Documentof The World Bank FOROFFICIAL USEONLY ReportNo: 28544-BR PROJECTAPPRAISALDOCUMENT ON A PROPOSEDADAPTABLEPROGRAMLOAN INTHEAMOUNTOFUS$572.2MILLION TO THE FEDERATIVEREPUBLICOF BRAZIL FORA BOLSAFAMfLIA PROJECT INSUPPORTOFTHEFIRSTPHASEOFTHEBOLSAFAMfLIAPROGRAM May 25,2004 This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CURRENCY EQUIVALENTS (Exchange Rate EffectiveMarch23, 2004) Currency Unit = Real Real 2,95 = US$1 FISCAL YEAR January 1 - December31 ABBREVIATIONSAND ACRONYMS APL Adaptable ProgramLending CCT ConditionalCashTransfer CEF CaixaEconBmicaFederal CFAA CountryFinancialAccountability Assessment CMC Core ManagementCommittee CPAR CountryProcurementAssessment Report DFID DepartmentFor InternationalDevelopment EBC ElectronicBenefit Card FMRs FinancialMonitoring Reports FMS FinancialManagementSystem FUNAI FundaqiIoNacional do Indio GPN GeneralProcurementNotice IBGE Instituto Brasileiro de Geografiae Estatistica ICB InternationalCompetitiveBidding ICCN Incentivo ao Combate&s CarenciasNutricionais IFPRI InternationalFoodPolicy ResearchInstitute INSS Instituto Social do Serviqo Social IPDP Indigenous People's DevelopmentPlan IPEA Instituto de PesquisaEconBmicaAplicada LCS Least-Cost Selection M A S Ministry of SocialAssistance MDG Millennium Development Goals MDS Ministkrio do DesenvolvimentoSocial e Combateti Fome MEC MinisteriodaEducaqb e Cultura MESA ExtraordinaryMinistry for FoodSecurity MS Ministtrio da Saude (Health) NCB National CompetitiveBidding NIS Numerode Identificaqb Social -Social IdentlJicationNumber PNAD PesquisaNacional por Amostra de Domicilios POF Pesquisade OrqamentosFamiliares PPA ProgramaPlurianual PSRL Public SectorReform Loan QCBS Quality and Cost-BasedSelection SEAIN Secretariade Assuntos Intemacionais SEPPIR Ministry of Policiesto PromoteRacialEquality SFC SecretariaFederalde Controle SIAFI SistemaIntegradode Administraqllo Financeira SOE Statements ofExpenditure SOF Source of Funds SOT Statements ofTransfers SENARC SecretariaNacional de RendadaCidadania - SecretariatforCitizen Income STN Secretariado TesouroNacional SWAP Sector-wideApproach TCU Tribunal de ContasdaUniZLo VIGISUS Disease Surveillance andControl Project Vice President: David de Ferranti Country ManagerDirector: Vinod Thomas Sector Director: Ana-Maria Arriagada Sector Manager: HelenaRibe Sector Leader: Maria Madalenados Santos Task Team Leader: Kathy A. Lindert FOR USEONLY BRAZIL First BolsaFamiliaProject CONTENTS Page A. STRATEGIC CONTEXT AND RATIONALE..................................................................... 1 1. Country.and sector issues.................................................................................................... 1 2. Rationale for World Bank involvement.............................................................................. 3 3. Higher level objectives to which the project contributes.................................................... 4 B. PROJECT DESCRIPTION..................................................................................................... 4 1. Lending instrument. Programobjective andPhases........................................................... 4 2. Project development objective and key indicators.............................................................. 8 3. Project components............................................................................................................. 9 4. Lessons learned and reflected inthe project design.......................................................... 12 5. Alternatives considered and reasons for rejection............................................................ 14 c. IMPLEMENTATION........................................................................................................... 14 1. Institutional and implementation arrangements................................................................ 14 2. Monitoring and evaluation of outcomes/results................................................................ 15 3. Sustainability..................................................................................................................... * . . 15 4. Critical risks and possible controversial aspects............................................................... 17 5. Loan conditions and covenants......................................................................................... 18 D. APPRAISAL SUMMARY................................................................................................... 19 1. Economic and financial analyses...................................................................................... 19 2. Technical ........................................................................................................................... 20 3. Fiduciary ........................................................................................................................... 20 4. Social................................................................................................................................. 21 5. Environment...................................................................................................................... 22 6. Safeguard policies............................................................................................................. 22 7. Policy Exceptions and Readiness...................................................................................... 22 Annex 1: Country and Sector or Program Background................................................................ 23 Annex 2: Major Related Projects Financed by the Bank andor other Agencies.......................... 35 This document has a restricted distribution and may be used by recipients only in the performance of their official duties I t s contents may not be otherwise disclosed . without World Bank authorization . Annex 3: Results Framework and Monitoring.............................................................................. 36 Annex 4: Detailed Project Description......................................................................................... 39 Annex 5: Project Costs.................................................................................................................. 64 Annex 6: Institutional and ImplementationArrangements........................................................... 65 Annex 7: Financial Management and DisbursementArrangements ............................................ 69 Annex 8: Procurement.................................................................................................................. 77 Annex 9: Economic and Financial Analysis................................................................................. 83 Annex 10: SafeguardPolicy Issues............................................................................................... 90 Annex 11: Project Preparation and Supervision........................................................................... 98 Annex 12: Documents inthe Project File................................................................................... 100 Annex 13: Statementof Loans and Credits ................................................................................ 101 Annex 14: Country at a Glance................................................................................................... 105 Annex 15: Letter of Development Policy................................................................................... 107 .. 11 BRAZIL BR BOLSA FAMILIA 1ST APL PROJECT APPRAISAL DOCUMENT LATIN AMERICA AND CARIBBEAN LCSHS-DPT Date: May 25, 2004 Team Leader: Kathy A. Lindert Country Director: Vinod Thomas Sectors: Other social services (60%);Health Sector Manger/Director: Helena G. Ribe (20%); Primary education (20%) Themes: Other social protection and risk management (P);Other human development (P) Project ID: P087713 Environmental screening category: C Lending Instrument: Adaptable Program Loan Safeguard screening category: Limited impact Project Financing Data [X] Loan [ ] Credit [ ] Grant [ ] Guarantee [ ] Other: For Loans/Credits/Others: Total Bank Loan Amount (US$m.):1 572.2 Proposed terms: Fixed-Spread Loan (FSL) Financing Plan (US$m) Source Local Foreign Total FEDERATIVE REPUBLIC 6187.4 (of which the IBRD 7.5 (all of which 6194.9 OF BRAZIL & loan will reimburse 551.5 in comprises part of the INTERNATIONAL BANK expenditures for conditional IBRD loan providing 1.8 cash transfers under the for technical assistance FOR RECONSTRUCTION BFP and provide 13.4 for and project management AND DEVELOPMENT1 technical assistance and and 5.7 as capitalization project management) of the front-end fee) Borrower: Federative Republic of Brazil Responsible Agency: Ministry of Social Development and Eradication of Hunger Address: Ministério de Desenvolvimento Social e Combate ŕ Fome Esplanada dos Ministérios, Bloco C. Brasília, DF, Brazil, CEP. 70.058-900 Tel: (55-61) 313-1630 Estimated disbursements (Bank FY/US$m) IBRD Fiscal Year 2005 2006 2007 Annual 168 227.8 176.4 Cumulative 168 395.8 572.2 Project implementation period: 2.5 years Expected effectiveness date: 08/2004 Expected closing date: 12/31/2006 1The total IBRD loan is US$572.2 million. The BFP conditional cash transfers to beneficiaries (grants) to be reimbursed by the loan are funded with government resources that are largely earmarked for that purpose. Does the project depart from the CAS in content or other significant respects? Ref. PAD A.3 [ ]Yes [X] No Does the project require any exceptions from Bank policies? Ref. PAD D.7 [ ]Yes [X] No Have these been approved by Bank management? [ ]Yes [ ] No Is approval for any policy exception sought from the Board? [ ]Yes [ ] No Does the project include any critical risks rated "substantial" or "high"? Ref. PAD C.5 [X]Yes [ ] No Does the project meet the Regional criteria for readiness for implementation? Ref. PAD D.7 [X]Yes [ ] No Project development objective Ref. PAD B.2, Technical Annex 3 The broad objectives of the BFP are to reduce poverty and inequality and promote human capital investments among poor families through the provision of direct monetary transfers to poor families and incentives for investing in human capital. As a Sector-Wide Approach, World Bank assistance would broadly support these same objectives. As the World Bank's financial contribution is relatively small (US$572.2 million, or 9.3% of total conditional cash transfers under the BFP), it would be most accurate to state that overall reduction of poverty and inequality is the ultimate objective of the Bank's assistance under the proposed project. The proposed project would seek to strengthen the BFP's ability to achieve these broader objectives of reducing poverty and inequality and promoting human capital development by supporting: (a) a consolidation of conditional cash transfer programs and reductions in gaps and duplications in coverage; (b) a strengthening of the system for identifying the target population; (c) the development of a monitoring and evaluation system for the BFP; and (d) a strengthening of the basic institutional functioning of the program. Project description [one-sentence summary of each component] Ref. PAD B.3.a, Technical Annex 4 The proposed project will include 5 components: (a) conditional cash transfers (grants to beneficiaries); (b) strengthening the system for identifying the target population; (c) the development of a monitoring and evaluation system; (d) institutional strengthening; and (e) project management. Which safeguard policies are triggered, if any? Ref. PAD D.6, Technical Annex 10 Indigenous Peoples (OD 4.20). Significant, non-standard conditions, if any, for: Ref. PAD C.7 Board presentation: None Loan effectiveness: 1. Effectiveness Conditions: · Adoption of an Operational Manual for the Project satisfactory to the World Bank, including: institutional arrangements; operational procedures (including a system for monitoring and verification of conditionalities, eligibility procedures); procurement and contracting procedures, sample standard bidding documents and filing instructions (including for QCBS and NCB); disbursement and financial management procedures, including template FMRs; auditing procedures; and supervision, project administration, monitoring and progress reports, evaluation plans; and Performance Indicators. · MDS will appoint for the PIU two procurement specialists, a financial manager and two accountants with qualifications and experience satisfactory to the World Bank. · Draft Terms of Reference for audits approved by the World Bank. Covenants applicable to project implementation: 1. Milestones for moving from 8% to 9% disbursement percentage for Component 1 (reimbursing against SOTs) ("Stage 2 milestones" from Table 2): · New Cadastro Único questionnaires developed · The instruments for implementation of eligibility criteria for BF beneficiaries revised · Normative acts and operational guidelines for the Cadastro Único published on MDS webpage · Terms of reference for design of MIS for BFP prepared · Inventory of existing inter-governmental partnerships for BFP completed · Terms of reference for Quality Control Reviews developed · Terms of reference for baseline beneficiary survey prepared · Matrix of key functional and institutional responsibilities for BFP published · Beneficiary information booklets (cadernetas) designed 2. Milestones for moving from 9% to 11% disbursement percentage for Component 1 (reimbursing against SOTs) ("Stage 3 milestones"): · Data collection strategy designed and piloted · All detectable duplications in the Cadastro Único database eliminated · Cadastro Único users (MDS and 30% of municipalities) with access to Cadastro Único data · Adequate Management Information System (MIS) for BFP developed and operational · Quality Control Reviews for Bolsa Família processes developed and piloted in at least 10 municipalities · Baseline beneficiary survey developed and piloted in at least 10 municipalities · System for verifying human capital conditionalities developed, piloted in at least 5 municipalities · Beneficiary information booklets (cadernetas) distributed 3. Covenant: · The Borrower shall carry out the IPDP. A. STRATEGIC CONTEXT AND RATIONALE 1. Countryand sector issues Brazil's Government is committed to realizing the country's large potential for welfare improvements through simultaneous actions on the economic and social fronts. Deeper social progress is essential to improve the quality o f life o f the least advantaged groups. The importance of achieving visible and rapid social gains while maintaining economic stability is further highlightedby the ambitions, highexpectations, and hopes that rest on this particular administration. Rapid achievements on the social side would consolidate broader support for economic responsibility and difficult reforms, thus strengthening the foundations for growth and for even faster social progressinthe future. Despite impressive progress over the past decade, challenges remain. Education indicators still lag due to Brazil's low starting point, particularly among the poor. Disparities remain in health care funding across regions, and in utilization rates across socio-economic groups. Poverty has remained fairly constant since the mid-1990s even if the poverty profile has changed, with the effects of the recent economic slowdown being felt primarily in metropolitan areas in the Southeast. In fact, poverty levels in metropolitan areas increased sharply, with the Southeast share of poverty increasing from 17-26% from 1998-2001. Poverty incidence among youth also increased, and poverty rates continue to be especially high for those with low levels of education, as well as for indigenous and afro-descendent (Quilombola) populations. Related to these disparities, Brazil is still one o f the world's most unequal societies: with a Gini coefficient of 0.59 for incomes, the richest 1% o f the population receives 10% o f total monetary income-the same share as the poorest 50%. Analysis shows that income inequality in Brazil stems primarily from unequal access to education and a highskill premium for labor, along with a highly regressive pension system. Higherquality and more equitable access to education (including secondary and tertiary education) are key to reducing poverty and inequality inthe long run. However, the positive impacts of education reform take time to materialize. Reforms to social assistanceprograms and a more equitable pensionsystemcan achieve complementary results sooner and substantially reduce inequality and alleviate poverty. ReformAgenda: Integrating Social AssistancePrograms. As part o f its dual focus of maintaining macro stability while achieving more equitable growth and rapid social progress, the Government recently initiated sweeping reforms to improve its social safety net. These reforms seek to integrate several federal programs, including Bolsa Escola, Bolsa Alimentaqilo, Cartilo AlimentaqBo, and Auxilio-Ghs. Although each of these programs maintained its own emphasis (promoting schooling, health care, food consumption, or compensating for adjustment), they all provided cash transfers to roughly the same target group of poor families. Each program was managed with separate administrative structures, beneficiary selection processes, and banking-sector contracts for payments issuance. These separate structures created inefficiencies and administrative duplications, resulted in considerable gaps and duplications in coverage, andmissedimportant synergies fromjointly promotingeducation, health and nutrition. Given these inefficiencies, the reforms seek to integrate these programs into a single, improved conditional cash transfer (CCT) program called the "Bolsa Familia Program (BFP)." The objectives of this safety net integration reform include: (a) consolidating and rationalizing existing federal conditional cash transfer programs; (b) promoting efficiency in the use of public resources; (c) improving the system for identifying the target population; (d) leveraging synergies from jointly promoting education, health and nutrition incentives; (e) strengthening monitoring and evaluation; and (f) leveraging opportunities to promote vertical integration inthe social safety net between federal and sub-national programs. 1 These reforms, and the design of the new Bolsa Familia program, were supported by the First Integrated HumanDevelopment Policy Sector Reform Loan (PSRL) and technically by the programmatic Brazil Social Assistance (BRASA) ESW program(see below). The BolsaFamiliaProgram. With this platform, the objectives ofthe new BFP itself include: (a) reducing poverty and inequality today, through the provision of direct monetary transfers to poor families; and (b) reducing poverty and inequality tomorrow, by providing incentives and conditions for investments in human capital on behalf o f beneficiary families, and by linking beneficiary families to complementary servicesthat could helpthem invest and grow out of poverty inthe future. The BFP defines two target groups. The priority target group is those "extreme poor" families with per capita monthly incomes below R$50 (US$17). The program also targets those "moderately poor" families with per capita monthly incomes below R$100 (US$34)' but higher than R$50 (US$17). Payments are preferably made to the mothers, as per the BolsaFamilia law. The program will provide cash transfers ranging from R$15-95 (US$5-33)2 to target families. The exact transfer amount will depend on income levels and household composition (see Annex 1). These amounts were set with the goal o f minimizingthe number of people benefitingfrom the previous programs that could either be excluded or face reduced benefits under the integrated program. Monthly transfers are expected to average about R$71 (US$24) per beneficiary family, significantly higher than average benefits of about R$24 (US$8) in total for the previous program^.^ On a per capita basis, this represents about 19% of the poverty line used by the World Bank, 12% of IPEA's poverty line, and 6% of the minimumwage. While these transfers could create adverse incentives to work (causing, for example, a reduction intotal adult hours worked), available evidence suggests they do not (see Annex 1). The transfers would be conditional on all relevant family members complying with key humandevelopment conditionalities (see Annex 1) as part o f an important policy shift towards focusing assistance on the family (rather than on individuals). Previous programs provided support and conditionalities only for some individuals within the family. For example, under Bolsa Escola, school attendancewas only requiredfor up to three school-aged kids, but not for any more than three children. The same was true for Bolsa Alimentaqiio. In addition, synergies are expected from the simultaneous promotion of investments inhealth and education within a family. In some areas, the supply of adequate or the availability of culturally appropriate health and education services may be lacking. The BFP i s developing a policy to establish co- responsibility requirements for beneficiary families inthese cases. The interaction between targeted CCT programs and the provision of more universal"supply-side'y services for health and education is mutually complementary: (a) on the one hand, the provision o f quality health and education services is clearly a crucial requirement for conditional transfers to work; as such, efforts are being made to simultaneously invest in such services; (b) on the other hand, even if health and education services were universally available, CCTs would be needed to help alleviate common demand-side constraints (direct and indirect costs) that prevent poor households from taking up these services. Investments inboththe BFP, and inthe supply and quality o f health and education services, are thus crucial elements of a balanced humancapital development strategy. An inter-ministerial social committee has been established to help ensure coordination among the social sector ministries (and the Ministry o f Education, 'Brazildoes nothaveanofficial poverty line andresearch institutionsusedifferentpovertylines that leadto differentpovertynumbers. The poverty line used in the World Bank`s Poverty Assessment is set at aroundR$80 per month (US$27) and is adjusted for each region using a regionalprice index. This results in a nationalpoverty rate of 28.7%. IPEA's extremepoverty line was establishedat R$62, yielding an extreme poverty rate of 14.6%. PEA calculatesthe full poverty line as twice the extremepoverty line, or R$124, resulting in apovertyrate of 33.6%. This value is close to that beingusedbythe Government intheir documents. * US dollaramounts are providedfor referenceonly, usingan exchangerateof 2.9. d'Avila Viana, Ana Luiza (February 2004). "Programa BolsaFamilia." Preparedfor the World Bank in preparationfor the Shang'hai ScalingUp events. 2 for example, has also established a special secretariat to focus on the issues of inclusion and diversity in education, including coordinating links with the BFP). In addition, an important policy emphasis for the BFP is to foster bridges between Bolsa Familia beneficiaries and other complementary services so as to help promote their "emancipa@o" (development) and to help them escape from poverty (porta de saida). Indeed, BolsaFamilia has opted notto establish rigid time limits for beneficiaries, but rather to help reduce their dependency on social assistance by fostering their development throughthese links(as discussed inAnnex 1). Bolsa Familia is rapidly becoming one o f the Government's major social initiatives. It surpassed its target o f 3.6 million families in its first three months of operation in 2003 (reaching 3.615). The program targets coverage of 11.2 million families (or about 44 million people) by 2006, which will represent a significant scaling up of social assistanceas compared with the (overlapping) coverage o f its predecessors. Meetingthese targets will require a significant but managed increase in financing, reaching a total of R$31 billion (US$11.5 billion) over the PPA (multi-year plan) period from 2004-07. The planned budget is R$5.8 billion (about US$2 billion) for 2004, which i s already over R$1.5 billion higher than spending on the separate programs in 2003, mainly due to the higher average benefits paid to Bolsa Familia beneficiaries as compared to those on the previous programs. In terms of shares of total federal primary expenditures and GDP, the program will rise from 1.1% o f primary expenditures in 2003 to 2.5% by 2006 and from 0.2% of GDP in 2003 to 0.5% by 2006. This rise in spending needs will likely be offset by expected savings from the recently-approved pension reforms, estimated at 0.4% o fGDP (or about R$8 billion) peryear duringthat period. The shift in resources from regressive pensions to progressive conditional cash transfers under the BFP i s an important part o fthe Government's strategy to reduce poverty and inequality inBrazil. Efficiency gains will arise from mergingthe programs, and will also partially offset these higher costs (thus allowing for greater coverage). The program is also seeking better complementarity and a pooling o f resources with state and local programs for expanded coverage to further expand the pool of available resources. This includes various types of partnerships, including: (a) some that seek vertical integration with existing local bolsa-type programs (that would allow for a pooling o f federal and local resources to either expand the number of beneficiaries covered or increase the benefit levels); and/or (b) some that seek to link beneficiaries of the BFP with complementary services (as discussed inAnnex 1). 2. Rationale for World Bankinvolvement The proposed Project will support one of the Government's flagship social initiatives: the Bolsa Familia Program. First, it would help the Brazilian authorities to carry out and improve the effectiveness and efficiency o f the BFP over the course o f the project period, reducing overlaps and filling gaps in coverage. Second, the World Bank loan would also help to ensure that the program receives priority and is protected from possible budget cuts duringthe project period. Third, the proposed project would, at the same time, contribute to the financing o f Brazil's critical development priorities overall. Fourth, the loan would provide critical additionality intechnical design and implementation. The activities supported by the loanto strengthen the system for identifying the target population, monitoring, evaluation, and institutional arrangements seek to provide important value added with respect to the entire program. In addition, performance incentives (see below) to pace disbursements against the productive activities of the BFP with technical improvements to the program, as well as on-going World Bank involvement in supervision throughout the project period, would help augment the impact derived from the technical components. As such, the World Bank's participation with a loan size that i s only a small share (less than 10%) of the BFP 3 over the project period would nonethelessallow for a scale of impact on the content and performance of the entire program. This first ambitious program-based sector investment operation (Program-Based SWAp) builds on lessons learned in smaller SWAP components in previous projects, in Brazil and elsewhere. It also paves the way for the use o f SWAps in Brazil, as set out in the FY04-07 CAS, in support o f high priority government programs where similar Bank policy and technical additionality can have a strategic and catalytic impact on a sector or large program. The proposed project will draw on the significant technical assistance that the World Bank has been providing to the Government for the conceptualization and design of these reforms since March 2003, following discussions between President Lula and President Wolfensohn. This support has been provided underthe umbrella of the Brazil Social Assistance Program (BRASA,EW-P078828-ESW), a programmatic ESW/AAA program, including assistance on the overall design of an integrated program, issues pertaining to the system for identifying and selecting beneficiaries, and monitoring and evaluationo fthe program. The proposed project would also build on and complement the on-going PSRL program in the Human Development (HD) sectors, which emphasizes actions to enhance equity, improve the quality and efficiency of social programs, and increase accountability in the social sectors. It will also entail significant synergies with the proposed HDTechnical Assistance Loan (TAL), which would assistthe Government indeveloping institutional capacity to monitor the progress and impact of its social policies. Specifically, the proposed TAL would provide support to the BrazilianInstitute of Geography and Statistics (IBGE) to strengthentheir household survey system for the purposes of monitoring and evaluating social policy, including the generation o f data that would be highly relevant for monitoring and evaluatingthe performance of the BFP. The TAL would also support a strengthening of monitoring and evaluation in education and health, which entails valuable inter-sectoral linkages for the BFP. More generally, the World Bank has acquired considerable technical expertise around the world in strengthening targeting systems, developing monitoring and evaluation systems for conditional cash transfers, and in promoting the types o f institutional innovations envisaged for the BFP. The Bank team consistsof specialists who bringconsiderable international experience ineach of these areas. 3. Higherlevelobjectivesto which the projectcontributes The proposed project is at the center of one of the core pillars of the Bank's CAS - achieving a more equitable Brazil- and is closely aligned with corporate priorities of achieving the Millennium Development Goals (MDGs). The BFP will directly contribute to the reduction of poverty (MDG 1) and inequality inthe short- and long-term. Incoordination with other government programs, many of which are Bank supported, it will also contribute to reducing malnutrition (MDG l), achieving universal education (MDG 2), reducing child mortality (MDG 4), and improving maternal health (MDGS) through the demand-side incentives it provides for investments ineducation, nutrition and health (for pregnant women andyoung children). B. PROJECTDESCRIPTION 1. Lendinginstrument,ProgramobjectiveandPhases The project will adopt three key features interms of the lending instrument: (a) a programmatic approach; (b) a Program-Based Sector-wide Approach (SWAp); and (c) mechanisms to pace disbursements against conditional cash transfers under the BFP (Component 1) in line with technical improvements. These three 4 features will help foster an effective link between the implementation of the BFP while helping strengthen the program through crucial technical assistance. ProgrammaticApproach (APL). The World Bank proposesto support the Bolsa Familia Program (BFP) with a programmatic series o f operations (Adaptable Programmatic Lending, APL). This programmatic approach would allow for a long-term, on-going partnership between the Government and the World Bank on one o f the Government's main flagship social programs (see Annex 15). It would also permit a deepening of the technical strengthening of the program over time. The overall development objectives of this multi-phased program would be to strengthen the BFP's ability to reduce poverty and inequality by providing poor families with income transfers and incentives to invest inthe human capital of their children. With these overall objectives, the programmatic series of operations would be divided into two phases: 0 Phase 1 (mid-2004 to end-2006). This first project (with a proposed loan of US$572.2 million) would cover the transition phase, focusing on consolidating the four main conditional cash transfer schemes and improving the basic architecture of the BFP. As discussed below, the main objectives o f this phase include strengthening the program's effectiveness in reducing poverty and inequality and promoting human capital development by: (a) consolidating conditional cash transfer programs and reducing gaps and duplications in coverage; (b) improving the system for identifyingthe target population; (c) developing a monitoring and evaluation system for the BFP; and (d) strengthening the basic institutionalfunctioning ofthe BFP. 0 Phase2 (2007-08). The secondproject (with a proposed loan o fUS$520.2 million or higher) would buildon these foundations to further consolidate the safety net and deepen technical improvements aiming to transform Brazil's safety net in general and the BFP in particular into a modern social assistance and social service system. Such improvements would depend on the institutional and reform opportunities that arise in Phase 1, but could include: (a) supporting additional efforts to consolidate the safety net (for example with possible further integration of other transfer programs into the BFP); (b) further improvements in the system for identifying the target population; (c) on- going contributions to monitoring and evaluation (including the implementation of the follow-up beneficiary survey); (d) further improvements in service delivery, including possible mainstreaming o f approaches to link BF beneficiaries to complementary services and develop tailored empowerment (emancipaqh) strategies for improving their self-reliance. The key achievements o f the first phase - which would also constitute trigger indicators for moving to the secondphase -are described inTable 1below. 