? Challenges of Expansion to Urban Areas 6.35 The challenges of expanding the effort to urban areas include: * It is more difficult to target in urban areas because occupations are much more heterogeneous than those in certain rural areas. * Selecting program beneficiaries in urban areas is difficult because many of the forms of child labor deemed harmful take place in the streets. Data for children vending items, collecting trash, selling drugs, and engaging in prostitution are difficult to obtain. 51 * Establishing a cash transfer level in urban areas presents a challenge because the types of activities in which children are involved are much more heterogeneous. In the proposal for the urban component of the program, the government has recognized that the level must differ from that in rural areas given that wages are higher in urban areas. * The types of labor in urban areas present another set of questions in terms of physical and developmental harms. The risks and exposures that street children confront daily, and the public nature of their work and their hardships, have elevated concerns about tackling this problem. * For children living in the streets, exposure to drugs and violence adds to other harmful conditions, such as traffic accidents, extreme weather, poor sanitation, and psychological distress induced by ostracization and discrimination. These factors may require additional methodologies and support, such as counseling and rehabilitation. * Children working in prostitution are also more vulnerable to violence, substance abuse, and HIV infection, which may require additional measures such as health care and education on the risks. - Strategies to involve families in the program may be more complicated given the potential weak link between children and their families. - Monitoring and evaluation may be more difficult in urban areas for the same reason. 52 BIBLIOGRAPHY A Face Jovem da Exclusdo. Perfil das Crian,as e Adolescentes em Situacdo de Rua em Brasilia. 1998. Carlos Henrique Araujo, 4 Temas CODEPLAN. Brasilia, D.F. A Implanta,do da Jornada Ampliada na Regiao Sisaleira da Bahia. Estrategia de Manuten,do da Crianqa na Escola (Registro de Experiencia). 1999. Solange Lamego, Movimento de Organiza,cao Comunitaria (MOC). Agente Jovem. Programa Agente Jovem de Desenvolvimento Social e Humano. Reverter a Exclusdopor Meio da Promo'do Humana e do Desenvolvimento Social. 1999. Outubro. MPAS/SAS. Akabayashi, Hideo, and George Psacharopoulos. 1999. "The Trade-off between Child Labor and Human Capital Formation: A Tanzanian Case Study. Journal of Development Studies 35, June :120-140. Alderman, Harold, Peter F. Orazem, and Elizabeth M. Pattern. 1996. "School Quality, School Cost and the Public/Private School Choices of Low-Income Households in Pakistan." Impact Evaluation of Education Working Paper Series No. 2. World Bank, Washington, D.C. Araujo, Carlos Henrique. 1998. "A Face Jovem da Exclusao. Perfil das Crian,as e Adolescentes em Situacao de Rua em Brasilia. 4 Temas CODEPLAN." Brasilia, D.F. Baland, Jean-Marie, and James A. Robinson. 2000. "Is Child Labor Inefficient?" Journal of Political Economy, 108 (August) :663-679. Baker, Gary, and Felecia Knaul. 1991. Exploited Entrepreneurs: Street and Working Children in Developing Countries, Child Hope USA, Working Paper No. 1. Barreto de Oliveira, Francisco Eduardo, Kaizo Iwakami Beltrao, and Maria Tereza de Marsillac Pasinato. 1999. Reforma Estrutural da Previdencia: Uma Proposta para Assegurar Proteqido Social e Equidade. Texto para Discussao, N 690 IPEA. Rio de Janeiro, December. Basu, Kaushik. 1999. "Child labor: Cause, Consequence, and Cure, with Remarks on International Labor Standards." Journal of Economic Literature 37 (September): 1083-1120. Becker, G. S. "The Theory of Allocation of Time." 1965. Economic Journal, September :493-517. Buarque, Cristovam. 1999. "A segunda Abolicao-Cristovam Buarque. Um manifesto- proposta para a erradicacao da pobreza no Brasil." Sao Paulo, Paz e Terra. Buckley, Stephen. 2000. "The Littlest Laborers." The Washington Post, March 16, p. A01. Card, David, and Alan B. Krueger. 1992. "Does School Quality Matter? Returns to Education and the Characteristics of Public Schools in the United States," 100 (February) :1-40. Carneiro, Pedro Manuel. 