Document of The World Bank IMPLEMENTATION COMPLETION AND RESULTS REPORT (IBRD-70370) ON A LOAN

Transcription

1 Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Document of The World Bank Report No: ICR IMPLEMENTATION COMPLETION AND RESULTS REPORT (IBRD-70370) ON A LOAN IN THE AMOUNT OF EUR MILLION (US$ MILLION EQUIVALENT) TO THE FEDERATIVE REPUBLIC OF BRAZIL FOR A LAND-BASED POVERTY ALLEVIATION PROJECT I Sustainable Development Department Brazil Country Management Unit Latin America and Caribbean Region June 29, 2009

2 CURRENCY EQUIVALENTS (Exchange Rate Effective December 31, 2008) Currency Unit = Real (R$) R$1.00 = US$ US$ 1.00 = R$ FISCAL YEAR January 1 December 31 ABBREVIATIONS AND ACRONYMS BB BNB CAS CdaT CEDRS CF CMDR CONTAG CPS ESALQ FEALQ FECAMP FETAG ICB IICA INCRA MDA MIS NCB NEAD NTU PNAD PPA PRONAF RPAP RPRP SAT SIC SOE STU TJLP UNICAMP Bank of Brazil Bank of the Northeast of Brazil Country Assistance Strategy Land Reform and Poverty Alleviation Pilot Project (Cédula da Terra) State Sustainable Rural Development Council Crédito Fundiário Municipal Rural Development Councils National Confederation of Agricultural Workers Country Partnership Strategy School for Agrarian Studies Luiz de Queiroz Foundation for Agrarian Studies - Luiz de Queiroz Economics Foundation of University of Campinas (State of São Paulo) State Federations of Agricultural Workers International Competitive Bidding Inter-American Institute for Cooperation on Agriculture National Institute for Colonization and Agrarian Reform Ministry of Agrarian Development Management Information System National Competitive Bidding Center for Agrarian Studies (Ministry of Agrarian Development) National Technical Unit National Household Survey Government Multi-year Development Plan National Program to Strengthen Family Agriculture Rural Poverty Alleviation Program Rural Poverty Reduction Program Land Acquisition Subproject Community Investment Subproject Statement of Expenditure State Technical Unit Long-term Interest Rate University of Campinas (State of São Paulo)

3 Vice President: Pamela Cox Country Director: Makhtar Diop Sector Manager: Ethel Sennhauser Project Team Leader: Jorge A. Muñoz ICR Team Leader: Jorge A. Muñoz

4 Brazil Land-based Poverty Alleviation Project I CONTENTS Data Sheet A. Basic Information B. Key Dates C. Ratings Summary D. Sector and Theme Codes E. Bank Staff F. Results Framework Analysis G. Ratings of Project Performance in ISRs H. Restructuring I. Disbursement Graph 1. Project Context, Development Objectives and Design Key Factors Affecting Implementation and Outcomes Assessment of Outcomes Assessment of Risk to Development Outcome Assessment of Bank and Borrower Performance Lessons Learned Comments on Issues Raised by Borrower/Implementing Agencies/Partners Annex 1. Project Costs and Financing Annex 2. Outputs by Component Annex 3. Economic and Financial Analysis Annex 4. Bank Lending and Implementation Support/Supervision Processes Annex 5. Beneficiary Survey Results Annex 6. Stakeholder Workshop Report and Results Annex 7. Summary of Borrower's ICR and/or Comments on Draft ICR Annex 8. Comments of Cofinanciers and Other Partners/Stakeholders Annex 9. List of Supporting Documents MAP

5 A. Basic Information Country: Brazil Project Name: Land-Based Poverty Alleviation Project I Project ID: P L/C/TF Number(s): IBRD ICR Date: 06/30/2009 ICR Type: Core ICR Lending Instrument: SIM Borrower: FEDERATIVE REPUBLIC OF BRAZIL Original Total Commitment: USD 202.1M Disbursed Amount: USD 193.3M Revised Amount: USD 0.0M Environmental Category: B Implementing Agencies: Ministerio de Desenvolvimento Agrario Cofinanciers and Other External Partners: B. Key Dates Process Date Process Original Date Revised / Actual Date(s) Concept Review: 12/15/1998 Effectiveness: 10/16/2001 Appraisal: 12/19/1998 Restructuring(s): Approval: 11/30/2000 Mid-term Review: 11/07/2003 Closing: 09/30/ /31/2008 C. Ratings Summary C.1 Performance Rating by ICR Outcomes: Risk to Development Outcome: Bank Performance: Borrower Performance: Moderately Satisfactory Moderate Moderately Satisfactory Moderately Satisfactory C.2 Detailed Ratings of Bank and Borrower Performance (by ICR) Bank Ratings Borrower Ratings Quality at Entry: Moderately Satisfactory Government: Moderately Satisfactory Quality of Supervision: Moderately Satisfactory Implementing Agency/Agencies: Moderately Satisfactory Overall Bank Overall Borrower Moderately Satisfactory Performance: Performance: Moderately Satisfactory i

6 C.3 Quality at Entry and Implementation Performance Indicators Implementation QAG Assessments Indicators Performance (if any) Potential Problem Project No at any time (Yes/No): Problem Project at any Yes time (Yes/No): DO rating before Closing/Inactive status: Moderately Satisfactory Quality at Entry (QEA): Quality of Supervision (QSA): None None Rating D. Sector and Theme Codes Original Actual Sector Code (as % of total Bank financing) Central government administration General agriculture, fishing and forestry sector Other social services Roads and highways 20 5 Water supply 8 10 Theme Code (as % of total Bank financing) Land administration and management 17 5 Other rural development Participation and civic engagement Social safety nets 33 E. Bank Staff Positions At ICR At Approval Vice President: Pamela Cox David de Ferranti Country Director: Makhtar Diop Gobind T. Nankani Sector Manager: Ethel Sennhauser John Redwood Project Team Leader: Jorge A. Munoz Luis O. Coirolo ICR Team Leader: Jorge A. Munoz ICR Primary Author: Anna F. Roumani Jorge A. Munoz ii

7 F. Results Framework Analysis Project Development Objectives (from Project Appraisal Document) The project seeks to reduce rural poverty in 14 states of Brazil by increasing the incomes of about 50,000 poor rural and peri-urban families by extending the community-based approach to land acquisition and participation in complementary, demand-driven community subprojects. Revised Project Development Objectives (as approved by original approving authority) (a) PDO Indicator(s) Indicator Indicator 1 : Value quantitative or Qualitative) Baseline Value Original Target Values (from approval documents) Formally Revised Target Values Poverty alleviation impact as measured by household income growth Annual, gross family income of R$2,057.00, Not established at well below Federal appraisal Government poverty line. N/A Actual Value Achieved at Completion or Target Years (a) FEALQ (2005): Family income increased from R$1,656 (2003) to R$4,064 (2005) Date achieved 10/16/ /30/ /31/ /31/2008 (a) FEALQ '05: Gross income/fam/yr up 145% ; (b) BCR '09: Gross Comments income/family/yr aver. R$9,005. Net after payment land loan = R$5,216; and (c) (incl. % FECAMP '06 (on CdaT, indicative for CF): Aver. gross family income up 220% achievement) vs control 176%. Number of families benefited Indicator 2 : Value quantitative or Qualitative) Zero. (15,267 families already benefited under CdaT pilot in five NE states) 50,000 families in 14 states N/A 40,102 families fully settled; another 8,300 families approved and waiting financing; and 22,000 families with proposals registered. Date achieved 10/16/ /30/ /31/ /31/2008 Comments Achieved 80.2% of target (96.8% if 8,300 families with proposals approved are (incl. % counted, which will be 100% financed by Gov. of Brazil) achievement) Economic efficiency and financial viability of community subprojects as Indicator 3 : measured by IRR and repayment rate by beneficiaries of land loans Value N/A Appraisal IRR: N/A Actual IRR of iii

8 quantitative or Qualitative) Range 24% - 58%, Project IRR: 35% 12.6%. Land loan repayment rate in 11 states averaging 97.6% Date achieved 10/16/ /30/ /31/ /31/2008 Comments (incl. % achievement) PAD analysis was farm-model based. Final evaluation in 2010 will collect such data. ICR analysis based on CdaT income (net of other effects), assets, value of agric. production. Results indicative for CF: IRR 12.6%; NPV US$143.0 m. Indicator 4 : Cost effectiveness of the community-based approach as measured by financial costs associated with land purchase, subproject investments and other financing. Value quantitative or Qualitative) N/A PAD estimated a 40% cost saving by using community-based approach. N/A No data available on costs of other Govt. land reform programs Date achieved 10/16/ /30/ /31/ /31/2008 Comments (incl. % achievement) (a) Project approach not comparable to INCRA expropriation model due to lack of INCRA data and different target farms; and (b) project admin. costs are low, 90% of total costs passed to beneficiaries (b) Intermediate Outcome Indicator(s) Indicator Indicator 1 : Value (quantitative or Qualitative) Baseline Value Original Target Values (from approval documents) Formally Revised Target Values Actual Value Achieved at Completion or Target Years Family Farms established for about 50,000 beneficiary families in 14 states Zero 50,000 additional Family Farms N/A 40,102 family farms established (8,300 additional families approved) Date achieved 10/16/ /30/ /31/ /31/2008 Comments Achieved 80.2% of target. Around 39,000 families in Northeast (96%), and (incl. % 1,700 families settled in the four Southern States achievement) On-farm investment subprojects (SIC) implemented covering about 50,000 Indicator 2 : families. Value (quantitative or Qualitative) Zero 50,000 families with on-farm investments financed and installed. N/A 40,102 families benefited from onfarm (SIC) investments. Date achieved 10/16/ /30/ /31/ /31/2008 Comments (incl. % About 80.2% of appraisal target achievement) Indicator 3 : Evaluation studies to determine impact on beneficiaries. Value FECAMP studies (of One baseline and N/A (a) FEALQ baseline iv

9 (quantitative or Qualitative) CdaT) conducted in 1999 and four re-surveys survey of CF (2003) and interim re-survey (2005); (b) BCR survey of CF (2009); and (c) FECAMP resurveys of CdaT ( ). Date achieved 10/16/ /30/ /31/ /30/2008 Comments 100% of evaluation targets in slightly different configuration to appraisal plan. (incl. % Final evaluation of CF planned for achievement) G. Ratings of Project Performance in ISRs No. Date ISR Archived DO IP Actual Disbursements (USD millions) 1 06/29/2001 Satisfactory Satisfactory /21/2001 Satisfactory Satisfactory /20/2002 Satisfactory Satisfactory /26/2002 Satisfactory Satisfactory /21/2003 Satisfactory Satisfactory /26/2003 Satisfactory Satisfactory /18/2003 Satisfactory Satisfactory /18/2004 Satisfactory Satisfactory /01/2004 Satisfactory Satisfactory /29/2005 Satisfactory Satisfactory /29/2005 Satisfactory Moderately Unsatisfactory /17/2006 Satisfactory Satisfactory /09/2006 Satisfactory Satisfactory /29/2007 Satisfactory Satisfactory /03/2007 Satisfactory Satisfactory /12/2008 Satisfactory Satisfactory /14/2008 Moderately Satisfactory Moderately Satisfactory H. Restructuring (if any) Not Applicable v

10 I. Disbursement Profile vi

11 1. Project Context, Development Objectives and Design 1.1 Context at Appraisal The Land-based Poverty Alleviation Project, known in Brazil as Crédito Fundiário (CF), sought to address land access for the rural poor in 14 states of the Northeast and Southern regions. The project was preceded by two successful Bank-supported pilots: the first implemented as a component within the Ceará Rural Poverty Alleviation Project (Ln BR) in , and the second as the free-standing Land Reform and Poverty Alleviation Pilot Project (Ln BR), known in Brazil as Cédula da Terra (CdaT), implemented from 1997 to 2002 in five states of the Northeast Region. The pilots tested a community-driven approach to land reform under which eligible beneficiary groups negotiate directly with willing sellers to purchase suitable properties and then obtain financing for the purchase of the land, and grants for complementary on-farm investments and technical assistance to settle and launch basic productive activities. The project was intended to provide Government with another option for redistributing land to the rural poor at a time when the public and political rhetoric on landlessness was increasingly strident, matched by violence in the countryside Poverty in Brazil has a strong rural and regional dimension and poverty rates are much higher than in other countries with a similar per capital GNP. Limited access to land and extreme inequality of land ownership in Brazil are central factors contributing to rural poverty. 1 Brazil has one of the most unequal distributions of land ownership in the world. At the time of project appraisal, the Agricultural Census of 1996 showed 4.5 million rural households with insufficient land for subsistence and while these data did not show a direct relationship, evidence in Brazil suggested that the vast majority of the country s 16.5 million rural poor were likely to fall into that cohort. It was clear that strategies for rural poverty reduction needed to emphasize policies and programs to reduce asset inequality, with diverse interventions directed at different population groups At the time of Project appraisal, the Northeast Region was home to 60% of all Brazilian poor and about 69% of the country s rural poor. The region has a relatively poor natural resource base, variable climatic conditions including periodic drought, skewed land distribution, lack of basic socio-economic infrastructure and poorly-functioning rural financial markets. The project expanded geographic coverage to include five more Northeast states plus four states in the South/Southeast, a wealthier region overall where the quality of land and natural resources are generally better and agriculture is more capital intensive, but where poverty is similarly concentrated in rural areas and associated, inter alia, with limited access to land Government s Strategy: The 1990s saw increasing pressure on the Brazilian Government to address landlessness. Traditional, government-administered mechanisms were not meeting the challenge in terms of coverage, speed, cost-effectiveness, or sustainability. Landlessness moved to the forefront of Government s political and social agenda in response to rural unrest and calls for action from diverse and vocal grassroots organizations with growing national scope and influence. Government-administered land distribution in Brazil had 1 2 By 2000, barely 3% of the population owned over two-thirds of all arable land in Brazil. Participating states (14): (a) Northeast: Bahia, Ceará, Pernambuco, Piauí, Maranhão, Paraíba, Rio Grande do Norte, Sergipe, Alagoas and the northern part of Minas Gerais; and (b) South/Southeast: Paraná, Rio Grande do Sul, Santa Catarina and Espírito Santo. 1

12 traditionally focused on expropriation with compensation, and land settlement, both commonly associated with long delays, high costs, and political and legal conflicts. Total families benefited over the years were low relative to needs, beneficiaries waited a long time to get settled and costs per beneficiary were high. 3 Government also realized that rules governing land expropriation (including the minimum size of properties eligible for expropriation) were effectively excluding about 40 million hectares potentially accessible to poor, landless families under the tested pilot mechanism. Government was actively seeking more cost-effective, less conflictive and complementary alternatives Government s initial response included upgrading sector institutions, creating a Federal Ministry for Land Reform (subsequently re-named Agrarian Development, MDA), and steadily increasing the Federal land reform budget. Government s Multi-year Development Plan for 2000/2003 strongly endorsed efficient land reform within the national development strategy. In expanding the successful five-state experience under Cédula da Terra to 14 states in the Northeast and South/Southeast regions under Crédito Fundiário, Government envisaged a programmatic series of three operations 4 targeting about one-quarter million families over a decade, of which the new CF operation would serve about 20% or 50,000 families over a period of 4 years (from 2000 to 2004). Rationale for Bank involvement The Bank s involvement in further scaling-up this community-based land reform mechanism was justified as follows. First, the Bank s Country Assistance Strategy (CAS) for Brazil (20160-BR, March 30, 2000) identified the reduction of poverty and inequality as central objectives of the Bank s assistance efforts and recommended priority attention to the Northeast Region. The CAS recommended expanding the community-based, market-driven methodology under CdaT to the rest of the Northeast and other parts of the country, e.g., the South/Southeast. Second, the community-based approach to land reform under CdaT, complementing other, more traditional approaches, had positive results in terms of speed, lower cost per family, beneficiary participation and preliminary indications of impact and sustainability. At the time of appraisal, several thousand poor families with properties identified for purchase were already waiting to participate in the program. A land reform program building on the experiences of CdaT was seen as the natural, expanded continuation of initiative (see Box 1). Third, economic changes in Brazil had made land reform more feasible and politically palatable. Economic distortions historically promoting land concentration were alleviated; agricultural credit subsidies were cut, inflation had declined, and the tax on unproductive land was increased. Holding land as a financial hedge became less attractive, increasing the supply of land and reducing its price. In the Northeast, large tracts of land were available at low prices from owners, and from banks holding land as collateral for defaulted farm debt. Finally, global evidence showed that land reform could create a 3 Studies by FAO and the Ministry of Agrarian Development (MDA/NEAD) suggested post-adjudication costs under the traditional land reform approach in Brazil increased land expropriation costs up to three-fold. Despite legal requirements, families settled rarely repaid the cost of land expropriation and complementary investments (and thus did not get full ownership titles to the land which remained under the tutelage of the Agrarian Reform and Colonization Institute, INCRA), and the total cost to government for compensating expropriated landowners increased when they successfully pursued legal remedies. 4 In December 1998, COFIEX, the Brazilian Federal Committee for External Financing, approved a US$2 billion three-phase program, half of which would be financed by Bank loans, with a first loan of EUR 218 million for this operation. 2