5 IConsolidatingConditionalCash e Coverageof at least2/3 of extremepoorfamilies by program Transfers e Pre-reformtransfer programseliminated(former beneficiarieseither convertedto BolsaFamiliaor droppedfrom programs) Strengtheningof System for CadastroquestionnaireandBF eligibility criteriarevisedandpublished Identifyingthe Target Population Expansionandrecertificationdata collectionstrategy designedandimplemented All detectableduplicationsinthe CadastroUnico databaseeliminated e Periodicityfor updatingCadastroUnico andrecertifyingBolsaFamiliabeneficiaries formally established Recertificationprocesscommencedinat least25% of municipalities e CadastroUnico users (MDSandat least 50% of municipalities)with accessto data Developingmonitoringand MISfor BFPfully operationalandgeneratingadequatereports evaluation system e Results of at leastone roundof Quality Control Reviews for BolsaFamiliaprogram published Baselinebeneficiarysurvey conducted InstitutionalStrengthening Matrix of key functionaland institutionalresponsibilitiesfor BFPpublished Strategy for operatingprograminremoteareas developed Beneficiaryinformationbooklets (cadernetas) designed, publishedanddistributed Systemfor verifying humancapitalconditionalitiesoperating Interms of financingmechanisms, Component 1ofthe proposedloanwould reimburse the Governmentfor a small share (less than 10%) of productiveexpenditures on conditionalcash transfers made underthe BFP, which are funded with government resources that are largely earmarked for that purpose. The Government i s pursuingfiscal reforms to address this situationand the World Bank i s supportingsuch measuresthrough the fiscal policy dialogue and relatedESW and lendingprogram (as discussed in the CAS). The Brazilian Government requestedthat the World Bank loan, in support of Component 1, would reimburse a share of these expenses. Giventhe current earmarkingofrevenuesinBrazil, government revenueswouldbeapplied to make the payments for expenditures under Component 1, and the Bank would then reimburse expenditures as specified in the legal agreement. The Bank funds disbursed would be part of the regular federal budget financing, with the Bank as one of the sources of overall finds, which are spent consistent with federal budget policies and procedures. The Governmenthas indicatedto the Bank its intentionto use the increased funds available for development priorities. Annual federal budgets are approved by the BrazilianFederal Congress and are derived from the budget plan (LDO) and the Multi-Year Plan (PPA), which is a transparent four-year framework for federal expenditures, developed by the Brazilianauthorities with the participationof civil society, the World Bank, and other development partners. The Government engageswith the Bank inadiscussion ofthe development expenditures. Aside from the Bank's financial contribution, the loan would provide critical additionality in technical design and implementation. As discussed below, the Government has requested that the World Bank provide technical assistance and partneringto support key improvements in its system for identifyingthe target population, monitoring and evaluation, and institutionalreforms under three technical components. Performanceincentives (see below) to pace disbursements against the productiveactivities ofthe BFP with technical improvements to the program, as well as on-going World Bank involvement in supervision throughout the project period, would help augment the impact derived from the technical components. 6 Hence while the Bank's financial participation would represent only a small share of the program over the project period, its support seeks to strengthenthe scale and performance o f the entire BFP. The financing arrangements for these three technical components (and for a small component to support project management) would conform to traditional lending arrangements for technical assistance. This would ensure an on-going close association of the Ministry of Social Development and the World Bank for the implementation o fthe technical assistanceactivities. PacingDisbursementswith TechnicalImprovements The proposed project would also build in certain mechanisms to pace disbursements in-linewith technical improvements. While the total amount o f financing for Component 1 (US551.5 million) would represent 8.9% of the total cash transfers budget for the BFP duringthe project period, disbursements and cost-sharing percentages would be linked to key milestones of technical improvements. Specifically, loan proceeds for this component would be made with monthly disbursement percentages that will increase as the actions listed in Table 2 are met. The disbursements will start at the monthly rate of 8%, will be increasedto 9% when the stage 2 milestones in Table 2 have been met; and will be increased to 11% when the stage 3 milestones inTable 2 have been met. The pacing of disbursements for Component 1 with technical improvements to the program (supported by Components 2-4) has several objectives. First, it would allow the World Bank to partner with the BFP in a pragmatic way, recognizing the important role ofthe program inthe short-run and also the needfor technical improvements inthe program's systems. Second, it would promote stronger synergies between support o f program implementation and on-going safety net consolidation (Component 1) and the specific support to technical activities to strengthen the basic architecture of the program (Technical Components). Third, it would create incentives for the program implementing agency to achieve, or exceed, the key milestones for systems improvements and for the Finance and Planning ministries to ensure support for these crucial technical activities (e.g., if milestones are reached, the loan's financing percentage would increase and disbursementswould accelerate). Assuming that the stage 2 and 3 milestones are met by end 2004 and end 2005 respectively, the pace of disbursements is such that loan proceeds for Component 1 would be disbursed by end 2006, as indicated in Table 4 below. In the unlikely event that stage 2 milestones are not met, disbursements would continue to be made at a 8% disbursement percentage; similarly, in the event that stage 3 milestones are not met (but stage 2 milestones have been met), disbursements will be made at a 9% disbursement percentage. If disbursementsremain at either o f these lower disbursement percentages, they would continue into 2007. The full level of support would still bejustified sincethe pre-reform CCT programs (such as Bolsa Escola and Bolsa Alimentagiio) - and the newly created BFP that replaces them - are among the top third of social assistance programs in developing countries and would thus be worthy o f support even under the unlikely worst-case scenario in which the proposed improvements are not implemented. Moreover, the risks of this scenario materializing are very unlikely as the milestones were designed to serve as reasonable indicators of timely progress in strengthening the systems of the BFP and they have the full commitment of the Ministry of Social Development (MDS). In addition, since the triggers for moving to the second-phase APL project represent cumulative progress with technical improvements encompassed by the two sets of milestones, the potential withdrawal o f the second-phase APL operation provides a sufficient disincentive for the unlikely scenarioof non-compliance. 7 Table 2 -PERFORMANCE MILESTOP S (Disbursement Conditions) STAGE 2 MILESTONES STAGE 3 MILESTONES (indicative target date: by December 2004) (indicative target date: by December 2005) Strengthening 0 New CadastroUnico questionnaires Datacollectionstrategy designedandpiloted of Systemfor developed All detectableduplicationsinthe CadastroUnico Identifying the Instrumentsfor implementingthe eligibility databaseeliminated Target criteriafor BF beneficiariesrevised CadastroUnicousers (MDS and30% of Population Normativeact; and operationalguidelines for municipalities)with access to CadastroUnico the CadastroUnico publishedon MDS data webpage Developing Terms ofreference for designof MISfor the Adeauate ManagementInformationSystem(MIS) monitoring and . I BFP prepared for BFP developedandoperational evaluation Inventoryof existinginter-governmental Quality Control Reviews for BolsaFamilia system partnershipsfor BFP completed processesdevelopedandpilotedinat least 10 Terms of reference for Quality Control municipalities Reviews developed Baselinebeneficiary survey developedandpiloted 0 Terms of reference for baselinebeneficiary in at least 10municipalities survey prepared Institutional 0 Matrix ofkey functional and institutional Systemfor verifying humancapital Strengthening responsibilitiesfor BFPpublishedonMDS conditionalitiesdeveloped, piloted in at least 5 webpage municipalities 0 Beneficiaryinformation booklets Beneficiaryinformationbooklets(cudernetus) (cudernetus) designed distributed J 2. Project development objective and key indicators As discussed elsewhere, the broad objectives of the BFP are to reduce poverty and inequality and promote human capital investments among poor families through the provision o f direct monetary transfers to poor families and incentives for investing in human capital. As a Sector-Wide Approach, World Bank assistance would broadly support these same objectives. As the World Bank's financial contribution is relatively small (US$572.2 million, or 9.3% of total conditional cash transfers under the BFP), it would be most accurate to state that overall reduction o f poverty and inequality is the ultimate objective o f the Bank's assistanceunder the proposed project. The proposed project would seek to strengthen the BFP's ability to achieve these broader objectives of reducing poverty and inequality and promoting human capital development by supporting: (a) a consolidationo f conditional cashtransfer programs and reductions ingaps and duplications in coverage; (b) a strengthening ofthe systemfor identifyingthe target population; (c) the development of a monitoring and evaluation system for the BFP; and (d) a strengthening o f the basic institutional hnctioning of the program. Key performance indicators for tracking progress towards these objectives are listed in Table 3 below. 8 Table 3 -Key Performance Indicators for Monitoring Progress Over Period Covered by Phase 1Project OUTOME INDICATORS b At least2/3 ofextremepoorfamiliesreceivingBFincometransfers b At least40% oftotaltransfersgoingto families inthe bottomquintile b At least80%ofprimary-ageschool-agechildreninextremelypoor beneficiaryfamiliesattendingschool b At least95% ofbeneficiarychildrenwith healthcards INTERMEDIATE RESULTSINDICATORS P Pre-reformprogramseliminated(former beneficiarieseither convertedto BolsaFamiliaor droppedfrom programs) b TransactionscoFtsfor transfer paymentsreduced b NewCadastroUnicoquestionnairesdevelopedandprinted b Expansionandrecertificationdatacollectionstrategydesignedandoperational b EliminationofalldetectableduplicationsintheCadastroUnicodatabase b Periodicityfor updatingCadastroUnicoandrecertifyingBolsaFamiliabeneficiariesformally established 9 CadastroUnicousers (MDS andat least50% ofmunicipalities)with accessto data b AdequateMISdevelopedandoperating 9 Qualitycontrolsurvey for BFprocessesimplemented b Baselinebeneficiarysurveyconducted b Matrix ofkey functionalandinstitutionalresponsibilitiespublished b Strategyfor operatingprograminremoteareas developed b Beneficiaryinformationbooklets(cudernetus) designed, publishedanddistributed b Systemfor monitoringhumancapitalconditionalitiesoperating b Trainingprovidedto at least 1/2ofmunicipalitiesfor their roleinimplementingBFprogram 3. Project components The proposed first phase APL project (mid-2004 to end-2006) would consist of five components: one component that would reimburse the Government for expenditures on conditional cash transfers under the BFP; three mutually-reinforcing technical components to help improve the quality and basic architecture of the program; and a small component for project management. ComDonent 1: Conditional Cash Transfers (Grants) (US$6,174 million in project costs; US$551.5 4. Institutional I 0.5 1.3 1.o 2.8 million in loan financing). This 0.2 I 0.3 0.3 0.8 component would provide support to 5. ProjectManagement FrontEndFee: I/ 5.7 0.0 0.0 5.7 the Government in consolidating and rationalizing conditional cash transfers, and in reducing duplications and gaps in coverage o f the BFP, as part of a broader and balanced strategy for economic and human capital development. For this component, the World Bank would reimburse the Treasury a share of the expenses on conditional cash transfers made under the BFP duringthe project period. The category of disbursements would be cash transfers (BF grant transfers). Disbursement percentages would increase as key technical improvements ((`milestones") are made, as discussed above. (See Annex 7 for details on financial management and disbursement arrangements). Component 2: Strengtheningthe System for Identifying the Target Population (US$4.4 million in project costs including contingencies). This component would provide assistance to improve the mechanisms used to identify the target population of the BFP, which constitute a key element of the basic 9 architecture of the program. With a more accurate system for identifyingthe target population, the program will achieve better redistributive efficiency, with a higher share of its resources going to the poor, and will be able to cover a larger share o f the poor. It will thus serve as a more efficient instrument for reducing poverty and inequality inBrazil. As with its predecessors (pre-reform programs), the BFP selects beneficiaries from a national database of potentially poor households called the "Cadastro Unico." While the construction o f the Cadastro Unico sought to integrate household registries across the (previous) multiple programs in a very short period of time, there are several aspects of the system that could be improved. To do this, Component 2 of the proposed project will support activities to strengthen the system for identifyingthe target population of the BFP and overhaul the Cadastro Unico system inseveral key areas: (a) improving the strategic framework for the overall system including conducting a study to identify other programs that could potentially use the Cadastro for eligibility determination; (b) revising the eligibility criteria for the BFP and the Cadastro questionnaire and data collection strategy, taking into account different local realities reflecting urban vs. rural settings (including population and poverty density) and diverse cultural, ethnic and racial characteristics o f the populations; (c) reducing administrative errors in the Cadastro Unico; and (d) strengthening the technical and technological capacity o f the Cadastro Unit in MDS to monitor, verify and use the Cadastro Unico, as well as the capacity o f municipalities and other potential cooperating institutions for carrying out data collection and beneficiary selection. Specific activities to be financed are listed and described inAnnex 4. Interms of general timing of activities and results of this component, most activities to be supported under the component assume that preparatory activities would be undertaken in 2004 and early 2005, so as to allow for the rolling out of the new Cadastro system in2005-06. This would mirror the upcoming needs of the BFP in terms of its planned large-scale expansion in 2005-06 as well as the need to recertify existing households in the Cadastro in that period (since by that time, their information would have expired). As discussed above, improvements to the Cadastro Unico - including designing and printing the new questionnaires, revising and publishing the new BFP eligibility criteria, designing an improved data collection strategy for expansion and recertification, eliminating all detectable duplications in the Cadastro database, and granting access for MDS and municipalities to Cadastro Unico data - would constitute performance milestones for increasingthe disbursement percentage for Component 1, as well as triggers for moving to Phase 2 o f the proposed APL program as they would signal key achievements in strengthening the basic architecture ofthe program. Component 3: Developing a Monitoring and Evaluation System (US$7.0 million in project costs includingcontingencies). This component would support the design and implementation of a monitoring and evaluation (M&E) system. Such a systemwould constitute a key element o f the basic architecture of the program, and one that the Government has made a top priority since the launching o f the BFP. To support this, Component 3 of the proposed project will support: (a) the development of a strategy for the overall M&Esystem and support for capacity buildinginMDS inthis area; (b) the strengthening ofBolsaFamilia's management information system (MIS) which would provide up-to-date information on program activities and outputs; (c) the development o f instruments to monitor program processes to provide feedback on the quality of service delivery and program implementation and to detect problems so that they can be corrected early, including (i) inter-government processes; (ii) annual Quality Control Reviews to monitor BFP processes (including verification of health and education conditionalities, the selection of beneficiaries, and payments); and (iii) support for mechanisms for citizen oversight ("social control") respecting socio-cultural and ethnic specificities of prospective beneficiaries; and (d) the development, execution, and analysis of qualitative instruments and quantitative surveys to monitor and evaluate key outcomes and the impact o f the program at both the "territorial" (municipal and sub-municipal) and family levels. Specific activities to be financed are listed and described inAnnex 4. 10 Since the design and operation of a solid monitoring and evaluation system constitutes a key element of the basic architecture of the BFP, such improvements would constitute performance milestones for increasing the disbursement shares for Component 1,and triggers for movingto Phase 2 ofthe proposed APL program, as noted above. Component 4: InstitutionalStrengthening (US3.2 million in project costs includingcontingencies). This component would support the strengthening of the basic institutional functioning of the BFP in terms Of: 0 Strengthening of the operational institutional framework, including the legal architecture of the program (developing needed regulumentos, operational guidelines, norms, etc. for the BFP and the CadastroUnico); 0 Strengthening program management, including: (a) clarifying and strengthening institutional roles for the basic functioning o f the program itself, including developing a matrix o f responsibilities, clarifying information flows for the Cadastro Unico, and developing and implementing a system for verifying human capital conditionalities; (b) providing training and technical assistance on the system for verifying human capital conditionalities; (c) developing a strategy for operating the program in remote areas; (d) defining possible mechanisms for inter-governmental partnerships to help inexpanding the program, by increasing the value o f benefits, and/or expanding coverage, and linking BFP beneficiaries to complementary services; and (e) developing an inter-municipal information network to share good practices, experiences with implementing conditional cash transfers and the BF Programacross municipalities; 0 Developing a solid communications and dissemination strategy with consideration o f culturally- and ethnically diverse publics, including: (i) establishinga "hotline" number with trained operators to answer questions on the program and the Cadastro Unico (for the general public, beneficiaries, other officials, etc.); and (ii)designing, printingand disseminating information booklets (cadernetus) for beneficiaries (including information about benefits and conditionalities); and 0 Identifying and evaluating social welfare innovations in "empowerment" and "bridging" approaches. This sub-component would support the identification and evaluation of initiatives to help BFP beneficiaries to develop tailored empowerment strategies ("emuncipuqi?oyystrategies) and to gain access to complementary programs and services (bridging services). These initiatives could involve various approaches, including: (a) social worker support to help beneficiaries develop these empowerment (emuncipuqiio) strategies and bridging access to complementary services; and/or (b) the development o f "one-stop shops" where beneficiaries can gain access to information about a variety of social services (including the BFP). Some similar experiences have also been developed at the local level in Brazil and in other countries. The sub-component would thus support: (a) international study tours and exchanges to share information on these various "empowerment" and "bridging" approaches; (b) consultancies and workshops to identify and evaluate such approaches; and (c) training and capacity building for beneficiaries, trainers (multiplicadores) and program managers(gestores). Since strengthening the basic institutional functioning o f the BFP constitutes a key element of the basic architecture of the program, such improvements would constitute performance milestones for increasing the disbursement shares for Component 1, and triggers for moving to Phase 2 o f the proposed APL program, as noted above. Annex 4 provides additional details on the component. Component 5: Project Management (US$0.7 million in project costs including contingencies). The project will support a strengtheningo f the implementation capacity of MDS. To provide such support, the 11 proposed Project will finance incrementalcosts related to management and supervision of Project activities. These costs include consultants that would support strategic, technical and operational aspects of the project, as well as equipment and travel expenses. Annexes 4 and 6 provide additional details. 4. Lessons learned and reflected inthe project design The choice of project design benefits from lessons learned inBrazil's relatively long and rich history with conditional cash transfers (both at the federal and at the local levels) and on international experience. As discussed above, it also builds on the findings of an on-going programmatic sector study (the BRASA), particularly the in-depth work on safety net integration reforms and targeting. Among the main lessons learnedfrom these experiences are: (a) Instrument. Experience to date -albeit still very limited-reveals that as an approach, SWAps offer an enormous opportunity for leveraging up the Bank's own financial contribution to provide impact on an entire sectoral program (rather than just at the project level). By relying on the Borrower's fiduciary framework (in this case for financial management only), the SWAP approach is responsive to client requests for a streamlining o f Bank requirements by eliminating the need for maintaining parallel records and procedures only to satisfy the Bank's own requirements. In addition, task manager experience with an ongoing SWAP in Brazil indicates that the end result of streamlining fiduciary systems and requirements i s that the Bank's supervision process becomes focused more on providing technical advice and addressing sectoral issues that may arise, as opposed to focusing on procedural requirementsthat are best handled by fiduciary specialists. (b) Importanceof strong Borrowercommitment, ownership and leadership. There is no doubt that the BFP has both high-level and widespread commitment. Indeed, it is the Government's flagship social initiative. Support for the consolidation of conditional cash transfers as a way to build on previous successes and bring them to a new level of performance has been widespread, cutting across political parties, branches o f government (with strong executive support and rapid passage of the Bolsa Familia Law by the Congress), academic circles, civil society and even the media. It is generally believedthat such integration will improve the efficiency and equity o f these instruments. Internationally, providing conditional cash transfers to poor families has been shown to be an operationally feasible and politically acceptable approach to social assistance. Researchhas shown that concerns about cash subsidies being "assistentialist" can be overcome by: (a) linking the assistance to desirable behaviors such as sending children to school and for health care visits; and (b) giving the transfers to mothers whose decisions regarding the intra-household allocation of resources often favor children's nutrition, health and education. Both of these conditions are features o f the BFP. Other concerns about possible "welfare dependency" can be alleviated by efforts to link beneficiaries to other complementary servicesthat could help them "grow" out o f poverty (as will be developed under the BFP, with the support of the proposed project). (c) Importanceof accurate selection of beneficiaries. Internationally, the impact of transfer programs on poverty and inequality has been considerably higher with an effective system for identifying and selecting beneficiaries. As discussed above and in Annex 4, recent evaluations (including one conducted under the BRASA programmatic ESW) suggest that several aspects of Brazil's Cadastro Unico - the household registry from which Bolsa Familia beneficiaries are selected - should be improved. Given its importance as a critical feature of the basic architecture of the program, the proposed project will support key activities to strengthenthe system for identifyingthe target population of the BFP (and the Cadastro Unico system). 12 (d) Impact on educational attainment. Conditional cash transfer programs have improved educational indicators and outcomes. An ex-ante evaluation by Bourguignon, Ferreira and Leite (2003)4 found that Brazil's Bolsa Escola Program (one of the predecessorsto the BFP) significantly increasedthe share of children in school and decreased the share that were only working. Using similar methodologies, simulations in Annex 9 suggest that the BFP could significantly increase total educational attainment and reduce repetition rates. Mexico's PROGRESA program (a conditional cash transfer program now called Oportunidades) has had significant impacts in improving educational attainment of the poor, raising total years of schooling by an additional 0.66 years. It has also contributed to increasing enrollment and reducingrepetition and drop out rates. (e) Impact on health and nutrition. Conditional cash transfer programs have also improved health and nutrition indicators, but care must be taken in avoiding the use o f anthropometric indicators for eligibility purposes. Mexico's PROGRESA program significantly improved nutrition and helped make young children more robust to illness. IFPRI's evaluation of the Bolsa AlimentaqBo Program (one of the predecessorsto the BFP) shows it had a significant impact on food consumption and dietary diversity. Nonetheless, despite these positive impacts, the program had a surprisingly negative impact on child growth. Since most o f the mothers selectedto participate inthe program were former beneficiaries of a program ("ICCN") that previously distributed milk and oil exclusively to underweight children, it is possible that beneficiary mothers assessed that the continued eligibility of their children would be negatively affected by their achieving normal weight. This kind o f strategic behavior had been widely reported by health workers monitoring ICCN beneficiaries. The IFPRI evaluation also found that the Bolsa AlimentaqBo program had a small positive effect on the growth o f children who had not previously benefited from the ICCN program. The implications o f these results underscore the importance o f not using anthropometric indicators for eligibility purposes and investing in significant public information campaigns to promote the positive messages of growth promotion. In fact, the BFP includes growth promotion as a condition of receiving benefits for very young children, and does not base eligibility on evidence o f malnutrition. ( f ) Empowerment strategies and bridging to complementary services. International experience suggests that efforts to link ("bridge") beneficiaries o f transfer programs to other complementary services and to help them develop empowerment (emancipa@o)strategies can be an effective way to help them "grow" out o f poverty and reduce their dependence on social assistance. As discussed above and in Annex 4, the proposed project would support the identification and evaluation of these approaches. (g) The need to establish a solid monitoring and evaluationsystem. Any new, large-scale social sector program needs to have a system to monitor program activities (MIS), implementation processes, beneficiary satisfaction and service delivery, and program outcomes and to evaluate program impact. Such a system would include formal information mechanisms (such as surveys and audits), but also citizen oversight mechanisms (social control) to promote transparency and accountability. As discussed above and in Annex 4, the proposed project supports the development o f such a system for the nascent BFP. (h) Institutional and implementation capacity. Programs o f this nature require sophisticated administrative capacity. Although Bolsa Familia i s a new program operating under a new ministry, it builds on the foundations established by the pre-reform programs (e.g., maintaining the channeling of payments to beneficiaries in a fairly direct manner through Brazil's extensive banking system). In addition, many o f the staff involved in the Ministry o f Social Development in general and the BFP in `Bourguignon, Franpois, Francisco H.G. Ferreira, and Phillippe G. Leite (2003). "Conditional Cash Transfers, Schooling, and Child Labor: Micro- SimulatingBrazil's BolsaEscolaProgram." World Bank EconomicReview, Vol. 17.No. 2. 13 particular have extensive prior experience managing or working on the federal pre-reform programs or similar local CCT programs. Institutional arrangements are discussed in Annexes 1, 6, and 7. As discussed above and in Annex 4, the proposed project supports the strengthening of institutional capacity both at the federal and sub-national levels (via training, systems strengthening, consultancies, etc.). 5. Alternatives considered and reasonsfor rejection The approachadoptedfor the proposedproject has many advantagesover other possible instruments: 0 Programmaticvs. Single Operation. The programmatic APL approach allows for a long-term and on-going partnership between the Government and the World Bank to support Brazil's flagship social program; 0 Investment vs. Adjustment Operation. The Government has requested the use of an investment operation rather than an adjustment instrument because it allows for a closer, continuous partnership to accompany on-going program implementation. Adjustment operations are more suitable for supporting policy reforms (such as the reforms that fostered the integration of CCT programs, as supported by the HD PSRL), but less appropriate for accompanying program strengthening and implementation on an on-going basis. The specific design o f the proposed investment operation -a programmatic SWAP that paces disbursements against the productive activities supported by the BFP with technical improvements to the program through performance incentives - would better meet the needs o fthe Brazilian Government and the BFP. Joint Package vs. Separate SWAP and TA Operations. Numerous synergies arise from combining technical and financial support in a single instrument, since (a) the technical would contribute to the on-going strengthening o f the program; (b) pacing disbursementswith progress on the technical aspects (performance-based disbursement incentives) would reinforce links between program implementation and technical improvements; and (c) the larger magnitude of financial support would help provide leverage for achieving the technical improvements supported by the project. The combined package offers multiple opportunities for impact in terms of aligning substanceand incentives. 