1999. Child Labor in Brazil. World Bank Background Paper. Washington, D.C., October. Cartilha do PETI 2000. Brasilia, janeiro. MPAS/SAS 53 Cem Perguntas e Respostas Que Voce Precisa Saber Sobre a Bolsa-Escola. 1999. Mania de Educa9ao, Missao Crian,a. Brasilia-DF, novembro. Child Labor in Brazil. 1997. Brasilia: Presidente da Republica Camara de Politica Social. Dimenstein, Gilberto. 1994. "Little Girls of the Night." NACLA Report on the Americas. May/June, Vol. 27, No. 6. Duryea, Suzanne. 1998. "Children's Advancement Through School in Brazil: The Role of Transitory Shocks to Household Income." Washington, D.C.: Inter-American Development Bank, July. Emerson, Patrick M., and Andre P. F. de Souza. 2000. Cornell Department of Economics Working Paper, June. Escudero, Walter Sosa, and Mariana Marchionni. 1999. Household Structure, Gender, and the Economic Determinants of School Attendance in Argentina, October. Fassa, Anaclaudia, Luiz A. Facchini, Marinel M. Dall'Agnol, and David Christiani. 2000. "Child Labor and Health: Problems and Perspectives." International Journal of Occupational and Environmental Health, Vol. 6, No. 1, January- March :55-62. Friedlander, D., and P. K. Robins. 1995. "Evaluating Program Evaluations: New Evidence on Commonly Used Non-experimental Methods." American Economic Review, September :923-937. Green, Duncan. 1998. Hidden Lives: Voices of Children in Latin America and the Caribbean. London: Save the Children. Greene, W. H. Econometric Analysis. 1997. New Jersey: Prentice Hall. Grootaert, Christian, and Harry A. Patrinos. 1999. Policy Analysis of Child Labor: A Comparative Study. New York: St. Martin's Press. Gustafsson-Wright, Emily. 2000. "A Summary of a Qualitative Assessment of the Brazilian Child Labor Eradication Program: PETI in Bahia and Pernambuco." World Bank Background Paper. Washington, D.C., December. Gustafsson-Wright, and Hnin Hnin Pyne. 2000. "Gender Note on Child Labor and Street Children in Brazil." World Bank Background Paper. Washington, D.C., July. Haddad, Lawrence, John Hoddinott, and Harold Alderman. 1997. Intra-Household Resource Allocation in Developing Countries. Baltimore: Johns Hopkins Press. Heckman, J., and V. J. Hotz. 1989. "Choosing Among Alternative Non-experimental Methods for Estimating the Impact of Social Programs: The Case of Manpower Training." Journal of the American Statistical Association, December :862-874. Heckman, J., and J. A. Smith. 1995. "Assessing the Case for Social Experiments." Journal of Economic Perspectives, Spring :85-110 Ilahi, N., P. Orazem, and G. Sedlacek. 2000. "The Implications of Child Labor for Adult Wages, Income and Poverty: retrospective Evidence from Brazil." World Bank, unpublished working paper. Washington, D.C., August. ILO-IPEC. 2000. "Relat6rio de Avalia9ao." Brasilia, August. Innocenti Digest. 1999. "Child Domestic Work." UNICEF. Investigaqdo dos Comprometidos do Trabalho Precoce na Saude de Criancas e Adolescentes: Um Estudo de Caso em 3 Estados do Nordeste Brasileiro, Sumario 54 Executivo. Ministerio do Trabalho e Emprego, Secretaria de Inspecao do Trabalho, Departamento de Segurania e Sauide no Trabalho. Jacoby, Hanan G., and Emmanuel Skoufias. 1997. "Risk, Financial Markets, and Human Capital in a Developing Country." Review of Economic Studies 64 (July) :311-335. Jensen, Peter, and Helena S. Nielsen. 1997. "Child Labour or School Attendance? Evidence from Zambia." Journal of Population Economics, 10 (October) :407- 424. King, Elizabeth M., Peter F. Orazem, and Elizabeth M. Paterno. 1999. "Promotion With and Without Learning: Effects on Student Dropout." Impact Evaluation of Education Reforms, Working Paper No. 18. World Bank, Washington, D.C. Lam, D., and S. Duryea. 1998. "Effects of Schooling on Fertility, Labor Supply, and Investments in Children, with Evidence from Brazil." Journal of Human Resources, March :160-192. Lavinas, Lena. 1998. "Programas de Garantia de Renda Minima: Perpectivas Brasileiras." IPEA. Lavinas, Lena, and M. Bittar. 