13 sustainable source of income for beneficiaries. Studies advocated land acquisition by the rural poor using market-based transactions between willing parties, with self-selection driving participation. 1.2 Original Project Development Objectives and Key Indicators (as approved) As stated in the PAD, the Project Development Objective was to reduce rural poverty in fourteen states of Brazil by increasing the incomes of about 50,000 poor rural and peri-urban families through extension of the community-based approach to land acquisition and participation in complementary, demand-driven community subprojects Key performance indicators as approved were: 5 number of families benefited; economic efficiency and financial viability of community subprojects as measured by the Internal Rate of Return and the repayment rate by beneficiaries of the land loans; the poverty alleviation impact as measured by household income growth; and cost-effectiveness of the community-based approach as measured by financial costs associated with land purchase, subproject investments and other financing. 1.3 Revised PDO (as approved by original approving authority) and Key Indicators, and reasons/justification The PDO was not revised. 1.4 Main Beneficiaries The primary target group comprised 50,000 poor rural and peri-urban families in 14 participating states. Targeted beneficiaries were poor rural people without land or with insufficient land for subsistence. 1.5 Original Components The project had five components: (a) Component 1: Land Purchase Subprojects SAT (EUR million, 40% of total cost 6 ) was to be implemented with Federal Government counterpart (budget) funds, including the 5 PDO indicators listed in the main text of the PAD differ from those shown in Annex 1 Project Design Summary which are: income of families participating in the project in comparison with control groups and pre-project income levels; net economic benefits of increased agricultural production; and, pilot program tested and evaluated (referring to the newly-participating Southern states). 6 The loan was denominated in EUR. At signing in July 2001, the IBRD loan amount of EUR million represented approximately US$202.1 million. In September 2004, there was a partial cancellation of EUR million, resulting in a revised loan amount of EUR million, representing US$ 144 million at the time. In June 2009, total disbursements stood at EUR million, representing US$ million approximately. Given considerable fluctuations between EUR, US$ and R$ during the life of the project, cost estimates in US$ can lead to confusion. Thus, to the extent possible, the ICR cites cost figures in EUR and R$ only. 3

14 National Land Fund (Lei Complementar No. 93), and financed community-based land purchases for a targeted 50,000 rural families. The Federal Government and participating states would conclude a formal agreement on the annual allocation of resources for land purchases. Key criteria governing land acquisition are detailed in Annex 2. (b) Component 2: Community Investment Subprojects SIC (EUR million, 49% of total cost) would provide matching grant funds for on-farm investments identified by the respective association as priorities for community wellbeing and productivity growth. These included family housing, basic socio-economic infrastructure (access roads, fencing, water supply, energy, health posts) and productive activities (small-scale agro-processing, communal tractors and small irrigation schemes), and related technical assistance. See Annex 2. (c) Component 3: Community Development Support and Strengthening (EUR 13.3 million, 3% of total cost) would boost the capacity of community associations by providing technical assistance, training, best practice exchanges between associations (e.g., on operation and maintenance of subprojects) and between states, and seminars. This component would also finance information dissemination about the project, its guidelines and mechanisms to increase transparency, awareness and participation. See Annex 2. (d) Component 4: Project Administration, Supervision and Monitoring (EUR 26.8 million, 6.1% of total cost) supporting field supervision and monitoring as well as incremental operating costs of the National Technical Unit (NTU) and State Technical Units (STU) for overall coordination of project activities. The Ministry of Agrarian Development (MDA) would delegate project implementation to the participating states but retain responsibility for approving annual operating plans (POA) and resource allocations to each state. (e) Component 5: Impact Evaluation and Dissemination (EUR 8.3 million, 1.9% of total cost) financing studies designed to continuously calibrate overall design and implementation of what was envisaged as a phased land reform program. The National Confederation of Agricultural Workers (CONTAG), its state-level federations (FETAGs) and local unions were expected to actively participate in the evaluation program. The MDA would coordinate impact evaluation and dissemination Financing Package: Total available financing for a community association the sum of a loan for land acquisition, matching grants for productive investments, an initial cash grant for immediate settlement needs, and technical assistance had a ceiling of between R$12,000 and R$18,000 per family, depending on the State and region 7. The number of community members determined the size of the total package for the community association. One innovative feature of this package is the direct link between the loan for land acquisition and the grant for everything else, providing a powerful incentive for beneficiaries to negotiate a good price for the land. The lower the negotiated land price, the lower their level of indebtedness and the higher the amount of resources available for productive investments. In practice, land costs amounted to between onethird and one-half of the total financing package, resulting in low debt-to-equity ratios. 7 The three States in the South (Paraná, Santa Catarina, and Rio Grande do Sul) where the project was considered an experimental pilot had higher financing ceilings - ranging from R$18,000 up to R$40,000 in some municipalities, with most in the R$20,000 to R$35,000 range - due to substantially different agronomic, economic, and institutional conditions there (see sections 2.1.7(a) and 2.2.5(f)). 4

15 1.5.3 Project cost: Total, planned project cost was EUR million. The Federal Government would finance EUR million (40%), participating state governments EUR 21.8 million (5%), community associations EUR 21.8 million in cash or kind (5%), and the Bank Loan would finance the remaining 50% of costs or EUR million. The Federal Government s financing of land purchases constituted its counterpart funding of the Project. 1.6 Revised Components Components were not revised. 1.7 Other significant changes Cancellation of Loan funds: A partial cancellation of EUR million in August 2004 resulted in a revised Loan amount of EUR million, resulting mainly from the depreciation of the Real during the project implementation period up to that date Extensions of the Closing date: (a) September 30, 2004 to August 31, 2006: to recoup delays between appraisal and effectiveness, permit full disbursement of the (reduced) Loan and implementation of planned activities under projected improved implementation conditions. An Action Plan stressed: reinforcing CONTAG s role, greater information dissemination, ensuring that beneficiary associations maximized the use of funds available, and promoting wide-ranging discussions on ways to improve execution. (b) August 31, 2006 to December 31, 2007: to permit full disbursement of the Loan based on an Action Plan to bring total settled families to 47,000 (94% of the appraisal estimate) and rapid implementation in the first four months of 2006 equivalent to 75% of all families settled to end (c) December 31, 2007 to December 31, 2008: In late 2007, the sudden Federal Government decision not to seek Bank financing for continuation of the program left the project s implementing agencies unprepared to deal with a large number of potential beneficiary families (about 30,000) waiting to participate in the program. This last extension was granted to allow for a smooth transition from Bank-financed to Government-financed activities and to complete the formal institutionalization of the program, including passage of a decree that would permit the direct transfer of federal resources to beneficiary associations. After considerable internal debate with project stakeholders and within Government, the decree was passed on December 3rd, 2008, providing a solid foundation for sustainable continuation of the program after project closing Reallocation of Loan Funds: At the Borrower s request, in August 2006 (at the time of the second extension of closing date) a total of EUR 2.7 million from Category 7 Unallocated was reallocated as follows: EUR 1.7 million Category 3 Consultants Services for Part E of the Project; and EUR 1.0 million to Category 4(b) Administrative Costs for Supervision. 5

16 2. Key Factors Affecting Implementation and Outcomes 2.1 Project Preparation, Design and Quality at Entry Soundness of the background analysis: The background analysis drew extensively on evaluation studies conducted following inception of this program in 1996, and other studies sponsored by MDA on special, related issues. 8 Smaller specialized studies were also commissioned and seminars conducted in 1998 and 1999 to evaluate Cédula da Terra. 9 Seminars sponsored jointly by MDA and the Bank had broad participation including international and national authorities, NGOs and labor unions which discussed the Cédula da Terra project, evaluation results, lessons learned and a range of technical and operational topics applicable to the design and implementation of the proposed project, all informed the proposed project preparation, design and quality at entry. Assessment of project design: Project design was essentially sound, but with some operational deficiencies which required adjustments over time (see para 2.1.8). The Project drew on lessons learned from the two successful land reform pilots. In a major breakthrough, given its previous opposition to the community-based approach to land reform, the National Confederation of Agricultural Workers (CONTAG), which represents state-level federations (FETAGs) and over 4,000 rural workers unions, with more than 20 million members nation-wide, became a critical project partner in the mobilization and assistance to beneficiaries The project kept all core design features of the Cédula da Terra project, scaled up coverage to 14 states, and mainstreamed certain adjustments. Basic guiding principles included: decentralized institutional framework driven by community demand; cost-effective and nonconflictive methodology using self-selection of beneficiaries and lands to be purchased; automatic and immediate access to on-farm investments to increase land productivity and avoid the damaging delays in accessing public support services typical of other programs; and, systematic evaluation and further piloting permitting ongoing adjustment to the methodology, as needed (see Annex 2 for more details on operational aspects of the project) The long delay between project appraisal in December 1998 and Board approval in November 2000 was largely the result of strong political opposition by some groups in Brazil to the ongoing Cédula da Terra project, active from July 1997 to December The Cédula da Terra pilot project was subject to two Inspection Panel requests, in December, 1998 and September, 1999, respectively, but in both cases the Panel did not recommend investigations. During the Panel s assessment of the first request, it did not encounter any situation that constituted prima facie direct material harm to the project participants on the contrary, the projects that the Panel visited showed marked improvements on the living conditions of the 8 Evaluations initiated by the University of Campinas (FECAMP) in 1998/99 visited over 50% of lands purchased under the five-state CdaT pilot and surveyed 222 representative households to compare economic and financial impacts between community-based and the traditional land reform approach. An expanded sampling effort was also conducted in late Detailed case studies in the States of Ceará and Bahia covered quantitative models of production systems adopted, and qualitative assessment of project implementation; analysis of the project s financial structure and measures to ensure timely land loan repayment; potential impact of community-based land reform on land markets and prices; analysis of the demand for land access; and, analysis of the total target population for land reform. 6

17 participants. The Panel did note, however, beneficiaries concerns about technical assistance, complementary finance, and assistance to improve their management skills. In the second request for inspection, with essentially similar arguments to the first case, the Panel concluded that the Request was not eligible for an investigation. Nevertheless, these requests caused considerable delays and prompted Bank Management to agree with Government on certain adjustments to the implementation of Cédula da Terra and in the design of Crédito Fundiário (the two projects overlapped by about 14 months). The adjustments were noted positively by the Inspection Panel at the time. Specifically, three important innovations were introduced: (i) loan terms were made more flexible, in particular the repayment period was extended from 10 to 20 years, (ii) eligible lands were limited to those farms not subject to expropriation which meant the community-based land reform program no longer competed with, but rather was a complement to the traditional expropriation-based approach; and (iii) increased resources were devoted to technical assistance to beneficiaries Box 1 below shows that the Cédula da Terra pilot phase and the Crédito Fundiário expansion phase of the community-based land reform program are substantially the same in terms of the micro-design and operational modalities (i.e., the central role of beneficiary associations in land acquisition, investment sub-projects, and overall implementation), but with CF incorporating lessons learned from CdaT and offering more technical assistance to land settlements Design adjustments: Lessons learned and adjustments made in the latter stages of Cédula da Terra directly influenced the design of Crédito Fundiário: (a) Land loan terms were adjusted to 20 years and three years respectively, at six percent, with a 50% rebate on the nominal interest rate for timely payment by associations located in difficult areas (e.g., semi-arid zones). 10 (b) The project could not finance the purchase of lands legally subject to expropriation from 1999 onward, to differentiate community-based mechanism as a complementary not substitute instrument, i.e., one among several agrarian reform options available to Government. (c) A more strategic, inclusive approach to training was devised. Newly-formed associations were trained to increase their organizational skills, social cohesion and ability to work cooperatively. (d) Approval of on-farm investment proposals was streamlined. Funds (including the R$1,300 settlement grant) were deposited immediately to the association bank account following land purchase and released upon State Technical Unit (STU) approval of community subproject proposals (SICs). (e) The allocation for technical assistance (contracted directly by beneficiaries) was increased to eight percent of land cost during land purchase and subsequent production planning over the first two years, to ensure increased productivity and incomes. (f) Differentiated financing packages were introduced by State and region to account for varying circumstances including agro-climatic conditions, land prices, and other factors. 10 The adjustment was applied retroactively to CdaT beneficiaries to prevent two classes of borrowers developing. The two project overlapped by 14 months. 7

18 (g) Land purchase proposals could only be approved if there were adequate resources held at participating Financial Agents for financing the on-farm investments (SIC), to avoid an accumulation of settlers with land waiting for the basic resources to facilitate settlement. Box 1: Cédula da Terra (CdaT) and Crédito Fundiário (CF) Projects Item Cédula da Terra Crédito Fundiário Implementation July, 1997 December, 2002 October, December, 2008 (Effectiveness - Closing) Coverage (States) 5 NE states: Maranhão, Ceará, Pernambuco, Bahia and Minas Gerais 10 NE states: Maranhão, Piauí, Ceará, Rio Grande do Norte, Paraíba, Pernambuco, Sergipe, Alagoas, Bahia, Minas Gerais. 4 SE/S states: Espírito Santo, Paraná, Physical targets met 15,267 families settled on 609 properties covering 398,732 hectares Santa Catarina, Rio Grande do Sul 40,102 families settled on 2,114 properties covering 831,602 hectares Average ha per family 26.0 ha 20.7 ha Average cost of land per ha and per family-sat R$191 / R$4,984 R$454 / R$9,414 Average cost of on-farm investments per family- SIC R$ 6,941 R$ 9,255 Total cost per family (SIC & SAT) R$ 11,925 R$ 18,670 Financing package R$12,000 per family including settlement grant of R$1,300 Land loan terms Eligible lands Evaluations conducted 10 years with 3 years Federal Govt. long term interest rate (TJLP) Initially unrestricted, but changed in 1999 to include only lands <15 módulos fiscais (by law not subject to expropriation) 12 FECAMP: 1999 baseline, 2000 and 2003 follow-ups, and 2006/07 final evaluation (note: 2003 and 2006/07 follow ups were financed by Crédito Fundiário) R$12,000 R$18,000 per family 11, including settlement grant of R$1,300. Added credits: R$2,000 for semi-arid; R$1,000 for agroecological investments 20 years with 3 years 6% with 50% rebate for timely payment in difficult zones (semi-arid) Only farms NOT subject to expropriation, i.e., < 15 módulos fiscais FEALQ 2003 baseline, 2005 interim follow-up, and final planned for 2009/2010. Borrower Completion Report See footnote 7 for Southern States. 12 Módulo Fiscal (MF): It is a federally-determined unit for each municipality in the country which defines the minimum size for an efficient family productive holding (in hectares). The MF unit in the municipalities of Paraná vary between 5 ha and 35 ha, whereas in the municipalities of Maranhão and Piauí the MF unit varies between 15 ha and 75 ha. Thus, eligible farms under this project could have a limit as low as 75 ha in some municipalities in Paraná, or as high as 1,125 ha. in parts of Maranhão or Piauí. In practice, the project operated on farms that ranged from an average of 40 ha in Santa Catarina (South) to 609 ha in Maranhão (Northeast). 8

19 2.1.7 Project design issues: Project design was essentially sound, but it had some deficiencies and inconsistencies. (a) The inclusion of Southern States was not adequately justified or documented at appraisal, and it was simply treated as a pilot. The inclusion of these states seems to have responded more to political needs. For example, CONTAG, a new partner in project implementation, had a major presence in the South. Further, from a Federal perspective, it was important to keep a geographic balance but it was never expected that there would be a lot of activity. The financing earmarked was quite modest and the region was subject to higher financing limits per family. As noted in Section below, weak implementation in these States was a direct result of essentially three factors which distinguished them from the Northeastern States: (i) land is considerably more expensive, reflecting much better agronomic conditions and more sophisticated factor and output markets, as well as a more bimodal tenure structure (juxtaposition of very large enterprises with individual landholdings which discouraged the type of communal settlements common in the Northeast); (ii) farmer organizations are more developed and there is a stronger tendency in favor of individual landholdings; and (iii) there are other financing options for land acquisition (Banco da Terra). (b) The PAD assumed land purchases under the proposed project would likely occur well within traditional agricultural areas where services were available and markets nearby, in contrast to conventional land reform projects which often are located at or near the agricultural frontier. However, the PAD (Annex 7) also stated that most subprojects would be in scattered or remote areas, suggesting the opposite. (c) The Project targeted very poor rural people and assumed that most candidates would purchase lands located close to their previous homes or at least in the same geographic region to avoid relocation over large distances, but this inevitably meant that many settler families in some States would end up on lands of only fair quality in semi-arid areas with high infrastructure/other needs and questionable sustainability. (d) The availability of funds for technical assistance did not equate with availability of the services themselves especially in remote areas far from urban centers; project design included a generous allocation of eight percent of the land purchase cost for technical assistance (TA) and agricultural extension services, based on lessons from the CdaT. The issue of technical assistance should have received more structured planning and specifics at appraisal, given its importance to the sustainability of farms. 13 (e) The PAD did not discuss arrangements to ensure that beneficiary repayment information would be tracked by the participating financial agents, as part of the Project MIS, and be accessible for analysis. (f) The implementation period of only 43 months (February 28, 2001 to September 30, 2004) to disburse over EUR 200 million projected in the PAD proved to be grossly inadequate. Nine of the 14 participating states essentially started from scratch without prior experience of the methodology. Also, the inclusion of the Southern states as pilots introduced a different set of issues. 13 Deficient TA services are discussed in both the FECAMP (2007) and FEALQ (2006) evaluations. 9