0 Program-Based SWAP Approachvs. TraditionalInvestmentOperation. The use of a Program- Based SWAP approach instrument for the financing of the BFP is appropriate since: (a) it supports integrating the main CCT programs across the "sector" o f social assistance (comprehensive policy reforms); (b) the project will link World Bank disbursementsto a highpriority Government program (ownership); and (c) the instrument will reduce transactions costs associated with traditional investment lending. Through the Program-Based SWAP approach, the World Bank's participation with a loan size that is only a small share (less than 10%) o f the BFP over the project period would neverthelessallow for a scale o f impact on the content and performance o f the entire program. C. IMPLEMENTATION 1. Institutional and implementation arrangements The Borrower will be the Federative Republic o f Brazil. The proposed project will be implemented by the newly-created Ministry o f Social Development (MDS). MDS was created on January 23, 2004 with the 14 objective o f further integrating the management of Brazil's social programs (as part of a broader integration strategy that also covered the integration of conditional cash transfer programs into the BFP). The MDS integrated three previous entities, including the Extraordinary Ministry for Food Security (MESA), the Ministry of Social Assistance (MAS), and the former Executive Secretariat for the Bolsa Familia Program (which was previously under the Presidency). The MDSstructure includes inter alia an Executive Secretariat(SECEX) underthe Minister,which oversees two special departments, a Directorate o f Special Projects (DPE) and a Secretariat for Planning, Budget and Procurement (SPOA), and which supervises the activities of five secretariats, including the Secretaria Nacional de Renda de Cidadania (SENARC); a secretariat for other social assistance programs (e.g., BPC- LOAS, PETI, Agente Jovem); a secretariat for food security; a secretariat for articula@o (policy); and a secretariat for monitoring, evaluation and capacity building. The SENARC oversees the operations of the Bolsa Familia Program. The BFP i s also guided by the decisions o f the President, supported by the recommendations of management board (Comitf? Gestor do Programa Bolsa Familia), which is responsible for policy coordination and promoting partnerships between the various sectors and levels of Government and with civil society. Institutional arrangements for various functions of the program also involve a number of other actors (see Annex 1). To promote a mainstreaming of project activities and to contribute value added to the overall BFP, the proposed project will build on the organizational and leadership structures of MDS. Specifically, it will be headed by the Executive Secretary (SECEX) and managed by a Core Management Committee (CMC, Nucleo de Gestrib), which will be comprised ofthe managersfrom the relevant unitsof MDS: SECEX, DPE, SENARC, the Secretariat of Monitoring and Evaluation, and the head o f the PIU (see below). Functionally, this Core Management Committee will be divided into two areas: a Technical Secretariat and a Project ImplementationUnit(PIU). See Annex 6 for details. 2. Monitoring and evaluation of outcomedresults The Ministry of Social Development is committed to monitoring and evaluating results of the BFP. From the outset of the program, the Government has prioritized the development o f a solid M&E system for the program. Component 3 will support this priority (see above and inAnnex 4). The proposed Integrated HD TAL will also support complementary efforts, such as the implementation o f the household budget survey (POF) in2005, which will also provide relevant data for monitoring and evaluatingthe BFP. Project results and outcomes will be monitored via on-going project supervision using a combination of household survey and administrative data (see Annex 3). The pacing o f disbursements under Component 1 with technical improvements (milestones) also provides incentives for monitoring intermediate results during project implementation. 3. Sustainability Three factors shape the sustainability o f the BFP: Government commitment; the credibility and quality of the program; and fiscal sustainability. First, as discussed above, the BFP is Brazil's flagship social program and it plays a central role in the Government's development strategy. Overall commitment to the BFP is extremely high, both in terms of support from the highest levels o f government and in terms o f broad support in the different branches of government (as evidenced, for example, by the rapid passage o fthe law establishing the program). 15 Second, while rapid expansion is aimed at fulfilling the high expectations attached to the program and the need for quick results, investments in the quality of the program are also crucial for ensuring the sustainability of the program. A solid system for accurately identifying the target population is the foundation of the program's credibility. The proposed project seeks to strengthen this system (the Cadastro Unico) under Component 2. Institutional strengthening - including a clarification and strengthening of the institutional roles for core program functions (such as the system for identifying the target population and the verification of conditionalities) and a solid communications strategy - is also central to improving and maintaining the program's credibility. Component 4 will support efforts to strengthen these functions. Quality control processes are also crucial for building credibility and providing program managers with neededfeedback to identify and correct problems as they arise. The development of such processes will be supported by Component 3. A solid impact evaluation can also promote sustainability of the program by demonstrating the program's effectiveness and results. This has been the case for conditional cash transfer programs in other countries where a solid demonstration and quantification o f positive outcomes has resulted in the resilience o f programs across government administrations. Component 3 will support the design and implementation of a baseline impact evaluation survey, which would then be followed up by an evaluative survey inthe proposed secondAPL project. Third, with respect to fiscal sustainability, the program is expected to increase in size and resource needs. As shown in Table 5 below, the program is expected to reach 11.2 million families by the end of the project period (2006), up from 3.6 million families at the start of 2004. A significant share of these beneficiaries will be converted from existing pre-reform programs, so the marginal increase is smaller than if the beneficiaries were entirely new. Nonetheless, the programwill require a significant but managed increase in resources, from R$4.3 billion in 2003 (even less for the pre-reform programs in 2002) to R$9.4 billion by 2006. Interms of shares o f total federal primary expenditures and GDP, the program will rise from 1.1% of primary expenditures in 2003 to 2.5% by 2006 and from 0.2% of GDP in2003 to 0.5% by 2006. This rise in spending needs will likely be offset by expected savings from the recently-approved pension reforms, estimated at 0.4% o f GDP (or about R$8 billion) per year duringthat period. The shift in resources from regressive pensions to progressive conditional cash transfers under the BFP is an important part of the Government's strategy to reduce poverty and inequality inBrazil. Efficiency gains will arise from merging the programs, and will also partially offset these higher costs (thus allowing for greater coverage). The program is also seeking better complementarityand a pooling o f resourceswith state and local programs for expandedcoverage to further expand the pool of available resources. 2002 2003 2004 2005 2006 Beneficiaries: Million families covered Bolsa Familia ma. 3.6 6.5 8.7 11.2 Budgets: Billion RS; Pre-reform BolsaFamiliab R$2.4a R$4.3" R$5.85 R$7.50 R$9.39 Budgets: Billion 17% Pre-reform BolsaFamiliab US$0.8" US$1.SC US$2.0 US$2.6 US$3.2 16 4. Critical risks and possible controversial aspects A reviewo f critical risks and mitigation measures is listed inTable 6 below. The BFP is an important part of Brazil's development strategy, one that has the potential to reduce poverty and inequality today (through the provision of redistributivetransfers) and tomorrow (by tyingthese transfers to incentives for investments inhuman capital). Nonetheless, the potential ofthe BFP needs to be placed inthe context of Brazil's well- rounded strategy for development, which promotes simultaneous actions on both the economic and social fronts, including crucial investments in the supply of quality education and health services as well as other empowerment (emuncipuqiib) efforts and social programs, which all complement the BFP. Table 6 -Crii Risks RiskMitigationMeasures RiskRatingwith Mitigation To PI Possibledelays or problems inadaptingto anew ministerialstructureand institutional arrangements resultingfrom the integrationofmultiple programs. Associatedriskscould involve: (a) the risk of failing to meet high expectationsdue to possibleinitial implementationproblemsas institutional arrangements are sortedout; and(b) the possibility of vague definition of inter-institutionalroles amongthe various agenciesinvolved. Internationalexperiencewith transfer programs Risk mitigationmeasures include: (a) strengthening S untilCadastro suggests that controlsystems shouldbe inplace to the system for identifying the target populationand improvements ensuretransparencyandobjectivity, particularly selectingbeneficiaries; (b) publicizing beneficiarylists implemented duringelectionperiods. Giventhat municipal on the internetso as to promote socialcontrol (as is elections are to be heldinlate 2004, the Government currently beinglaunched); (c) regularre-certification is takingmeasuresto minimizethe potentialfor ofhouseholdsfor eligibility purposes(updatingthe political manipulationat the municipal level in Cadastrodata); and (d) quality control reviews for the Cadasteringandbeneficiaryselectionas the program BFPandthe Cadastro. Components2-4 will support expands. these measuresto help improvethe quality and credibility ofthe program. Risk of budgetcuts for the BFP. Earmarkingof governmentresourceswill help M mitigatethisrisk. Disbursementmechanismfor proposedproject(reimbursingagainst expenditureson the program) will also help protectthe budget for the BFP. Low municipalcapacityto implementCadastro data All Componentsseek to strengthenmunicipal S inthe interim, collectionfor selectionof beneficiariesand/or capacity. Since there are over 5,000 municipalities, Masproject verificationof conditionalities. the projectwill likely focus first on strengtheningthe progresses capacity of some municipalitiesto ensure quality, and then focus on multiplying these results inother municipalities. Riskthat programmight fail to reach extremepoor, Designof programto prioritize extremepoor (with per M evenwith technical improvements. capitaincomes2 visits (1996) 64.2% 97.2% I n.a. n.a. %>7 visits (2000) ma. ma. 46.0% ma. Per cap. health visits n.a. n.a. 2.4 (2001) Sources: Educationdata: PNAD2001,2002. Health data: MSDATASUS; 1996figures from PesquisaNacionalsobreDemografiae Saude. As shown in Table 6 above, the aims of the conditionalities appear to be appropriate, given the apparent correlation between poverty and relatively weak outcomes of key education and health indicators (with much worse indicators for the poor). For example, children in the poorest quintile are less likely to enroll, more likely to repeat, and complete on average two years less of schooling than their counterparts in the richest quintile.Only half of all poor children under five are fully up-to-date for their vaccines. Malnutrition is also fairly high among the poor. Only a small share o f pregnant poor woman appear to be having more than two pre-natal visits. 30 HumanCapitalInvestments: Supply- and Demand-SideInteractions. The interactionbetweenthistype of targeted CCT program and the provision of more universal "supply side" services for health and education is mutually complementary: (a) on the one hand, the provision o f quality health and education services i s clearly a crucial requirement for conditionaltransfers to work; as such, efforts are being madeto simultaneously invest in such services; (b) on the other hand, even if health and education services were universally available, CCTs would be neededto help alleviate common demand-side constraints (direct and indirect costs) that prevent poor households from taking up these services. Investmentsinboththe BFP, and in the supply and quality of health and education services, are thus crucial elements of a balancedhuman capital development strategy. An inter-ministerial social committee has been established to help ensure coordination among the social sector ministries (and the Ministry o f Education, for example, has also established a special secretariat to focus on the issues of inclusion and diversity in education, including coordinating linkswith the BFP). InstitutionalArrangements. The Bolsa Familia Program is managed by SENARC in the newly-created MDS. The BFP also involves three types of inter-actions: inter-governmental, inter-sectorial, and between the Government and civil society. The roles and degree of involvement of each of these actors varies by specific aspect o fthe program (see simplifiedmatrix inTable 7). First, overall programoversight and articulationis managed at the federal level by MDS. SENARC is responsible for overseeing the operations of the program, including managing the Cadastro Unico (see below) and the payment o f benefits. It is guided by the decisions of the President, supported by the recommendations o f management board (the Comitg Gestor do Programa Bolsa Familia), which is responsible for policy articulation, promoting partnerships between levels and sectors o f government and with civil society, overseeing formal audits and social controls. Second, as discussed below, beneficiary identification is conducted via a data collection and eligibility process that is managed by the SENARCMDS (including the Cadastro Unico unit in the SENARC), and involves the municipalities (for data collection) and the Caixa Econamica Federal (CEF, to manage the national database).23See Table 7 and discussion below for more details. Third, the institutional arrangements for financial flows and payment of transfers are fairly direct and simple, building on the long experience of cash transfer programs in usingthe country's extensive banking system. As discussedinAnnex 7, each month, the M D S sends a Program FinancingProposal (PPF) to STN for approval. This PPF i s established based on the list o f eligible beneficiaries from the Cadastro U r ~ i c o . ~ ~ Once the PPF i s approved by STN, the MDS issues an ordem bancaria and instructs the STN (National Secretariat of the Treasury) to transfer funds to a specific BFP account in the Caixa (CEF). Eligible beneficiaries (on the payroll) then withdraw their monthly benefits using electronic benefit cards (EBCS).~~ The EBCs are distributed to eligible beneficiaries according to the list established by the MDS. Upon entering the program, beneficiaries must pick up these cards at the nearest CEF agency (or lottery, post office or municipality). They are notified through the municipalities and must pick up their cards and withdraw funds within a 90-day period o f time. Experience from the pre-reform programs shows that approximately 3-5% o f Bolsa Escola beneficiaries did not withdraw their benefits within the 90-day time period, mainly due to communications or transport difficulties in extremely remote rural areas. This share 23CEF is a federal savingshedit union organization. Apart from bankingservices, CEF has traditionally provided payments issuance services for federal assistance programs. CEF operates its open 2,047 agencies nationwide, and is linked with 8,900 lottery points and 2,000 banking correspondents. This broad network is supposed to guarantee its presence in all Brazilian municipalities, one of the reasons for its central role in managingthe Cadastro Unico databaseandpayments issuancefor social programs. 24As discussed in this Annex, some of these beneficiaries include those currently receiving benefit amounts from the pre-reform programs (Bolsa Escola, Bolsa Alimentaqb, Cartb AlimentaqZLo, and Auxilio Gas) but who are gradually being migrated to the (higher) unified benefit schedule under the BFP. They are all being validated as eligible for the BFP and are subject to the same set of conditionalities as the families alreadyon the BFP. 25Beneficiariesare requiredto havetheir own bank accounts. 31 was reportedly much higher under the Bolsa Alimentaqilo Program (up to one third o f beneficiaries that did not withdraw their benefits, according to an operational audit by the Tribunal de Contas da Unico, TCU). The Operations Director in SENARC in MDS monitors the balance o fthe BFP account inthe Caixa (CEF). Undisbursedamounts arising from the expiration o f benefits that are not claimed within the 90-day period are returnedfrom the Caixa (CEF) to STN and credited to the MDSfor use by the BFP. Fourth, MDS is responsible for verification of health and education conditionalities, which is still definingmechanisms to be used for this purpose. These mechanisms will likely involve both inter-sectoral and inter-governmental actors, for example with municipalities gathering and forwarding compliance records to the Ministries of Health and Education (which could help foster a link to the "supply side" for these services) and to M D S . The development and strengthening o f such a system for monitoring conditionalities would be supported by Component 4 of the proposed project. MDS will also verify compliance via annual Quality Control Reviews (with random samples of municipalities and beneficiary families), as discussedfor Component 3 o fthe proposed project inAnnex 4. Fifth, the institutional arrangements for monitoringand evaluatingprogram performance and impact are still being configured (and this will be supported by Component 3 o f the proposed project, as discussed in Annex 4). MDS has the responsibility for monitoring and evaluation at the federal level, and developing such a system is one of the top priorities of the program. Municipalities will also be responsible for evaluating the performance and the impact of the program in their ownjurisdictions. Formal audits of the program will be a federal responsibility, but civil society will also be given an important role in promoting social controls. Further information i s included inAnnex 4. Inter-GovernmentalPartnershipsas a Strategy for Expansion and Vertical Integration. One of the recent innovations o f the BFP has been the emphasis on rationalizing the safety net across levels of government. Bolsa Familia has sought to forge partnerships with states and municipalities that also operate their own cashtransfer programs (local Bolsa-typeprograms) or are willing to addtheir ownresourcesto the program in order to further expand benefits or coverage. For example, the three state governments are currently already co-financing the program (Rio Grande do Sul, Roraima, and Silo Paulo), six others are in advanced negotiations with the federal Government for similar co-financing arrangements, and one state is completely merging its own program into Bolsa Familia. Numerous other states are also negotiating the provision of complementary services to Bolsa Familia beneficiaries with the federal Government (see below). Similar types of partnerships are being forged with multiple municipalities. These vertical partnerships constitute an innovative way to allow for a further integration o f the safety net, and an expansion or scaling up o f the BFP beyondthe limits o f federal fiscal resources. They are further discussed inAnnex4. Beneficiary Identification, Selection, and Expansion. As discussed above and in Annex 4, the BFP makes use of the Cadastro Unico - a national registry of potentially poor households -for beneficiary selection. With respect to institutional responsibilities for the Cadastro Unico, data collection and beneficiary registry are decentralized to the municipalities, but operation and maintenance of the database are centralized at the federal level, with oversight provided by the Ministry o f Social Development (the Cadastro Unico unit within MDS), and system management and operation conducted by the Caixa EconBmica Federal (CEF). 32 r e e e e m m . . e e e . e U P . . . . e e . e e As discussed for Component 2 in Annex 4, eligibility criteria for the BFP are currently based on self- reported income. International experience suggests that the use of unverified incomes for identifying and selecting beneficiaries i s generally problematic due to measurement errors and incentives for under- reporting and manipulation. The experience with Brazil's Cadastro Unico i s no exception. The income questions inthe current questionnaire were poorly defined (reference periods, coverage of types of income), and, without any form o f verification, there is simply too much room for error and under-reporting. Given these problems, a broad consensus has developed to shift to a proxy-means tested system. However, the Bolsa Familia Law specifies target groups interms of income cutoffs (for eligibility criteria). Nonetheless, it does not specify how income is to be measured. Since self-reported income is unreliable, welfare (and eligibility) could be measuredusing: (a) predicted income (predicted basedon proxy variables); or (b) a two- step process whereby self-reported income is the first "filter," and then it is "verified" or "validated" using proxy indicators (as discussed inAnnex 4). Inthe first phase ofthe Bolsa Familia reforms (end 2003-2004), most existingbeneficiaries were converted from the pre-reform programs (BE, BA, PCA, some of which also have AG, see Table 4) into the BFP. The goal of the reforms was to minimize losses to existing beneficiary households. This process also involved cleaning and validating the existing Cadastro Unico database to detect and correct for any obvious errors (missing information, duplications, etc.). Since the coverage o f families mainly consisted of conversion families, there was no massive expansion in terms of numberso f new families receiving benefits (other than the planned expansion for some 900,000 new families in capital cities in 2004) in this phase. The second phase (2005-06), however, will involve significant expansion o f the BFP (see Table 3). This expansion - along with the need to re-certify26existing households inthe Cadastro (since by then their information will have become out o f date) - will imply a new wave of widespread data collection for the Cadastro Unico. Actions to strengthen data collection efforts for the Cadastro Unico and eligibility criteria for the BFP will be supported by the proposed project and are described in Annex 4 (Component 2). Empowerment("Emancipa@o")Efforts:LinkingBFBeneficiarieswith Complementary Services. An important policy emphasis for the BFP is to foster bridges between Bolsa Familia beneficiaries and other complementary services so as to help promote their empowerment ("emuncipu@o") so as to help them escape from poverty (portus de suidu). Indeed, the BFP has opted not to establish rigid time limits for beneficiaries, but rather to help reduce their dependency on social assistance by fostering their development through these links. Such bridges could include (a) links to education and health services, including culturally-appropriate services (the additional demand generatedby these conditionalities could also serve to stimulate supply o f these services); and (b) connections to other complementary services (such as job placement or training services, micro-credit, emergency assistance, etc.). While a strategy for buildingthese links is still being developed (and indeed would be supported by the proposed project, as discussed for Component 4 in Annex 4), some states and municipalities are already linkingBolsa Familia beneficiaries to complementary services offered at these sub-national levels. 26The BolsaFamilia law does not specifythe periodicity requiredfor re-certification. Discussionswith programofficials suggest that apolicy for re- certificationwill beestablished, andthe periodicitywould likely be 1-2years. 34 Annex 2: Major RelatedProjectsFinancedby the Bankand/or other Agencies BRAZIL: BRBolsaFamilia Latest Supervision Sector Issue Project (PSR) Ratings (Bank-financedprojects only) ~~ Implementation Development Bank-financed A. PROJECT UNDER Progress (IP) Objective SUPERVISION (DO) Iealth Sector Family HealthProject S S (Loan 7105-BR) Iealth Sector DiseaseSurveillance Control- VIGISUS (Loan 4394-BR) S S B. PROJECTCOMPLETED First Programmatic Human IumanDevelopment Development Sector Reform Loan S S (Loan 7159-BR) locialProtection Social Protection Special Sector Adjustment Loan (Loan 4430-BR) HS HS C. IBRD PLANNEDPROJECTS Iuman Development HumanDevelopment Technical Assistance Loan (P082523) IumanDevelopment HumanDevelopment Programmatic Sector Reform Loan I1(TBD) Iealth Sector Health Surveillance Modernization Project -VIGISUS I1(P083013) IumanDevelopment Programa Sectorial de Promocibn del Closed Capital Humano (BR-0360) Closed ocial Protection Programa de Reforma y Proteccibn Social (1174/OC-BR) 35 Annex 3: ResultsFrameworkandMonitoring BRAZIL: BRBolsaFamilia Table 1-RESULTSFRAMEWORK PDO OutcomeIndicators Use of OutcomeInformation LONG-TERM IOGRAMDEVELOPMENT OBJECTIVESFORBOLSA FAMILIA HASEAPL (2004-2008) Strengthenthe BFP's > - At least3/4 of extremepoor families receivingBF income YR 0 (pre-refodpre-program): ability to reduce transfers latafrom POF '03 createsbaseline povertyandinequality 9 At least90% ofprimary-ageschool-agechildreninextremely YR 2-3: follow-up data collection by providingpoor poor beneficiary families attending school uillallow to monitorobjectives families with income > At least97% of beneficiarychildrenwith healthcards YR 4-5: adjustthe programbased and incentivesto invest > At least% of childrenaged0-12monthshaveat leastthree mfeedbackfromyears2-3 inthe humancapitalof weightingsmarkedon their healthcards YR 6: follow-up datacollection their children. > At least% ofpregnantwomen makingahealthvisit beforethe end w i l l allow further monitoringof o f the fourth monthof pregnancy. Jbjectives 1LOJECTDEVELOPMENT OBJECTIVESFORFIRSTPHASEOF APL 1004-2006) Strengthen > At least 2/3 ofextremepoor families receivingBF transfers YR 1: Gauge degree of: effectivenessofthe > At least40% oftotal transfers goingto bottomquintile :onsolidation of programs, safety net by: > 9 At least80%ofprimary-ageschool-agechildreninextremely reductionsin gaps andduplications Consolidating poor beneficiary familiesattending school incoverage, improvementsin > CCT programs > At least 95% of beneficiarychildrenwith healthcards zfficiency, the system for Reducinggaps identifyingthe targetpopulation, > andduplications anddevelopmentofM&E system Improvingsystem YR 1:Determine if strategyfor for identifying establishingbasic architectureof > target population BF programneedsto be adjusted DevelopingM&E YR 2: Adjust programarchitecture as needed IntermediateResults ResultsIndicatorsfor EachComponent Use of ResultsMonitoring ComponentOne: ComponentOne: Conditionalcash 9 At least2/3 ofextremepoorfamiliesreceivingBFtransfers Insufficient consolidationmay flag transfers consolidated 9 Pre-reformprogramseliminated(former beneficiarieseither lackof funding or lack ofpolitical > convertedto BFPor droppedfrom programs) will to phaseout pre-reform Transactions costs for transfer paymentsreduced programsfor those familiesthat will not be convertedto BFP ComponentTwo : ComponentTwo: System for Identifying New Cadastro Unico questionnairesdevelopedandprinted Inaccuraciesinthe identification TargetPopulation >> Expansionandrecertificationdata collectionstrategy designedand of the target populationmay flag Strengthened operational inadequatedatacollectjon 9 Eliminationofall detectableduplicationsintheCadastroUnico strategies inCadastroUnicoandor database impreciseapplicationofeligibility 9 Periodicityfor updatingCadastroUnicoandrecertifyingBolsa criteria > Familiabeneficiariesformally established Cadastro Unico users (MDS and at least 50% ofmunicipalities) with accessto data ComponentThree: Component Three: ComponentThree: Architecturefor > Adequate MISfor BFP developedand operating Lack ofprogressmay flag lack of monitoringand k Quality ControlReviewsofBFprocessesimplemented appropriate intra-andinter- evaluationsystem > Baselinebeneficiarysurvey conducted institutional structures. Will established inform on-goingrefinementof M&E strateev. ComponentFour: Component Four: ComponentFour: Strengthenbasic P Matrix ofkeyresponsibilitiesfor BFPpublished Clarification ofkey functionaland institutional > Strategy for operatingprograminremoteareas developed institutional roleswill help inform functioningof Bolsa 9 Beneficiaryinformationbooklets(cadernetas)designed, published the strategiesfor beneficiary FamiliaProgram and distributed selectionandM&E. 9 Systemfor monitoringhumancapitalconditionalitiesoperating > Implementationtraining providedto at least 1/2 of municipalities 36 e .. e . . L, m r- e . . . . 4 . . . e . 0 . Arrangementsfor resultsmonitoring Overall Outcome Indicators. Overall outcome indicators will primarily be monitored using a set of household surveys: 0 Household Budget Surveys (POF). This survey instrument collects comprehensive data on welfare measures (consumption and income, which could be used for estimating poverty and inequality), subjective measures of poverty, and anthropometric measures (for malnutrition). It also is the only survey that includes quantitative' questions regarding a comprehensive list of federal social programs (pensions, unemployment insurance, and, importantly, conditional cash transfer programs). The survey i s nationally-representative(and representative at some sub-national levels) with a sample of about 66,000 households. The POF was carried out by IBGE in 2002-03, and included quantitative questions on the coverage and receipt o f benefits from the pre-reform programs (Bolsa Escola, Bolsa Alimentaqlo and Auxilio Gas). This provides an excellent opportunity for establishing baseline indicators for the Bolsa Familia Program, such as coverage, gaps, duplications, and distributional incidence. The data are being processed and should be available for analysis by end May 2004. With the support of the proposed Integrated HD Technical Assistance Loan (TAL, under preparation), the next POF would be carried out in 2005. This would allow for the evaluation o f overall changes in living conditions (poverty, inequality, and malnutrition), as well as many o f the outcome indicators for the BFP. 0 Beneficiary Surveys. The beneficiary surveys to be conducted under the proposed APL program (Component 3) would provide further data on outcome indicators (many outcome indicators, including those for education and health). The baseline information would be collected under the first project (in 2005) and a follow up survey would be conducted under the second operation (in 2007) to allow time to gauge impact. See Annex 4 for more information. Component 1. The results indicators for Component 1 (see Table 1) would be monitored as part of on- going supervision, relying primarily on administrative data from the program. Data on coverage of the program are readily available from payments information. The pre-reform programs are gradually being phased out as beneficiaries are "migrated" to the BFP (or dropped if deemed ineligible). This process is expected to be completed by 2006. With respect to savings on administrative costs due to the consolidation o f four programs into one, data will come from administrative information from MDS and Caixa (available information suggests that making a single payment instead of four separate payments will result in about a significant reduction intransactions costs). Technical Components 2-4. The results indicators for these components are all based on technical improvements. These would be monitored as part o f on-going supervision, relying primarily on administrative data from the program. 