1999. Special Tabulations of the Minimum Income Programs, December. IPEA Lavinas, Lena. 1999. ",Renta Minima o beca escolar? Paralelos entre las experiencias intemacionales y las iniciativas brasileflas." De Igual a Igual: El Desaflo del Estado Ante Los Nuevos Problemas Sociales. Fondo de Cultura Econ6mica de Argentina, S.A., pp.181-213. Lavinas, Lena. 2000 "Avaliacao de Dois Programas Bolsa Escola-Belo Horizonte e Recife." Levison, D., and K. S. Moe. 1998. "Household Work as a Deterrent to Schooling: An Analysis of Adolescent Girls in Peru." Journal of Developing Areas, Spring :339-356. Levy, Victor. 1985. "Cropping Pattern, Mechanization, Child Labor, and Fertility Behavior in a Farming Economy: Rural Egypt." Economic Development and Cultural Change, 33 (July) : 777-791. Lobato, Ana L. (org.). 1998. "Garantia de Renda Minima - Ensaios e Propostas." Brasilia, IPEA. Lusk, Mark. 1988. "Street Children Programs in Latin America." Journal of Sociology and Social Welfare, Vol. 16, No. 1, March. Manual de Supervisdo do PETI Padrdes Minimos de Qualidade. 2000. Brasilia, janeiro de (1999). Mercado de Trabalho Rural, Informacdo e Aplica,do. 1999. Servi9,o Nacional de Aprendizagem Rural (SENAR) e Organizacao Internacional do Trabalho (OIT). Brasilia. Mercado de Trabalho, Conjuntura e Analise. 1999. Ano 4-outubro, Num. 1 1. Ministerio do Trabalho e Instituto de Pesquisa Economica Aplicada (IPEA). Mincer, Jacob. 1974. Schooling, Experience and Earnings. New York: National Bureau of Economic Research. Moura, A. S., A. Butto, P. B. Fontenelle, and V. L. Jatoba. 2000. "Trabalho Infantil no Brasil: Entre a Cana e o Sisal." Recife, Brazil, May. 55 Moura, Alexandrina Sobreira. 1999. Programa de Erradica,co do Trabalho Infantil na Zona da Alata de R-ernambuco, Brasil. Recife, Brazil, June. Neri, Marcelo Cortes, Emily Gustafsson-Wright, Guilherme Sedlacek, Daniela Ribeiro da Costa, and Alexandre Pinto. 2000. "Microeconomic Instability and the Children's Human Capital Accumulation: The Effects of Idiosyncratic Shocks to Father's Income on Child Labor, School Drop-Outs and Repetition Rates in Brazil." World Bank Background Paper. May. Washington, D.C. OEI/UNICEF. 2000. -Saindo da Rua Para a Conquista de Cidadania." (A Study about Innovative Methodologies used in the Process of the Social Reinsertion of Street Children and Adolescents in Brazil.) Paes de Barros e Rosana Mendonca. 1999. Uma Avaliaqiao do Custos e dos Beneficios da Educa,do Pre-escolar. IPEA, November. Paes de Barros, Ricardo, and Eleanora Cruz Santos. 1991. Consequencias de Longo Prazo de Trabalho Precoce. Relat6rio Interno/N 06, Coordenadora Regional/RJ IPEA, July. Parsons, Donald O., and Claudia Goldin. 1989. "Parental Altruism and Self-Interest: Child Labor Among Late Nineteenth-Century American Families." Economic Inquiry 28 (October) :637-659. Patrinos, H. A., and G. Psacharopoulos. 1997. "Family Size, Schooling and Child Labor in Peru - An Empirical Analysis." Journal of Population Economics, October :377-386; 387-405. PETIManual Operacional. 1999. Brasilia, dezembro. MPAS/SAS. Pesquisa Nacional por Amostragem de Domicilios, PNAD. 1995. Instituto Brasileiro de Geografia e Estatistica - IBGE. Pesquisa Nacional por Amostragem de Domicilios, PNAD. 1997. Instituto Brasileiro de Geografia e Estatistica - IBGE. Pesquisa Nacional por Amostragem de Domicilios, PNAD. 1998. Instituto Brasileiro de Geografia e Estatistica - IBGE. Pesquisa Nacional por Amostragem de Domicilios, PNAD. 2000. Instituto Brasileiro de Geografia e Estatistica - IBGE. Programa de Erradica,do do Trabalho Infantil. Area de Assistencia Social. Serie Programas de Assistencia Social. MPAS/SAS Programa de Erradica,co do Trabalho Infantil. 2000. Area de Assistencia Social. Serie Programas de Assistencia Social. MPAS/SAS Politica Nacional de AssistMncia Social. 1997. Ministerio da Previdencia Social(MPAS)/Secretaria de Estado de Assistencia Social (SAS)Agenda Social. MPAS/SAS. PROGRESA (Programa de Educaci6n, Salud y Alimentaci6n). 1999. "Evaluaci6n de Resultados del Programa de Educaci6n, Salud, y Alimentaci6n.-Primeros Avances." Brazil. Psacharopoulos, G. 1997. "Child Labor versus Educational Attainment: Some Evidence from Latin America." Journal of Population Economics, October, 377-386. Ravallion, Martin, and Quentin Wodon. 2000. "Does Child Labor Displace Schooling? Evidence on Behavioral Responses to an Enrollment Subsidy." Economic Journal. Relatorio Final. 