20 2.1.9 Participatory processes: Responding to a growing controversy among some national civil society organizations over this new land reform approach being pursued by the Government through the Cédula da Terra project, throughout appraisal (and following effectiveness) Government and the Bank increased efforts to convey a more vigorous, consistent and inclusive strategy for communicating the objectives and operational rules of the project. The project team also discussed the community-based approach involving land purchase with representatives of rural organizations active in the South/Southeast regions, where some of the strong opposition originated. As noted, CONTAG helped prepare the Operational Manual. Their state affiliates (FETAGs) and municipal-level rural workers unions became directly involved in project implementation at the local level. Seminars and local workshops with broad stakeholder representation explained the project rationale, objectives and experiences Objectives and components: Poverty reduction was the over-arching objective of the program since its inception. With initial evaluation indicating that Cédula da Terra was a success, this project defined its primary objective as rural poverty reduction and increased incomes for a scaled up 50,000 families through tested, complementary activities, i.e., land purchase and onfarm investments using community-based and participatory mechanisms. This objective was rational and backed by direct evidence that it was achievable. Project components were a good fit with objectives Institutional arrangements and capacity: The Project retained the institutional and administrative structure of its pilot predecessor. There was a National Technical Unit (NTU) coordinating and overseeing the implementation through State Technical Units (STU). While all the STUs in the Northeast had considerable experience with a community-driven development (CDD) methodology, only five of the 14 states had direct experience in implementing the land reform program under Cédula da Terra. The learning curve for the other nine states proved to be steep. In fact, streamlining procedures is one of the key challenges for the future. 14 See Annex 2. Adequacy of Government s commitment: The Government which prepared the project was fully committed due to strong evidence from the pilot phase that the program could be a fast, cost-effective and non-conflictive mechanism to redistribute land to the rural poor. Elections in 2002 brought in a new Administration which had an even stronger philosophical commitment to land reform although some organized social groups continued to oppose the project s market-oriented elements (i.e., negotiating land purchases at market prices) and gained greater influence within the new Government. The inclusion of CONTAG as a project partner was a bold and prudent move which strengthened the Project s bona fides by forming a bridge between Government and civil society, and by giving the project visibility and support through its immense capillarity in the countryside. In the end, the transition to the new Government administration was relatively smooth. Risk assessment: Risks cited in the PAD were correctly projected but mitigation measures were not realistic in some cases. Additionally, several potentially important risks were not identified: 14 The beneficiary/stakeholder survey conducted for the Borrower Completion Report (NEAD 2009) rated the reduction of bureaucratic requirements to accelerate implementation as the top priority for improving the program. 10

21 (a) Allocating a percentage of land purchase cost to technical assistance could not mitigate the shortage of such services in parts of the Northeast and especially in remote areas. The availability of technical assistance to support beneficiary settlement and productive activities and boost their implementation capacity is a complex issue linked to other risks; (b) The PAD assumed correctly that beneficiary associations would select lands near their traditional and customary areas, but this was erroneously assumed to equate to: (i) lands being of generally good quality; and (ii) located near basic socio-economic services; (c) Assuming that beneficiaries would not move into the fragile agricultural frontier was incorrect in that many potential beneficiaries already lived there and, as correctly assessed by the PAD, chose to stay and not-relocate to distant, unfamiliar areas. Project design assumptions overstated the extent to which STU controls and oversight of land purchases would/could mitigate this; (d) The risk of disruption to the timely flows of funds from the Federal to State levels was correctly rated as Substantial, but mitigation measures were insufficient and should have included, inter alia, regular, high-level Bank-Borrower consultations on budget and the flow of funds; (e) The risk of delays resulting from the readiness of the states and institutions to participate in the project was not identified at appraisal. Actions such as getting the STUs to become operational, signing of agreements between the federal and state governments, getting the banks to participate in the financing program, etc. would have required time but not provided for at the project design stage; (f) A risk not discussed in the PAD was the logistical and budgetary implications for the Bank s ability to supervise a Federally-financed project implemented in 14 States, in thousands dispersed rural communities, and in two quite different regions of the country; and, (g) The risk of land prices in the South hampering demand should have been rated as Substantial to High. The known characteristics of land-holdings and high land prices severely limited the effectiveness of the project in the southern states. 2.2 Implementation The closing of Crédito Fundiário capped a successful 12-year partnership between the Federal Government, CONTAG, and the Bank to develop, scale up, and evaluate an innovative, complementary land reform mechanism which could settle poor rural families relatively quickly, cost-effectively, without generating conflict, and with good farm sustainability prospects which is likely to have lasting implications not just for Brazil, but for other developing countries facing similar challenges as well. Government s commitment to further expand the program is reflected in the issuance of Decree No of December 2, 2008 which institutionalized the Project mechanisms and conditions, including direct transfer of funds to community associations At Closing, some 2,114 farms, covering 831,602 ha., were acquired and settled by 40,102 beneficiary families, 80.2% of the appraisal target (with 70% of appraisal loan estimate, or 90% of the revised loan amount). The team considers this outcome moderately satisfactory because it did not reach 100% of the beneficiary target and it took double the original implementation period (8 instead of 4 years). But it is worth noting that there are another 8,300 families with 11

22 fully-approved land acquisition proposals awaiting the release of financing, bringing the total number of direct project beneficiaries to 48,402. Moreover, the program has generated demands for land acquisition from 22,170 families which are currently awaiting analysis at the state levels. Domestic resources have since been financing the continuation of the program. Most noteworthy, the land loan repayment rate of project beneficiaries, as reported by Banco do Brasil and Banco do Nordeste, stands at an impressive 97.6%, with loan over-dues averaging just 2.4% However, project implementation was not smooth over time or similar across states quite the opposite. First, implementation proceeded slowly between 2002 and 2006, largely as a result of sustained fiscal stringency in the wake of the new Fiscal Responsibility Law of 2002 which tightened rules on budgetary allocations, but also due to the complexities of launching implementation in 14 states, and the arrival of a new Government administration. But, implementation and disbursements accelerated considerably in the last three years. Second, geographically, over 60% of project beneficiaries (and 65% of loan disbursements) concentrated in the Northeast states of Piauí, Maranhão and Bahia. Meanwhile, the three states in the South had only 1% of beneficiaries (and 0.5% of disbursements), largely due to considerably different market conditions and options available to beneficiaries. And third, implementation was smoother in some states, such as Bahia and Espírito Santo, and more problematic in others, such as Ceará and the Southern States. For example, in Maranhão which in terms of reaching targets did really well, the State experienced operational problems towards the end of the project. Some beneficiary communities complained about limited attention by the State Technical Unit, delayed disbursement of community grants, and concerns about the contracting process for investment subprojects. These issues resulted in the Bank s agreement with the National Technical Unit in September 2007 not to approve new land acquisitions and instead implement a remedial action plan in that State. This uneven performance also justifies the team s moderately satisfactory rating of project outcomes. The main factors affecting overall project implementation are detailed in section below Costs: Total project cost was EUR million shared as follows: IBRD loan EUR million, Federal Government of Brazil EUR million, State Governments EUR million, and Beneficiaries million. This represents 75.7% of the original appraisal estimate (in EUR) or approximately 96.9% of the revised total estimate after the September 2004 partial cancellation of EUR million from the IBRD Loan (minus a corresponding amount of counterpart contribution). The undisbursed balance of EUR 16.5 million was cancelled after Closing. Actual total costs by component deviated from appraisal estimates with all components under-expended compared to original projections (see Table (a), Annex 1) Main factors affecting project implementation (a) Effectiveness: Project Effectiveness was delayed for 11 months after Board Approval due to difficulties in complying with Effectiveness conditions (largely related to the signing of various implementation agreements among agencies). Other issues causing delay included excessive bureaucracy at both the Federal and State levels and difficulties experienced by some States in regularizing their fiscal obligations with the Federal Government without which the Federal/State Project Agreements needed to participate could not be signed Also relevant, the Cedula da Terra pilot was still under implementation in five Northeast States until December 2002, to some extent diverting attention from the NTU and Northeastern STUs from the Effectiveness needs of the new Project. 12

23 (b) Project launching: Following Effectiveness, political, institutional and financial issues absorbed over two years of the Project s planned execution period, including: (i) creating or making changes in existing State Technical Units (especially in States which had not previously participated); (ii) establishing State Sustainable Rural Development Councils (CEDRS); (iii) States negotiation and signing of contracts with the Federal Government and preparation of State Development Plans; (iv) information dissemination campaigns; and (v) States reaching agreement with participating banks for financing land purchases. In the Southern States, agreeing on alternative forms of Project coordination 16 and special issues arising from the structure of landholdings, high land prices, and new State Governors in three of the four states, hindered project initiation. (c) Fiscal conditions: The Project suffered continuous, and at times intense, budget constraints at the Federal level reflecting above all, the impact of the new Fiscal Responsibility Law (FRL) of 2002 and constant pressure on Government to increase the fiscal surplus. Not only were annual budgets tight and budget funds released in an untimely manner, but the practice of contingenciamento Government s right under the FRL to withhold a portion of a ministry s approved budget which might or might not be released at a later time, sharply reduced levels of funding for the Project and affected timely project execution. Contingencies reached up to 40% of the Project s budget allocation in any one year, and in some years, absolute budget limits were imposed on the Project. 17 The persistent problem was securing adequate budget allocations to permit the entry/internalization of Project Loan funds for the community investment subprojects. In contrast, counterpart resources in the form of Federal Government funding for land acquisition subprojects (SAT) were never an issue (see 2.2.4). (d) Change of Government: Newly-elected Federal and State Governments in early 2003, with new institutional executives and philosophy, took time to settle in. Significant time elapsed while the new Government formulated and reached consensus on its own national land policy. This was made more complex by the closer alignment with and presence within the new Government - including the Ministry of Agrarian Development in charge of the project - of influential social movements which had come to dominate the agrarian reform discourse and who were opposed from the outset to the community-based land reform approach relying on negotiated, market transactions to acquire farms. (e) Bureaucracy, commitment and processing capacity: Excessive bureaucracy affected many aspects of project implementation. Further, some STUs were variously affected by weak commitment at senior levels, quality of professional staffing, implementation capacity and multiple responsibilities including physical supervision of settlements. Leveraging of partnerships to facilitate implementation did occur, but many of the STUs tasks could not be delegated given the tight fiduciary controls needed over SAT and SIC processing. The Bank team provided STUs with detailed and continuous guidance on ways of streamlining procedures, easing workloads, promoting efficiency and advocating for improved office facilities. 16 In Rio Grande do Sul (RS) for example, the Rural Workers Union proposed itself as STU. The idea of municipal consortia to run the Project in certain parts of the State was also floated. 17 The Ministries of Agrarian Development and Agriculture were always included among those affected by contingencies and budget resources for investments were very low. 13

24 (f) Performance of Southern Pilot: As noted elsewhere, the pilot in the Southern States, with the exception of Espírito Santo, made little headway. Inclusion of these States was largely experimental and the resources allocated to them were limited. Project implementation in these States was hampered by conditions on the ground which proved unsuitable for application of the project methodology: elevated land prices, making the Project financing package unattractive; limited supply of land to settle groups; and strong competition from the also Federally-funded Land Bank (Banco da Terra) which had an active presence in the South, especially in Rio Grande do Sul, promoting much higher loan limits and individual contracts which were not supported under this Project Mid-term Review (MTR): The MTR in 2003 comprised a series of joint missions to eight of the 14 participating states, meetings in Brasília with NTU staff, and a seminar in Brasília with 11 of the 14 states, and including CONTAG and its affiliates. Three of the participating Southern states had only just initiated land purchases and their contribution to the MTR was minor. No project performance study was produced for the MTR. The economic and financial benefits of the project s financial package were already well-documented in the Federal University of Campinas (FECAMP, 2001 and 2003) evaluations of the CdaT Project, its Implementation Completion Report of 2003, the ESALQ study of 2003, and monthly Project data panels issued by the NTU. Since the MTR coincided with key policy adjustments under consideration by the new Government Administration, its recommendations were slowly addressed by the established consultative processes and through supervision at the Federal and State levels (see next section). Financial performance was rated Unsatisfactory by the MTR and it was clear that a substantial extension of the closing date was needed to make headway, as well as a possible cancellation of Loan funds Project at Risk Status and Actions Taken: The Project was declared At Risk in 2004 due mainly to its weak financial/disbursement performance indicative of slower than expected physical progress. MDA and the NTU, in consultation with the Bank, put a series of measures in place, described below Measures to improve execution: MDA and the NTU, in collaboration with the Bank, established the conditions for improved Project execution: (a) Given huge demand and following high-level consultations between the Bank and the new Government, the budget obstacle was alleviated by a Decree changing the management mechanism of the federal National Land Fund (Fundo da Terras, FT) to permit the financing of SICs. Federal Government counterpart funds for land purchase SATs were already being sourced from this fund and greatly exceeded Project needs. The change resulted in a much stronger budget position for the Project from onward, along with a marked improvement in physical performance and disbursements. 18 (b) The Project was formally integrated into the Federal Government s National Agrarian Reform Plan (PNRA) for settling 530,000 families by end-2006, of which 400,000 would be under traditional, expropriation-based model, and 130,000 families under the broad National Credito Fundiario Program (PNCF). The PNCF had three financing windows, or lines of action, only two of which were supported by this project: (i) The Rural Poverty Reduction (CPR) line, targeting 50,000 families (which constitutes the core of this project); (ii) Nossa Primeira Terra 18 Some slippage was evident in 2007 and The NTU delayed its analysis of individual State Project Accounts and SOEs and consequently its release of resources to the States and participating banks, for SATs and SICs. 14

25 (Our First Land) targeting 4,000 young families (some of which received financing from the World Bank loan); and, the Consolidation of Family Agriculture Program (CAF), which was exclusively Government-financed through a Land Bank. (c) The Bank s strategy from this period to Closing involved the maintenance of regular, high level dialogue with the Government to ensure budget availability and release to the Project. Regular, regional strategy meetings with the STUs and authorities continued to discuss budgets and action plans and to leverage field support for project implementation. These meetings rallied the participation of 180 entities, including CONTAG and its State FETAGs, about 130 NGOs and service providers, and local authorities. (d) Overall, the institutional structure in each participating State was strengthened via support from the Inter-American Institute for Cooperation in Agriculture (IICA) to better handle Project demand, supported by further improvements in management information systems (MIS) to improve planning, administration, transparency and operations. In States with especially strong demand, e.g., Piauí, Maranhão, Rio Grande do Norte and Pernambuco, social and labor organizations, NGOs and technical assistance providers were mobilized to support the associations preparation of SAT and SIC proposals. (e) Given the enormous geographical dispersion of project sites in 14 states, standard project supervision budgets did not allow for detailed follow up in every state. Bank supervision thus encouraged regional meetings of STUs where common issues could be addressed. Also, given the similarity in methodology and target populations between the Crédito Fundiário project and the ongoing Bank-financed Rural Poverty Reduction Projects in nine of the Northeastern states, supervision missions often combined both projects. In most Northeast states the implementing agencies were the same for both projects. 2.3 Monitoring and Evaluation (M&E) Design, Implementation and Utilization Monitoring and evaluation design represented the continuation of innovative approaches developed under the CdaT pilot. First, project monitoring was to use a Management Information System (MIS) capable of storing and producing data on physical performance, financial management, and project management. STUs and Financial Agents (participating banks) would establish their own systems storing key project information, to be linked to the NTU s MIS. The NTU was then to aggregate all information from these satellite systems. Second, the NTU would conduct impact evaluation studies based on panel data sets and control groups, with support from CONTAG and other civil society organizations and using aggregated MIS data. The impact evaluation program was designed and field tested under the CdaT pilot. This project financed further stages of that evaluation program and launched a Crédito Fundiário-specific evaluation program which will be completed in late 2009 or early Monitoring: In practice, monitoring did not reach appraisal goals. The breadth of information collected by the STUs, most of which had good MIS systems and capability, was far greater than what was transmitted to and aggregated by the NTU. Thus, for example, organized data on community investment subprojects and on training programs were not centralized by the NTU. For preparation of this ICR, data had to be obtained directly from individual States. Similarly, land loan repayment data were not systematically transmitted to the NTU, requiring laborious ex-post data collection, on a State by State basis. Also, the NTU s monitoring efforts focused heavily on information about the land acquisition component, at the expense of other project activities. 15

26 2.3.3 Evaluation: The Project financed: (a) a baseline study (2003) and interim progress study of the same beneficiary cohort (2005) of Crédito Fundiário beneficiaries by the Luiz de Queiroz Agrarian Studies Foundation (FEALQ); (b) a beneficiary/stakeholder survey by MDA for the Borrower Completion Report (MDA, 2009); and (c) two further studies under the evaluation program initiated in 1999 for Cédula da Terra beneficiaries by the Economics Foundation of the University of Campinas (FECAMP, 2003 and 2006). Moreover, MDA is commissioning a follow up study to the FEALQ 2003 and 2005 studies, which will be completed by As a middleincome country, Brazil implements its national programs (such as this land reform program) following its own timing and priorities, which do not necessarily follow the Bank-financed project. One practical implication of this is that, unfortunately, the results of this third FEALQ study were not available for inclusion in the ICR. Accordingly, for the purposes of assessing the Project s achievement of its PDO, the results of FEALQ 2003 and 2005 have been used as primary sources, with complementary information from the BCR (2009) survey. The Bank project team strongly recommends that the Bank sponsor a detailed impact evaluation of this project in late 2010 taking full advantage of the third round of studies currently under preparation. (Study results are discussed in more detail in Section 3.2 and Annex 5 below.) The impact evaluation of any redistributive land reform program is technically complex and requires long-term data-collection and institutional commitments. Given that the bulk of the farm settlements and investments under the Crédito Fundiário phase of the program took place in the period there has not been sufficient time to carry out a detailed long-term impact evaluation of direct project beneficiaries. Detailed results of impact evaluations of Cédula da Terra beneficiaries are based on a considerably longer time horizon. Therefore, this ICR also includes the results of CdaT studies. Although the subjects of CdaT evaluations are not this project s direct beneficiaries, the study conclusions can be indicative (and possibly predictive) of the longer-term impacts of Crédito Fundiário. As noted earlier, CdaT and CF are two phases of the same program, utilizing essentially the same methodology and targeting poor rural families with identical profiles Dissemination: Several dissemination efforts were carried out during the project. The Secretary of Agrarian Development of MDA gave a presentation on Crédito Fundiário to Southern African land reform stakeholders at an event in South Africa in Further, both the FECAMP and FEALQ study results were published by MDA as books in 2008 and discussed with stakeholders at seminars and special events within Brazil. Additional publications of study results are being planned including by FECAMP. For its part, the Bank has sponsored many international events related to the project. In some respects, over the past decade the Cédula da Terra Crédito Fundiário program in Brazil has been the Bank s flagship program on redistributive land reform. This program has always had a high international profile. Since 1998, the Government and the Bank have hosted several official study tours to both Cédula da Terra and Crédito Fundiário project sites, including delegations from Zimbabwe, South Africa, Malawi, Honduras, Guatemala, Paraguay, and Bolivia. The project s international relevance was particularly important to the Borrower. A case study on both Cedula da Terra and Crédito Fundiário was also presented at the Bank-supported Shanghai Poverty Conference Scaling up Poverty Reduction in The results of the program were discussed and disseminated during ESSD Week in 2005 and the International Land Policy Conference held in Washington, D.C. in Finally, the experience of the Cédula da Terra/Crédito Fundiário program in Brazil is the 16