38 Annex 4: DetailedProject Description BRAZIL: BRBolsa Familia This annex provides details on eachofthe five components: (a) conditional cashtransfers (Component 1); (b) strengthening the system for identifying the target population (Component 2); (c) developing a monitoring and evaluation system (Component 3); (d) institutional strengthening for improved service delivery(Component 4); and (e) project management(Component 5). Component 1- Conditional Cash Transfers (US$6,174 in project costs; US551.5 million inloan financing) This component would provide support to the Government in consolidating and rationalizing conditional cash transfers, and in reducing duplications and gaps in coverage o f the BFP, as part of a broader and balancedstrategy for economic and human capital development. For this component, the World Bank would reimburse the Treasury a share o f the expenseson conditional cash transfers made under the BFP during the project period. The category o f financing would be conditional cash transfers (BF Grant Transfers). As in any investment loan, the World Bank would disburse against verification o f project expenditures and conformity with the loan conditions. (See Annex 7 for details on financial management and disbursement arrangements). The proposed project would also build in certain mechanisms to pace disbursements in-line with performance of the technical aspects. While the total amount o f financing for Component 1 (US$55 1.5 million) would represent 8.9% of the total cash transfers budget for the BFP, disbursements and cost-sharing percentageswould be linkedto key milestones of technical improvements. Specifically, loan proceeds for this component would be made with increasing disbursement percentages that will increase as the actions listed in Table 2 in Annex 3 are met. The disbursements will start at the rate of 8%, will be increased to 9% when the stage 2 milestones in Table 2 in Annex 3 have been met; and will be increasedto 11% when the stage 3 milestones inTable 2 inAnnex 3 have been met. The pacing of disbursements for Component 1with technical improvements to the program (supported by Components 2-4) has several objectives. First, it would allow the World Bank to partnerwith the BFP in a pragmatic way, recognizing the important role of the program in the short-run and also the need for technical improvements inthe program's systems. Second, it would promote stronger synergies between support of program implementation and on-going safety net consolidation(Component 1) and the specific support to technical activities to strengthen the basic architecture o fthe program (Technical Components). Third, it would create incentives for the program implementing agency to achieve, or exceed, the key milestones for systems improvements and for the Finance and Planning ministries to ensure support for these crucial technical activities (e.g., if milestones are reached, the loan's financing percentage would increase and disbursementswould accelerate). The main indicators for monitoring the results of this component include: (a) the share o f extreme poor families receivingBFP transfers; (b) the gradual phasing out o f pre-reform programs (such that the former beneficiaries are either converted to the BFP or dropped from the pre-reform programs); and (c)the reduction in administrative (including transactions) costs for transfer payments (due to the merging o f four separate programs into a single integrated program). 39 Component2 - Strengtheningthe system for identifyingthe target population2' (US4.4 million in project costs) OBJECTIVESOF COMPONENT This component seeks to improve the mechanisms used to identify the target population of the Bolsa Familia program to guarantee its republican, federative, and democratic characteristics. By doing so, the program will achieve better redistributive efficiency, with a higher share o f its resources going directly to the poor, and will be able to cover a larger share of the low-income population. RESULTS FRAMEWORK KEY MILESTONESCOMPONENT AND FOR The main indicators for gauging the outcomes ofthis component include: (a) the percent oftotal transfers going to the poor; and (b) the share of the poor covered by the program. These outcomes could be monitoredthrough the collection of household survey data. Infact, baseline data for these outcomes were collected in 2002-03 for the pre-reform programs under the Household Budget Survey (POF) and will soon be available for analysis. While the component will support numerous specific activities to help overhaul the Cadastro Unico system used by the BFP to target benefits to the poor, a number of key milestones (outputs) will be monitoredto gaugethe progress towards meeting these goals, notably: 0 By December 2004: (a) new Cadastro Unico questionnaires designed; (b) instruments for implementing eligibility criteria revised; and (c) normative acts and operational guidelines for the Cadastro Unico publishedon MDSwebpage. 0 By December 2005: (a) data collection strategy designed and piloted; (b) all detectable duplications in the Cadastro Unico database eliminated; and (c) Cadastro Unico users (MDS and 30% o f municipalities) with access to Cadastro Unico data. THECADASTRO UNICO The accurate identification o f poor households is an important input into the success o fthe BolsaFamilia Program as a tool.for poverty reduction. The target population for the Bolsa Familia program is poor families (with per capita incomes less than R$100 per month), with priority given to the extreme poor families (with per capita incomes less than R$50 per month). Ensuring that the transfers - and the incentives for human capital investments - actually reach these families i s crucial to the program's success. As with its predecessors,the BFP selects beneficiaries from a national database of poor households called the "Cadastro Unico." The Cadastro Unico was created in 2001 as an initiative to construct a single beneficiary registry database to serve its many means-tested conditional cash transfer programs. Until that time, each program operated its own system for identifying and selecting beneficiaries (all via unverified means testing) and some o f these databases are still not fully compatible with the Cadastro Unico (e.g., the Bolsa Escola registry, which still includes 3.6 million families). The creation of a unique database (the Cadastro Unico) and social identification number (NIS) responded to the need to improve efficiency and coordination, and reduce duplication of administrative costs across the numerous safety net programs. The N I S has the same numerical structure as the PIS (identification o f workers in the formal ''Formore details, see: de la Briere, Benedicte and Kathy Lindert (2003). "Reforming Brazil's Cadastro Unico to Improvethe Targeting of the Bolsa Familia Program." World Bank and DFID. Draft ESW. This section also draws on the conclusions of two workshops held inBrasilia in November 2003 (one on household targeting systems, the other on proxy means testing), on a World Bank-BolsaFamilia workshop held in December2003, andon further dialogue and consultationsbetweenthe World Bank andthe Bolsa Familiaprogramteam. 40 private sector) and the PASEP (identification of public sector employees) and no new NIS is created if one of these two numbersalready exist for an individual.28 The main screening mechanism used with the Cadastro Unico is unverified household incomes (unverified means testing). Data collection and beneficiary registry are decentralized to the municipalities, but design, operation and maintenance of the database are centralized at the federal level, with oversight provided by the Ministry of Social Development (MDS), and system management and payments conducted by the Caixa EconGmicaFederal (CEF).29 There are several advantages to having a single national database for determining and monitoring eligibility for social transfer programs, including inter alia: (a) preventing or reducing duplication of benefits (e.g., people receiving benefits from more than one municipality or program); (b) reducing duplication o f administrative costs across programs; and (c) monitoring time in the program and graduation criteria. The Cadastro Unico representsa major step forward inadvancing these goals. Nonetheless, several aspects of the Cadastro Unico should be improved, including: (a) clarifying its objectives and institutional structure; (b) distortions due to the use of a priori registry quotas; (c) distortions due to the use of self-reported income for determining eligibility; (d) problems with the identification of households; (e) the static nature of the database; (f)the lack o f a system for quality control; (g) a lack of data access by key potential users (programs, municipalities, MDS); and (h) software and implementation problems. PRINCIPLES OF AN IMPROVED SYSTEM The introduction of the BFP offers an opportunity to revamp Brazil's Cadastro Unico system so as to improve the identification o f target populations for social transfers and the coordination between complementary programs with different levels of governments. International experience3' suggests several principles of good practice for target household identificationsystems: 0 Transparency. This principle applies to procedures for gaining entry into the registry (e.g., the Cadastro Unico) and for determiningeligibility for the program (e.g., the Bolsa Familia program). Household identification systems should also adopt internal controls (cross-checks, process quality control) and social controls to ensure transparency and consistency. 0 On-goingaccess to the CadastroUnicofor a dynamic safety net. With this principle, systems for the identification of target households should avoid the static nature o f the quota-basedsurvey system: any family should be allowed to register at any time (with the explicit understandingthat registry does not guarantee benefits). An improved system should also allow for on-going updates(demographic, location) for those who do benefit, and require regular re-certification. 0 Minimization of leakages to the non-poor. All programs entail some degree of leakages. In fact, some degree of "leakage" to the non-poor is acceptable if: (a) the leakage goes to the very near poor; and (b) the costs o f reducing this leakage (in terms o f the financial costs of improved target accuracy and the indirect costs that could result from accidentally excluding eligible poor families due to barriers to entry) outweigh the benefits of reducing it (interms o fthe savings from While these cross-checksare now routinely performed, other cross-checkswith databases such as the NIT, which registers the self-employed, are still not inplace, generatingsome duplication of identification. 29CEF is a federal savingshedit union organization. Apart from banking services, CEF has traditionally provided payments issuance services for federal assistance programs. CEF operates its open 2,047 agencies nationwide, and is linked with 8,900 lottery points and 2,000 banking correspondents. This broadnetwork is supposed to guaranteeits presenceinall Brazilian municipalities, one of the reasonsfor its central role in managingthe Cadastroh i c o database andpayments issuancefor social programs. Castailed4 Tarsicio, Kathy Lindert, Luisa Femandez, and Bknkdicte de la Brihe (2003). "Implementing Household Targeting Systems: Lessonsfrom LatinAmericaandthe United States." World Bank. Draft ESW. 41 allocating transfers to non-poor households that could instead be used to cover more poor households). Combined geographic and proxy-means tested household targeting systems in other countries, have generatedimpressive redistributive outcomes at fairly low 0 Outreach to the poor. Specific efforts should also be made to ensure that the poor know of their potential eligibility for the programand can accessthe application and interviewprocess. 0 Cost Efficiency. Efforts should be made to minimize the cost of interviewing families, while ensuring the integrity of intake efforts. Although it is not possible to ascertain the poverty status of families apriori before interviews and data collection (and indeed, that was the problem with quota-basedsurveyingunder the current system), certain tools, such as self-selection mechanisms and geographic targeting (poverty maps), can help minimize the cost of interviewing large numbers of ineligible families. National experience from some states and municipalities and the Cadastro Unico itself suggests the following additional principles for Brazil: 0 National Identification System. An inter-governmental working group is evaluating the remaining cross-checks necessary to minimize duplications and maximize the issuance of a NIS for each individual; 0 Shared Management of Social Databases Across Levels of Government. While legal norms to guarantee access for state and municipal agencies are already in place, their operationalization is still lagging. IMPROVING THE SYSTEM FORIDENTIFYING THE TARGET POPULATION OF THEBFP AND THE CADASTRO UNICO With the above principles in mind, this component seeks to improve the mechanisms usedto identify the target households of the BFP. The component will mainly focus on the strengthening o f the Cadastro Unico, Brazil's target household identification and registry system. It also makes reference to specific actions that can be taken for improving the system for identifyingthe target populationof the BFP that are (somewhat) independent of the Cadastro Unico (since the BFP is only one o f the users of the Cadastro). The discussion and ensuing matrices distinguishwhether the proposed reforms and activities apply to the Cadastro generally, or to the BFP specifically. It i s important also to note that work under another World Bank supported project (the HD Technical Assistance Loan) is supporting the construction o f micro-area poverty maps (in IBGE), which could become an important instrument for targeting in Brazil (either on their own, or as an input intothe CadastroUnico, as discussedbelow). In terms of general timing, most actions and activities to be supported by the project assume that preparatory activities would occur in 2004 and 2005 with the goal o f piloting, then rolling out the new systemas the program expands (and existingbeneficiaries are recertified) in2005-06. 2A. Improving the Strategic and InstitutionalFramework for the Cadastro Unico This sub-component seeks to clarify the strategic and institutionalframework for the Cadastro Unico. In terms of timing, these issues should be addressedby the end of 2004. "Coady, David, MargaretGrosh, andJohnHoddinott (November2002). "The TargetingofTransfersinDevelopingCountries:Reviewof Experience and Lessons." The World Bank. Social Safety Net Primer Series. Castafleda, Tarsicio, Kathy Lindert, Luisa Femandez, and Bknkdictede la Britre (2003). "Implementing Household TargetingSystems: Lessons from LatinAmerica andthe United States." World Bank. DraftESW. 42 Issues. First,the uses ofthe Cadastro should be clarified. While the Cadastro was originally designedto serve multiple conditional cash transfer programs, these have now been merged into a single federal program (Bolsa Familia). Yet there are numerous potential other uses for the Cadastro, including state and local programs (provided these governments have access to the data, as discussed below), and even other federal means-tested programs. Second, there is a need to strengthen institutional arrangements, clarifying responsibilities of the different agencies involved, including identifying: conceptualization, data collection, data management, management of the system for identifying possible target populations, systemsmanagement for registry and payments database, formal and social quality controls, maintenance, etc. Third, the current Cadastro system mixes the functions o f identifying potential target populations and registration of beneficiaries to the detrimentof the effectiveness of the system. The primary function of the Cadastro should be to collect household data on key variables for the purposes of screening these households against a predetermined set of criteria to determine eligibility. As a secondary function, a sub-set of these data could then be stored in a program-specific registry (for example for payments and conditionality monitoring) for each "user program." Remarks for an Improved System. First, a "user strategy" should be designed to identify potential users of the Cadastro. There would be many advantagesto expanding the scope of the Cadastro to serve as a broader tool for social policy at many levels of government, including the cost efficiency gains of usinga single instrument for multiplepurposesand the incentives this would provide to municipalitiesto maintain their Cadastro records up-to-date. Second, the primary objective o f the Cadastro Unico should be to serve as system for identifying the target population. This implies interviewing and registering more households than will be eligible for any or any specific program (and communicating information about role of Cadastro to avoid raising expectations will be crucial, see below). A secondary objective would be for the Cadastro to provide registries for payments and monitoring for specific programs (e.g., creating program-specific sub-databaseswith historical and current information for beneficiaries and non- beneficiaries). These sub-databases should be tailored to the needs of each "user program.'' Third, the new system should embrace clear institutional and financing roles for each aspect o f Cadastro, preferably with centralized design (with consultations), decentralized data collection (properly financed), and centralized database management. MainActivities. Inlight of the above issues and recommendations, this sub-component would support a number of activities designed to strengthen the overall strategic and institutional framework for the Cadastro Unico, including: 0 Developing strategic planning capacity and providing steering guidance for the component; and 0 Conducting a study to review possible programs by different agencies and levels of government that are means-tested and that could benefit from using the Cadastro Unico. The study should also identify possible modifications that would be neededto the Cadastro (e.g., the questionnaire or the data collection strategy) to serve these programs. In addition, the Institutional Strengthening Component (Component 4) will contemplate the following complementary activities to strengthen the Cadastro's institutional structure, including: 0 Conducting a study to review existing legislation governing the Cadastro Unico and to propose appropriate updates as necessary, clearly specifying functions, content, user programs, responsible agencies, funding, and outputs; 0 Conductingan institutional and financing assessment of the existing Cadastro system and develop a proposal for reforms and clarifications; 0 Developing normative acts (decree, reglamento) and operational guidelines regulating the Cadastro system, and spellingout general operational rules to be enacted; and 43 0 Conducting workshops / seminars for consultations on the strategic and institutional framework of the Cadastro Unico. 2B. Revising Eligibility Criteria (Bolsa Familia Program), New Questionnaires (Cadastro Unico)and Data Collection Methods This sub-component seeks to (a) strengthenthe definition and implementationofthe eligibility criteriafor the Bolsa Familia and revamp the Cadastro Unico questionnaire accordingly; and (b) strengthen the data collection strategy for the Cadastro Unico. The eligibility and questionnaire aspects should be addressed bythe end o f2004. The aim would also be to havea solid data collection strategy inplace by 2005, when most existinghouseholds in the system would need to be re-certified (since most data would be out-of- date by that time) and when the BFP would begin large-scale expansion (carrying over into 2006 and 2007). A new wave of data collection is thus likely to begin in 2005, and the new system should be in place to support that. Some piloting of new approaches could occur in 2004 (particularly in large urban areas not previously covered bythe Cadastro). Issues with Respect to Eligibility Criteria and Questionnaires. There are several weaknesses under the current program system. First, the use of unverified incomes for identifying and selecting beneficiaries i s generally problematic due to measurement errors and incentives for under-reporting and manipulation. The experience with Brazil's Cadastro Unico is no exception. The income questions inthe current questionnaire were poorly defined(reference periods, coverage o f types o f income), and, without any form of verification, there is simply too much room for error and under-reporting. Second, the current Bolsa Familia Law does not provide any guidance for explicit prioritizing among the poor (moderate poor vs. extreme poor; families with children vs. individuals or couples with no children, etc.). With fiscal limitations, clear and transparent guidelines should be established to determine the ranking of eligible families to enter the program. Third, the Cadastro Unico questionnaire and its associated manuals, while it did collect proxy information other than self-reported incomes, would benefit from substantial strengthening (concepts, definitions, structure o fthe questionnaire, manuals, etc.). Recommendationsfor Eligibility Criteria and Questionnaires. First, the BolsaFamilia Program (and the Cadastro questionnaire) should shift to a proxy-means tested system (eligibilidade baseado em uma cesta de indicadores multi-dimensionais). However, the Bolsa Familia Law specifies target groups in terms of income cutoffs I (for eligibility criteria). Proxy Means Testing Nonetheless, it does not Proxy-means tests use a relatively small number of household characteristics to calculate a specify how income is to score that indicates how well off the family is (providing an estimate of its "probability of be measured. Since self- being poor"). These characteristics should ideally be easily observed, but not easily reported income is manipulatedby families, andthey shouldgenerallybe correlatedwith overallwelfare. Typical unreliable, it could be variables used in other countries include: geographic location, demographics and household composition, housingquality, ownership of durables, education of adults, availability of public measured using: (a) infrastructure services, and employment. Locality-level poverty rates (from poverty maps) predicted income could also be used as proxies, reflecting the probability that the family is poor due to its (predicted based on proxy location in a particular geographic micro-area. These variables are then weighted using variables); or (b) a two- statistical techniques to generate an overall points score. Those families with scores below certain thresholds are then eligible for benefits. Specific variables and weights are generally step process whereby identified through analysis of household survey data and consultations with policy makers. self-reported income is Proxy means tests are generally better at accurately identifying those living in long-term, the first "filter," and then chronic poverty than those whose situations have recently worsened due to shocks (transient it is "verified" or poor). Since most variables used for proxy-means testing change more slowly than incomes, "validated" using proxy updatingandre-certificationofbeneficiariescan be less frequent (say everytwo years). indicators. These Proxy means testing is currently being used in a number of countries with relatively strong indicators should be redistributiveoutcomes (low leakagerates) and low costs. Another advantage of proxy means identified on the basis of systemsis the multi-dimensionalityof the "poverty measure." 44 empirical analysis of household surveys. They should be determined at the national level (for transparency), but should also build in variables to take into account geographic differences (e.g., urban, rural, regional). Second, Bolsa Familia should give priority to the extreme poor (before the moderately poor) andto families with children (over those without). The current Bolsa Familia Law does not address the issue of eligibility of couples or individuals without children. Bolsa Familia has been giving priority to families with children, but this policy should be documented in a normative act for transparency. Finally, the Cadastro Unico questionnaire should be revamped to improve its structure (maximum 2-3 pages), content (including proxy variables), and concepts and reference periods. IssuesRegardingData CollectionMethods. There are several problems with the current systemof data collection under the Cadastro Unico. First, the current system is static. It does not allow for on-going entry of households into the registry. Nor does it have a clear policy on the periodicity of updating and re-certification. Information for all households in the Cadastro i s rapidly becoming out-of-date. Second, the current practice of using a priori quotas to limit number o f households, that are interviewed and registered ineach municipality, results inexcluding potentially poor households before data are collected. There are several problems with quota-basedregistries, including: (a) a lack o f transparency for deciding which households get interviewed; (b) opportunities for political manipulation; (c) potential for replicating existing inequalities at local level. Third, there are many inefficiencies in data collection methods. More guidance i s needed to municipalities on the menu of possible methods to be used for collecting data (which would be tailored to local circumstances). Remarks for an Improved System of Data Collection. First, a policy should be established to clarify the periodicity of updates and re-certification. Demographic changes should be updated as they occur. Full information (re-certification) should be conducted at a minimumevery two years. Second, to avoid the static nature of the current system, the new system should allow for on-demand applications for registry in the Cadastro Unico on an on-going basis (with information provided to applicants about the fact that entry into the Cadastro does not guarantee benefits). Third, the use o f apriori quotas for the registry should be abandoned, and an information campaign should be conducted to avoid raising expectations among registrants in the Cadastro (to inform people that registry in Cadastro does not guarantee eligibility in any specific program). Fourth,data collection methods should be tailored to local realities, depending on density o f poverty, urbankural and cultural-ethnic groups (informed by poverty map). A menu of approaches should be developed, including on-demand vs. survey methods.32 Fifth, minimum quality standards should be developed to certify municipal databases, particularly in cases where municipalities already had their own registries (standards that would allow for the acceptance of these municipal registries as inputs into the Cadastro). Sixth, a data collection strategy (for the roll out beginning in 2005) should transparently determine which municipalities would be updated and/or expanded first. This plan should be objective, based as much as possible on priorities identified by poverty maps. Practically speaking, it will probably also ,reflect somewhat the pattern of partnerships @actuaGo"es)beingestablished between the federal Government and specific states and municipalities. Main Activities. In light o f the above issues and remarks, this sub-component would support a number of activities designedto (a) strengthenthe eligibility criteria for the BFP, (b) revamp the questionnaire for the CadastroUnico, and (c) strengthen the systemfor datacollection for the Cadastro Unico, including: Conducting studies using household surveys (POF, PNAD) and data in existing Cadastro to develop and test an appropriate set of Proxy Means Indicators (variable and weightsQ3; 32For more details, see: de la BriBre, Benedicte and Kathy Lindert (2003). "Reforming Brazil's Cadastro Unico to Improvethe Targeting of the BolsaFamiliaProgram." World Bank and DFID. Draft ESW. 33The goal would be to identify the variables that best predict welfare (or income) using a random sample of households (from the POF or PNAD) and using regressionmodels. Given the heterogeneityof living conditions inBrazil, these models should includeregional effects and a distinctionbetweenurban, peri-urban and rural areas as well as a distinctionfor diverseethno-/cultural-populations. 45 Developing and piloting a new Cadastro questionnaire based on above studies and taking into account the diverse geographic, cultural, ethnic and racial characteristics of the population (maximum 2-3 page q~estionnaire)~~; Reviewing strategies for collecting data in a handful o f municipalities; developing a menu of approaches (on-demand, survey approaches, with possibility o f using GPS systems as a data collection instrument) depending on poverty density, urbanization, and diverse ethnidracial populations; and classifyingmunicipalities according to this menu; Developingcriteria for minimumstandards and a processfor certifying local databases; Assessing implementation costs (municipal level costs o f interviewing, logistics for data collection), developing a strategy for providing financial incentives for data collection, piloting incentive scheme; Developing manuals and training strategy and materials for interviews (including manuals for questionnaire with definitions o f concepts and variables), supervision, administration (describing all municipal activities for planning registry work, costing activities, contracting out if needed, managing data, and quality controls and transmission of databases), data entry and validation software, including specialized training materials to take into account the needto be sensitive to the diverse cultural, ethnic and racialcharacteristics ofthe population; Printing and distributingnew questionnaires (see Part 2 - questionnaires/eligibility above) and manuals; Developingchannels and mechanisms for on-demand applications and appeals; Piloting new data collection strategy and new questionnaires in a handful o f municipalities o f differing characteristics and evaluatingpilot; Conducting training on data collection methods and questionnaires, particularly in municipalities that are poorer or have weaker administrationcapacity; and Rollingout new data collection system (with new questionnaires). The corresponding sections of the operational guidelines clarifying (a) eligibility issues for the BFP (criteria, measurement, definitions of families, exit criteria, etc.) and (b) policies for updates and re- certification; data collections strategies; process and criteria for certifying minimum standards for municipal databases, will be drafted as an activity of the Institutional Strengthening Component (Component 4). 2C. ReducingAdministrativeWeaknessesinCadastro UnicoSystem This sub-component seeks to strengthen the accuracy of the Cadastro Unico by developing systems controls and checks to reduce administrativeweaknesses present inthe current / original system. Many o f these cross-checks could already be implemented for the existingCadastro databasefor use in 2004 (some already are). They should also be built into the design of the new overall system (functions and technology). Issues. Weaknesses in identification o f individuals and a lack o f internal and cross-system controls have resultedin administrative errors, duplications and ghost registrants inthe Cadastro Unico. As part of the Bolsa Familia reforms, some o f these errors are being cleaned up. However the new system needs to be strengthened so as to detect and correct for these errors on a systematic (automated) basis. In addition, the new focus of the BFP on the family rather than the individual as the assistance unit presents the additional challenge of tracking the identities both of individuals, and o f individuals linkedto family units (to ensure that individuals don't benefit inmore than one family unit). 34A Cadastro questionnaire should be developed on the basis of information from the various programs that would use it (including but not limited to the BFP) so that it collects the variables neededfor these specific programs(which might differ ineligibility measures and criteria). 