1999. Congresso Nacional - Comissao Mista Especial destinada a estudar as causas estruturais e conjunturais das desigualdades sociais e apresentar 56 solucoes legislativas para erradicar a pobreza e marginaliza,co e reduzir as desigualdades sociais e regionais. Brasilia, D.F. Rizzini, Barker, and Cassaniga. 1988. "From Street Children to All Children, Improving the Opportunities of Low Income Urban Children and Youth in Brazil." Paper presented at the Johann Jacobs Foundation Conference. Germany. Rocha, S. 1999. "Child Labor in Brazil and the Program for Eradication of Child Labor: The Case of Pemambuco." World Bank, background paper, July. Washington, D.C. Rocha, Sonia, and Joao Sab6ia. 1999. Programas de Renda Minima. Linhas Gerais de uma Metodologia de Avalia,co a partir da Experiencia Pioneira do Paranoa no Distrito Federal. IPEA. Rocha, Sonia. July, 1999. Child Labor in Brazil and the Program for Eradication of Child Labor. The Case of Pernambuco. Rio de Janeiro, Brazil. Rocha, Sonia. 1998. "Minimum Income Programs - How Do They Apply to the Brazilian Nuclei?" World Bank Report, March. Washington, D.C. Rosenzweig, M. R., and R. Evenson. 1977. "Fertility, Schooling, and the Economic Contribution of Children in Rural India: An Econometric Analysis." Econe netrica, 45(5) :1065-1079. Saboia, Joao m d Ana Lucia Saboia. 1999. "Trabalho e Criancas e Adoloscentes no Brasil nos Anos 90 - Uma Analise Desagregada por Estado e Microrregiao." Rio dt axnero, Brazil, November. Sant'Ana, Silvio Rocha, Moraes, Andrea. 1997. "Avalia,co do Programa Bolsa Escola du GDF." Fundacao Grupo Esquel Brasil. Scheper-H-ughes, Nancy, and Daniel Hoffman. 1997. "Brazil's Street Kids Won't Just go Away." The Urban Age, Vol. 5, No. 1. Schultz, T. Paul. 1978. "Fertility and Child Mortality over the Life Cycle: Aggregate and Individual Evidence." American Economic Review, May :208-215. Schultz, T. Paul. 2000 "Preliminary Evidence of the Impact of PROGRESA on School Enrollments from 1997." Sedlacek, Guilherme, and Nadeem Ilahi. 1999. "The Potential Economic Benefits of the PETI-Bolsa Escola Programas in Brazil." World Bank Background Paper, November. Washington, D.C. Sedlacek, Guilherme, Emily Gustafsson-Wright, and Nadeem Ilahi. Brazil: An Assessment of the Bolsa Escola Programs. (forthcoming) World Bank, Washington, D.C. Soler, Salvador. 2000. "Crian,as e Adolescentes em Situa,co de Rua: Uma Leitura preliminar de Metodologias e Procedimentos de Monitoramento e Avaliasao Utilizados no Brasil. UNICEF, February. Subs idios ai Elabora,do do Programa de Combate ai Explora,do Sexual de Crianqas e Adolescentes de 7 a 14 anos. (versaopreliminar, nao autorizada). 2000. MPAS/SAS, January. Takahashi, Miki, and Caroline Cederlof. 2000. "Street Children in Central America: An Overview." World Bank, Washington, D.C. Thomas, D., R. F. Schoeni, and J. Strauss. 1995. "Parental Investments in Schooling: The Roles of Gender and Resources in Urban Brazil." Santa Monica, CA: RAND Corporation. 57 Tzannatos, Zafiris. 1998. "Child Labor and School Enrollment in Thailand in the 90's." Social Protection Discussion Paper, No. 9818. World Bank, Washington, D.C. UNICEF. 1993. "Street and Working Children." Innocenti Global Seminar. Summary Report. Waiselfisz, Julio J. 1999. Resumen Ejecutivo del informe "Programa de Becas Escolares del Distrito Federal" de Brasil. De Igual a Igualk El Desaflo del Estado Ante Los Nuevos Problemas Sociales. Fondo de Cultura Econ6mica de Argentina, S.A., pp. 214-227. Waiselfisz, Julio J., Miriam Abramovay, and Carla Andrade. 1998. Bolsa Escola - MelhoriajEducacional e Reduqdo da Pobreza. Projecto Conjunto: UNESCO, UNICEF, POLIS. Brasilia, mar,o. Welch, Finis. 1966. "Measurement of the Quality of Schooling." The American Economic Review, 56 (March) :379-392. Yap, Y. 2000. "A Descriptive Look at Child Labor in 1999 and its Trends during the 1990s: The Case of Brazil." World Bank Background Paper. Washington, D.C., October. Yap, Y. 2000. "Evaluating the Impact of PETI on Child Labor Supply and Schooling Demand in Rural Northeastern Brazil: The Case of Pernambuco, Bahia and Sergipe." World Bank Background Paper. Washington, D.C., August.