27 subject of a chapter of a book comparing redistributive land reform experiences worldwide just published by the Bank Safeguard and Fiduciary Compliance Environment: Compliance with environmental safeguards was generally satisfactory. Contracts for SAT land purchase were signed only after State environmental agencies prepared findings and authorized purchase. In cases involving fragile eco-systems, the environmental agency could recommend against purchase and an association would be obligated to find another property. Preservation areas were by law, identified on each property and registered. Productive subprojects in such areas were only approved by the STUs when environmental protection measures were presented in subproject proposals. Training programs for beneficiaries also had environmental content. Even so, impact evaluations noted a tendency for settlements in some states to be located in remote, fragile areas with natural resource problems, implying that the STU s analysis of land purchase proposals did not always rigorously apply the natural resource and environmental criteria required for approval (see Annex 5) Financial Management: Overall, project management systems and controls in place were judged reasonably adequate and reliable for managing and monitoring the Project. However, a Financial Management Supervision (FMS) in 2006 gave the Project a Marginally Satisfactory rating due to the NTU having financial and project management systems not integrated or communicating electronically. The follow-up FMS in late 2007 also rated the project Marginally Satisfactory with a risk of Moderate, mainly due to procedures used to transfer resources to community associations, an issue raised previously in audit reports and aide memoires. The Borrower responded positively, installing a new SYSCOM integrated program permitting the automatic issuance of project FMRs and PMRs, and conducting an FM training workshop for all 14 participating STUs. Subsequent FMS in November 2008 and January 2009 focused on audit performance and on financial arrangements for project closing in December 2008 and the grace period through end-april Audit: Audit performance was generally satisfactory. Financial Statements were either Unqualified or Qualified Exception while Special Opinions through 2005 were Unqualified (and no longer utilized thereafter). An audit report covering the period from January 1 st, 2008 through April 30th, 2009 is due by October 30, Qualified Exception opinions were the result of issues similar to those raised by FMS, e.g. separate systems for FM and project management. Records indicate strong follow-up by the Borrower and the Bank to resolve issues raised by auditors Procurement: The Project received two specialized Procurement Post-review (PPR) missions over the project period. 21 The PPR of May 2006 found no major procurement issues. 19 Agricultural Land Redistribution: Toward Greater Consensus (2009), Hans P. Binswanger-Mkhize, Camille Bourguignon, Rogier van den Brink, editors. Washington, D.C. : The World Bank. 20 This re-visits the issue of candidate groups tending to select land near their areas of origin which in many cases, as is typical of poor rural people in the Northeast, were in more fragile, water-deficient zones. 21 In earlier years, Procurement and FM issues were supervised by the core technical team rather than Bank Procurement and Financial Specialists. This changed in

28 The most recent PPR of March 2009 reviewed 636 contracts (goods, non-consultant and consultant services, and works) totaling R$11.3 million, about 10% of all community subproject contracts implemented. While all subprojects inspected by the PPR were found to be completed and functioning normally, the PPR noted that: (a) the same suppliers had been selected repeatedly to quote prices; (b) in some cases, direct contracting was used although this method was not permitted under the Operational Manual; (c) selection of individual consultants did not observe the three-cv comparison requirement; and (d) procurement was, more often than not, being prepared by individual consultants providing technical assistance to associations rather than by the associations themselves The PPR concluded that: (a) geographical spread and access difficulties may have prevented associations from requesting price quotations from a wider range of suppliers; (b) for this same reason, it may not have been possible to obtain more than one price quotation; (c) sole sourcing of individual consultants may be justified on the basis of the Guidelines (para. 5.4(b)); but that (d) the associations involved missed the opportunity to enhance their ability to carry out future projects more independently. The PPR mission report noted that community subprojects executed in remote localities with difficult access imply the need to purchase and contract directly at pre-determined values, given limiting factors of location and the factual impossibility of conducting even minimally competitive procedures in specific cases. The PPR mission made a series of recommendations which were later discussed with MDA Post-completion Operation/Next Phase Initial overtures by MDA to the Bank regarding a follow-up operation during 2006 and 2007 were subsequently overtaken by the Federal Government s sudden decision in late 2007 to continue the program without financing from the Bank. As noted earlier, full institutionalization of the program required passage of a decree allowing the direct transfer of federal resources to beneficiary associations. After considerable internal debate with project stakeholders and within Government, the decree was passed on December 3rd, 2008, providing a solid foundation for sustainable continuation of the program after project closing This institutionalization is viewed as a major achievement of the program and a good example of the high potential benefits of a long-term partnership between a committed client and a responsive Bank focused on addressing a complex, politically sensitive, and risky development challenge. Continuation of the program is grounded on solid foundations. Institutional and physical sustainability are in the hands of the States and beneficiaries. The STUs now have considerable experience after receiving considerable training and guidance under the Project. Even so, adherence levels are unpredictable and the future of the program may rest with a few, 22 It is possible also, that direct contracting was prompted in 2008 by an accumulation of subproject proposals to be executed in a very short time period given that the Closing Date of end-december 2008 was fast approaching. Aide Memoires show that STUs were urged from late 2007 onwards to reduce bureaucracy and accelerate processing, while maintaining standards, for this reason. The Bank team assisted the NTU in preparing a set of guidelines for explaining procurement procedures to the STUs and associations in more understandable language. 23 In regard to the stated prohibition on direct contracting, the PAD (Annex 7), Loan Agreement (Section I, Part B 4) and Operational Manual (Annex 2) explicitly allowed for that modality. The PAD foreshadowed the issues, stating: This procurement procedure is appropriate because most subprojects would be small and/or implemented in scattered and remote areas and therefore it will be difficult to obtain competitive proposals. The procurement provisions of the Operational Manual were not amended during implementation. 18

29 highly-committed States and/or new States interested in participating. In terms of physical sustainability, settler communities were given basic training in the O&M of their on-farm subprojects and in the case of water supply, evaluation studies and supervision missions noted that on settlements visited, beneficiaries were commonly paying user fees. 3. Assessment of Outcomes 3.1 Relevance of Objectives, Design and Implementation Project objectives were and remain highly relevant for Brazil. Redistribution of land to the rural poor continues to be a major development challenge for the country, particularly in the regions where the project operated. The project s basic approach to land distribution (i.e., selfselected community associations negotiate land purchases financed by long-term loans and receive complementary matching grants for settlement and investment expenses) proved to be a successful mechanism to address land inequality. As noted elsewhere, the program showed considerable speed, cost-effectiveness, and sustainability in addressing one of the most difficult and potentially explosive rural development challenges for any country. The program s results and lessons learned merit wide dissemination by the Bank. Moreover, the project shows that even with a conceptually fairly simple model, implementation of a land reform program is a daunting challenge with multiple political, technical, financial, and operational obstacles. Nevertheless, strong borrower commitment and sustained Bank support were rewarded by encouraging results. Evaluation studies to date validate the methodology and objectives based on income, wellbeing, social capital and poverty targeting outcomes. 3.2 Achievement of Project Development Objectives The Project made important advances in reducing rural poverty. Moreover, key steps recently taken by Government (i.e., issuance of decree institutionalizing project operational modalities, commissioning a full impact evaluation, processing of land acquisition proposals already in the project s pipeline) provide strong indications that the sustainable continuation of the project is very likely. The project s main achievements as embodied in the PDO were: Project Development Objective: Reduce rural poverty in 14 states of the Northeast and South/Southeast regions by increasing the incomes of about 50,000 poor rural and peri-urban families through extending the community-based approach to land acquisition and participation in complementary, demand-driven community subprojects The project financed the acquisition of 2,114 farms, covering 831,602 ha, and the settlement of 40,102 beneficiary families, 80.2% of the appraisal target (at 70% of appraisal loan estimate, or 90% of the revised loan amount). In addition, there are another 8,300 families with fully-approved land acquisition proposals awaiting the release of financing, bringing the total number of direct project beneficiaries to 48,402. Most noteworthy, in December, 2008 the loan repayment rate of project beneficiaries, as reported by Banco do Brasil and Banco do Nordeste, stood at an impressive 97.6% (see below) The complete results of the Project impact evaluation will be available only in early However, the evolution of a sample of project beneficiaries from 2003 to 2005 was studied by FEALQ (2005) using a panel format. The average income of project beneficiaries increased 145%, from R$1,656 in 2003 to R$4,064 in 2005, and the number of families producing on-farm and marketing their production doubled from 25% to 50% in the same period. The study noted an 19

30 increase in social capital of project beneficiaries as well. It concluded that income gains could be linked directly to land access and agricultural production. This study also found that the combination of activities supported by the project enabled families to access and leverage the benefits of other programs, especially the Federal PRONAF-A credit line, health and education programs. In 2005, about 89% of project beneficiaries were working on their properties, up from 57% in 2003, agricultural production conducted on-farm had increased from 37% in 2003 to 82% by 2005 and some 59% of production was being processed and having value added, up from 20% in 2003 (see Annex 5 for more details on the scope and methodologies of the various studies cited in the ICR) The Borrower Completion Report (BCR - MDA/NTU 2009) analyzed 19 beneficiary groups in five states. Incomes from agricultural production on-farm, while still modest, averaged R$9,005 per family annually or a net income of R$6,495 per family/year. The survey did not attempt to re-create a baseline so that the incremental gain cannot be measured, and the age of the settlements surveyed is not known. After repayment of the loan used to acquire the land and where relevant the PRONAF loan families net income averaged R$5,216 per family/year or a net R$190/month per family agricultural worker (see Annex 5). Since project performance in the Southern States was modest, planned evaluation studies were not conducted and thus the project s poverty reduction impact there is not known The Project also financed two additional stages of the FECAMP evaluation program focused on Cédula da Terra beneficiaries, with both control group and pipeline comparators (FECAMP, 2006). The study shows that average total family income of CdaT beneficiaries increased 220% from pre-project levels through 2006, compared to 176% for the control group in the same period. The study shows that for early 2006, the program-financed land acquisition was directly responsible for 88% higher incomes for CdaT beneficiaries (R$4,471) compared to preproject pipeline cohort subjects (R$2,370). FECAMP s econometric analysis also shows farm income of the CdaT group evolving at a higher rate than income from paid labor, indicating that there was a major effort to re-direct activities towards agriculture and reduce the precarious temporary labor characteristic of the pre-project situation Repayment of land loans: A critical indicator of beneficiaries confidence, motivation, and perception of moving out of poverty is their capacity and willingness to repay their land loans. In this regard, project performance was outstanding. The overall repayment rate as of December 31 st, 2008 in 11 States was 97.6%. The performance of Espírito Santo (86%), a Southern State, contrasts sharply with that of the poorer Northeast States No information was provided on Paraná, Santa Catarina and Rio Grande Sul due to the negligible amounts of loans under the project in those States. 20

31 Table 1: Land Loan Repayment Rates, 11 States (December 31, 2008) State Beneficiary Debt Outstanding Beneficiary Over-dues as at Percentage R$ R$ Alagoas 12,271, , Bahia 31,862, , Ceará 17,199, , Maranhão 38,600, , Minas Gerais 11,464, , Paraíba 23,262, , Pernambuco 19,727, , Piauí 44,913,579 1,037, Rio Grande do Norte 39,004, , Sergipe 12,157, , Espírito Santo 19,580,973 2,718, Total: 257,888,250 6,249, % Source: Bank of Brazil, Bank of Northeast Brazil Integration of programs: An important outgrowth of the project has been the extent to which agrarian reform beneficiaries can take advantage of other anti-poverty programs and services. For example, families have guaranteed access to the National Program to Strengthen Family Agriculture (PRONAF) under a special credit line for agrarian reform beneficiaries (PRONAF-A) for developing and consolidating productive activities. MDA-sponsored programs such as marketing support, government purchasing of family agriculture products, minimum price supports, crop insurance, and specialized PRONAF lines for women and youth are also accessible. The Federal Luz para Todos rural electrification program also gives priority to agrarian reform settlements, as do certain health and education programs. FEALQ (2005) noted that project beneficiaries access to income transfer programs grew from 41% in 2003 to 61% by These benefits from the integration of such programs have also been confirmed by the FECAMP study. Since the project-financed investments were never intended to provide for all the requirements of agrarian reform settlements, these findings are particularly encouraging. 3.3 Efficiency The results of the CdaT program were used to inform the economic and financial viability of the investment sub-projects supported by the CF project at appraisal. The investments under the two projects are essentially the same in this aspect as follows: (a) land acquisition (same methodology and similar incentives to beneficiaries); (b) type of broad investment sub-projects (housing, electricity, water supply, small-scale agroindustries, basic marketing infrastructure); (c) broad cost of indicative sub-projects (standard subproject designs and costs, financing package ceilings only adjusted by region); and (d) broad income streams from indicative sub-projects (farm models for CF were based on actual CdaT experiences) Since a large chunk of these CF investments were actually concluded after , the time period available does not permit a comprehensive impact evaluation of investments by the Crédito Fundiário beneficiaries. Such an evaluation will be available only in However, given the similarity between the CdaT and CF phases of the program in terms of the beneficiary 21

32 profile, and the nature of the investments, the impact evaluation in of Cédula da Terra beneficiaries, which were financed by and constituted an integral part of the evaluation efforts of the Crédito Fundiário project, has been used to provide an indication of the outcome. The analysis adopts a conservative approach. Assets considered in this estimation do not include assets owned as a direct result of project interventions, such as the value of the land obtained through the CdaT credit program for land purchase (SAT) - or the value of housing obtained through CdaT grant-based investments (SIC). If these assets were incorporated, the measured impacts would be significantly larger. (a) Economic: First, gross annual income estimates are all positive. Beneficiaries income increased in real terms by 46% over their gross income in the year of settlement, as a direct impact of the project. Second, results are positive and significant for all estimates of asset growth: assets of beneficiaries increased by 100% (not including land and housing acquired under the project) over the average total asset value owned pre-settlement. Even with debt discounted from total asset value, net assets remain positive and substantial. Productive assets (farm machinery, tools and livestock) are the main source of impact. Third, the impact on agricultural production value is highly significant, indicating an average positive effect of about R$2,200/year, mainly from agricultural production on project-financed land. The project s impact on intermediate indicators related to agricultural production, were also positive and significant; beneficiaries increased planted area by about 20 hectares, had about 60% more technical assistance, and had about R$1,800 in additional credit. Use of mechanical and/or animal traction also increased. (b) Financial: The average additional income obtained by beneficiaries due to the CdaT project, i.e., the impact of the project on income, was R$1,463 gross per family/year, indicating that they have the conditions to repay their land loans, payments of which average around R$ per year. In the case of Crédito Fundiário, repayment data from December 2008 clearly supports this analysis since 97.6% of SAT loans in 11 participating states, including 10 states of the Northeast, were current in payments. This surrogate variable clearly supports the contention that the investments under CF are financially viable. (c) Cost-effectiveness: Using very conservative assumptions including a benefit stream of only 20 years (benefits are likely to accrue much longer given increased agricultural production, and land/labor are still under-utilized), and an over-valuation of costs (see Annex 3), the analysis shows a positive internal rate of return (IRR) of 12.6% and a net present value (NPV) of R$143.0 million, using a discount rate of 6%. 25 (d) International comparisons: The Cédula da Terra / Crédito Fundiário program in Brazil served as a model for other Bank-supported redistributive land reform operations in Latin America, so a comparison of results with the experiences of Guatemala (1999 to 2006) and Honduras (2001 to 2007) is useful. 26 Considering that Brazil s overall income levels are considerably higher than in Guatemala and Honduras, the results of the Crédito Fundiário program fare very well in terms of the cost-effectiveness of creating rural employment for landless farmers through distributive land reform. The average cost of land acquired under Crédito Fundiário was approximately US$195/ha compared to US$1,020/ha in Guatemala; the 25 This compares to a project overall IERR of 35% and NPV of US$ million estimated at the appraisal of Crédito Fundiário. However, the methodology used at appraisal, using representative farm models, is not strictly comparable to the analysis presented here using data from beneficiaries of the Cédula da Terra phase of the program. 26 Guatemala: Land Fund Project (ICR 34615, March 21, 2006); and Honduras: Access to Land Pilot Project (PACTA) (ICR , June 21, 2007). 22