46 Remarks for an Improved System. First, the new system should adopt a single social identification number (NIS) that can be used to cross-check identities (a) within the Cadastro system at any point in time and with past data; and (b) between the Cadastro and other relevant databases. Second, the new systemcould also use "soundex number" method of checking matchesand near-matchesof a specified set of characteristics (including the NIS) within the Cadastro and between the Cadastro and other databases. Third, the new system should build in automated checks to cover key ccflags,"including: duplications, identity issues, address issues, ghost registrants (checking for example against obituary registry), missing information, and zeros (e.g., zero income). Fourth,the new system shouldbe able to link individuals and families. Fifth, the new system should find an institutional way to help people get needed documentation. Main Activities. In light of the above issues and remarks, this sub-component would support a number of activities designed to detect and correct for administrative errors in the Cadastro Unico database, including: 0 Reviewing existing database with respect to possible duplications; reviewing possible cross- checks within the Cadastro and between systemsthat could facilitate confirmation of identityand reduction of ghost registrants; and developing specified targets for duplications, missing information, people without IDnumbers, etc.; 0 Conducting cross-checks on the consistency o f the Cadastro Unico with data in the Population Census 0 Developing and piloting strategy and system for: (a) adopting the NIS and helping potential applicants obtain needed documentation and NIS numbers; (b) runningautomated within-system "soundex style" cross checks and other forms of verification; (c) running automated cross- systems checks; and 0 Developing a strategy and materials for training (videos, materials, courses, etc.) on the new identity systems and checks and conducting training of (a) local officials for verifying identity, helping individuals obtain documentation, and runningcross-checks duringregistry and interview process; (b) central officials for runningbetween-systems cross-checks on a periodic basis. 2D. ImprovingInformation Technology(IT) and the TechnicalEfficiencyof the Cadastro Unico System This component seeks to strengthen the IT capacity of the Cadastro Unico unit in MDS (and in its collaboration with the CEF) and in municipalities (for data entry, validations, transfers, management, verifications and cross-checks, data warehousing, use by user programs and formulation of demands to the databasemanager, the CEF). Issues. Numerous IT problems hamper the efficient functioning o f the Cadastro system, including: (a) problems with uploading data from municipal databases; (b) lack o f real time data access at various agencies and levels o f Government (including in MDS, municipalities, other potential users), (c) data losses intransfers; (d) lack o f internal automated systems checks; (e) lack of ability o f the Cadastro Unico database to converse with other databases for systems checks; (f) inability o f system to accept updates or maintain registrant history across updates / time; and (g) a general lack o f user-friendliness that greatly hampers the ability o f municipalities and federal programs to access and use the data for program management. While issues (a) to (f) are dealt with inthe discussion of the new management contract with CEF, the Bolsa Familia unit in MDS and some municipalities need support to increase their data user capacity for program management. Remarks for New System. Following a thorough assessment o f the hardware, software, and communications needs o f the whole Cadastro system (all involved agencies, including the federal, state, 47 and local agencies), the component would support activities to: (a) clarify and define the functional needs and demands of the users (mainly MDS and municipalities); and (b) supervise the piloting of the new Caixa system. The component would also seek to enhance IT capacity in MDS (for managing the Cadastro Unico data and for using it for social assistance programs management, including the Bolsa Familia Program), and at the state and municipal levels through training and increased hardwarehoftware capacity. Main Activities. In light of the above issues and remarks, this sub-component would support a number of activities designed to improve the IT and efficiency and capacity o f the Cadastro Unico agency users, including: 0 Conducting a needs assessment on hardware and software for improved system for the MDS (Cadastro Unico Unit),states and municipalities; 0 Supervising the pilot test of the new Cadastro Unico system in the Caixa (CEF) and contracting an outside expert to review pilot results; Purchasingnew software and hardware for MDSandmunicipalities; and 0 Conducting IT capacity building and training at federal (MDS and CEF) and sub-national levels (agencies involved in Cadastro Unico system either as producers, managers or users of the data), as part of an integral "municipal-support kit" (that would include the training, manuals, and the equipment and software). 48 . C 0 . . m . . . 3 . 5 8 . . . . 0 . . * . . . . . I e C i L e e L I I : E f C ? e e L L L L 2 C ? e i , ? 5 I! L E I e e e 0 Component 3 -Monitoringand Evaluationof the BolsaFamilia Program (US$7.0 millioninprojectcosts) OBJECTIVESOF COMPONENT This component seeks to support the Bolsa Familia Program (BFP) in designing and implementinga results- based monitoring and evaluations system aimed at: (a) monitoring the effectiveness of program delivery, (b) monitoringprocessesand instituting quality control reviews aimed at promoting public accountability and legitimacy, and (c) monitoring outcomes and assessing the program impacts at both the territorial and household levels. RESULTSFRAMEWORKKEYMILESTONESCOMPONENT AND FOR While the component will support numerous specific activities to help develop a results-basedM&Esystem, a number of key milestones (key results) will be monitored to gauge progress towards meeting these goals, including: 0 By December 2004: (a) terms of reference for design of the management information system (MIS) prepared; (b) inventory o f existing inter-governmental partnerships for monitoring and evaluation completed; (c) terms o f reference for designing and conducting Quality Control Reviews developed; and (d) terms of reference for baseline beneficiary survey prepared. 0 By December 2005: (a) adequate MIS developed and operational; (b) Quality Control Review for core BF processes developed and piloted in at least 10 municipalities; and (c) baseline of survey deignedand piloted inat least 10municipalities. DEVELOPINGA MONITORINGEVALUATION AND SYSTEMFORTHEBFP The monitoring and evaluation system proposed by the BFP includes four main sub-components: (3.A) Management Information System to improve effectiveness o f program delivery; (3 .B)Processes monitoring and quality control to ensure accountability and legitimacy; (3.C) Monitoring of outcomes and impact assessment at the municipal and household levels, and (3.D) Design and oversight of overall M&E system by external advisors. 3.A. ManagementInformation System The objective o f this sub-component is to assistthe BF team in developing a MIS to gather and process real time information that is crucial for project management and implementation supervision. Once inplace, the system will be able to instantaneously supply information about various implementation indicators to the project managementteam and policy makers alike. A list of such indicators i s found in Table 2 below. The MISdeveloped under this sub-component will gather and process information heldby Caixa Economica Federal (CEF), municipal governments and the education and health ministries. Regular newsletters and reports with detailed information on program delivery will be prepared and sent to key policy makers at federal, state and municipal levels as well as representatives o f indigenous peoples and Quilombola communities. A web site will also be developed to provide real time information on program delivery and outreach status. This sub-component would support the following activities for the setting up of an MIS for the BFP: (a) the design, training o f staff and implementation o f the M I S system; (b) MIS software and hardware equipment; (c) system maintenance and updating; (d) operational costs for project period; (e) training and dissemination workshops for MIS users (including federal, state andmunicipal government staff and policy makers). 52 3.B. Monitoring of Processesand Quality Control This sub-component seeks to strengthen the efficacy and to promote transparency, accountability, and legitimacy of the BFP with activities in three main areas: (a) monitoring of inter- and intra-governmental processes; (b) quality control reviews to monitor core BFP processes at a downstream level (local, beneficiaries) to ensure accountability, legitimacy and transparency of program execution by establishing effective incentive mechanisms for decentralized implementation, and (c) the promotion of citizen oversight ("social control") for the implementationofthe BFP by local stakeholders. Monitoring of Inter- and Intra-Governmental Processes. For this area, the project will support a series of activities aimed at: (a) creating a continuously updateable inventory of programs implemented by states and municipal governments, as well as by civil society, (b) assembling an inventory o f current legal agreements (puctuaq6es)establishing partnerships between state, municipal and federal governments for decentralized program implementation, (c) compiling existing local partnerships (puctuaq6es)between civil society and local governments for social program implementation, and (d) mapping inter-dependencies between line ministries at the federal Government level which may either hinder or facilitate program implementation. Specific activities for this area include: (i) designing and carrying out the inventory o f existingprograms and partnerships (puctuaqo"es),as well as maintaining the inventories updated; (ii) workshops to disseminate the results ofthe inventory; and (iii) studies insix differentmunicipalities to validate inventory. case Quality Control Reviews for Monitoring of Core BFP Processes. For this area, the project will support the design and execution of random annual Quality Control (QC) Reviews at the municipal and beneficiary level to verify: (a) beneficiary identification procedures and eligibility35;(b) compliance by BF beneficiaries with human-capital conditionalities and the system for verifying and reporting on this; (c) receipt of payments by BF beneficiaries; and (d) beneficiary satisfaction and feedback on the program. The QC Reviews will also seek to establish how quickly eligibility information gets out o f date due to changes in economic and demographic circumstances (which could inform policies on updating and recertification). The QC Review process will be defined during project implementation, but the current scheme envisages that a random sample of numerous municipalities would be drawn each year. From each municipality, a further random sample of beneficiary families will be selected. Specific activities for these QC Reviews include: (a) designing the QC Review instruments, sampling strategy, and implementing periodic field surveys in 50 municipalities per year; (b) conducting workshops and disseminating publications with results of the QC Reviews; and (c) developing program re-design strategies to address implementation issues identifiedby the QC Reviews. The terms o f reference for these Quality Control Reviews would be designed in2004 andannual reviewswould be carriedout in2005 and2006. Citizen Oversight (Social Control). Citizen oversight can also serve as an important mechanism for monitoring the BFP. The Government has already taken an important step to support citizen oversight by publishingthe names o f beneficiaries in a public manner. In addition, this component will support the following activities designed to strengthen citizen oversight: (a) a mapping and typology of existing manifestations o f social control mechanisms in a sample o f municipalities (in a diversity o f cultural, ethnic and racial settings); (b) workshops to disseminate the results of this social control mapping; and (c) conducting in depth case studies to describe and analyze manifestations of social control. In addition, citizen oversight (social control) will be supported by the other components and sub-components via the publication of policies, procedures, study/survey results, the development o f an information hotline to respond to questions about the BFP, etc. 35These controls would seek to verify information collectedin the Cadastrowith re-interviewsand verification of information usingboth the current and the original referenceperiodsof the original interview. This would allow for officials to gauge the accuracy of information collected at the time of the original interview, and to determine how quickly such information becomes outdated (e.g., changes in household composition and socio- economic circumstances), which would provide feedback as to the quality of information inthe Cadastro(and identify sources of errors) and as to the frequencywith which such information needs to be updated(re-certified). 53 3.C. Monitoringof Outcomesand ImpactAssessment This sub-component will support the monitoring of outcomes and assessment o f program impact at two levels: (a) territorial-level outcomes, and (b) family-level outcomes. These two levels of analysis are mutually reinforcing, since, for example, the analysis of territorial-level outcomes can provide important contextual information for the analysis of family-level outcomes, and vice versa. A combination of complementary qualitative and quantitative methods will be usedto monitor and evaluate outcomes for both levels of analysis. Territorial-Level Outcomes. The objectives of territorial-level analysis are (a) to examine the potential impact of the BFP at a broader socio-economic level; (b) to assess the functioning and interactions of the BFP with other complementary programs and services; and (c) to provide broader contextual information that could also be useful in assessing impact at the household level. The unit o f analysis would be a territorial reference base (e.g., sub-municipality, municipality, with a diverse representation of cultural and ethnic populations, and/or regional area). The indicators to be monitored include a variety of economic indicators, social indicators, the supply of public services and complementary programs, and social capital and other social themes (see Table 2 below). These indicators will be monitored and analyzed using spatial analysis techniques. Data sources will include existing data, imputations between existingdata sources, and the collection of new data. First, the monitoring and evaluation of outcomes at the territorial level will use existing and newly collected data at the municipal level. Existing data sources include the national population census, the national educational census carried periodically by the Ministry of Education, and administrative data from the Ministry of Health and other line ministries. Second, imputation methods (such as those developed by Lanjouw et. ai. inthe World Bank for poverty mapping) can be usedto further expand the spatial database o f indicators at the municipal level by combining census and sample data from large national surveys (e.g., PNAD, POF, etc.). Third, In addition to supporting the conformation o f a municipal level data base from existingdata sources, this component will also support data gathering o f some key municipal level indicators that may not be available elsewhere. Finally, these quantitative data sources will be complemented by the use of focus groups (e.g., o f beneficiaries and non-beneficiaries, local program managers, etc.) and other qualitative methods to monitor territorial impacts on social aspects, such as citizenship (cidadania). Family-Level Outcomes. The objectives of family-level analysis will be to: (a) monitor the living conditions of BFP beneficiaries; and (b) evaluate the impact of the BFP. The indicators to be monitored include the economic and social conditions o f the family, the family's access to other public services and programs, and indicators to gauge the beneficiaries' knowledge of and satisfaction with the BFP (beneficiary satisfaction survey), as shown in Table 2. A combination of qualitative and quantitative methods will be usedfor the analysis of household-level outcomes. First, several participatory qualitative instruments will be used to monitor and evaluate the impact of the BFP at the family level. Anthropological and ethnological case studies will be conducted. These will be complemented by the use o f focus groups, life histories (Zinhas de vida) and other methods, which can greatly enhance our understanding of the impact of the program on key social indicators and the interactions that the program has with other aspects o f living conditions (including citizen participation, complementary services, etc.) especially for traditionally disadvantaged groups. Second, this sub-component will support the collection of baseline survey data from a sample of program benefi~iaries.~~ This will facilitate the monitoringof changes inliving conditions for these BFP beneficiaries 36As explained in Annex 3, another opportunity for analyzingsome of the neededquantitative outcome indicatorscomes from the HouseholdBudget Surveys (POF), These surveys collect comprehensive information on welfare measures (income, consumption) and anthropometric data (for malnutrition), as well as quantitative information on key social programs, including conditional cash transfers. The first nationally-representative 54 (see Table 2 for example indicators). While the design of the survey will be defined during project implementation, the current concept is that program participants (from diverse geographic, cultural, racial and ethnic settings) included inthe sample will be interviewedright after receivingtheir magnetic cards, but just before being able to make their first cash withdrawal. This approach will have two advantages: (a) families will already know of their eligibility for the program, and would thus understand that their answers to the baseline survey would not affect their chances of being accepted into the program; and (b) data would be collected before the paymentscan begin to have an impact (hence it would be a baseline survey). Data for the baseline survey will be collected on an on-going basis (as families are registered into the program), beginning in 2005. A follow up survey will be conductedtwo years later (in 2007), and would be supportedby the second-phase APL project. Data would thus be available for "pre-program" (baseline) and "post-program" impact analysis and monitoring. In terms of impact evaluation methodologies, it would also be desirable to survey a "control" group of similar non-participant households (so as to generate comparison data of those "with" and "without" the program). The express delineation of a "control" group - that would be denied program benefits despite eligibility - raises ethical and political concerns. Indeed, officials from MDS have expressed such concerns. An interestingalternative would be to include in the survey a sample of non-participant families that were registered in the Cadastro Unico but were not selected into the BFP because they did not meet eligibility criteria. Surveyingthese excluded householdswill allow for two valuable types of analysis: (i) assessment an of the efficacy of the registration process through the Cadastro Unico, by verifying if excluded households are in fact less deprived than beneficiaries, and (ii) assessment of the impact of the program on household an level socio-economicindicators via the method of regressiondiscontinuity design (see Hahn, Todd and Van der Klaauw, 2001). The baselineand follow up surveys will interview these non-participantfamilies.37 3.D. Design and Oversight of the Overall M&E System This sub-component will support the creation of an external advisory group of national and international experts to assist the BFP in the design and implementation of the overall M&E. Specific activities to be supported by this component include: (a) hiring of national and international consultants to assist with the design and oversight of M&E system; and (b) workshops with national and international consultants to discuss and disseminate the overall design of the M&E system with the program management team and policy makers alike. POF was carriedout in2002-03, and hencepresents a "natural" baselinefor many indicatorsfor the BFP. A follow up POF will likely be carried out in2005 with the support ofthe proposedIntegratedHDTAL. These POFsurveyswill providesome(but not all) ofthe indicatorsneeded. "However,extracareneedsto betakento ensurethatnon-participantwill provideaccurateinformationby assuringthemthattheir chancesof eventuallygettingthe programwill not be affected by their answersto the interview. 55 . ... ... Y 3 . . 8 z... a L...... . . . a m a . . . . 0 . . 0 . . . . . . 0 . 0 . . . . . . . 62 E L m B 5" C a . . . . . .... T T ..i a ie de .I CC ; c t P TE 'i tC .-e C $ 4c W C 9C C cI ! ca zc Component4 -InstitutionalStrengthening (US$3.2 millioninproject costs) OBJECTIVES OF COMPONENT This component would support the strengthening ofthe basic institutional functioning ofthe BFP interms of: (a) strengthening the overall and operational framework for the program; (b) strengthening program management; (c) developing a solid communications and dissemination strategy; and (d) piloting innovative approachesto service delivery. RESULTS FRAMEWORKKEY MILESTONESCOMPONENT AND FOR The main results of this component include the accomplishment of a set of activities such as: the publication of a matrix o f key functional and institutional responsibilities; the publication of the BF beneficiary information booklets (cudernetus); an operating system for the verification of beneficiary compliance to human capital conditionalities; and training on BF implementationprovided to at least half of BF municipalities. A number of key milestones (outputs) could be monitored to gauge the progress towards meetingthese goals, they include: e By December 2004: (a) matrix of key functional and institutional responsibilities published; (b) beneficiary information booklets (cadernetus)designed. e By December 2005: (a) system for verifying human capital conditionalities developed and piloted in at least 5 municipalities; (b) beneficiary information booklets (cudernetus)distributed. MAIN ACTIVITIES SUPPORTED BY COMPONENT 4.A. StrengtheningInstitutionalFrameworkfor BFP A key step towards strengthening the overall and operational institutional framework of the BFP is the definition o f the normative and legal architecture o f the program. Activities inthis area would include the completion and publication o f neededregulumentos, or norms, on the design and functioning of both the BFP and the Cadastro Unico. For example, the specific definition of BFP conditionalities for beneficiary family members example should be clarified. The periodicity o f re-certification o f family data in the Cadastro Unico also needs to be defined. The definition of proxy variables to predict or validate income for eligibility criteria also needs to be legally established. Guidelines to register and include indigenous and Quilombola families inthe Cadastro Unico and the BFP should be clearly defined. These regulations could be documented and disseminated in the operational guidelines (regulamentos) of the BFP. The design of the overall operational framework and functions of oversight o f its implementation could be carried out by a committee, a group of consultants or advisors, such as those set up for the other components. These system design and oversight activities would be supported by technical and capacity buildingassistance. 4.B. StrengtheningProgramManagement The project will also support the strengthening o f program management in two key areas: (a) the institutional arrangements for the core BFP functions; and (b) mechanisms for inter- and intra- governmental partnerships. First, the institutional arrangements for core BFP functions needto be clarified. This would include the definition of a matrix o f key institutional responsibilities among all actors involved (MDS, SENARC, CEF, municipalities, line ministries, etc.). This would greatly improve program implementationand build 60 institutional accountability and credibility. The project will also support the definition and clarification of institutional roles and information flows for the CadastroUnico. Inaddition, a (decentralized) systemand information flows for monitoring beneficiary compliance with health and education conditionalities would be defined. Capacity building and training on the verification o f human capital conditionalities would also be supported. The project would also support the development of a strategy for operating the program inremote areas. Second, mechanisms for inter- and intra-governmental partnerships will be defined. The BFP is currently pursuingnumerous partnerships with states and municipalities. These partnerships take several different forms. Some seek to integrate local cash transfer programs with the BFP in order to expand the number o f beneficiaries or increase the level o f benefits (or both). Others seek to link BFP beneficiaries with complementary services and programs at the local level. The project will also support the development of an Information Network (Rede) across municipalities (to share good practices, experience, etc.). At the federal level, the BFP is also currently defining intra-governmental roles. The project will also support a study to explore the possibility of integratingother programs with the BFP. 4.C. Developing a Communications and Dissemination Strategy The institutional strengthening component will support the development of a solid communications and public information strategy. This will help promote the program, generate awareness o f the program, beneficiary compliance with conditionalities, and support the credibility and legitimacy of the program. It will also serve as an instrument to support citizen oversight (social control). The sub-component will support the following activities: (a) the design o f an information strategy that will consider the information needs of specific groups including culturally, ethnically, and racially distinct groups; (b) the development of a "hotline" number with trained operators to answer questions on the BFP and on the Cadastro Unico; (c) publications and editingservices; and (d) the design, publication and dissemination of beneficiary information booklets (cudernetus),including information about benefits and conditionalities. The latter would become an important tool for promotingprogram take-up and conditionality compliance. 4.D. Promoting the Empowerment (Emancipagcio)of BFP Beneficiaries This sub-component would support the identification and evaluation of initiatives to help BFP beneficiaries to develop tailored empowerment strategies("emuncipa@`o'' strategies) and to gain access to complementary programs and services (bridging services). These initiatives could involve various approaches, including: (a) social worker support to help beneficiaries develop these empowerment (emuncipu@o) strategies and bridging access to complementary services; and/or (b) the development of "one-stop shops'' where beneficiaries can gain access to information about a variety of social services (including the BFP). Some similar experiences have also been developed at the local level in Brazil and inother countries. The sub-component would thus support: (a) internationalstudy tours and exchangesto share information on these various "empowerment" and "bridging" approaches; (b) consultancies and workshops to identify and evaluate such approaches; and (c) training and capacity building for beneficiaries, trainers (multiplicudores) andprogram managers(gestores). 61 m .e M . . . . .. . .... 0 . . . . 0 . . 0 . . . . . m 0 e, .e Y . . . 0 . 0 . Component5 -Project Management (US$0.7 million) The project will support a strengthening of the implementation capacity o f MDS. To provide such support, the proposed Project will finance incremental costs related to management and supervision of Project activities. These costs include consultants that would support strategic, technical and operational aspects of the project, as well as equipment and travel expenses. The arrangements for project implementation are discussed inAnnex 6. 63 Annex 5: Project Costs BRAZIL: BRBolsa Familia Local Foreign Total Project Cost By Component and/or Activity us us us $million $million $million 6,174.0 0.0 6,174.0 1. Conditional CashTransfers38 2.8 1.2 4.0 2. Systemfor Identifyingthe Target Population 3, Developing a Monitoring and Evaluation 7.0 0.1 7.1 System 4. Institutional Strengthening for Improved 2.9 0.3 3.2 Service Deliverv 0.7 0.0 0.7 5. Project Management Support Total Baseline Cost 13.4 1.6 15.0 Physical Contingencies 0.3 0.1 0.4 Price Contingencies" -0.3 0.1 -0.2 Total Project Costs 13.4 1.8 15.2 Front-end Fee - 5.7 5.7 Total Financing Required 1 6,187.4 7.5 1 6,194.9 The total World Bank loan is US$572.2 million. This includes US$551.5 million that would be reimbursed to the Treasury for a share of expenditures on conditional cash transfers under the BFP, which are funded with government resources that are largely earmarked for that purpose. 39Price contingencies are generated by the COSTAB system on the basis of assumptions regarding projected inflation and exchange rate movements. 64 Annex 6: Institutional and Implementation Arrangements BRAZIL: BRBolsaFamilia 1. Implementationperiod The project i s expected to be implemented over a period of two years and four months (,,;st phase A with a completion date o fJune 30,2006 and a closingdate o fDecember 31, 2006. 2. Borrowerand ImplementationAgencies for the BolsaFamiliaProgram The Borrower will be the Federative Republic of Brazil. The BFP i s beingmanaged and implementedby the SecretariaNacional de Rendade Cidadania(SENARC, Secretariat for Citizen Income) inthe newly- createdMinistryo f Social Development (MDS).MDSwas created on January 23,2004 with the objective of further integrating the management of Brazil's social programs (as part of a broader integration strategy that also covered the integration of conditional cash transfer programs into the BFP). The MDS integrated three previous entities, including the Extraordinary Ministry for Food Security (MESA), the Ministryof SocialAssistance (MAS), andthe Executive Secretariat for the BolsaFamiliaProgram (which was previously under the Presidency). The MDS structure includes inter alia an Executive Secretariat (SECEX) under the Minister, which oversees two special departments, a Directorate of Special Projects (DPE) and a Secretariat for planning, budget and procurement (SPOA), and which supervises the activities o f five secretariats, including the SecretariaNacional de Renda de Cidadania (SENARC); a secretariat for other social assistance programs (e.g., BPC-LOAS, PETI, Agente Jovem); a secretariat for food security; a secretariat for policy formulation (articulaGZo) ; and a secretariat for monitoring, evaluation and capacity building. The SENARC overseesthe operations o f the Bolsa Familia Program. The BFP is also guided by the decisions of the President, supported by the recommendations of management board (Comiti?Gestor do Programa Bolsa Familia), which is responsible for policy coordination and promoting partnerships between the various sectors and levels of government and with civil society. Institutional arrangements for various functions o fthe program also involve a number of other actors, as discussed inAnnexes 1(general) and 7 (financial). 