33 average cost of land per beneficiary family in Brazil was US$4,058, compared to US$4,686 in Guatemala and US$2,780 in Honduras, respectively. The average total cost of the package per beneficiary (including land and investments) under Crédito Fundiário was US$8,047, compared to US$7,522 in Guatemala and US$7,480 in Honduras. The ratio of debt incurred by beneficiaries compared to the value of land-plus-investments was lower in Brazil (1:2) than in Guatemala (1:1.6), but higher than in Honduras (1:2.7). 3.4 Justification of Overall Outcome Rating Rating: Moderately Satisfactory By and large, the Project achieved its main objective of reducing rural poverty through an effective land redistribution mechanism, but it did so over a considerably longer period of time and fully met only 80% of its target. 27 Two project outcomes were of particular importance. First, the loan repayment rates by project beneficiaries are outstanding, indicating excellent targeting and strong beneficiary commitment to the program; and second, the Government (and civil society stakeholders, particularly CONTAG) have shown strong commitment to the continuation of the program. In particular, passage of a Decree institutionalizing the operational modalities of the program in December 2008 provides a solid foundation for the long-term sustainability of the program. However, the satisfactory outcome is downgraded to Moderately Satisfactory due to the fact that at Closing only partial impact results were available to support conclusions. Since the project is the expansion phase of a longer program, the ICR had to partially rely on existing comprehensive impact evaluation data from Cédula da Terra. These positive results can be considered indicative for Crédito Fundiário. Lastly, a comprehensive impact evaluation of Crédito Fundiário is currently under way and results should be available in The Bank project team strongly recommends that the Bank sponsor further evaluations of the program taking advantage of the latest round of results in Overarching Themes, Other Outcomes and Impacts (a) Poverty Impacts, Gender Aspects, and Social Development Poverty Targeting: The poverty targeting of benefits maintained the accuracy established under the pilot phase of the program, partly the result of self-selection for participation. Both FECAMP (2007) and FEALQ (2005) confirmed that beneficiaries at entry were exceptionally poor with incomes well below the National Household Survey (PNAD) poverty line, low educational levels and high levels of illiteracy (see Annex 5) Gender: While women have been important beneficiaries of the program and there was no restriction which limited their entry, the vast majority of direct beneficiaries were male. However, female-headed households comprised about 15% of direct beneficiaries, up from 8% under the CdaT. Women s modest direct participation may stem from the fact that female-headed households were among the poorest and in a project where demand for participation greatly outstripped supply, traditional attitudes may have prevailed. 27 If the approximately 8,300 families whose land acquisition proposals are already approved and ready for financing are included, the project achieved 96.8% of its target. 23

34 3.5.3 Youth: The Project settled about 2,500 young people under the Nossa Primeira Terra (NPT) program targeting youth aged 18-28, the goal being to insert youth from farm families into the rural economy. It was quite common for settlements financed initially under the NPT window of Crédito Fundiário to show significant attrition due to difficulties in maintaining social cohesion on settlements comprising mostly young families. Subsequent project information campaigns to attract young people resulted in many of them entering as members of groups of mixed age which expanded from 15% of all settlements in 2003 to 42% by Social capital formation: FECAMP (2007) included a formal, quantitative analysis, using control and pipeline groups, of how the political, legal and institutional environment created by the project methodology effectively contributed to forming and strengthening social capital among its beneficiaries and how that social capital in turn, impacted on the wellbeing of beneficiary families (Costa and Romano, 2007). The study concluded that the social capital of beneficiaries had increased, specifically in regard to their relationship with public authorities, and to a greater extent than the pipeline or control groups. (b) Institutional Change/Strengthening Institutional development is assessed as substantial, with some caveats. The Federal Government s decision to institutionalize the program s methodology and finance 100% of its investment costs reflects confidence not only in the methodology but also in MDA/NTU s institutional structure and capacity to implement it. At the State level, however, capacity was variable despite project-financed training and supportive partnership arrangements. Fluctuating commitment on the part of some State Governments also played a role in how well the STUs performed The Project created some 2,114 community associations via the land acquisition process. The institutional structure and incentives promoted by the project strengthened community associations. The institutional relationships between beneficiaries and local power groups helped them to establish greater participation and control over actions by the State and promoted transparency in decision-making and information dissemination. Preliminary analysis of the impact of this social capital on certain indicators of wellbeing demonstrated that project-fostered social capital has positively affected beneficiaries income, assets and food security. (c) Other Unintended Outcomes and Impacts (positive or negative) One unexpected positive outcome was the extent to which project investments were targeted on a territorial basis. In the Northeast, Minas Gerais and Espírito Santo, the project supported proposals in 78 Rural Territories under the Territorial Citizenship programs benefiting 24,590 families equivalent to 62% of total beneficiaries in these States. Also, the specific focus on youth, as a result of the Government s new NPT program, was an innovative variation of the program which yielded important findings. 28 Institutional development of the Financial Agents is difficult to assess directly. Experience indicates that while their extensive branch networks were, prima facie, an advantage, their cumbersome procedures and excessive documentation requirements were sometimes a material hindrance to the Project s more rapid progress. 24

35 3.6 Summary of Findings of Beneficiary Survey and/or Stakeholder Workshops Key results (income, social impacts and targeting) of beneficiary surveys are summarized in Sections 3.2, 3.5 and Annex 5. The following are some additional, relevant findings from the survey process The FEALQ (2005) opinion survey of Crédito Fundiário beneficiaries concluded that the project was reaching its targeted public. Some 66% of families had physically settled on-farm compared to just 8% in 2003 and families accessing technical assistance had increased from 30% to 65%. Housing quality had markedly improved: 99.5% of settlers interviewed had houses of brick/cement, up from 48% in Importantly, income gains could be linked directly to land acquisition/access and agricultural production in a stable environment. However, FEALQ also found that investments in clean water were needed, even though supply and quality had improved. Further, access to and use of technical assistance and rural extension had risen but many beneficiaries were not being reached and quality was variable. An important issue was the greater degree of technical sophistication required from TA services once families moved into marketable production, and lack of capacity and preparation for this phase on the part of service providers The BCR (2009) conducted interviews with institutional stakeholders - STUs, State and Municipal Rural Development Councils, and technical assistance providers and beneficiaries, in five states. 29 They rated the project 3.8 to 4.0 (Good and Very Good) on a scale of 5.0 (Excellent) for its capacity to generate employment and income, and promote social inclusion. Of the 19 settlements interviewed, 12 considered the project Excellent and seven viewed it as Very Good or Good. The survey revealed broad stakeholder agreement that the methodology promoted inclusion, improved the quality of life of poor rural families and could increase incomes. Further, some social movements formerly opposed to the project had muted their criticism because they now recognized that the methodology was having positive impacts. By project closing some of them had increased their participation in/engagement with, its implementation FECAMP (2007) concluded that program beneficiaries under the Cédula da Terra phase after nine years of settlement showed a maturity stemming from a process of differentiation based on individual skills. Maturity was evident in the consolidation of leadership and productive routines which combined individual efforts with the exploitation of small, collective areas, both essential to ensure repayment of land loan obligations. In social capital terms, the individual still predominated over the collective. Despite challenging conditions in many cases for the conduct of agriculture, families persisted in seeking ways to produce and market, and repay their loans. The study called attention to the environmental sustainability of some settlements, compromised by the low technological level of properties and the need for technical assistance. The study concluded that the stability of about 70% of CdaT settlers for over nine years was a positive outcome and boded well for longer-term sustainability. 29 States included were: Piauí, Rio Grande do Norte, Paraíba, Minas Gerais and Espírito Santo. The study did not attempt to assess incremental achievements over time. 25

36 4. Assessment of Risk to Development Outcome Rating: Moderate The sustainability of farm settlements varies. Settlements in areas with good-quality land, adequate water and with access to basic socio-economic services and infrastructure including roads are sustainable. The sustainability of settlements in less-endowed areas, e.g. semi-arid with water issues and lack of infrastructure, may be questionable. The rotational aspects of farm settlement (i.e., families unable to cope or with other obstacles prompting their attrition) are a common phenomenon, but they tend to get resolved through continuous demand from poor rural families wishing to acquire land and substituting for those who desist. Economic and financial sustainability is likely for those settlements already producing at levels above subsistence, but much depends on families ability to access technical assistance and additional sources of credit The longer-term sustainability of Crédito Fundiário itself, given the Federal Government s full institutionalization of the program, appears very likely. Government s issuance of a Decree sustaining key elements of the methodology, and the institutionalization of that methodology within Government s National Agrarian Reform Program indicates commitment. Factors which may bear on longer-term sustainability include the Federal and State elections in 2010, the generalized economic crisis, and the Bank s departure from the program. 5. Assessment of Bank and Borrower Performance 5.1 Bank Performance (a) Bank Performance in Ensuring Quality at Entry Rating: Moderately Satisfactory The Project represented the logical scale-up of a tested and evaluated methodology deemed important by the Federal Government on the grounds of emerging evidence of costeffectiveness, ability to complement existing agrarian reform schemes, impact on poverty and lack of conflict. The Bank responded promptly and effectively to the Government s interest in expanding the program. Project design was improved based on lessons learned and solid evaluation results of the Cédula da Terra pilot phase. The consultation process was proactive and the inclusion of CONTAG as an active implementation partner was an innovative and politically astute step which strengthened the institutional make-up of the project. Also, the Bank s insistence on maintaining the key features of the model (e.g., self-selection of beneficiaries, transfer of investment funds directly into beneficiary accounts, establishing limits to financing packages) was appropriate. On the other hand, some design issues were not adequately addressed, justifying a less than satisfactory rating for the Bank s performance at entry (as noted in para ). In particular, the rationale for including the Southern States was not properly justified. Despite, or perhaps because of, the two year delay between appraisal and approval, the Project was not ready for launching in late 2001 due to several operational issues related to the complex institutional arrangements with participating states. Although improvements were made based on the Cédula da Terra experience, the allocation of resources for technical assistance to beneficiaries was neither sufficient, nor did it ensure the physical availability of such services in all project sites. Likewise, the Bank team could have paid greater attention to the monitoring arrangements in place at the time of approval, particularly those related to tracking beneficiary repayment of land loans and provision of adequate technical assistance. Further, as noted in , many risks cited in the PAD were correctly projected, but mitigation measures were not always realistic. Some potentially important risks were not identified, such as the risk of delays resulting from the readiness of the states and institutions to launch implementation. And last but 26

37 not least, the preparation team was over-optimistic, based on the encouraging results from Cédula da Terra, about the speed of project implementation. The project took eight years (resulting in three extensions of the closing date) instead of the four years planned at appraisal. (b) Quality of Supervision Rating: Moderately Satisfactory Supervision of this Project was less than fully satisfactory and challenging due to its national scale, logistical issues, and modest supervision budgets, slightly higher than for a typical one-state CDD project. The Bank team simply underestimated the logistical and budgetary implications for the Bank s ability to supervise a Federally-financed project implemented in 14 states and thousands of geographically dispersed rural communities. The system shows 58 staffweeks were expended post-effectiveness to supervise a single project with the physical equivalent of 14 separate state projects, which in retrospect is quite low compared to regional norms and especially low given the decentralized, community-based nature of the methodology and the geographical dispersion (even within a single state) of the investments financed. The system also shows another 22 staff-weeks (under Lending) reflecting the intensive efforts by the Bank team to resolve institutional and related issues between Board approval and Effectiveness (see 2.2.5). Given budgetary constraints, supervision missions to individual states were thin and far between. A few States - generally those with better performance and where the STUs demonstrated strong commitment - received more than two-three visits over the full course of project execution. The Bank standard of two supervisions per year/per project was not feasible in every participating state, physically or financially. Various measures were adopted by the Project team to compensate and were helpful (e.g., encouraging multi-state meetings of STUs, combining Crédito Fundiário missions with supervision missions of the Rural Poverty Reduction Projects in the Northeast) but these could not substitute completely for more numerous visits to the States and especially to beneficiary settlements. 30 Also, more intensive focus was needed on States which were not doing so well. The Northeast is likely to remain the primary focus of Government s land re-distribution efforts under this program and thus more intensive capacityand commitment-building in weaker-performing states would have been an investment in the program s future sustainability and expansion Bank supervision performance was less than satisfactory also due to overly optimistic Project ratings, especially prior to Satisfactory ratings for DO however, were justified because evaluation results utilizing accepted methodologies were already showing positive outcomes for beneficiary incomes and wellbeing. Procurement oversight prior to 2006 was conducted through random sampling/analysis of contracts by an experienced member of the Project Technical Team. Supervision missions, to the extent possible, included fieldwork and verification of the physical status of sampled SICs. Supervision reporting was comprehensive, and reflected a consistent problem-solving approach designed to assist the STUs and beneficiaries. 30 The number of staff weeks shown by the system (see Annex 4) may not have picked up the time spent on supervising the Project via its joint supervision with the Rural Poverty Reduction Projects in the Northeast. 31 To leverage quality oversight: (a) Bank missions to the States were always accompanied by managers and senior technicians from the NTU/MDA, the National CF Program and CONTAG; (b) annual seminars coordinated by the Bank and MDA brought STUs and key stakeholders to Brasília or Recife for collective review of progress and issues; (c) supervision of the Northeast Rural Poverty Reduction Projects often included supervision of the CF since the STUs in most Northeast States coordinated both projects; and (d) the Bank s regional office in Recife maintained constant telephone and contact with all involved STUs. 27

38 The Bank team maintained excellent working relationships over a long period with MDA/NTU and all decentralized stakeholders. The Bank s consistent support for the impact evaluation program left an important research archive on an innovative and successful agrarian reform methodology. 32 (c) Justification of Rating for Overall Bank Performance Rating: Moderately Satisfactory The Bank correctly judged the timeliness, political acceptability and methodological soundness of the Project but may have under-estimated its logistical, budgetary and institutional implications. Some aspects of project design were not analytically sound and the complexities of launching full project execution suggest that important institutional/other arrangements and decisions were still pending at Effectiveness. Supervision presented challenges for which the Project Team tried hard to compensate but which in key aspects, were un-resolvable. Even so, the Bank s consistent approach throughout to high quality impact evaluation, combined with intensive and successful efforts in the final year to ensure that the Project methodology was formally adopted by the Federal Government in direct response to the latter s announced intention to assume full financial ownership of the Project, were singular achievements which merit special commendation. 5.2 Borrower Performance (a) Government Performance Rating: Moderately Satisfactory The commitment of MDA leadership was solid from project inception through Closing, but the trajectory of the Project and MDA s ability to influence the budget process and leverage greater support for the Project represented a continuous struggle. MDA played an important role in boosting the decentralized approach, bringing in social organizations as local partners, seeking integration through the settlements with other rural support programs and policies to leverage a framework of support and ensuring that the STUs received training in key operational and administrative elements of the project. Senior MDA managers involved in the project regularly accompanied Bank supervision missions and the Ministry s relationship with the Bank was consistently supportive. (b) Implementing Agency or Agencies Performance Rating: Moderately Satisfactory National Technical Unit: The NTU had experienced and capable managers committed to the Program. MDA always maintained an NTU supervisor in each State with adequate facilities, and NTU professionals frequently accompanied Bank supervision missions. On the down side, however, the NTU was linked to the Federal Government through the institution of administração direta which intensified the bureaucracy affecting every stage of the Project. Processing and approvals especially of procurement - were extremely slow and this affected, inter alia, the timing and contracting of the final impact evaluation of CF (now planned for late 2009). Also, while individual states had good MIS in place, evidence suggests the NTU was lax 32 The ICR recommends translation and publication in English of the MDA/FECAMP final evaluation report, Estudos de Reordenamento Agrario Avaliação de Impacto do Projeto Cédula da Terra, Fundação Economia de Campinas, FECAMP-UNICAMP, 1 st Edition, Brasilia 2007, synthesizing the results of well over a decade of field work and analysis utilizing global standards. 28

39 in ensuring that comprehensive project data was aggregated, transmitted and accessible centrally. Further, the NTU s Borrower Completion Report, while including a field survey of beneficiaries and stakeholders, lacked most of the essential data, analysis and insights which the Bank team had requested well before Closing. Important information had to be rounded up piecemeal, ex post. Thus, taking all this into account, the rating for the NTU is Moderately Unsatisfactory State Technical Units: The Project was Federally-financed, coordinated and evaluated, but day to day execution of both the SAT and SIC activities rested with the STUs. Performance varied considerably between States, as did their commitment to the project. Most of the Northeast STUs were simultaneously coordinating the Rural Poverty Reduction Projects, resulting in massive workloads and in many cases, bureaucratic back-ups affecting timely processing. Some States which had not participated in the previous project launched activities with insufficient technical personnel for field work, and/or inadequate systems. Even so, several exceedingly poor Northeast States e.g., Piauí, Maranhão and Bahia, showed exceptional performance, accounting for 66% of total families settled. The issue of a minimum uniformity of institutional and operational capacity, commitment and readiness at the State level merited greater attention at appraisal. On balance the rating is Moderately Satisfactory CONTAG: CONTAG was a critical partner in project implementation but it is not considered an official implementing agency. However, the quality of its performance merits comment. CONTAG and its State and local affiliates made a strong operational contribution to the project at all levels. Such a partnership merits replication in other projects with decentralized execution. CONTAG s activities were complemented in the South by FETRAF-Sul (Federation of Family Agriculture Workers). These organizations supported information campaigns, mobilization and organization, participated in the Federal, State and Municipal Councils and in Project impact evaluation, enhanced overall transparency and acted effectively as interlocutors between public institutions and organized civil society. They gave the project visibility and credibility in the countryside. CONTAG s performance is rated Satisfactory. (c) Justification of Rating for Overall Borrower Performance Rating: Moderately Satisfactory The hierarchical aspect of project management, variable commitment and capacity levels as well as implementation context, make generalization difficult. The factors mentioned above reflect a complex situation which, in aggregate, merits an overall rating of Moderately Satisfactory. 6. Lessons Learned (a) The community-based approach to land reform, relying on negotiated land purchases at market prices and community implementation of investment subprojects, can be an agile and effective mechanism to redistribute land to poor rural families. From identification to purchase the process has been condensed into about six months and is generally conflict-free. Selfselection can effectively pinpoint the rural poor, including the entrepreneurial poor, those more likely to settle and produce successfully. The vast majority of beneficiaries have household incomes and characteristics consistent with targeting rules and this is not compromised by scaling up. Communities are adept at managing the trade-offs between land cost, size and on-farm investment funding within the family financing package provided. Independent evaluation demonstrates the financial and economic viability of the methodology. In semi-arid zones, financial estimates are reached where water is adequate. 29