3. ProjectManagement As discussed elsewhere, the proposed project would adopt a program-based sector-wide approach (Program-Based SWAp) in order to associate disbursementsunder the World Bank loan with sector-wide reforms that integrate the main federal conditional cash transfer programs into the Government's flagship social program, the BFP. By taking advantage o f Brazil's sound institutional and fiduciary framework, the Program-Based SWAp is expected to increasecommitment, impact and flexibility. Consistent with this approach, and to promote a mainstreaming o f project activities and to contribute value added to the overall BFP, the proposed project will build on the organizational and leadership structures of MDS. Specifically, it will be headedby the Executive Secretary (SECEX) and managed by a Core Management Committee (CMC, Nucleo de GestZo), which will be comprised o f the managers from the relevant units o f MDS: SECEX, DPE, SENARC, the Secretariat o f Monitoring and Evaluation, and the Coordinator o f the PIU (see below). Functionally, this Core Management Committee will be divided into two areas: a Technical Secretariat and a ProjectImplementationUnit (PIU). 65 FIGURE1 PROJECTMANAGEMENTWITHINTHE STRUCTUREOFMDS(FOR - COMPONENTS2-5) EXECUTIVE SECRETARIAT NUCLEODE GESTAO SECRETARIAT DPE I SENARC (-2- COMP 2 PLANNING & PROCUREMENT FINANCIAL SUPERVISION MANAGEMENT TechnicalSecretariat. The Technical Coordinator for the project (likely to be the Director of the DPE; and serving on the CMC) will be supported by a technical team and will be responsible for establishing the technical directives and activities of the project, ensuring the solidity o f their content, and ensuring that they are consistent with the objectives o f the project and the directives established by the CMC. This technical team would consist of full-time consultants, including: (a) a technical assistant to assist the Technical Coordinator with overall technical coordination; and (b) an advisory committee for each of the three technical components, consisting of technical specialists and technical assistants. This technical team will be responsible for: (a) ensuring that the content o f the activities to be implemented under the project is consistent with the directives of the CMC and the directions o f the BFP; (b)overseeing technical design and implementation; (c) planning and monitoring the technical project components; (d) preparing technical specifications and terms of reference for contracting consultant services; and (e) working with ad hoc technical consultants/advisors to be hired for eachtechnical component. Project Implementation Unit (PIU). The Coordinator of the PIU will be designated by the Executive Secretariat of MDS and will likely also be the Director of Operations of the SENARC. The PIU will be responsible for: (a) on-going planning and supervising o f the project; (b) verifying that the operational procedures, as established in the Loan Agreement and in the Operational Manual, are being complied with; (c) administration of project activities, including agreements, contracts, procurement and the Procurement Plan; (d) preparing reports on financial performance and disbursement; (e) tracking project funds; (f) contracting independent technical audits; (g) providing operational guidance to the units in 66 charge of implementation of project activities; (h) supervisingcompliance with these guidelines; and (i) preparing progress reports on the project (including technical and operational activities, and compliance with operational guidelines) and submitting these to the World Bank every four months. To accomplish the requirements o f the project described above the PIU will be supported by the Operations Department in SENARC and will be staffed by MDS officials and with following key personnel (in addition to the PIUCoordinator): 0 Overall Project Operations Support: The PIU Coordinator will be assisted by a technical assistant (full-time consultant) for overall operations coordination. An IT specialist (full-time consultant) will also be hiredto support the entire project team. 0 Project Coordination,Planningand Supervision: The team will include a technical specialist and a technical assistant (both full-time consultants) for planning and supervision. This team would be responsible for overseeing and monitoring the project, preparing annual and updated project plans and preparingproject progress reports. 0 Procurement: Two procurement specialists (one full-time consultant and one ad hoc consultant) who will be responsible for: (i)overall procurement processing; (ii)annually updating the Procurement Plan; (iii) ensuring that the activities financed by the project follow the procedures established in the Guidelines of the Bank and in the Loan Agreement. (See Annex 8 for more details). 0 FinancialManagement: A financial manager (staff member) with the overall responsibility for financial management and administration of the project, who will liaise with STN for Component 1, oversee the preparation of the financial management sections o f the Operational Manual, and train staff on the operation of the SIAFI (Sistema Integrado de Administraqiio Financiera) system. This financial manager will be assisted by two accountants who will be responsible for: (i)preparation of Statements of Expenditures (SOE) for Components 2-5; ($preparations of consolidated FMRs to be sent to the World Bank (see Annex 7); and (iii) ensuring that the records o f the project will be kept for the annual audit by SFC (Secretaria Federal de Controle) and World Bank supervision. (See Annex 7 for more details). 4. ImplementationStrategy Implementationof Component 1. Component 1will support the implementation ofthe BFP, which will provide conditional cash transfers (grants) to poor families (as described in Annex 1). The BFP i s implemented by the SENARC in the MDS, in partnership with the Caixa EconGmica Federal (CEF) and the municipalities. Institutional arrangements for the BFP are discussed in more detail above and in Annex 1. Disbursement requests for Component 1 will be made directly by the Treasury (STN), as discussed in Annex 7. Implementationof Technical Components. The other components (Components 2, 3, 4 and 5) will be implemented as per the arrangements discussed above and according to the procurement guidelines outlined inAnnex 8 and the financial management arrangementsoutlined inAnnex 7. Disbursements. Disbursement arrangements for all components are discussed inAnnex 7 OperationalManual (OM). In order to reduce implementationdelays and define roles, responsibilities, mechanisms, schedules and accountability arrangements, the Operations Team will prepare a draft of an Operational Manual o f the project (enactment would be a condition o f effectiveness). The O M will include conditions necessary for implementation of the project such as: institutional arrangements; operational procedures (including a system for monitoring and verification o f conditionalities, eligibility procedures); procurement and contracting procedures, sample standard bidding documents and filing 67 instructions (including for QCBS and NCB); disbursement and financial management procedures, including template FMRs; auditing procedures; and supervision, project administration, monitoring and progress reports, and evaluation plans; and Performance Indicators. The OM will also include procedures for carryingout Component 1, including financial reportingand disbursementprocedures. 68 Annex 7: FinancialManagementandDisbursementArrangements BRAZIL: BRBolsaFamilia A financial managementrisk assessment ofthe BolsaFamilia Program (BFP) was conducted. As discussed inAnnex 6, the proposed project will be implemented by the Ministryof Social Development (MDS)at the federal level. The assessment was carried out in accordance with OP/BP 10.02 and the Guidelines for Assessment of Financial Management Arrangements in World Bank Financed Projects issued by the Financial Management Sector Board on October 15,2003. The overall objective of the assessment was to determine whether the Borrower has, or will have, acceptable financial management arrangements in place by loan effectiveness. These include, but are not limited to, capacity to (a) properly manage and account for all program's proceeds, expenditures and transactions; (b) produce timely, accurate and reliable program financial statements and reports, including Financial Monitoring Reports (FMRs), for general and Bank special purposes; (c) safeguard program's assets; (d) engage independent auditors acceptableto the Bank in a timely way; and (d) disburse Bank funds inaccordanceto applicable Bank rules andprocedures. The assessment was carried out jointly with the implementing MDS staff, Secretaria do Tesouro Nacional (STN, National Treasury Secretariat), and other staff members involved in the implementation o f the program. It included the (a) discussion with financial management staff in MDS and STN officials; (b) evaluation o f financial management systems inplace andor those to be usedfor program's monitoring, accounting and reporting; (c) review of staffing requirements, including training and qualifications; (d) discussion of flow o f funds arrangements and disbursement methodology; (e) review of internal control mechanisms inplace; (f) discussion with regard to reporting requirements, including the format and content of FMRs; and (g) review o f internal and external audit arrangements. The financial management assessment o f the BFP builds on the overall Country Financial Accountability Assessment (CFAA) which was carried out for Brazil in 2002. The CFAA assessed the public financial management system at the national level and concluded that ". ..Brazil's system provides reliable information. Adequate systems exist to manage and track the receipt and use of funds and there is a high level ofJisca1 transparency, both of which would support any program of adjustment lending. The risk to both Bank and countryfunds is low." The Program-Based SWAP approach for the proposed project to support the Federal Government's BFP will take advantage of these sound institutional and financial managementsystems, namely the SIAFI(Sistema Integrado de Administrag20 Financeira) system. I. FINANCIALMANAGEMENT ARRANGEMENTS All financial managementaspectsofthe BFP are beingmanagedby MDS staff incollaborationwith staff in STN, usingBrazil's systems and existing organizational arrangements. The BFP will have a specific "cost center" (Unidade Gestora), within the SIAFI system, where all sources - except for loan proceeds for Component 1 -- and uses o f funds will be recorded and from where FMRs will be produced in a regular, timely fashion. A satisfactory Financial Management System (FMS) for the project will build on existing government arrangements and embedded internal control mechanisms and procedures, as well as collaborationwith S T N and its staff. Staffing. As discussed in Annex 6, MDS oversees the implementation of the BFP including financial management aspects. In terms o f staffing and supervision for the project, financial management for the Project will be managedby the PIUwith the support of a financial manager and assistedby two accountants along with the staff o f the Operations Department of SENARC. The PIU Coordinator will liaise with STN, overseethe preparation of the financial management sections of the OperationalManual for the project, and 69 train staff inthe operation o f the SIAFI system. The Director of Operations o f SENARC (who is also the PIU Coordinator) i s responsible for cash flow management, including monitoring o f the BFP account with the Caixa EconBmica Federal (CEF). FinancialReportingand Supervision. The PIUCoordinator will ensure the timely productionof semi- annual Financial Monitoring Reports (FMRs). These FMRswill make use o f existingfinancial reporting information in the SUFI system and will consolidate information for all components of the project. The content and format was agreedwith the Borrower andthe FMRs generally cover the following topics: 0 FMR 1A - Sources4' and Uses of Funds(by disbursement category), cumulative and for the period, including a variance analysis; 0 FMR 1B - Uses o f Funds by Program Activity/Component, cumulative and for the period; actual expenditures, including a variance analysis; 0 FMR2 - PhysicalProgressReport(monitoring indicators); 0 FMR3 -Procurement Report; FMRs will disclose that loan proceeds will reimburse the Treasury for a percentage of conditional cash transfers, which will have the Government as sole financier. In addition, program activities and processes will be monitored via annual Quality Control Reviews (covering Cadastro Unico processes, payments, conditionality compliance), as discussed in Annex 4 (Component 3.B). Financial management supervision will be carried out every six months and will include: (a) reviewing semi-annual FMRs; (b) reviewing o f annual audit reports and management letter and following up in a timely manner on any issues; (c) participating in project supervision missions as appropriate; and (d) updatingthe financial management rating inthe Project StatusReport (PSR). Accounting Procedures and Policies. Annual Financial Statements o f the BFP will be prepared each fiscal year in accordance with consistently applied accounting standards, acceptable to the World Bank (accounting on a cash basis will be utilized for the production of both semi-annual FMRs and annual program financial statements). These statements will be audited, by independent auditors satisfactory to the World Bank (see below), in accordance with acceptable auditing standards. A section covering Financial Procedures will be included in the Operational Manual and will provide guidance with regard to financial and administrative routines, procedures and policies. Internal and External Audits. An annual external audit will be undertaken according to Terms of Reference acceptable to the World Bank. The Auditors' report will follow the World Bank's new audit policy, as per the guidelines issued by the Financial Management Sector Board o f June 30, 2003 and will express an opinion on the financial statements of the BFP. A management letter, identifying any internal control weaknesses and areas o f improvement will also be produced. Second-semester FMRs' cumulative financial reports, may be accepted by the World Bank as the basis for the annual financial statementsof the program on which auditors' will express their independent opinion. The auditors' report will be submitted to the World Bank no later than six months after the closing o f the fiscal year. In addition, a group of permanent staff are being assigned to perform the function o f a "Coordenaqao Geral de FiscalizaqEo" of the BFP which will essentially act as an internalcontrols department. 40Thesources anduses offunds will identifythe Govemment as the only sourceof fundingfor Component 1. 70 11. DISBURSEMENTS FLOWOFFUNDS AND ARRANGEMENTS Flow of Fundsfor the BolsaFamiliaProgram As discussed above, the BFP is managed by MDS. Each month, the MDS sends a Program Financing Proposal(PPF) to STN for approval. This PPF is establishedbasedonthe list ofeligiblebeneficiariesfrom the Cadastro U n i ~ o .Once the PPF is approved by STN, an ordem bancaria is issued and funds are ~ ~ transferred from a STN account with the Central Bank to a specific BFP account in the Caixa (CEF), see Figure 1below. MDS monitorsthe balance of the BFP account inthe Caixa (CEF). Eligiblebeneficiaries (on the payroll) then withdraw their monthly benefits from pooled BFP account using electronic benefit cards (EBCS)!~ The EBCs are distributedto eligiblebeneficiariesaccording to the list establishedby the MDS. Upon entering the program, beneficiaries must pick up these cards at the nearest CEF agency (or lottery, post office or municipality). They are notifiedthrough the municipalitiesand must pick up their cards and withdraw funds within a 90-dayperiodoftime. Experiencefrom the pre-reformprograms shows that approximately 3-5% of Bolsa Escola beneficiaries did not withdraw their benefits within the 90-day time period, mainly due to communicationsor transport difficulties inextremely remote rural areas.43 This share was reportedly much higher under the BolsaAlimentaq2o Program (up to one third of beneficiaries that did not withdraw their benefits, according to a TCU operationalaudit). Undisbursed amounts arising fromthe expirationofbenefitsthat are not claimedwithin the 90-day periodwill be returnedfromthe Caixa (CEF) to STN and credited to the MDS for use by the BFP. The OperationalManual will specify the number of days that the CEF hasto returnfinds inthe eventthat they are notclaimedby the beneficiaries. The MDS will enter into a new contract with the Caixa (CEF) where reporting requirements, as well as penalties, will be clearly spelled out, and CEF will provide a payroll-type list of disbursements made to beneficiaries underthe BFP. Arrangements for Component 1(ConditionalCashTransfers) Overall Financing Instrument. An additional feature of the proposed loan would be the use of a program-based sector-wide investment approach (Program-Based SWAP). A Program-Based SWAP makes sense in this instance because it would associate World Bank financing with one of the Government's highest priority programs, which involves integratingthe main federal conditional cash transfer programs as a comprehensive sector policy. By taking advantage of Brazil's sound institutional and fiduciary framework, the Program-Based SWAP is expected to increase commitment, impact, and flexibility. The Bankwill carefully monitorand supervisethe proposedProgram-BasedSWAPoperation, like all other investmentoperations. Interms of financingmechanisms, Component 1 of the proposedloanwould reimbursethe Government for a smallshare (less than 10%) of productiveexpenditures on conditionalcash transfers madeunderthe BFP, which are funded with government resources that are largely earmarked for that purpose. The BrazilianGovernment requestedthat the World Bank loan, in support of Component 1,would reimburse a share of these expenses. Given the current earmarking of revenues in Brazil, government revenues would be appliedto make the payments for expenditures under Component 1, and the Bank would then reimburse expenditures as specifiedin the legal agreement. The Bank funds disbursed would be part of the regular federal budget financing, with the Bank as one of the sources of overall funds, which are 'IAs discussed in Annex 1, some of these beneficiaries include those currently receiving benefit amounts from the pre-reformprograms (Bolsa Escola, Bolsa AlimentapBo, CaMo Alimentapb, and Auxilio Gas) but who are gradually being migratedto the (higher) unified benefit schedule underthe BFP. They are beingvalidated as eligible for the BFP and are subject to the same set of conditionalitiesas the families already on the BFP. 42Beneficiaries arenotrequiredto havetheir own bank accounts. 43Component4 ofthe proposedproject will support the developmentof a strategy for improving paymentsand operations inremoterural areas., as discussed inAnnex 4. 71 spent consistentwith federal budget policies and procedures. The Governmenthas indicatedto the Bank its intention to use the increasedfunds available for development priorities. Annual federal budgets are approved by the Brazilian Federal Congress and are derived from the budget plan (LD) and the Multi- Year Plan (PPA), which is a transparent four-year framework for federal expenditures, developedby the Brazilian authorities with the participation of civil society, the Bank, and other development partners. The Government engageswith the Bank ina discussionof the development expenditures. Specifically, the World Bank loan for this component would disburse against SOTs containing a list of transfers made to CEF, and verifying the disbursementpercentages, funding calculationsand conformity with the loan conditions. SOTs are a disbursement mechanism through which the World Bank will reimburse STN for advances made to CEF ("BF Grant Transfers") for disbursementsof the conditional cash transfers ("BF Grants") to I beneficiaries, as shown in Figure 1. As discussed below (under Loan I Figure 1- Component 1Flow of Disbursementsand Flow of Funds, and Funds ProgramRisk Assessment), the annual financial audits will contribute, among other ex-ante and ex-post control mechanisms embedded in the program design, to the verification of eligibility of expenditures using SOTs , (PooledBFPaccount) mechanism. Specifically, in their annual report, auditors will verify that all funds ("BF Grant Transfers") received by the CEF were credited to Annualaudits the pooledbeneficiaryaccount andthat verify that BFGrantsI at least 85% of these amounts were IBRD reimbursesSTN BFGrantTransfers disbursed by CEF to beneficiaries in a % of SOT Z 85% the form of BF Grants. In the event Government Expendituresfor Component 1. As mentionedabove, for Component 1, the World Bank would reimburse the Treasury for a specific share (see below) of expenses on conditional cash transfers under the BFP during the project period. Governmentexpenditures for Component 1 inthis case would be equal to total spending on conditional cash transfers by the BFP. Table 1 below presents the planned BFP expendituresfor these transfersfor the project period(first row). Table 1-ExpectedGovernmentExpendituresand ExpectedPaceof LoanDisbursementsfor 2004 2005 2006 Total Overall World Bank (4 months) period Share Expected Federal Government spending on - BFP Transfers, US$mn* 639 2,458 3,077 6,174 n.a. Component 1 (indicative disbursement 156*** II 221 II 174 1I 551 II 8.9% o ftotal schedule, see variable cost shares below)** I *These expenditures exclude allocations in the BFP budget for administrative costs. They include amounts being paid to beneficiaries that currently receive benefit amounts from the pre-reform programs (Bolsa Escola, Bolsa AlimentaqSLo, Cartb Alimentaqllo, and Auxilio Gas) but who are gradually being migratedto the (higher) unified benefitschedule under the BFP. They are all being validatedas eligible for the BFP and are subjectto the same set of conditionalitiesas the families alreadyon the BFP. **The US$551.5 million for Component 1 (grants) would be reimbursed to STN against a share of expenditures on conditional cash transfers underthe BFP duringthe projectperiod. ***Includes retroactive financing (see below). 72 Variable Cost Shares to Pace Disbursementswith Technical Improvements. While the total amount of financing for Component 1(US$551.5 million) would represent 8.9% o f the total cash transfers budget for the BFP, disbursements and cost-sharing percentageswould be linked to key milestones of technical improvements. As discussed below, Component 1 of the loan would be disbursed on the basis of the transfers made by S T N to the Caixa (CEF) -- upon MDS instructions and triggered by Ordens Bancarias issued by MDS -- to provide grants (conditional cash transfers) to the beneficiaries of the Bolsa Familia Program (SOTs concept). Specifically, loan proceeds for this component (SOTs for BF Grant Transfers) would be made with increasing disbursement percentagesthat will increase as the actions listed inTable 2 in Annex 3 are met. The disbursements will start at the rate of 8%, will be increased to 9% when the stage 2 milestones in Table 2 in Annex 3 have been met; and will be increased to 11% when the stage 3 milestones inTable 2 inAnnex 3 have beenmet. Ifstage 2 milestones are not met, disbursementswould continue to be made at a 8% disbursement percentage; in the event that stage 3 milestones are not met (but stage 2 milestones have beenmet), disbursementswill bemade at a 9% disbursementpercentage. LoanDisbursementsand Flow of Funds. As discussedabove, for the BFP itself, funds flow from STN to the CEF upon the issuance o f an Ordem Bancaria by MDS, with all transfers being registered in the SIAFI system. Once these transfers are made, the CEF then pays out the conditional cash transfers to a list of eligible beneficiaries established by MDS. As discussed above, the World Bank will reimburse, on a monthly basis and upon request from STN, against statements o f transfers (SOTs), recorded in SIAFI, made by the S T N to the CEF ("BF Grant Transfers") upon issuance o f an Ordem Bancaria by MDS. Once this transfer is made, this advance for the grant to be made is considered as expensed and the expenditure eligible for World Bank financing (SOT concept). When requesting reimbursement from the World Bank, STN will not be required to reconcile with actual expenditures (e.g., payments made to beneficiaries with STN transfers to Caixa; or for expenditures by beneficiaries using their transfers), given the high number and small amount of the transfers to beneficiaries. As mentioned above, the auditors will certify intheir annual report that 100% ofthe funds received bythe CEF (BF Grant Transfers) for the BFP were credited to the pooled beneficiary account and that at least 85% of funds transferred to the CEF via SOTSwere disbursed by the CEF to beneficiaries (as BF Grants). Inaddition, unaudited semi-annual FMRswill reportonthe amounts transferred to CEF through SOTs vs. actual payments to beneficiaries made by the CEF. Existing controls over the beneficiaries N I S (Social Identification Number) and amount of the benefit allow for the tracking o f individual benefits. This control is done through monthly withdrawal reports, produced by Caixa, where amounts withdrawn are compared to the monthly list of beneficiaries and respective benefits, which were the basis for the transfer to the BFP's account with the Caixa. The technical verification of the compliance by beneficiaries ofthe grant conditions (health and education conditionalities) would be made on an on-going basis by municipalities and by the Federal Government via annual Quality Control Reviews for a random sample of beneficiaries (see Annex 4, Component 3.B.). These annual Quality Control Reviews will also cover payments (ensuring that payments are actually collected and received by intended beneficiaries) and Cadastro information, which will provide an additional lever for fiduciary controls. SFC, as part of their routine work, i s already contributing to the overall quality of the BFP by doing spot checks of the program's internal control procedures at the municipal level. These spot checks also include meetings and verifications with small samples of BFP beneficiaries. 73 The Government has opted not to use a Special Account for Component 1. Monthly SOTS,supporting the withdrawal application (Form-1903), will be prepared by STN and submitted to the Bank for reimbursement purposes. The MDS will consolidate and prepare semi-annual FMRs and annual financial statements for the program(for all components of the project). Arrangements for Components2-5 (Technical and ProjectManagementComponents) Disbursements under Components 2-5 will be made on the basis of transactions and withdrawal applications will be prepared and submitted by MDS (the PIU) on a regular basis (monthly) for replenishment o fthe Special Account (which will be established exclusively for Components 2-5). Disbursement claims (Form- 1903) will have as supporting documentation, a Special Account bank reconciliation, a Special Account bank statement, Statements of Expenditure (SOEs) and/or Summary Sheets as applicable, and for all payments made against contract above SOEPrior Review thresholds adequate supporting documents such as invoices and/or receipts. The Borrower has opted not to use direct payments to suppliers, from the Loan Account, as stipulated inthe Decree Law 4.992 of February 18,2004. Advances from the Special Account to UNagencies will be made against reasonable cash flow forecasts for eligible expenditures in accordance to Borrower agreements with UNDP. Further advances will only be made upon receipt o f adequate supporting documentation for the previous advance. The Borrower may request disbursements to be made using the Statement of Expenditures (SOE) methodology for all payments against contracts not subject to World Bank's prior review, including: (a) contracts for goods costing less than US$350,000 equivalent; (b) contracts for consulting firms services costing less than US$200,000 equivalent; (c) contracts for individual consultants' services costing less than US$50,000; (d) training activities; and (e) incremental operating costs. Retroactivefinancingwill be available for disbursementcategories 1(Goods), 2 (Consultant Services), 3 (Training and Seminars), 4 (BF Grants) and 5 (Incremental Operating Costs) procured in a satisfactory fashion and paid after January 1, 2004 up to a maximum o f 19% of the loan amount. This retroactive financing percentage is consistent with the parameters included in the New Policy Framework for Eligibility o f Expenses, and consistent with the established Brazil country financing parameters. It is recommend that, once the loan is declared effective, such expenditures be immediately claimed from the World Bank, on a reimbursement basis. In order to ensure timely provision o f funds to finance the costs of the project, the Government will establish, for the purpose of this project, a Special Account inU S dollars, exclusively for Components 2 to 5, in a commercial bank under terms and conditions acceptable to the Bank. An estimated amount equivalent to 4 months o fthe Bank's share o f project expenditures for these components will be deposited inthe Special Account. The authorized allocation for the Special Account is US$1.5 million. 111.FINANCIAL MANAGEMENTRISKASSESSMENT ProgramRiskAssessment. The Risk Assessment Matrix below, presents the items of potential risk for the programfrom a FinancialManagement standpoint. 74 i.Countryspecific ii.Sub-nationallevel 1x1 1x1 TI A components quality control reviews accuracyof (Cadastro - Unico); TORSfor SFC iii.Entityspecific X New entity; World Bank assistance will be required iv.Project specific X Same as for Sub-national level Control Risb i.Implementing Agency x I FMarrangementswill rely on SIAFI and embeddedcontrols ii.FlowofFunds 1x1I IIPayments in remote areas: electronic benefits cards; random tests iii.Staffing ] x I I See implementation arrangements iv. Accounting procedures X v. InternaVExternalAudit X Internal/External audits to be conducted by SFC; TORSfor SFC vi.Reporting & Monitoring x New entity but SIAFI will be utilized; World Bank support vii. Information Systems X Same as above The overall Financial Management risk associatedwith the Project was rated as moderate. There are two main areas of concern and potentialrisks. The first involves controls over payments, including: The risk that people other than the intended beneficiaries would claim the electronic benefit cards. This is a problem to some extent inany welfare program inany country inthe world. The proposed project will help the BFP develop systems to limit this risk, including random-sample Quality Control Reviews (see Annex 4, Component 3.B) and a variety o f social control mechanisms (including publicly disclosing lists of beneficiaries at the local level). The reconciliation between the amounts transferred to the Caixa ("BF Grant Transfers") and actual amounts disbursed to beneficiaries ("BF Grants"), including the issue of unclaimed benefits during the 90-day period as discussed above. This risk is mitigatedby: (a) the fact that MDSmonitorsthe BFP account inthe Caixa, which will transfer undisbursed amounts back to the STN for further use by the BFP; and (b) annual audits will ensure that (i) of funds receivedby the CEF (BF Grant 100% Transfers) under the BFP were credited to the pooled beneficiary account; and (ii) at least 85% of the funds transferred to CEF via SOTS were disbursed by the CEF to beneficiaries (i.e., BF GrantsBF Grant Transfers 2 85%). The Operational Manual will clearly state the number of days that CEF has to do these. The second main area o f concern and potential risk involves the accuracy of, and the decentralized process for collecting data for, the Cadastro Unico registry that is used for identifying and selecting beneficiaries. As discussed in Annex 4, the technical components of the proposed project will greatly strengthen the accuracy of the Cadastro processes, which will help mitigate this risk. The proposed project will also support annual Quality Control Reviews of Cadastro information (reinterviewing a random sample of beneficiaries to track payments, compliance with conditionalities, and the accuracy of Cadastro information; see Component 3.B.in Annex 4). Financial Management Assessment Conclusion. The overall Financial Management Assessment conclusion is that the program's implementing agency (MDS), with the collaboration of STN, has satisfactory financial management arrangements in place, and meets OP/BP 10.02 minimum requirements. The financial management risk associated with the program has been assessed as moderate. The Bolsa Familia Financial Management System will be able to produce timely, relevant and reliable financial information, and to support the program's management inthe planning, implementation and monitoringthe Program's overall progress towards the achievement of its objectives. 