40 (b) Even with a simple, beneficiary-driven model, the political economy and operational aspects of land reform introduce significant risks which potentially can delay or derail the process. From the outset (even under the Cédula da Terra pilot phase), the project had to confront considerable political opposition from certain groups which led to considerable delays in project approval and the formalization of institutional arrangements. Operationally, a community-based approach to land reform meant that Federal and State agencies had to adjust their traditional public investment and procurement practices in innovative ways to be able to respond effectively to the highly dynamic nature of rural demands for land acquisition and investments. Both of these factors consumed a lot of time and resources, in Government and by the Bank. Accordingly, land reform projects more generally, even with simple designs, must be designed with conservative implementation time horizons and must allocate sufficient resources for preparation and supervision. (c) The impact evaluation of any redistributive land reform program is technically complex and requires long-term data-collection and institutional commitments. Given that the bulk of the farm settlements and investments under the Crédito Fundiário phase of the program took place in the period, there has not been sufficient time to carry out a detailed longterm impact evaluation of direct project beneficiaries. Detailed results of impact evaluations of Cédula da Terra beneficiaries are based on a considerably longer time horizon. (d) Partnerships with grassroots stakeholders such as CONTAG and its affiliates are essential in the design and implementation of a land reform project. These partners can make a major contribution helping to de-fuse potential controversy/conflict, forming a bridge between Government, civil society, and the Bank exploiting organizational capillarity, and supporting the project operational strategy. Positive synergies can emerge when such groups are assigned specific responsibilities such as information dissemination, mobilization, organization, implementation monitoring, evaluation support, and technical assistance to beneficiary groups. Their contribution to the project should not be purely ad hoc and should be monitored for quality and consistency. (e) Beneficiary attrition from a settlement should be seen as a natural occurrence and can have many underlying causes. Even so, project evaluation shows that there has been a vigorous process of substitution of new settlers on vacated lots - virtually matching the rate of desistence - indicating that the perceived value of the opportunity to settle, and under the regime of a repayable loan, has not diminished. Further, the settlements financed under the project are generally sustainable, at the very least as legally secure residential locations for poor rural people if not always as the primary source of productive income. (f) Technical support for the land acquisition process needs rigorous assessment of the enabling conditions for beneficiaries agricultural activities and access to productive and social services, especially in the Northeast. Beneficiaries tendency to select properties in their regions of origin may be counter-productive if that region is poorly-endowed and remote from essential services. The technical screening of land purchase proposals should not shy away from advising groups against purchase if analysis warrants such action. That said, evaluation studies show no evidence that the program has any tendency to allocate poor quality land to beneficiaries, i.e., the location of farms in terms of regions with better or worse agricultural potential are proportional to the distribution of key characteristics within the states (e.g., drought risk, access to roads, agroecological conditions). 30

41 (g) Differentiated, area-tailored financing of a linked package can provide incentives and room to maneuver to associations/families to negotiate with sellers for the lowest market price. The lower the land price paid, the more resources which remained for on-farm investments which were grant financed, i.e., non-repayable. Additional, targeted incentives added to the SAT/SIC package also proved an effective catalyst to more conservative land management practices on-farm. (h) While access to and use of technical assistance and rural extension has increased markedly over the course of the project, it is still not reaching many beneficiaries and their responses suggest its quality is variable. A major issue is the greater degree of technological sophistication required from TA services once families move into more complex activities associated with marketable production, and the risk that providers especially in more remote rural areas - are not technically-equipped to deliver the type of services farmers need to manage this next phase. Technical assistance in sound natural resources management is also a priority need. Sources/arrangements for technical assistance need operational analysis and planning during project preparation. (i) Initial SIC on-farm investments are insufficient and in fact are not designed - to lift families above subsistence-level farming and launch full-scale income generating activities, especially in semi-arid areas. Settlers require continuous investment and thus SIC funds need to be complemented as soon as possible after settlement by other credit lines for investments in perennial crops, cattle or goat-raising and, in the Northeast region, irrigation, for sustainable income growth. Project design/appraisal needs to analyze the sources and availability of alternative credit/financing and facilitate settlers access. The need to repay a land loan puts further pressure on establishing viable income generation activities, whether on or off-farm. (j) While the property acquisition and settlement process is commonly collective, there is a marked beneficiary tendency and indeed, preference, for individual as opposed to group or collective productive activities. The benefits of collective activity emerge at the processing and marketing stages of production. Perceived pressure to farm/produce collectively is cited as a key reason for families abandoning the settlement for other locales and occupations. Under a selfselection methodology, the purchasing association can be composed of families with no prior connection, increasing the potential for stress over collective activities. Specialized technical assistance is needed to foster forms of collaboration and cooperation which do not impose a collective straitjacket on individual endeavors. (k) The Project was important as a catalyst to beneficiaries leveraging and integrating the benefits of other Federal and State programs. A relatively high proportion of settler families had already accessed PRONAF-A credit within their first two years which, in combination with the basic grant-based start-up capital of the SIC, enabled them to move beyond subsistence to launch productive activities with market potential. Other programs in rural electrification (Luz para Todos), health and education were also accessed more rapidly by virtue of families status as land reform beneficiaries, as members of legal associations and as families with stable locations. (l) It proved critical that a PRONAF-A credit subproject be concluded before the SIC funds were fully applied since most SIC funds financed basic infrastructure and subsistence production, leaving little for more market-oriented productive activity. Ensuring SIC and PRONAF-A credit availability overlapped ensured continuity in constructing the economic base of the settlement and boosted its sustainability by generating enough income to maintain a family over time and to repay the land loan. 31

42 (m) Monitoring arrangements and performance under a Federally-financed project implemented at the State and local levels require close supervision to ensure that key data is collected, stored and transmitted regularly by outliers to the center for aggregation. Linked management information systems, i.e., state systems linked to a central system, need regular technical assessment by Bank specialists for performance and content quality, and data collected should inter alia, cover the performance of all project components and activities. (n) Working with small, poor communities in remote rural regions entails a series of constraints which may merit greater flexibility and creativity in the approach taken to applying Bank standards. Community subprojects executed in remote localities with difficult access imply the need - in the case of procurement - to purchase and contract directly at pre-determined values, given limiting factors of location and the impossibility of conducting even minimally competitive procedures in specific cases. (o) Projects with Federal financing and oversight and decentralized implementation via CDD methodologies require site-specific budgets. Supervision of multiple states over vast areas, including hands-on monitoring of remote and dispersed settlements is a challenge. Close Bank supervision of subprojects is essential in States where execution is fast-paced to ensure compliance with the Operational Manual. However, the natural tendency to focus on the more successful and committed states, especially if supervision resources are tight and logistics difficult implies neglecting states whose performance could improve with proactive attention from the Bank. Supervision should not have to choose between winners and losers. 7. Comments on Issues Raised by Borrower/Implementing Agencies/Partners (a) Borrower/implementing agencies 7.1 Borrower comments were incorporated in the report, as appropriate. 32

43 Annex 1. Project Costs and Financing (a) Project Cost by Component (in EUR Million equivalent) Components Appraisal Estimate (EUR millions) Actual/Latest Estimate (EUR Percentage of Appraisal millions) LAND PURCHASE FUND COMPLEMENTARY INVESTMENT SUBPROJECTS COMMUNITY DEVELOPMENT SUPPORT (STATE LEVEL TA/TRAINING) PROJECT ADMINISTRATION, SUPERVISION AND MONITORING IMPACT EVALUATION AND INFO DISSEMINATION Physical Contingencies Total Baseline Cost Price Contingencies Total Project Costs Front-end fee IBRD Total Financing Required (b) Financing Source of Funds Type of Cofinancing Appraisal Estimate (EUR millions) Actual/Latest Estimate Percentage of (EUR Appraisal millions) Borrower N/A EUR EUR State Governments N/A EUR EUR Community Associations N/A EUR EUR IBRD 34 N/A EUR EUR Financed 100% by Federal Government. 34 Note that a partial cancellation in September 2004 reduced the IBRD loan amount from EUR million to EUR million. 33

44 Annex 2. Outputs by Component The following summarizes the component implementation experience and its outputs, as well as explaining aspects of the project methodology, operational mechanics and experiences where relevant. 2.1 Component 1: Land Purchase Funds (EUR173.7 million, total cost plus contingencies, Federal Government-financed), for land purchases (known as SAT Land Acquisition Subprojects) by self-selected communities of poor rural people without land or with insufficient land. Land purchase entailed taking a loan financed from Federal Government budget sources representing counterpart funding for the complementary Bank-financed activities known as Community Investment Subprojects (SIC). Financial agents/banks appointed by the Ministry of Agrarian Development (MDA) administered and disbursed the funds to associations for approved land purchase proposals. Land purchases were collective financial obligations formalized by an agreement between the financial agent and the community association. 2.1 Land Purchase Cycle: (a) Community associations identified suitable lands and negotiated their purchase with willing sellers. Communities were often guided in this process by rural labor unions, NGOs and others; (b) Communities presented a proposal to the Municipal Rural Development Council (CMDR) including the owner s declaration of willingness to sell, request for land purchase financing and choice of local financial agent/bank; (c) CMDR analyzed eligibility of community members, general characteristics of the land for sale and fairness of the price in relation to local land markets, making a recommendation to the State Technical Unit (STU); (d) STU processed the proposal confirming that the land title was clear with no liens, no other conditions threaten the purchase and that the land price was consistent with markets; (e) STU authorized financial agent to enter into a loan agreement with the association; (f) Financial agent concluded the loan agreement with the association and disbursed loan funds; and, (g) Communities decided internally on the division of the land among members and corresponding payment obligations. 2.3 Criteria governing land acquisition: (a) Eligibility: Candidates for settlement had to: (a) be organized into a legal association; (b) have family income less than R$5,800/year and family assets less than R$10,000, not including housing; (c) have minimum five years of experience in agriculture; (d) not have previously benefited from another land reform program; and (e) not be employed by the public sector; (b) Land selection criteria: Criteria for acceptability of a property included: (a) ineligible for expropriation, i.e., area less than 15 fiscal modules (FM), or area superior to 15 FM but considered productive; (b) owner had legal and legitimate title to the property, and there were no liens or conditions which would impede its legal transfer; (c) not situated in an indigenous reserve or in areas protected by environmental legislation or bordering such areas; (d) a price consistent with the market and physical conditions - either already existing or which could be implemented through program financing - permitting its sustainable use; (c) Financing Package: At appraisal, the total value of financing allocated to each family varied from about R$12,000 to R$18,000 depending on land markets in each participating State See footnote 7 on the Southern States. 34

45 Additional credits could be allocated, e.g., families in semi-arid areas received an added R$2,000; another R$1,000/family was available for families who developed environmental recovery activities; and, associations of women and youth could obtain another R$1,000 for especially innovative subprojects and/or R$1,000 for agro-ecological subprojects; and, (d) Land Loans: The land loans were collective financial obligations formalized through contracts signed between participating banks (Banco do Brasil, Banco do Nordeste do Brasil) and the purchasing community association. Division of the property purchased and individual family payment obligations were agreed within the beneficiary association. Terms of land purchase financing were those of the National Land Fund, i.e., financing up to 20 years with three years of grace at a nominal interest rate not to exceed 12%. 36 Up to eight percent of the total land purchase cost could be used by the community association for technical assistance, 50% of which could be used to prepare the land purchase proposal itself. See Annex Financing terms for land purchase: Financing followed those under the National Land Fund (Lei Complementar No. 93 of 1998) including financing up to 20 years with three years of grace at a nominal interest rate not to exceed 12% per year. At the time of project appraisal, financing conditions for loans of the amount to be paid under the project (i.e., <R$15,000) provided for a real interest rate of 4% with a 50% rebate on the interest rate for associations located in difficult/poorest areas, e.g. semi-arid zones. The Federal Government and participating states concluded a formal agreement on the annual allocation of federal land account resources among the states. Up to 8% of the total land purchase cost was made available to associations for technical assistance; 50% could be applied to the cost of preparing the purchase proposal and the rest could be used for productivity enhancing technical assistance over a three-year period. 2.5 Results/Outputs: The project financed 2,114 separate land acquisition projects covering properties totaling 831,602 hectares in 14 states and benefiting 40,102 families. In addition, at Closing, another 284 land purchase proposals representing 8,300 families were approved, under contract and/or waiting for financing by the Financial Agents. These are being financed with national resources. Further, proposals representing another 22,170 families were registered by the STUs before Closing and were awaiting further analysis. Total cost of this component was EUR million, 80% of the estimate at appraisal. 2.6 Relative performance between participating states was sharply differentiated with the exceptionally poor States of Piauí and Maranhão far exceeding most other states in numbers of land purchase proposals financed and beneficiary families settled. Three of the four participating Southern states the exception being Espírito Santo - showed very modest performance due to high land prices and other causes. The relative speed and efficiency of implementation was mainly due to the massive supply of land and low prices in certain Northeast states; differentiated levels of State commitment to the Project; and, organizational capacity and adequate, wellprepared staff. All three explain the success of Piauí which settled 11,100 families. Whether a State had participated/not in the CdaT pilot does not appear to have been a critical factor in its performance under CF - Piauí had not done so. Maranhão had participated and performed exceptionally well under CF until poor follow-up to existing settlements and administrative issues 36 Financing conditions at the time of appraisal for loans of the amount to be paid under the project (under $15,000/family) were a fixed real interest rate of 4% with a 50% rebate on the interest rate in the poorest regions of the country. 35

46 suspended operations there in the last 15 months. The State of Ceará which had good performance under the previous project was unable to repeat this under CF. 2.7 The original idea was to allocate 10% of the budget to the four Southern States for a pilot. Once it became clear that demand was low due to high land prices, resources applied for SICs in this region were below one percent of the total. While each state presented an annual POA indicating the number of families to be benefited and the resources needed, release of resources to each State was based on demand and an STU s relative operational efficiency. It was common for a group of states to not reach their targets and for others to exceed theirs. The State of Piauí, which experienced exceptionally strong demand and showed strong operational capacity, benefited from this strategy and settled 11,100 families. Table 2.7.1: Performance of Participating States at end-project (Values R$ 000) Indicator AL BA CE MA PB PE PI Land Purchase Families 914 4,953 2,225 8,567 2,658 2,350 11,100 Area (ha) 11,342 87,953 65, ,566 59,086 48, ,820 Value SIC Value SAT Value: SAT+SIC Indicator RN SE MG ES PR RS SC Total Land Purchase Families 3, , ,106 Area (ha) 75,014 9,018 15,763 7, , ,602 Value SIC Value SAT Value: SAT+SIC Land Prices: Average prices of land per ha in 2007 and 2008 were dramatically higher in all four Southern states (Espírito Santo, Paraná, Rio Grande do Sul and Santa Catarina). The lowest average price was in Piauí (R$228/ha) - which also had the largest numbers of land settlements financed and highest in Parana (R$6,917/ha) which financed only three settlements in the project period. Average cost of land purchase per family shows the same pattern. The overall average per family for the 14 states in was R$8,744 with the highest cost per family in Rio Grande do Sul (R$34,831) and lowest in Piauí and Maranhão (R$4,438 and R$5,420 respectively). Land prices in the Southern states represented an absolute and insoluble barrier to targeted families effective participation in the project and in this regard, the planned pilot experience in that region was unsuccessful. Table 2.8.1: Status of Project including Nossa Primeira Terra at Closing (Values in R$ 000) Indicators Total Lands Purchased ,114 Families 2,710 4,310 6,006 7,309 8,828 7,226 3,717 40,106 Area (ha) 67,490 98, , , , ,090 73, ,602 Value SAT Value SIC Value SAT/SIC

47 Table 2.6.2: Physical Performance vs Family Demand, NE States State Demand as % of Total Eligible Families No. Families Attended Families Attended as % of Total Demand Alagoas Bahia , Ceará , Maranhão , Paraíba , Pernambuco , Piauí , Rio Grande do Norte , Sergipe Total: , Demand: As seen in Table 2.6.2, there was a significant variation in physical performance between different states, as measured in terms of family demand and families benefited. The States of Bahia, Ceará and Pernambuco had relatively low performance, with aggregate demand of 54.94% of all eligible families but attending only 25.31% of total families benefited. On the other hand, States such as Piauí, Maranhão and Rio Grande do Norte, with 27.41% of total demand, were able to attend an aggregate 62.50% of total families benefited. This demonstrates the need to exert greater effort in negotiating with participating State Governments to gain greater commitment to the program and consequently, stronger support for its operational execution in States where performance was lower than expected. 2.8 While beneficiaries complained about the time taken to process land purchase proposals due to complex bureaucratic requirements, Table shows that the average time from receipt of a proposal to its contracting declined sharply from due to STU s efforts, in collaboration with the Bank and other support entities such as IICA, to simplify procedures. Table 2.8.1: Processing Period by Year, No. of Proposals Average Year % Time to Received Contracted Contracted Contract (days) ,749 1, Municipalities benefited by the Project in the Northeast States, compared to total municipalities in each State show a high concentration of contracted operations in a reduced number of municipalities as shown below which shows municipalities attended in relation to the total in each State: State Alagoas Bahia Ceará Maranhão Paraíba Pernamb. Piauí RGN Sergipe %