75 Loan ConditionsRelatedto FinancialManagement Adoption of Operational Manual including procedures for the flow of funds and information to and from beneficiaries and financial managementby effectiveness; MDSto retainaPIUincludingadequatefinancial managementstaffbyeffectiveness; and Draft Terms of Reference for external audits approved by the World Bank by effectiveness. 76 Annex 8: Procurement BRAZIL: BRBolsaFamilia ProcurementInstitutionalArrangements The project will consist o f five components: (a) one conditional cashtransfer Component (component l), in which the Bank will reimburse US$551.5 million to the Brazilian Treasury for expenditures under the BFP during the project period, that will not involve procurement issues; and (b) three technical components and one project management component, for which loan amount i s US$15.0 million to help improve the basic architecture o f the program. Therefore, the proposed project will only require procurement for the technical and project management components. The MDS will carry out hiring of individual and consultants firms, workshops and training activities incooperation with UNDP. PROCUREMENT METHODS Procurement for the proposed project will be carried out in accordance with the World Bank's "Guidelines: Procurement Under IBRD Loans and IDA Credits", published in January 1995, revised JanuaryIAugust 1996, September 1997, January 1999 ,November 2003 and May 2004; and "Guidelines: Selectionand Employment of Consultantsby WorldBank Borrowers" published in January 1997, revised in September 1997, January 1999, May 2002 and May 2004; and the provisions stipulated in the Loan Agreement. The Bank's Standard Bidding Documents and Standard Request for Proposals will be used for the Bank-financed ICB goods and consultant services, respectively. For NCB procurement of goods, MDS will use the model standard bidding document issued by SEAIN (Secretaria de Assuntos Internacionas). The electronic preg8o may be usedas an alternative to shopping method for off-the-shelf items in all project components. The project procurement method for each category is summarized in Table A ofthis Annex. World Bank-Financed Goods and Services Under the technical and project management components, goods will be procured by the Ministry's procurement unit, and consist o f computer equipment, peripherals, software, printing services, communication equipment, and logistics for workshops and training. The following methods will be used for this project: InternationalCompetitiveBidding(ICB). Goods estimated to cost US$350,000 equivalent per contract or more will be procured through International Competitive Bidding (ICB) procedures. During preparation, packages o f computer equipmenthohare for some states and municipalities, and of printing services have been identified to be procuredthrough ICB. National CompetitiveBidding (NCB). Goods estimated to cost more than US$lOO,OOO equivalent and less than US$350,000 equivalent per contract may be procured under contracts awarded on the basis of NCB procedures. Printing and distribution of booklets to the beneficiaries o f Bolsa Familia Program, as well as other items that may be grouped to reach NCB thresholds, are expected to be procured through this method. Seconded services for operating the hotline, under Component 4, will be hired using NCB documents satisfactory to the Bank, as described in item Incremental Operating Costs. Shopping. Contracts estimated to cost less than US$lOO,OOO equivalent each, up to an aggregate amount not to exceed US$600,000, may be purchased on the basis o f price quotations from at least three eligible suppliers. Scattered printing services, small technical services, computer equipment and peripherals, 77 office equipment and logistics for training events and workshops have been identified to be procured underthis method. Table A: ProjectCostsby ProcurementArrangements (in US%thousand eauivalent) 1. Goods 3.3 0.4 0.1 3.8 (3.3) (0.4) (0.1) (3.8) 2. Consultants' Services 10.2 10.2 (10.2) (10.2) 3. Training & Seminar 0.5 0.5 (0.5) (0.5) 4. Conditional CashTransfers3 6,174.0 6,174.0 (55 1.5) (551.5) 5. IncrementalOperating Costs 0.2 0.5 0.7 (0.1) (0.4) (0.5) Front-end-fee 5.7 5.7 (5.7) (5.7) Total 3.3 0.6 6,191.0 6,194.9 (3.3) (0.5) (568.4) (572.2) WorldBank-Financed ConsultingServices The proposed project would finance consultant services by firms and individuals. It is expected that the project will finance several studies, design of information systems, data collection and analysis, focal groups, activities to the Cadastro System, manuals to the BFP, etc. The following methods apply: Quality and Cost-Based Selection (QCBS) procedure will be used for consultant services, and training contracts. At preparation about 90% of the consultant contracts with firms under the project were identifiedto be procuredthrough this method, for an amount of approximately US$6.7 million. Quality-Based Selection (QBS) will be usedfor assignmentswhich meet the criteria specified inClause 3.2 o f the Guidelines. Duringpreparation, only one contract was identified to be procuredthrough QBS: the design and analysis of the impact evaluation to be financed under Component 3 (Developing a Monitoringand Evaluation System). Individual Consultants will be hired for assignments which meet the criteria specified in Section V of the Guidelines, up to an aggregate amount o f US$2.7 million equivalent. A core group for providing technical and operational assistance for project management will be hired through this method, and includes Planning and Monitoring specialist, specialists to each area of intervention, Procurement and Financial Management specialists, and Informationand Technology specialist. 78 Incremental Operating Costs Incremental operating costs refers to (i) project management and supervision expenses tickets, per diem - and sundries; (ii)seconded services to operatethe hotline, including staff, equipment, and communication costs (to be hiredthrough NCB procedures), and (iii) operating and maintenance costs ofthe management informationsystem(MIS/SIG). These services will be financed inaccordanceto bi-annualplans approved by the Bank. Identification of suppliers of sundries and minor services will be made through market inquiries inthe local market as in shopping. Once the suppliers are established (generally those offering the lowest cost of lists of items), a purchase order will be issued usingregular administrative procedures. The market inquirieswill be updated periodically. Internal travel will be procuredthrough administrative methods of the MDS, generally consisting of the comparison of discounts and level of services offered by travel agencies. Table Al: ConsultantSelectionArrangements andPriorReview Consultant Services Expenditure QCBS QBS SFB LCS CQ Other N.B.F. Total Cost' Category A. Firms 6.7 0.8 7.5 (0.8) -- (7.5) Priorreview All(6.7) contracts estimatedto cost the equivalentto US$200,000 or more B. 2.7 2.7 Individuals (2.7) (2.7) Prior review All contracts estimatedto cost the equivalentto US$50,000 or more Total 6.7 0.8 2.7 10.2 (6.7) (0.8) (2.7) (10.2) PROCUREMENT IMPLEMENTATION AND SUPERVISION Advertising. A General Procurement Notice (GPN) for hiring of international consultant services, and purchaseof ICB goods must be publishedinthe UnitedNations Development Business, no later than June 2004. Thereafter, the Notice will be updated annually for outstanding consultant services and ICB goods. Inaddition, detailed consultant services and assignmentswill be advertised, asthey becomeavailable, inat least one national newspaper o f a large circulation. MDS may also seek "expressions of interest" by 79 contacting embassies, professional organizations, or firms that it knows or that are registered inDACON (paragraph 1.15 of the Guidelines for Selection and Employment of Consultants). In this case, the information required would be minimum, limited to make ajudgment on the firm's suitability. Sufficient time (not less than 15 days) will beprovidedfor responses, before preparingthe short lists. Procurement Records. Detailed procurement records, reflecting the project's supply of goods and consultant services, including records of time taken to complete key steps inthe process and procurement activities related to supervision, review and audits, will be maintained by the Operations Coordination at Operations Department. These records will be maintained for at least two years after the project's closing date. The records for goods will include public notices, bidding documents and addenda, bid opening information, bidevaluationreports, formal appealsby bidders and outcomes, signed contracts with related addenda and amendments, records on claims and dispute resolution, and any other useful information, including a procurement plan ina format acceptableto the Bank. Scheduling of Procurement. A procurement plan for the tasks to be carried out during project implementation was prepared and submitted to the Bank during negotiations. The procurement plan will consist of: (i)procurement plan for the project's goods, including contract packaging, applicable procedures and process scheduling; and (ii) a consultant and training providers selection process plan for the project's training and consultant services, including contract packaging, applicable procedures, selection criteria. The procurement plans and consultant selection process plan will be updated periodically, and shall be submitted to the Bank in January of each year. Such updating shall include (i) list of contracts completed, under execution, under procurement, to be procured inthe upcoming calendar semester and, tentatively, in the subsequent semester; (ii)costs of completed and under execution contracts, estimated costs for upcoming contracts; (iii)schedule o f biddings. The working instructions will be inthe OperationalManual which will contain standard biddingdocuments, request for proposal, formats for reporting on procurement, forms o f contract, timetables, model of TOR and any other relevant informationrelatedto procurement. Prior Review and Loan Agreement reference. The Bank's procurement review will be made in accordance with Appendix 1 o f the Guidelinesfor Procurement and Appendix 1 of the Guidelinesfor Selection and Employment of Consultants. A review process similar to individual consultants will apply to training and workshops. The following biddingprocesseswill be subject to the Bank's prior review (refer to Table B below). The adoption of the alternative o f specifying the said thresholds in the procurement schedule of the Loan Agreement is recommended: 0 all ICB tender processes, irrespective of the estimated amount of the contract; 0 all consultant contracts estimated to cost the equivalent to US$200,000 or more. In this case the following documents will be reviewed: terms of reference, requests for proposal, short lists, terms and condition o f contracts as well as evaluation reports and recommendation for award; 0 all individual consultant contracts estimated to cost the equivalent to US$50,000 or more, in which case the qualification, terms o f reference, evaluation justification, and draft of the contract will be forwarded to the Bank and will be awarded only after the Bank's concurrence. A review process similar to individual consultants will apply to training and workshops. 80 Table B: Thresholdsfor ProcurementMethodsand PriorReview Works No works expectedunderthe project >350k 100-350k Shoppingor All ICB Regular supervision Goods preg%o< lOOk only Ex-Post Review. All bidding processes regardless o f their value are subject to ex-post procurement review. All documentation included but not limited to: terms of reference, bidingdocument and request for proposal, bids or proposal received, correspondence on all bids either prior or following of award of contract and any subsequent amendments should be maintained until at least two years following the closing ofthe project. Frequencyof procurementsupervision missions:The Bank will carry ex-post and supervision missions on procurement twice a year. Participation of UN Agencies. The Government requested the Bank's accord to carry out the procurement and contracting o f some of the activities of Components 2, 3, 4 and 5 through UNDP. The fees involved for these intermediationservices are not eligible for Bank-financing. PROCUREMENTRISKASSESSMENT There is a CPAR for Brazil dated March 2004. Government procurement at the federal level i s considered robust and average risk. The review indicates that the judicial protests or recurso continue to be a major difficulty in government procurement. Another finding indicates that the use o fpregiib and other modern procurement methods will add efficiency and transparency to government procurement. The overall procurement risk assessment of the Project is ranked as high as presented in the Capacity Assessment (in the project files), since the MDS was only recently created and has no significant experience with the operations o f the World Bank (even though the Capacity Assessment indicates that the overall organization of the MDS is acceptable). The procurement function should be strengthened by givingMDSthe resources neededas indicatedinActivity No. 2 ofthe Action Planbelow. 81 Confirm proposeduse o f a UNagency with MDS To reinforce the procurement implementing capacity. 2 MDS to retain two procurement specialists with To reinforce the Byproject the following profiles: (a) fulltime procurement procurement implementing Effectiveness manager with at least 5 years experience inBank- capacity financed procurement; and (b) part time senior procurement specialist (ad hoc). 3 MDS to draft: To provide the Borrower By Effectiveness 0 an QCBS document; and with a view o fthe as part o f the 0 anNCB goods document, using the complexity o f Bank Operational documents developed by SEAIN procurement; to ensure a Manual quick start up onkey activities; to establish a work flow within MDS 4 MDS to update the procurementplan prepared Reinforce the procurement Procurement Plan duringproject preparation planningcapacity updated 5 MDS shows readiness to carry out Preglo Ensure consistency and ByEffectiveness Eletrdnico promote modern, as part o f the transparent and efficient e- Operational procurement Manual 6 MDS to prepare filing instructions for procurement To improve filing o f ByEffectiveness documentation, and to give guaranteesthat the procurement as part o f the system will allow prompt monitoring o f documentation and to Operational procurement and contracts. facilitate supervision Manual 7 MDS to maintain and develop sections on To ensure compliance with Manual by Loan procurement for the Operational Manual the agreed procedures Effectiveness Table D: Allocationof LoanProceeds 1. Goods 3,500,000 100 2. Consulting Services 10,100,00 100 3. Training and Seminars 500,000 100 4. BolsaFamiliaGrants 55 1,478,000 8% untilthe first conditions have beenmet, 9% untilthe second conditions have beenmet, and 11% thereafter 5. Incremental Operating Costs 500,000 100%untiland aggregate o f US$200,000 is reached, and 70% thereafter Unallocated 400,000 Front-end Fee 5,722,000 Total Project Costs 572,200,000 82 Annex 9: EconomicandFinancialAnalysis BRAZIL: BRBolsa Familia OBJECTIVESOFANNEX As discussed inAnnex 1and elsewhere, the Bolsa Familia Program provides increasing income transfer amountsto two differenttarget groups basedon householdper capita income and household composition: 0 Priority i s given to those with monthly per capita incomes (PCI) less than R$50 (the "extreme poor"), which represent about 14.5% of the national population. These families receive a base transfer o f R$50, regardless of composition, plus a variable transfer of R$15 for each child between ages 0-15 years old. The maximum transfer i s R$95 per family. 0 The next priority is givento those with monthly PCI between R$50-100 (the "moderately poor"), which represents another 18.5% of the national population. These families receive a variable transfer of R$15 for each child between ages 0-15 years old. The maximum transfer is R$45 per family. Inthis group, households with no children are not eligible. In both cases, beneficiaries must comply with certain conditionalities related to minimum school attendance o f children ages 7-15 and health care requirements (visits, vaccines) for pregnant women and young children (see Annex 1 for details). As with other conditional cashtransfers, there are manypotential benefits o fthe BFP, including: 0 A reduction inimmediate poverty and inequality due to the redistributive cashtransfers; 0 Gains in education due to both an income effect and a price effect associated with the conditionality o f minimum school attendance required of school-aged children in all beneficiary families (see Annex 1); and 0 Gains in health status due to both an income effect and a price effect associated with health- relatedconditionalities for pregnant women and young children (see Annex 1). Conceptual and data limitations prevent the monetary quantification of many of these benefits. Conceptually, while it is possible to quantify the magnitude o f poverty and inequality impacts that could result from these transfers, it is not possible (or desirable) to quantify these benefits inmonetary terms. In addition, though conceptually feasible, data limitations also prevent the quantification o f the impact of the program on health status. Given these limitations, a full-fledged cost-benefit or cost-effectiveness analysis is not possible. Instead, this Annex will: (a) quantify the potential impact of the program on poverty, inequality, and educational attainment; and (b) present evidence of the impact of similar programs - both from other countries and from pre-reform programs in Brazil. The final section will also present an assessment of the financial sustainability o fthe program. SIMULATING THE POTENTIALIMPACT OF THE BOLSAFAMiLIAPROGRAM As discussed above, this section will attempt to quantify the potential impact of the Bolsa Familia Program on educational attainment, poverty and inequality. These impacts will be estimated using ex ante simulation methods, both for the specification parameters being used in the program today and for alternative scenarios. As discussed in Bourguignon, Ferreira, and Leite (2003), "ex ante methods are valuable tools because it i s easier to experiment on computers than on people. These methods are essentially prospective, since they rely on a set o f assumptions about what households are likely to do when faced with the program. They also permit direct counterfactual analysis o f alternative programs for 83 which no ex post data are available." The methodology used involves simulations that take into account behavioral responses, following the methodologies used inBourguignon, Ferreira, Leite (2003) for earlier ex ante evaluations of Bolsa Escola (but usingthe program parameters for BolsaFamilia). Data Source and Main Data Limitations The analysis is based on the National Household Survey (PNAD 2002). Some information is also supplemented from the PNAD 2001 (which includes an extensive supplement on education). According to the PNAD 2002, 12.4 million households (54 million people)44would be potentially eligible for the program, which is very close to the target included in the Government's PPA for 2007. This represents about a third of the country's population. Around 48% of the transfers would go to households in the BrazilianNortheast, 28% inthe Southeast, and the remaining24% spread between the other three regions (North, South, and Center-West). Some limitations arise due to data constraints. First, we are not able to simulate the impact of health conditionalities or health/ nutrition benefits due to a lack of health and anthropometric data in the PNAD 2002.45 Another key assumption inthe analysis is that the programwould be perfectly targeted (with no leakages to the non-poor). Also, the model ignores the issue of how the decision about a child's time allocation is made within the household, bypassing the discussion of unitary versus collective decision- making models of the household. Moreover, the decision to send a child to school is assumedto be made after all adults within the household have made their occupational decisions, and it i s assumed not to affect those decisions. The issue o f siblings in the same household and the simultaneity of the corresponding decision are not addressed. Finally, the model is appliedto all school age children within a household and the composition o fthe household is assumedto be exogenous. PotentialImpacton EducationalAttainment The rationale behind providing transfers conditional on school attendance is based on the premise that poor children are less likely to enroll in and attend school because: (a) their families cannot afford the direct costs of schooling (formal and informal fees, transport costs, uniforms, books, supplies, etc.); and/or (b) the children need to work to help supplementfamily income. The idea i s to provide an income transfer linkedto school attendance that could help compensate some o f these direct and indirect costs. Data from the PNAD 2002 support this rationale. First, enrollment rates are lower among the poor than the non-poor. While an overall 93% of Brazilian children ages 5-15 years old report enrollment in only 89% of those inthe poorest quintile report enrolling as compared with over 98% o f those inthe richest quintile. Second, child labor is also more than twice as high among the poor as the non-poor. In 2002, some 12.4% of children aged 5-15 in the poorest quintile reported working, as compared with only 5% among those inthe richest quintile. Most o f those children who report working were also studying, however, the PNAD 2002 shows significantly lower hours per day spent at school for the poor: while the majority (72%) of students inthe poorest quintile report attending school less than four hours per day, the majority (60%) ofthose inthe richest quintile report attending more than four hours per day. Interestingly, a significant share of those that were not studyingwere also not working (90% of those in the poorest quintilethat were not studyingalso reportnot working). A possible explanationfor this is that "Consideringthedesignoftheprogram,another1.5millionfamilies(or 11%ofhouseholdswithpercapitaincomes(PCI)betweenR$S0-100) would not be eligible for the (variable) transfer sincethey haveno children. Datafrom the POF2002-03, which do includesome of these necessarydata, are not yet available for analysis. 46Inresponseto the questioninthe PNAD2002 "Are you studying?" 84 they lacked the neededincome to pay for transportation, books, materials, uniforms, or other demand-side constraints (such as a lack of interest or illness). Indeed, the PNAD 2001 suggests that such demand-side constraints account for about 46% of the reasons reported for not attending school. Alternatively, a lack of available schools or space in schools could explain some o f these cases. Indeed, the PNAD 2001 suggeststhat such supply-side constraints account for about 20% o fthe reasons reported for not attending school. Clearly, any demand-side program such as the conditional Bolsa Familia income transfer program should be complemented by significant investments in the supply and quality of schooling in Brazil. Third, reflecting lower enrollment rates, inadequate daily attendance, fewer hours in school, and higher repetition rates:' total educational achievement for poor children i s far lower than their wealthier counterparts. The PNAD 2002 indicates that the average number o f years completed for children inthe poorest quintile is a full two years lower than the attainment of those inthe richest quintile(3.4 years for the poorest vs. 5.4 years for the richest). Using the methods explained at the end of this Annex, we simulated the potential impact of the BFP on schooling decisions and educational attainment. The results are significant (see Table 1): 0 While data do not permit analysis of the impact on daily attendance, enrollment would increase by 0.59%, which i s small mainly because about 89% o f poor children were already enrolled (this would increaseto 0.82% ifthe variable transfer were increasedto R$30 per child). 0 The repetition rate would decrease by 6% and average years of delayed schooling (due to repetition) would fall by about 5%. These would increase to 9% and 7% respectively if the variable transfer were increasedto R$30 per child. 0 The average o f total years o f schooling would increase by 3.4%. For those (11% of the poorest quintile) that previously were not enrolled in school, the Bolsa Familia transfer would result in one extra year o f schooling. Using Mincer standardregressions for adults ages 25-64 years old in the work force, an additional year of schooling could result in an 8% increase in wages earned. (The average of total years o f schooling would increase by 5% ifthe variable transfer were raised to R$30 per child). Table 1-PredictedEducationalImpacts Model estimates and changesafter the Bolsa Familia Transfer 1 Ln(Household A after BF transferwith income) coef. A after BF transfer R$30 reais per child I Enrolment rate 0.25558 0.59% 0.82% Repetition rate -0.42345 -6.23% -8.90% DelayedSchool -0.59678 -4.79% -6.80% Education 0.33757 3.43% 5.00% Source: World Bank calculations, PesquisaNacional por Amostragemde Domicilios 2002 The impact of the Bolsa Familia program is presented in the transition matrices below (Table 2). With the transfer and the imposition o f conditionalities, the share o f children not going to school falls by a full percentagepoint, from 5.2% (overall -this figure is higher for the poor) to 4.5%. ''The poor are more likely to repeat grades than the non-poor. Repetition rates are 37% for children ages 7-15 in the poorest quintile, as comparedwith 6% for those inthe highest quintile. They are even higher for children ages 10-15 (54% for those inthe poorestquintile, but only 8% for those inthe richest). 85 Not Goingto School Going to Schooland Working Goingto School (only) Total Not Going to School 83.7% 2.7% 13.6% 5.4% Goingto Schooland Working 98.2% 1.8% 12.5% Goingto School (only) 100.0% 82.1% Total 4.5% 12.4% 83.1% 100.0% PotentialImpactonPovertyand Inequality The potentialimpact ofthe BFP (with and without the enforcement o f conditionalities) -as well as several alternative specifications - on poverty and inequality was simulated taking into account possible behavioral responses, such as children's participation in schooling. Specifically, these impacts were simulated for the various scenarios described inTable 3 below. I Families with Basetransfer Household per Variable transfer capita income smaller than R%100 O<=HPIC<=50 50< HPIC<=100 Bolsa Familia: R$50 0 Can get R$15 for each child from 0-15 years Scenario 1 old, upto R$45 total variable transfer. Bolsa Familia: R$50 0 Can get R$15 for each child from 0-15 years Scenario2 old, up to R$45 total variabletransfer. School aged kids 7-15 must attend school and stay out ofthe labormarketto receivethe transfer. Scenario3 R$50 0 Can get R$30 for each child from 0-15 years old, up to R$90 total variable transfer. School aged kids 7-15 must attend school and stay I out ofthe labor marketto receivethe transfer. Scenario4 R$25 25 Can get R$15 for each child from 0-15 years old, upto R$45total variable transfer. Scenario5 R$25 25 Can get R$15 for each child from 0-15 years old, up to R$45 total variable transfer. School aged kids 7-15 must attend school and stay out ofthe labormarket to receivethe transfer. The results of these scenarios are presented inTable 4 below. For Scenario 1- the current Bolsa Familia specifications but without conditionalities of school enrollment/attendance enforced - the estimated impacts are quite impressive, particularly for the extreme poor. Under this scenario, extreme poverty would be almost halved, falling from 14.5% to 8.1%. Full poverty would fall by less, from 33.3% to 29.9%, mainly due to the fact that the "moderately poor" (those with household incomes per capita between R$50-100) would receive a much smaller transfer (only the small variable transfer, which i s not large enough to raise them above the poverty line) and because 11% of the moderately poor would be ineligible for this variable transfer since they do not have children. Nonetheless, the depth o f overall poverty (Pl) would fall significantly, as both the moderate poor and the extreme poor would see their incomes rise closer to the poverty line. Moreover, the overall severity of poverty would fall by almost a 86 half (from 9.4% to 5.4%). Inequality would also fall significantly, with the Gini Coefficient dropping from 0.595 to 0.578 (which is indeed significant given the resilience of Brazil's inequality over time). In fact, as shown in Table 4, these existing specifications o f the BFP outperform alternative scenarios in terms of their potential impact on poverty and inequality. The impacts of the BFP on poverty and inequality would be slightly less impressive if conditionalities were enforced (Scenario 2), since families whose children do not end up enrolling in school would be penalized (not eligible for the transfers). However, the conditionalities would generate other positive impacts, such as the reductions in child labor and the increases in educational attainment (as discussed above). The slight muting o f potential impacts on poverty and inequality due to the enforcement of conditionalities could be mitigated somewhat if the variable transfers were increased (Scenario 3), or if the basetransfer was reduced, butalso available to the moderately poor (Scenarios 4 and 5). Table 4 -PotentialImpactof I the BolsaFamilia Program on Poverty and Inequality Povertyreductionafter thetransfer I 2002 - actual Scenario 1 Scenario 2 Scenario3 Scenario4 Scenario! 8.1% 8.8% 7.5% 10.0% 10.0% 2.5% 3.0% 2.4% 4.1% 4.1% 1.1% 1.5% 1.1% 2.4% 2.4% 29.92% 32.60% 31.40% 31.55% 31.58% IO,e5% 1150% 10.22% 11.88% 11.91% 5.37% 5.94% 5.04% 6.79% 6.81% 0.6040 0.6180 0.5985 0.6398 0.6405 0.6880 0.6940 0.6643 0.6982 0.6965 2.0842 2.0966 2.0759 2.1016 2.1021 0.5787 0.5819 0.5767 0.5838 0.5840 Meanper capita income R$ 327.15 331.10 330.28 331.66 329.97 329.94 ~ PopulationinMillions 167.72 167.72 167.72 167.72 167.72 167.72 Totaltransfer per year In mllllons of Reds 7,967.72 6,308.47 9,079.81 6,682.16 6,613.72 Average transfer per household RS - 63.13 42.12 60.62 33.84 33.44 N& I ScenibI mc~moulofm~tnn~hri.~~dndwndtlan~loflh~childnnbahavlour . n d 2 S ~ n ~ B 2 , h ~ l m a v n t o f t h l I n n ~ b t l ~ w n d H ~ n r l ~ f t h ~ ~ l l d n n ~ r v i o u r 3Scm~bJ.mc~mountof~~tnn~krhurnd~Mnllofthcchiidran khlY(aUI Rt30rea~~psr~lYmd~il1ng~qual1oRt1IO(m~ts~dafR$%)wR~90(in~1e~dafRf45) 4 Scenrb4 meamountDIMCt n n s h r i.nmd and nd WndHionaIOfthe Children behaViOUrvlth dMLRn1ccillngand al least a I n n h iof R$25 Io, all POI hourholds 5Sccodo5 m ~ ~ m o u n t o f M ~ t n n ~ b r I ~ w n d H b n d o f 1 kb~hIYDU~*nhdiPRntulll~~llnd ~ i l d n n .t~a.1~V.n~~r~fR$25(ar.IIpM1houuhold. Source: World Bank calculations, PesquisaNacional por Amostragem de Domicilios 2002 EVIDENCE IMPACTFROMOTHERPROGRAMS OF Conditional cash transfer programs (CCTs) have becomewidespread in Latin America and the Caribbean (LAC). An interesting feature o f the implementationof CCTs in LAC i s that most were accompanied by rigorous impact evaluation studies in their pilot phases. In Brazil, a rigorous evaluation of the Bolsa Alimentaqgo program suggests that a small monthly transfer of about R$30($12) per family has increased food consumption by approximately 13%. The results also show significant improvements in diet quality due to a disproportional increase in beneficiaries' consumption of fruits and vegetables. Results on child malnutrition were mixed, however. While for one study subpopulation the results suggest a positive impact on child growth, for another subpopulation the estimated impact was negative. As the authors of the study conjectured, the subpopulationthat exhibited negative impact on child growth entered early on inthe program, when rules were still not clear, and there was still a perceptionthat beingmalnourished was a precondition to eligibility. This misperception might have generated a perverse incentive to slow the growthof beneficiary `*Olinto, Pedro, Saul Morris, Alinne Veiga, and Rafael Flores (2004). "The Impact of the Bolsa Alimentaqb Program on FoodConsumption, DietDiversityandChildGrowth." Mimeo. 