48 2.10 In the Northeast region as a whole, the Project attended 681 municipalities (about 38% of total municipalities nation-wide. Comparatively low numbers were attended in Bahia, Pernambuco, Sergipe and Paraíba while the highest were in Rio Grande do Norte and Piauí. This concentration in a reduced number of municipalities was even stronger when taking into account that the municipalities with more than 100 families attended corresponded to 8.1% of the total but 54.1% of all families benefited The overall operational performance of the Project, is shown in Table below: Table 2.9.1: Operational Performance, Year Aver. # Families per Settled Group Aver. # Ha per Family Aver. R$ per Family 38 Aver. Value of SAT per Family Aver. Ha per SAT Aver. Value of SIC per Family ,230 13, , ,594 13, , ,982 11, , ,996 7, , ,595 7, , ,008 8, , ,978 8, ,354 Total ,669 9, , Families Settled vs Land Size: As shown, the average number of families per settled group and average size of land area per family declined 22.4% and 20.5% respectively from These ratios are considered satisfactory in view of the following: (a) the average of 20 families per settling group has been found by project evaluation to be adequate for achieving efficient levels of organization with minimum scale of agricultural production which justifies and gives viability to collective activities for product processing, mechanization, production of inputs (mainly seeds and seedlings), transportation, storage and other aspects of economic and social progress/survival on the settlement; (b) an area of 20.7 ha per family, despite some opinions that it is insufficient, can also be considered adequate since it is higher than the average family farm in the Northeast region (12.9 ha/family), although somewhat below the average area of INCRA expropriation-based settlements which in the Northeast average about 29.0 ha per family Land Cost: The average cost of land purchase per family (SAT/family), as well as the cost per ha of areas acquired, showed considerable decline from , falling to about 57.5% of the average value in The trend rose again in 2007 to about 68.3% of the average of the first three years. This can be explained as much by the increased efficiency of beneficiaries in negotiating land prices, as by the incentives to obtain prices below those generally obtained in a given municipality, but also by families desire to generate increased resources from the overall financing package for on-farm SIC investments designed to facilitate family settlement and jump-start productive processes, by spending less on land. The impact on average SIC prices over time can be seen in Table above. Average SIC values rose 56.3% in compared to the average of the first three years Southern States: Data available on the Project experience in the Southern States shows significant variation in average numbers of families per land purchase subproject and average area per family, but the general trend is much lower than the Northeast. Similarly, average cost of

49 the SAT land portion and average cost per hectare are much higher. Given that the average size of the overall settlement package in the South was also much higher, the low average value of SICs is quite striking and either reflects lower settlement costs in general (the region is more fertile, wetter and agro-ecologically superior), or that even with larger packages of funding, land costs were prohibitive and left little for SICs. Table : Project Performance in the South and Southeast States, States Indicator Minas Gerais Espírito Santo Paraná Rio Grande do Sul Santa Catarina Families per land settlement subproject Area per family settled (ha) Value per family settled (R$) 18,134 22,676 37,036 37,630 31,834 Value of land purchase per family (R$) 10,483 16,748 28,289 32,625 29,321 Value of land purchase (SAT) per ha (R$) 664 2,994 6,134 4,088 4,561 Value of SIC investments per family (R$) 7,651 5,928 8,747 5,005 2, Land loan repayment: A critical indicator of beneficiaries confidence, motivation, and perception of moving out of poverty is their capacity and willingness to repay their land loans. In this regard, project performance was outstanding and consistent both with other Bank projects involving loans to small, poor borrowers and with the global research literature on this subject. The performance of Espírito Santo, a Southern State, contrasted sharply with that of the poorer Northeast States but with about 86% still paying on time, the result remains very satisfactory. Table : Land Loan Repayment Rates, 11 States, December 31, 2008 State Beneficiary Debt Outstanding Beneficiary Over-dues as at Percentage R$ R$ Alagoas 12,271, , Bahia 31,862, , Ceará 17,199, , Maranhão 38,600, ,229, Minas Gerais 11,464, , Paraíba 23,262, , Pernambuco 19,727, , Piauí 44,913, ,037, Rio Grande do Norte 39,004, , Sergipe 12,157, , Espírito Santo 19,580, ,718, Total: 257,888, ,249, Source: Bank of Brazil, Bank of Northeast Brazil 2.15 Component 2: Community Subprojects (EUR million, total cost plus contingencies, Bank-financed) financed complementary investment subprojects and technical assistance to establish and consolidate land settlement and improve productivity. Adhering to many of the features governing the CDD methodology of the Northeast Rural Poverty Reduction Projects, financing was in the form of matching grants to the associations for small-scale investments identified by the group as a priority. Subprojects could be of many types broadly categorized as infrastructure (water supply, electrification, local access road improvements), 39

50 social (school or health post renovation) and productive (small-scale communal agro-processing, communal tractors, minor irrigation schemes). Selection was demand-driven with a short negative list of ineligible types. The STU would appraise proposals on technical, environmental and institutional grounds. Financial agents (Banco do Nordeste, Banco do Brasil) would then disburse the funds for approved investments directly to the association s bank account. As noted in the main text, an important change resulting from experiences under the preceding Cedula da Terra project resulted in the SIC funds being deposited immediately in the association s bank account upon signature of the land purchase agreement. Funds were blocked in the account pending an STU s approval of a SIC proposal at which point they were released to the association Total available financing for a community association the sum of the land loan, grant for community investment, and an initial cash grant for settlement needs in the first year (about R$1,300) was subject to a ceiling ranging from R$12,000 to R$18,000 per family. Participating states could reduce this ceiling if they wished. Communities could decide how to manage the total amount available by purchasing more expensive land or investing more in productive facilities. Beneficiaries were expected to contribute a minimum 10% of the SIC subproject in cash, kind or labor. Minimum levels of cost-sharing were also expected from participating municipalities and State Governments. An Operational Manual specified these and other project rules and parameters. Subproject selection would be demand-driven with a short negative list of prohibited investments and proposals would use standard documentation and technical, environmental, economic and sustainability criteria set down in the project Operational Manual It is worth mentioning that project beneficiaries also had guaranteed access to the National Program to Strengthen Family Agriculture (PRONAF) under a special credit line for agrarian reform beneficiaries (PRONAF-A), which permitted them to develop and consolidate productive activities. SIC financing could not and was never intended to finance the full gamut of costs, even initial, involved in building a settlement. Beneficiaries also had access to other programs implemented by MDA such as market access support, government purchasing of family agriculture products, minimum price supports, crop insurance, and specialized PRONAF lines for women and youth. The Borrower Completion Report (2009) notes that project-financed evaluation studies show that beneficiary access to income transfer programs has grown: in 2003, the level of access was about 41% and by 2006 had increased to 61% Results/Outputs and Costs: SIC-funded investments covered a wide range of smallscale infrastructure, productive and social facilities, equipment and machinery among which the most prominent were: water storage and supply facilities/equipment; irrigation kits; housing rehabilitation and construction; agro-livestock installations; small-scale grains/other processing equipment and machinery; land preparation and planting equipment; manioc mills; animal traction and transportation equipment; animal assets (swine, cattle, goats, birds); tractors; electricity connections; and access road improvements The average cost per family for SIC investments rose slowly but steadily over the course of the Project, from R$7,602 in 2002 to a high of R$9,662 in 2008, an increase of about 27%. The average value of SIC investments varied quite sharply between participating states, the lowest being in the Southern States of Rio Grande do Sul and Espírito Santo (R$5,273 and R$6,283 respectively) due to the higher cost of land absorbing a greater portion of the total financing package available to each family. SIC value was highest in the States of Minas Gerais and Bahia (R$11,062 and R$10,787 respectively) due to lower land prices leaving more resources available for on-farm investments. Total cost of the component was EUR million, 70.6% of the appraisal estimate. 40

51 2.19 Component 3: Community Development Support and Strengthening (EUR 13.3 million, total cost plus contingencies) was designed to strengthen the effectiveness and quality of project operations by financing technical assistance, seminars and training courses for community associations, and information dissemination campaigns to educate potential beneficiary communities/stakeholders about the project, its objectives and guidelines and to promote awareness, transparency and participation. Total cost at Closing was EUR million, 89.0% of the appraisal estimate Results/outputs: The component financed a large number of diverse capacity-building exercises including: (a) Mobilization assistance for community associations, especially those lacking organization and social capital and not in condition to benefit from the project. It is important to understand that self-selection meant in many cases/settlements that members of a beneficiary association had not necessarily known each other prior to forming the association and thus mobilization and organization activities assumed great importance for future prospects and sustainability; (b) Specialized skills training involving on-demand and on-the-job guidance for the STUs and associations in monitoring techniques, financial administration, procurement, and specific technical training supporting subproject implementation, operation and maintenance; (c) Best-practice exchanges, mainly through annual seminars for STU managers and technicians from all participating states, as well as Federal authorities, an especially valuable exercise over the course of project implementation, complemented by smaller gatherings of representatives from several states as needed in the Bank s Brasilia and Recife offices; (d) Publicity campaigns, disseminating the availability, principles, eligibility and methodology/mechanisms of the project to promote transparency and rapid project execution. These activities were successfully conducted throughout the life of the project via diverse media radio, television, videos, public meetings, field days, seminars, and paper media specially designed for the targeted public in graphic, readily understood form. (e) Inter-American Institute for Cooperation in Agriculture (IICA): Special commendation is warranted for the important role played by IICA in project administration, especially in technical assistance to the NTU and STUs. All the FECAMP and FEALQ evaluation studies were contracted via IICA, as were professionals for the NTU at Effectiveness. IICA was an important catalyst to developing the networks needed to support the Project in the countryside. They organized and trained producer groups and assisted settler associations in preparing and transmitting land purchase proposals to the STUs Component 4: Project Administration, Supervision and Monitoring (EUR 26.8 million, total cost plus contingencies) financed field supervision and monitoring, and the incremental operating costs of the National Technical Unit (NTU) and State Technical Units (STU) for overall coordination of project activities. Total cost at Closing was EUR million, 97.1% of appraisal The project s institutional structure was a complex pyramid of functions and responsibilities, as follows: 41

52 Beneficiary Associations: the institutional backbone, organized explicitly to participate under the self-selection rubric. Municipal Rural Development Councils (CMDR): established in virtually all rural municipalities nation-wide, and with prominent membership of rural workers unions, verified beneficiary eligibility and reviewed land purchase proposals. State Technical Units (STU): usually reporting to State Planning Secretariats but with significant autonomy, signed Terms of Cooperation Agreements with MDA and prepared an Annual Operating Plan (POA) defining targets, priority regions for action, eligible public and operational strategy, upon which MDA decisions regarding annual budget allocations to each state were based. 37 State Sustainable Rural Development Councils (CEDRS): the principal decision-making bodies, approving State POAs and proposals for land purchase. National Technical Unit (NTU): responsible for coordination, monitoring, reporting, supervision and evaluation, subsidiary to the Ministry of Agrarian Development (MDA) which had principal oversight of the Project; and, National Confederation of Agricultural Workers (CONTAG): a formal project partner, supported many aspects of project implementation with its Southern counterpart, the Federation of Family Agriculture Workers (FETRAF-Sul) The STUs had multiple responsibilities: (a) verifying that title to a proposed property was clean without liens; (b) that no other condition threatened the effectiveness of a proposed land purchase; and (c) that the negotiated price was consistent with market conditions. They also appraised community subproject proposals (SICs) for compliance with project guidelines and eligibility criteria in the Operational Manual; assessed community participation in identifying, preparing and executing subprojects and quality of technical assistance; supervised the quality of overall project implementation through field supervision and through the MIS; implemented public information campaigns to disseminate information about the project; periodically reported progress on project indicators; prepared annual implementation and physical performance reviews; and, submitted Annual Operating Plans to the NTU, upon which budget allocations were based. The project used standardized subproject designs for the most commonly demanded types of subprojects and standard cost indicators to ensure reasonable costs for subprojects. Results and Outputs: The STUs conducted - using either their own technicians or contracted third parties (local extension agents, technical assistance providers, NGOs and/or local labor union representatives) over 14,000 individual visits to SIC subprojects to verify legitimacy, installation progress and completion Component 5: Impact Evaluation and Dissemination (EUR 8.3 million, total cost plus contingencies) was the responsibility of the Federal Government with MDA coordinating all 37 Most STUs in the Northeast were also coordinating the Bank-supported rural poverty projects and in five Northeast states had had stable, continuous responsibility since 1997, building operational and technical expertise. In the Southeast, three of the four states already had experience of community-driven methodologies under Bank-supported land management projects implemented over the previous decade, but required training and adaptation to internalize the market-based land reform project methodology. In the Southern States, the project was coordinated as follows: Espírito Santo Institute for Agro-Livestock and Forestry (IDAF), a subsidiary of the Secretariat of Agriculture. Supply and Fisheries; Rio Grande do Sul Delegacia Estadual do Ministério de Desenvolvimento Agrário, the only case where the project was under the direct responsibility of MDA; Paraná and Santa Catarina directly administered by the Secretaria de Agricultura. 42

53 related activities. High quality, participatory evaluation using control groups/panels, random sampling, baseline and follow-up studies has been a hallmark of this program since its inception in 1996 under the Ceará pilot. The Credito Fundiario project financed: (a) Beneficiary surveybased study of Crédito Fundiário beneficiaries conducted by the Escola Superior de Agricultura Luis de Queiroz (ESALQ) in 2003; and (b) FEALQ (same institution) follow-up study in 2005 of a representative sample of the same 2003 settlements to evaluate project impact on beneficiaries; (c) FECAMP impact evaluation studies in 2003 and 2006, as the continuation of earlier studies financed by the previous, Cedula da Terra Project. Total cost at Closing was EUR 1.30 million, 17.6% of appraisal. See details in Annex Table shows updated Project output indicators at Closing: Table : Project Performance Indicators - Planned and Actual 12/31/2008 Project Components and Activities Institution Unit Planned Actual Responsible A. Land Purchase: STUs/Assns - Beneficiary families Families 50,000 40,102 - Beneficiary individuals Individuals 250, ,530 - Subprojects B. Community Subprojects STUs/Assns - Start-up Grants Families 50,000 40,106 - Investments Subprojects # 5,000 C. Institutional Development Prep. of annual program of TA and Training STUs # per State 3 6 Training and TA Courses and seminars STUs provided: - For beneficiary associations # For STUs and SLI staff # 20 Na Publicity campaigns in each State: STUs - Presentation to Bank PY 1 PY 1 - Implementation PY 1 PY 1-6 D. Project Administration, Supervision and Monitoring - Supervision of subprojects STUs # visits 12,000 14,000 Annual Operating Plans (each state) STUs # per State 4 6 Special Account established MDA PY 1 PY 1 Operational Manuals STUs - Preparation PY 1 PY 1 - Review and Adjustments PY 1-3 PY 1-6 Monitoring Reports & Reviews - Monthly disbursement & MIS update STUs # per State Semi-annual reports STUs # per State External audits STUs # per State This represents total courses, workshops and seminars conducted in the project period. The NTU s detailed record of the substance and target cohort for each event is filed in IRIS. 43

54 - Physical Performance Reviews STUs # per State Implementation Review STUs PY 2 PY 3 - Mid-term Evaluation STU, MDA,WB PY 2 PY 3 E. Project Evaluation/Dissemination Information network and dissemination MDA - Implementation activities PY 1-4 PY 1-6 Evaluation Study MDA - Detailed Design Study PY 1 n/a - Baseline Study PY 1 PY 2 - Resurveys PY 1-4 PY 2-6 Studies (Completion Report) MDA PY 4 PY 6 44

55 Annex 3. Economic and Financial Analysis 3.1 The Land-Based Poverty Alleviation Program, known as Crédito Fundiário (CF), was effective in late 2001 and overlapped with the final stages of the Land Reform and Poverty Alleviation Pilot Project, known as Cédula da Terra (CdaT), which was implemented between 1997 and The CF was the subject of a baseline survey in 2003 and an interim progress and targeting analysis in 2005 by the Fundação de Estudos Agrários Luiz de Queiroz FEALQ, University of São Paulo. A second round of field surveys will be conducted in late 2009 and a final impact analysis of CF is expected to be completed by the first semester of In addition, the CdaT project has been extensively evaluated by the University of Campinas Economics Fooundations (FECAMP) starting with a baseline in 1999, and subsequent follow-up surveys and analyses in 2000, 2003 and a final evaluation in 2006 (see Annex 5). 3.2 While there are some technical differences between the pilot (Cédula da Terra) and expansion (Crédito Fundiário) phases of the program, and their time periods were different (albeit overlapping for about 14 months), their targeted populations/beneficiaries and methodology were virtually identical and the two projects have always been viewed as a continuous program. The baseline findings of FECAMP in 1999 on the CdaT s targeted population are mirrored by the FEALQ baseline of 2003 on CF. Both studies concluded that based on maximum income at entry, occupational history, asset poverty, agricultural experience and educational attainment, the program has been reaching its targeted cohort. Incomes of beneficiaries in both projects were well below the National Household Survey (PNAD) poverty line, educational levels were very low and illiteracy was high. 3.3 The PAD economic analysis was based on stylized, representative farm models of family farms expected to be established under the project. Given that well over half of all investment under the project was concentrated in the final 3 years (2006 to 2008) and will require a period of maturation to demonstrate benefits, the final impact evaluation of the project will not be available until Thus, the ICR did not have access to farm models or other definitive outcomes on which to base a rigorous analysis of Crédito Fundiário. The economic analysis below uses household survey data from CdaT beneficiaries collected in 2000 and Given the virtually identical beneficiary profiles and project implementation methodology, these results may be considered indicative for CF. The analysis presents the impact on income, assets and agricultural value of output of program beneficiaries, as well as a cost-benefit analysis and internal rate of return from household data collected in 2000 and The analysis (Romano et al., 2008), included both control and treatment groups totaling 730 households, of which 320 were CdaT beneficiaries from 104 settlements, and 410 were control households. The survey covered five states: Bahia, Ceará, Maranhão, Minas Gerais and Pernambuco (all of which were also included in CF). 3.4 The treatment group (beneficiaries) consisted of communities whose projects were approved up to 1999 and were first interviewed in 2000, soon after they were granted CdaT financing. The control sample, however, was not included until 2006 and was selected from groups of landless that applied in that year to the new phase of the program, under Crédito Fundiário. It was therefore necessary to obtain baseline information about the control group via recall questions, applied during the 2006 field survey. Only data that could be reliably collected via recall, such as assets, were requested on the control group s 2000 situation, while income information was only requested in relation to their situation in In 2006 all data was collected from both sample groups with the same instruments, which were specifically designed for this study, thus guaranteeing total compatibility between groups in the type of information obtained. 45