87 An ex-ante evaluation of Bolsa Escola (Bourguignon, Ferreiraand Leite, 2003) indicates that 60% ofpoor children ageing 10-15 not enrolled inschool would have entered school ifthey had accessto the program. The results indicate, however, that the impact on poverty would be a minor one. The authors predict that as conceived at the time, Bolsa Escola would reduce the incidence o f poverty by approximately 1 percentage point, and the Gini coefficient would fall byjust a half a point. This is because the per family transfers under Bolsa Escola were quite small (and much smaller than those under the new BFP). In Mexico, a thorough evaluation of the PROGRESA program has indicated that the program's interventions reduced poverty by about 10%. It is also estimated that the depth o f poverty was reduced by 30 percent, and the severity o f poverty by 36%. In addition, PROGRESA appears to have had a positive enrollment effect for boys and girls inprimary and secondary schools. At the primary school level, where enrollment rates before PROGRESA were already between 90% and 94%, the results revealed that PROGRESA succeeded at increasing the enrollment rate of boys up to 1.07% and of girls upto 1.45%. At the secondary school level, the increase in enrollment effects for girls ranged from 7.2 to 9.3 percentage points and for boys from 3.5 to 5.8 percentage points. The research findings also suggest an overall increase in educational attainment of about 10%. PROGRESA students are entering school at earlier ages, experiencing less grade repetition, and better grade progression. PROGRESA students have lower drop- out rates and higher school re-entry rates among those who had dropped out. The program is especially effective in reducing drop-out rates during the transition from primary to secondary school. The results also show reduced labor-market participation o f PROGRESA children for both boys and girls. Labor- force participation for boys shows reductions as large as 15 to 25% relative to the probability of participatingprior to the program. For girls, there are also significant reductions in child labor associated with PROGRESA. PROGRESA's evaluation studies also indicate that improved nutrition and preventative care in the program's areas have made younger children more resilient against illness. PROGRESA children aged 0- 5 exhibit a 12% lower incidence of illness than non-PROGRESA children. In addition, adult members in beneficiary households were found to be significantly healthier. On average, PROGRESA adult beneficiaries have 19% fewer days of difficulty with daily activities due to illness than non-PROGRESA individuals, 17% fewer days incapacitated and 22% fewer days in bed. Visit rates to clinics in PROGRESA communities grew faster than in control areas. In addition, there was a significant increase in nutrition monitoring visits, immunization rates and prenatal care. PROGRESA has also had a significant impact on increasing child growth and in reducing the probability o f stunting for children in the critical age range o f 12 to 36 months. Estimates imply an increase o f about 16% in mean growth per year. FINANCIAL SUSTAINABILITY OF THE BOLSAFAM~LIA PROGRAM With respect to fiscal sustainability, the program is expected to increase in size and resource needs. As shown inAnnex 1, the program is expected to reach 11.2 million families by the end o f the project period (2006), up from 3.6 million families at the start of 2004. A significant share o f these beneficiaries will be converted from existing pre-reform programs, so the marginal increase i s smaller than ifthe beneficiaries were entirelynew. Nonetheless, the programwill require a significant but managed increase inresources, from R$4.3 billion in 2003 (even less for the pre-reform programs in 2002) to R$9.4 billion by 2006. In terms of shares of total federal primary expenditures and GDP, the program will remain relatively small, rising from 1.1% of primary expenditures in 2003 to 2.5% by 2006 and from 0.2% of GDP in 2003 to 0.5% by 2006. This rise in spending needs will likely be offset by expected savings from the recently- approved pension reforms, estimated at 0.4% o f GDP (or about R$8 billion) per year during that period. The shift inresourcesfrom regressive pensions to progressive conditional cash transfers under the BFP is an important part of the Government's strategy to reduce poverty and inequality in Brazil. Efficiency gains will arise from merging the programs, and will also partially offset these higher costs (thus allowing 88 for greater coverage). The program is also seeking better complementarity and a pooling of resources with state and local programsfor expanded coverageto further expandthe pool of available resources. 89 Annex 10: SafeguardPolicyIssues BRAZIL: BRBolsaFamilia Indigenousand QuilombolaPeoplesDevelopmentPlan(Summary) Introduction The goal of the Bolsa Familia Program (BFP) is to reach 11.2 million poor families by 2006. The program's key objective is to associate to the income transfers, conditionalities that allow access to sustainable income earning opportunities for these families. Given the scope o f the Bolsa Familia Program, this requires that its implementation respects that ethnic and socio-cultural specificities of differentgroups of beneficiaries (e.g., indigenous andquilombolas). As part of preparing the proposed project to support the BFP, consultations were carried out with representatives of the indigenous and Quilombola (afro-descendents) peoples and a diversity of anthropologists specialized in indigenous and Quilombola issues. Representatives of the Government (including MDS) participated in the consultations with representatives o f indigenous and Quilombola peoples. The purpose o f the consultations was to collect recommendations on: (a) ensuring that indigenous and Quilombola groups benefit from the BFP in a culturally-appropriate way; and (b) monitoring and evaluating the potential impact of the BFP in these groups. The results of these consultations form the basis for this IPDP (summary), and the recommendations from beneficiaries and specialists are included in this document. To facilitate dissemination and disclosure, the IPDP was prepared in Portuguese, with an executive summary in English. Annexes include information on the differentsets of consultations. The complete IPDP is available inproject files. Finally, one consideration that permeated the aforementioned consultations as well as project preparation, and is worth reiterating in this summary is the fact that the Bolsa Familia i s a flagship program of the Federal Government o f Brazil. The BFP was already established by law (January 9, 2004), but the regulations (regulamento) for guiding the implementation of this law are currently being drafted. This context provides important parameters for both the opportunities and limitations in terms of the recommendations presented. Furthermore, it should be recognized that many of the potential issues highlighted during the consultations equally affect other groups living in remote rural areas and, as such, the recommendations made duringthe consultations should also benefitthese groups. Background It is estimated that there are about 220 different indigenous groups, speaking 180 languages, and totaling 370,000 people living under diverse conditions in Brazil. As for Quilombola peoples, the statistics are far less accurate, with estimates ranging from 1,264 communities (Fundaqiio Palmares) to 4,000 communities (ABA CRER) in 2003, located in rural areas of Amazonas, Maranhiio, Sergipe, Bahia, Pernambuco, Mato Grosso, S2o Paulo, Goihs, and Minas Gerais. However, only a small sub-set of these are officially recognized communities. LegalFramework Indiaenous Peoples. Up to the 1980s, the Brazilian Government had promoted linguistic and cultural integration of all the population into a presumed single national identity. To a large extent, the approach since the colonial period has been the establishment of a homogeneous Brazilian identity, subsuming into it all cultural specificities that came from different groups in one o f the most diverse societies in the 90 world. This stance changed with the 1988 Federal Constitution, Article 231, which moved from considering Indigenous Peoples as an "almost extinct category" to recognizing them as a differentiated ethnic group with subgroups with their own particular social organization, language, culture, customs, traditions, and beliefs. The 1988 Constitution establishes the rights o f indigenous peoples through: (a) recognizing their different social organization, customs, languages, beliefs and traditions, as well as the rightsto their land; (b) delegating to the Government the obligation to protect manifestations of local culture, including that o f indigenous and Afro-Brazilians; and (c) determining that fundamental (basic) education be available intheir native languages. Inaddition to the Constitution, indigenous rightsare further established inBrazil's EducationLaw (Leide Diretrizese Buses,LDB), by Decree 26/1991, inthe Resolutionofthe National EducationCouncil 03/99, by the National Education Plan (law 10.172/01), and in Law 10.558/2002. These laws ensure that indigenous peoples should have access to quality fundamental education that is culturally appropriate and bilingual, and that indigenous groups have the right to formulate their own pedagogical process. Finally, these laws also establish indigenous groups' accessto higher education. Similar to education legislation, there is extensive legislation granting indigenous peoples access to culturally-appropriate health care. Law 9.836/1999 establishes a sub-system for health care under the Sistemu Unico de Sazide (SUS). A series of other Decreesand Portarias from 1999to 2004 deliberate on the conditions of indigenous healthcare, andfurther differentiatethe rightsofthese groups. Finally, Brazil also has legislation granting and protecting indigenous rights to their land, as well as procedures for establishing indigenous territories. Quilombolas. Although Quilombos have had a formal recognition of their cultural, economic, and historic rights, these have not been easily translated into titling o f the land where they have lived and survived for several centuries. Since the Afonso Arinos Act o f 1951, racial discrimination is considered a criminal act, and according to Article 5 o f the Federal Constitutional of 1988, "racism is a crime for which bail i s not available." Paragraphs 1 and 2 o f article 212 read that "the state shall protect all expressions of popular Indigenous and African culture as well as all o f those that contributed to Brazilian civilization," whereas Article 68 of the transitory constitutional provision states that "the definitive ownership of the descendantsof the Quilombos to the land they are occupying will be recognized and the state shall make efforts to issue them titles to their land." The Ministry o f Policies to Promote Racial Equality (SEPPIR) and the Palmares Cultural Foundation, in the Ministry of Culture, in conjunction with other ministries (and an inter-ministerial commission), have been involved in promoting policy solutions related to Quilombos. Cultural Appropriateness and Indigenousand QuilombolaParticipationinBolsaFamilia Program BolsaFamilia is a federal programthat combines four cashtransfer programs-Bolsa Alimentaqgo, Cartgo Alimentaqiio, Bolsa-Escola and Auxilio Gas-into a single conditional cash transfer program. The objectives of the BFP include: (a) reducing poverty and inequality today, through the provision of direct monetary transfers to poor families; and (b) reducing poverty and inequality tomorrow, by providing incentives and conditions for investments in human capital on behalf of beneficiary families, and by linking beneficiary families to complementary services that could help them invest and grow out of poverty inthe future. The BFP defines two target groups. The priority target group is those "extreme poor" families with per capita monthly incomes below R$50 (US$17). The program also targets those "moderately poor" 91 families with per capita monthly incomes below R$100 (US$34)49but higher than R$50(US$17). These groups are likelyto include a significant proportion of the indigenous and Quilombola peoples. The BFP uses the "Cadastro Unico," a national database of potentially poor households, to identify beneficiaries. Current eligibility criteria are based on per capita income of family, though these will be revised to incorporate a set of proxy indicators (cesta de indicadores multidimensionais) since experience shows that the use of self-declared incomes is not very accurate. The program will provide cash transfers ranging from R$15-95 (US$5-33) to target families5' The exact transfer amount will depend on income levels and household composition. Monthly transfers are expected to average about R$71 (US$24) per beneficiary family. Payments are preferably made to the mothers, as per the Bolsa Familia law. The transfers would be conditional on all relevant family members complying with key human development conditionalities, such as school attendance, children's vaccinations and health care consultations, and pre-natal and peri-natal care. Important synergies are expected from the simultaneous promotion of investments in health and education within a family. In some areas (such as remote and/or indigenous and Quilombola areas), the supply of adequate or the availability of culturally-appropriate health and education services may be lacking. The BFP is developing a policy to define co-responsibility requirements inthese exceptionalcases. While BFP is designed to be inclusive o f all families identified under its pre-determined criteria, if not carefully considered, certain cultural differences among the indigenous and Quilombola may hinder their access to this program. Consultations with indigenous and Quilombola representatives and with anthropologists highlighted some areas for consideration in both program design and implementation. The consultations all emphasized that the Bolsa Familia Program should not be considered a comprehensive solution to the problems faced by these peoples. Key structural reforms and programs are needed. In addition, with respect to the program itself, the following are the key concerns and recommendations presentedduringthe consultations. Conceut of Family. A key aspect o f the BFP i s the identificationo f the family as a unit of eligibility as well as to determine the amount o fthe benefit. However, the concept o f nuclear family is not necessarily culturally appropriate for indigenous and Quilombola peoples and its application should be cautiously thought out during the revision o f the Cadastro Unico. While "misclassification" of a family is a potential problem, a greater concern is that, due to the ceiling in the benefit per family granted by the program, some families may feel encouraged to disintegrate into smaller units so as to receive more benefits. On the basis o f the consultation discussions, specific recommendations include: 0 Beneficiary families should be defined as production and consumption units, not by physical residency. The nuclear family i s the best reference for this definition but there should be flexibility; and The definition o f a beneficiary family should be done basedon consultations with the concerned community, community associations and/or organizations, or through specialized consultants. Eligibility criteria. As previously stated, the current eligibility criterion for the BFP is poverty as measured by the per capita income o f the family. The Government already recognizes that self-declared income is not an appropriate measure and i s inthe process o f developing a set o f proxy-means indicators 49Brazil does not have an official poverty line and researchinstitutions use different poverty lines that lead to different poverty numbers. The poverty line used in the World Bank's study is set at around R$80 per month (USS27) and is adjusted for each region using a regional price index. This resultsin anationalpoverty rate of 28.7%. IPEA'sextreme povertylinewas establishedat R$62, yielding an extremepoverty rateof 14.6%. IPEA calculates the full poverty line as twice the extreme poverty line, or R$124, resulting in a poverty rate of 33.6%. This value is close to that beingusedby the Government intheir documents. 50The US dollar amount used hereinand thereafter isjust as areference and it is basedonthe exchangerate of US$l =B2.90. 92 (cesta de indicudores multi-dimensionais) to determine eligibility. In order to ensure cultural appropriateness o f the indicators selected, it was suggestedthat the definition take into consideration the general situation o f indigenous and Quilombola communities. For this, consultations with representatives of these groups may be appropriate and the criteria established should be transparent to all. Another issue that was raised during consultations is the prioritization of beneficiaries on the part of the Government due to limitedresources duringthe next couple of years. Given the internal closeness of most of these communities, denying benefits to one potentially eligible family while granting benefits to another may lead to problems and even disintegrationwithin the community. Furthermore, although others supported a criteria-based selection for all BFP beneficiaries, indigenous and Quilombola representatives defended that all indigenous and Quilombola families should benefit from the program. Giventhe aforementioned concerns, the following specific recommendations were made: 0 Eligibility criteria should be broader than the family unit, and culturally appropriate to the conceptso f indigenous and Quilombola populations; 0 When selecting the new set of proxy-means indicators to determine eligibility criteria, the Government should consider including indicators that are more appropriate to the socio-cultural reality of indigenous and Quilombola populations; 0 Priority criteria should be defined to determine which municipalities / communities will receive benefitfirst; 0 All eligible beneficiaries within an identified indigenous and Quilombola community should receive the benefit to avoid creating divisions within a community [indigenous and Quilombola leaders defended that absolutely all families within a community should receive the benefit, regardlesso f eligibility]; and 0 The community should be consulted, if exceptions are to be made, and all decisions should be transparent. Cadastro Unico. Under the BFP, all potential beneficiaries are registered under a national household registry (the Cudustro Unico) which serve as the basis to determine family eligibility. The enrollment of indigenous and Quilombola families into the Cadastro Unico raises several challenges. Currently, the Government is still in the process o f fine-tuningthe Cadastro to improve the system of identifyingtarget populations. Municipalities generally collect and enter the data, and the Caixa EconGmica Federal (CEF) consolidates the Cadastro database, while the SENARC inthe MDS oversees the whole process. A pilot of the Cadastro for indigenous populations was carried out in Parani, with the National Health Foundation (FUNASA) in charge o f registering the indigenous groups. The results of that pilot are currently under evaluation. Nonetheless, several issues regarding the Cadastro Unico that could particularly affect indigenous and Quilombola peoples were raised during the consultations. First and foremost, there i s a concern that municipal governments may taint the registration process, as indigenous and Quilombola groups are not traditional members o f the local political electorate. Second, the Cudastro itself may not be a culturally appropriate instrument, as many o f the questions in the questionnaire are based on concepts that are biased against these cultures. Additionally, many of the terms used in the Cadustro may not be appropriate for these cultures. In addition to the specific recommendations made to the instrument itself, detailed inthe full IPDP report presentedto the Government, the following suggestions were made: 0 Register (cudustrar) all indigenous and Quilombola families inthe Cadastro and only then, using the agreed eligibility criteria, select beneficiaries; 93 Either develop a new registry instrument designed specifically for indigenous and Quilombola groups or revise the current instrument to be more culturally appropriate for these groups. Involve representatives from these groups inthe process; Use community health agents (Agentes Comunitarios de Safide) to register indigenous group^;^' Use Quilombola people to carry out the registration of the Quilombola peoples; Provide specific training to agents responsible for registering indigenous and Quilombola families so that they not impose their own cultural references inthe process; Disseminate information about the program, including eligibility criteria, benefits, and conditionalitiesto all groups including indigenous and Quilombola populations; and Processthe Cadastro Unico inthe Indigenous Sanitary Districts (Distritos Sanitcirios Indigenas, DSEI),rather than at the municipal level. Benefits. As discussed above, under the BFP, families receive a monthly stipend that varies between R$15 and R$95, dependingon the family's per capita income and the number o f children between 0 and 15 and pregnant family members. Payments under the BFP are made on a monthly basis. Beneficiaries withdraw benefits from a pooled account usingelectronic benefit cards (EBC) distributedby the CEF (the law stipulates that EBCs are preferably distributed to women in the household). In the case of indigenous and Quilombola beneficiaries, the following recommendations were made and incorporated into the project: The definition of the person responsible for receiving the resources should respect the family organization and traditions. This meansthat inmany instances, it may not be the female head of household; Access to benefits by beneficiary families should be guaranteed, such as transportation to the city to withdraw the resources;52 Provide training to staff from CaixaEconGmicaFederal so as to improve service to Quilombola peoples and reduce discrimination o f these groups; and Beneficiary families shouldhave autonomy inmanaging their resourcesand there should be some flexibility as to the timeframe to withdraw resources as monthly withdrawals may be difficult in remote areas (the program currently requires withdrawals to be made within a 90-day period). Conditionalities. The specific conditionalities for beneficiary families include: (i)children untilthe age o f one must have seven doctor visits per year; (ii)children between 0-7 years o f age must have all of their vaccines up to date; (iii)children between 7 - 15 must fulfill a school attendance requirement; (iv) pregnant women must have six pre-natal visits, one peri-natal visit, and vaccines up to date. Compliance with these conditionalities is to be verified through school enrollment and vaccination records. During the consultation process, it was agreed that the conditionalities should be maintained for all beneficiaries of the BFP, but that there should be flexibility in interpreting and enforcing the requirements for "InthepilotcarriedoutinParankthoseagentswereusedtocollectinformation. Theuseofcommunityhealthagentsisalsocommonpractice for the CadastroUnico. "Thisisaproblemaffectingallremoteruralcommunitiesand, assuch, alternativesarebeingconsideredbytheGovernment. As such, the proposedprojectwould support the developmentof a strategyfor operatingthe program in remoterural areas (including remote indigenous and Quilombolacommunities; see Component4). 94 indigenous and Quilombola populations. This is particularly true because of supply constraints and cultural differences. The following recommendations were made: 0 Beneficiaries should not be penalized for non-compliance when schools are not available or are culturally-inappr~priate;~~ 0 Interpretationof pre-natal care conditionality should be culturally-appropriatewithout stipulating the number of visits and allowing for the choice usingtraditional mid-wives and ensuring that the visits include appropriate vaccines and lab tests; and 0 Verification of conditionalities should be conducted by the communities themselves (or trained agents) so as to ensure culturally-appropriateinterpretation intheir application. ProgramStrategiesfor the Emancipapjo ofBeneficiaryHouseholds The BFP contemplates evaluating and identifying innovations in "emancipa@o" and "bridging" approaches and service delivery. In this context, the following recommendations were made during consultations and incorporated into the project: 0 EmancipaqLio approaches to avoid creating dependency should be designed so as to respect the culture and organization o f indigenous and Quilombola peoples, their autonomy o f decision, and their environment; 0 Complimentary programs and services should be planned parallel to the implementation of the BFP ina culturally-sensitive way; and 0 Ensurethat programeligibility and re-certificationcriteria aretransparent andthat indigenous and Quilombola communities understand them. Conclusions As ethnic minorities, indigenous and Quilombola peoples find themselves at a disadvantaged position at terms of opportunities and quality o f life prospects. While the BFP is designed to benefit all Brazilians that fit the eligibility criteria ("the poor"), such social and cultural differences may create entry barriers for these groups and put them at a disadvantage to benefit from the program. Consultations with anthropologists and beneficiaries pointed to six general areas o f concern: (i)the concept of family; (ii) eligibility criteria; (iii)Cadastro Unico; (iv) benefits; (v) conditionalities; and (vi) program exit. While there was consensus that the overall program design is applicable to indigenous and Quilombola groups, specific recommendations were made to improve their access to the program. In addition, the consultations all emphasized that the Bolsa Familia Program should not be considered a comprehensive solutionto the problems faced by these peoples. Key structural reforms and programs are needed. ProposedProjectActions to Address Recommendations The full IPDP was presented and discussed with the Government (which also participated in consultations), which agreed to consider the recommendations while fine-tuning the Bolsa Familia Program. Additionaily, in the context of the technical assistance components o f the proposed loan (US$15 million), the following recommendations have been incorporated into the project: 53Indigenous peoples are entitled by law to have access to indigenousschools. However, in the case of Quilombola there is no such guarantee andregularschools are often culturally inappropriate. 95 Ensurethat the updatingofthe CudastroUnico will consider cultural differences, including inthe selection of indicators that will be usedto determine beneficiary eligibility. Train agents responsible for registering indigenous and Quilombola families so as to ensure cultural sensitivity and knowledge inthe process. Ensure proper dissemination of BFP information, including eligibility criteria, operations, and conditionality, specifically tailored to indigenous and Quilombola publics; Continuously monitor and evaluate the BFP to ensure that indigenous and Quilombola communities have access and benefit from the program ina culturally appropriate manner and the that the program does not have unexpected negative effects onthe community; Provide technical assistance to the Govemment so that it develops a strategy to ensure that payments are accessible to groups living in reach remote areas, including indigenous and Quilombola communities; and Establish a monitoring mechanism that includes representation from indigenous and Quilombola peoples. Specific Monitoringand EvaluationIndicators The proposed project includes a component (Component 3) that seeks to support the whole BolsaFamilia Program (including the Bank's contribution to it) in designing and implementing a results-based monitoring and evaluations system aimed at: (i)monitoring the effectiveness of program delivery, (ii) monitoring processes and instituting quality controls aimed at promoting public accountability and legitimacy, and (iii) monitoring outcomes and assessing the program impacts at both the territorial and household levels. The Monitoring and Evaluation Component includes four sub-components, each with specific activities related to ensuring the appropriate participation of indigenous and Quilombola families in the program, monitoring their participation, and evaluating impact. Whenever possible, data collection and evaluation will be disaggregated by ethnic categories and qualitative research will specifically monitor implementation process by sub-groups, including indigenous and Quilombola. The following are specific monitoring and evaluation activities planned under the project, which will contemplate indigenous and Quilombola communities: Preparation o f regular newsletters and reports with detailed information on program delivery sent to representatives o f indigenous and Quilombola groups. Design and execution of random quality control reviews to check whether beneficiary identification for indigenous and Quilombola families follow adequate program guidelines, and whether conditionality compliance as reported is credible. Promotion o f citizen oversight and social control through the dissemination of detailed program informationto indigenous and Quilombola stakeholders. Design and implementation o f in-depth case studies to describe and analyze specific social control manifestations. 96 0 Analysis o f territorial level outcomes, including specificities of indigenous and Quilombola areas (quantitative data will be complemented by the use of focus groups, including in these communities). 0 Analysis of family level outcomes usinga combination of quantitative and qualitative methods. This analysis will enhance the understanding of the impact of BFP on key culturally-appropriate social indicators and the interactions that the programhas with other aspects of living conditions, specific to indigenous and Quilombola families. Cost Estimatesand FinancingPlan The Bolsa FamiliaProgram i s estimated to cost the Government of Brazil approximately US$6.5 billion over the proposed project period (mid-2004 to end-2006). The proposed project will finance US572.2 million (9.3% of the total program budget). O f this, US$551.5 million (Component 1) will be used to reimburse the Treasury for the direct financing of the BFP (hence the expenditure category is conditional cash transfers or grants). Another US$15 million (Components 2-5) will be used to provide technical assistance (and support to project management) to improve the effectiveness and efficiency o f the program. The Bank's participation with a loan size that is only a small share o f the program cost, nevertheless, will allow for a scale of impact on the content and performance o f the entire program. It is in this context that the proposed project will carry out a series of activities that will contemplate indigenous and Quilombola groups, to ensure that they benefit from the BFP in a culturally-appropriate way and to monitor the program's impact on those groups so as to prevent any negative effect. While costing out the proposed project, the preparation team identified all the specific activities (i.e. consultancies, workshops, services, etc.) which should consider and/or benefit indigenous and Quilombola communities, among other beneficiaries. From the US$15 million in technical assistance, approximately US$8.7 million i s allocated to those activities. The itemized list is included in project files and summarized below. Total, US$MillionIncluding Contingencies StrengtheningSystemfor IdentifyingTarget Population 0.8 DevelopingaMonitoring and EvaluationSystem 5.0 InstitutionalStrengtheningfor ServiceDelivery 2.9 Total ProiectCosts-TechnicalComuonents 8.7 97 Annex 11: Project Preparation and Supervision BRAZIL: BRBolsaFamilia Key institutionsresponsiblefor preparationof the project: Secretaria Nacional de Renda e Cidadania do MinistCrio de Desenvolvimento e Combate 6 Fome, in collaboration with the World Bank, and with assistancefrom DFIDand IFPRI. Bank staff and consultantswho worked on tl projectpreparation: Name Specialty Kathy Lindert Team Leader I Senior Economist Maria Madalenados Santos Sector Leader UlrichZachau Operations Director Marta ElenaMolares-Halberg LeadCounsel Joachim von Amsberg LeadEconomist Mary Sheehan LeadOperationsOfficer Efiaim Jimenez LeadProcurementSpecialist Luiz Gazoni Senior ProcurementSpecialist Zeze Weiss Civil Society Specialist