56 3.5 The analysis uses three different methods for dealing with selection bias: (i) a pipeline control sample (control group selected from families in the pipeline waiting to access the project); (ii) propensity-score matching (use of relevant characteristics for matching beneficiaries and control for comparison); and, (iii) double-difference analysis (comparison not only between control and beneficiaries but also between the baseline year and 2006) for indicators for which baseline data were available for the control sample. Impact estimation using propensity-score can be performed using a number of different methods, which can then be compared to analyze the sensitivity of results to different estimation methods. Five different matching methods were used, plus the Wooldridge parametric analysis, to obtain a series of impact estimates for all indicators analyzed. In general, all methods used showed similar results for all indicators analyzed, with few exceptions, thus supporting the conclusion that these estimates are not sensitive to different methodologies. The results obtained on the impact of Cédula da Terra are robust to variations in methodologies used. The numbers presented in tables are averages of these estimation methods. Impact on Assets 3.6 The impact on total assets is obtained with double-difference analysis. The results are positive and significant in all estimations, indicating that the program (under the CdaT phase) increased beneficiaries assets by an average of R$4,540, equivalent to a 100% increase over the average total asset value they owned before being settled, a considerable sum even if one considers that this is over a period of six to seven years. If debts are discounted from the total asset value, the impact of the CdaT on net assets remains significant and positive, indicating an average impact of around R$2,300. Assets considered in this estimation do not include assets owned as a direct result of the CdaT program, such as land (using SAT financing under the project) or housing (using SIC financing). If these assets were incorporated, the impact would be significantly larger. Table 1 shows the impact of the program by type of asset, indicating that productive assets, i.e., farm machinery, tools and animals, are the main source of impact on total assets. All six estimation methods used in the analysis are presented to show their overall similarity for each estimate of impact. Impact on Income 3.7 Estimates of the impact on income are obtained with single-difference analysis since the value of baseline income for the control groups was not available. Gross annual income estimates were all positive, and statistically significant in four out of six analytical methods. Considering the average for all estimated methods, the program (under the CdaT phase) increased beneficiaries annual income by R$1,463 (or US$669), equivalent to a real increase of 46% over beneficiaries average gross income in the year before settlement, which was R$3,194 (in 2006 values 39 ). This measure of income is net of other effects on income due to the use of a control sample of non-beneficiaries. Estimates of the impact on net income were not significant and the average for all analytical methods was R$505. Income from farm production and, particularly, from consumption of own-produced goods, was positively and significantly affected by the program in all sets of estimates, whereas income from work off-farm was negatively affected. Generally, benefited farmers decreased their work off-farm and increased work on-farm (see Table 2). 39 This information was obtained in the 2000 survey, including consumption of own-produced goods. The monetary correction value is done using the variation of an average inflation index (IPC) between 1998 and

57 Impact on Value of Agricultural Production 3.8 The impact on agricultural production value is highly significant in all analytical methods used, indicating an average positive effect of about R$2,200. This indicates that the source of increased income for CdaT beneficiaries is mainly agricultural production on their land. Nonbeneficiaries have more income from off-farm work and less from own production. The impact of CdaT on intermediate indicators related to agricultural production were also positive and significant: beneficiaries increased area planted by about 20 hectares, they also had about 60% more technical assistance, about R$1,800 more in credit, and about 10% more were using mechanical or animal support/labor (See Table 2). Internal Rate of Return and Cost-Benefit Analysis 3.9 The cost-benefit analysis carried-out for the Crédito Fundiário project is very conservative in its assumptions, since it incorporates a stream of benefits of only 20 years, although benefits likely will accrue for longer given that investments in agricultural production were increased considerably and land and labor are likely still under-utilized. The stream of benefits considers the impact on gross income, and the stream of costs includes grants, administrative costs and the total value of credit provided. This latter assumption is an overvaluation of the costs since the real cost of credit would be the subsidies given and not the total amount of credit, which assumes that no loans would be repaid. 40 The analysis indicates a positive internal rate of return (IRR) of 12.6% and a net present value (NPV) of almost R$143.0 million, using a discount rate of 6%. Financial Analysis 3.10 The annual loan payment per family indicated in the Borrower s Completion Report varied from R$190 to R$1,100, with an average of around R$ The additional income obtained on average by beneficiaries due to the Crédito Fundiário program, i.e. the impact of the project on income, of about R$1,463 in gross income terms, indicates that on average, beneficiaries have the conditions to pay their monthly loan installments. The loan repayment data (as of December 31 st, 2008) showing repayment rates between 98% and 99% (except in Espiritu Santo) supports this analysis (see Table 1 in main text, or Table in Annex 2). 40 The costs of the program include the land loans (SAT), the investment grants and technical assistance (SIC), and administrative costs. In a simple analysis, usually the costs of a subsidized credit program represent the amount used as the subsidy. If the market rate is 10% and the subsidized rate is 6%, the difference (4%) applied to the total amount loaned is the cost to government of that credit program (assuming repayment of the loans occurs at the same rate as they are repaid in the broader market). In the absence of calculations of this subsidy cost (a worthwhile future exercise) and obtaining information on interest rate costs (i.e., Brazilian Government payment of interest rates on the Bank loan for the grant portion of the program costs), the analysis has simply over-estimated total costs of the credit by using the total value of the credit given as costs of the program. Since there will be repayment of the loan (and evidence shows it is high 97.6% - total costs are over-estimated by using this definition of credit costs. On the other hand, the analysis under-estimates the costs to the Brazilian Government of the grant portion of the program by not including the interest rates paid by the Government to IBRD. Given all these caveats, the analysis is simplified by over-estimating costs of the program. This implies that most likely, the IRR would be greater than calculated here. 47

58 Table 3.1: Monetary value of impact of CdaT on assets (matching estimates using double-difference analysis for six different estimation methods) 41 Estimation methods/ Asset analyzed Method: Nearest Neighbour 1 Method: Nearest Neighbour 5 Method: Radius Method: Mahalanobis Method: Kernel Method: Wooldridge Assets (except land and housing in settlement) a Total gross assets 4,525.85*** 4,657.89*** 4,517.65*** 4,213.04*** 4,787.33*** 5,496.24*** Total net assets 2,252.99** 2,427.69*** 2,270.27*** 2,087.56** 2,548.75*** 3,169.81*** Productive assets e.g., farm machinery & tools, livestock 2,512.65*** 3,122.60*** 2,805.64*** 2,996.40*** 3,090.92*** 2,566.36*** a. Monetary values in Brazilian currency (R$) of 2006 * Statistical significance of 10% ** Statistical significance of 5% *** Statistical significance of 1% 41 (a) Nearest neighbor matching: the closest or some number of closer individuals, according to the propensity score, are chosen to pair between samples; (b) Radius matching: a circle of a certain score distance is selected and all individuals within it are used to pair with the beneficiary; (c) Kernel (and Local Linear): non-parametric and utilize a weighting system for all observations in the control group in order to construct a counter-factual corresponding to each treated observation. 48

59 Table 3.2: Monetary value of impact of CdaT on income, value of agricultural production and agricultural production intermediary indicators (matching estimates using single-difference analysis for six different estimation methods, R$ 2006) Estimation Methods - Income Source Analyzed Method: Radius Method: Mahalanobis Method: Kernel Method: Wooldridge Method: Nearest Neighbour 1 Method: Nearest Neighbour 5 Income a Gross annual income 1, ,461.49** 2,295.12*** 1,323.07* 1,464.53** Net annual income , Income from salaried work (off-farm) Income from sale of own agricultural production Income from consumption of own production * * * * * * 1,149.2 *** 1,736.99*** ** 1,379.23*** 1,022.77*** *** *** 1,154.74*** *** ** Agricultural practice Area planted (hectares) 20.14*** 19.27*** 20.49*** 20.45*** 20.08*** 15.95*** Frequency of technical 0.69*** 0.62*** 0.66*** 0.68*** 0.65*** 0.64*** assistance Quality of technical 0.64*** 0.67*** 0.79*** 0.81*** 0.76*** 0.74*** assistance (subjective evaluation) Value of credit received 1,768.70*** 1,799.43*** *** 1,812.42*** 1, *** 1,927.25*** If use pesticides c -- If use fertilizers c -- If use purchased seeds c -- If use mechanical or animal 0.123*** 0.100*** 0.090*** 0.140*** 0.094*** c -- workforce If irrigated land a. Monetary values in Brazilian currency (R$) of 2006 b. *, **, *** refer to statistical significance at 10%, 5% and 1% levels, respectively c. Indicators with binary values not estimated for Wooldridge method 49

60 Table 3.3: Cost-Benefit Analysis and Internal Rate of Return for CdaT: Stream of Net Benefits in 20-year period (R$ 2006) Year Gross income and Full Costs Year Gross Income and Full Costs ,813, ,473, ,660, ,846, ,282, ,219, ,617, ,592, ,898, ,966, ,487, ,339, ,860, ,712, ,234, ,998, ,607, ,945, ,980, ,981, ,353, ,934, ,726,902 IRR 12.6% ,100,118 NPV (+) 142,472,346 Discount Rate 6% 50

61 References: Romano, Neder, da Silveira, and Magalhães The Impact of a Market-Assisted Land Reform Program in Brazil: The Case of Cédula da Terra. Working Paper, unpublished. Da Silveira, José Maria, Avaliação de Impacto do Projeto Cédula da Terra. Coordenação-Geral de Planejamento, Monitoramento e Avaliação/SRA/MDA, Ministério de Desenvolvimento Agrário: Estudos de Reordenamento Agrário, Vol. 3, Brasília, D.F. (Nº ISBN: ) 51

62 Annex 4. Bank Lending and Implementation Support/Supervision Processes (a) Task Team members Names Title Unit Responsibility/ Specialty Lending Luis O. Coirolo TTL LCSRE Agric. Economist Joaquim von Amsberg Country Economist LCC5C Economist Edward Bresnyan Senior Rural Develop. Specialist LCSAR Agric. Economist Alberto Ninio Sr. Counsel LEGLA Lawyer Joao Barbosa-De Lucena Operations Officer LCSRE Agriculturalist Tulio Barbosa Operations Officer LCSRE Agric. Economist Jose Janeiro Financial Management Spec. LCSFM Fin. Management Enzo de Laurentiis Sr. Procurement Officer LCSPR Procurement Supervision / ICR Jorge A. Muñoz Lead Rural Development Specialist LCSAR TTL, Agric. Econ. Susana Amaral Financial Management Specialist LCSFM Fin. Management Tulio Barbosa Consultant LCSAR Agric. Economist Joao Barbosa-De Lucena Consultant LCSAR Agriculturalist Edward Bresnyan Senior Rural Develop. Specialist LCSAR Agric. Economist Raimundo N. Caminha Consultant LCSAR Agriculturalist Luis O. Coirolo Consultant SASDA Agric. Economist Mark R. Lundell Sector Leader LCSSD Economist Anna Roumani Consultant LCSAR RD Specialist Luciano Wuerzius Procurement Specialist LCSPT Project Proc ment Maria de Fatima Amazonas Operations Officer LCSAR Proj. Mgmt Spec. Isabela Micali-Drossos Legal Specialist LEGLA Lawyer Claudia Romano Consultant LCSAR Agric. Economist Alberto Costa Consultant LCSAR Anthropologist (b) Staff Time and Cost Staff Time and Cost (Bank Budget Only) Stage of Project Cycle USD Thousands (including No. of staff weeks travel and consultant costs) Lending FY FY FY FY FY Total: Supervision/ICR FY FY

63 FY FY FY FY FY FY Total:

64 Annex 5. Beneficiary Survey Results 5.1 Beneficiary surveys have underpinned all major studies conducted by this Project and its predecessors. The following summarizes the methodologies and findings of three sets of studies, including the questionnaire-based Borrower Completion Report. All studies are held in project files (IRIS4). A. ESALQ (2003) and FEALQ (2005): 5.2 A beneficiary opinion survey-based study conducted by the Escola Superior de Agricultura Luis de Queiroz (ESALQ) in 2003, complemented by data from the Management Information System (MIS) coordinated by the NTU/NEAD, analyzed Credito Fundiario beneficiary profiles from nine Northeast states and 174 of the 226 settlements implemented by the Project as at July Some 750 families were interviewed between October and November 2003, 18% of total families benefited by the Project at that point, and an average four beneficiary families were interviewed per settlement. The study enabled better understanding of the profile of beneficiaries at the time of initial settlement, evaluated the land acquisition process, living conditions of beneficiaries prior to settlement and the changes immediately after settling. Most interviewed families were not even living on the new property at the time due to the need to complete rental or other obligations elsewhere. Analysis covering inter alia, maximum income at entry, professional history and agricultural experience indicated that the project was reaching its target public. Among findings, the average income of families at entry in 2002 was R$1,455 per year or R$122 monthly per family, well below the National Household Survey (PNAD, 2002) poverty line. Educational levels were very low, with high levels of illiteracy (37% of men, 19% of women). However, data also showed that illiteracy could represent a barrier to accessing the project even though studies also demonstrate that illiterate beneficiaries achieve similar increases in income to beneficiaries with more schooling. FEALQ suggests that there may have been an inherent trade-off between targeting the very poorest and recovering the land loans, resulting in some bias towards candidates with more schooling. Further study is suggested. The Fundação de Estudos Agrários Luiz de Queiroz (FEALQ - same institution) returned in 2005 to a representative sample of the same settlements/families to evaluate project impact on the lives of beneficiaries in the interim period (a panel-based research format). FEALQ selected, via probability sampling, 91 settlements and 200 families, averaging about 2.2 families per settlement. Results follow. Income: The income of surveyed families was found to have increased 145% between 2003 and 2005 from R$1,656 to R$4,064 and the number of families producing and selling had doubled from 25% to 50%. 42 (This was viewed by FEALQ as a significant finding since the study was not based on an experimental design). The study also found that having a fixed, stable residential situation in the countryside provided better conditions for families to generate income off-farm. 42 Income was defined as self-consumed agricultural production; income from sale of agricultural products; salaries and per diems off-farm; pensions; income transfer programs; income from sale of non-agricultural products; and income from the rental of equipment. The increase by State surveyed was: Maranhão 128%; Piauí 164%; Ceará 39%; and, Pernambuco 152%. 54

65 The number of families with access to technical assistance in the same period had increased from 30% to 65%. Beneficiaries had settled onto their properties. Whereas in 2003 only 8% had settled physically, this had increased to 66% by 2005 showing that physical settlement on the land is a process and not immediate. Housing quality showed a marked evolution: some 99.5% of settlers houses were constructed of brick/cement, up from 48% in 2003, and satisfaction levels were high. The combination of activities supported by the Project had enabled families to improve their incomes, generated new forms of productive occupation and importantly, positioned families to access/leverage the benefits of other programs, especially PRONAF-A credit, health, education and culture. About 89% of beneficiaries surveyed were working on their properties, up from 57% in Agricultural production conducted on-farm had increased from 37% in 2003 to 82% and some 59% of agricultural production was being processed/value added, up from 20% in Settlement had generated new income opportunities and production was intensifying. Other findings: FEALQ (2005) also found: (a) need for strong focus on investment in clean water supply, even though supply and quality had improved; (b) beneficiary links between and membership in, local labor unions was quite low; (c) on-farm agricultural activities were predominantly individual, not collective areas were set aside for collective farming but not commonly used; (d) access to and use of technical assistance and rural extension had risen markedly but was still not reaching many beneficiaries. However, the quality of such services had declined only 52% of users considered such services of good quality. A major issue was the greater degree of technological sophistication required from TA services once families moved into more sophisticated activities associated with marketable production, and lack of preparation for this phase on the part of service providers. Natural resource management deficiencies were also noted by interviewees. Conclusions: FEALQ (2005) concluded that the outlook was generally positive, that there were no major problems for the future that could not be resolved, and the Project could be considered to have achieved its objectives. The Project was well targeted - the only doubt being access for illiterates - and demonstrated capacity to increase wellbeing and incomes, contrary to what studies reveal about other forms of land reform in Brazil where marked changes were not evidenced. However, the consolidation of settlements was seen as a much longer-term process. Income gains could be linked directly to land access and to agricultural production, or to greater stability of families and their more effective insertion in local labor markets outside the settlement. The greater possibility of increasing income, principally monetary income, was still associated with the expansion and intensification of agricultural income on-farm. At the time of the 2005 survey, income from agricultural production (average R$524/family per year) still represented only 14% of total family income. Increased income on-farm and from labor/other activities off farm is essential for these families who would no longer be considered poor by the targeting criteria for entry to repay their land loans. 55

66 B. Borrower Completion Report (MDA/NEAD, 2009): 5.3 The BCR (2009) conducted interviews with institutional stakeholders - STUs, State and Municipal Rural Development Councils, and technical assistance providers - in five states. 43 Case studies were also prepared covering 19 beneficiary groups categorized as recently-settled, long-settled, successful and problematic. The following were among the BCR s findings: There is broad stakeholder agreement that the methodology promotes inclusion, improves the quality of life of poor rural families and can increase income. Further, social movements formerly opposed to the market-based mechanism have muted their criticism due to the social recognition that it has positive impacts, and are increasing their participation in its management. Data collected from STU s, Councils and TA service entities rated the project 3.8 to 4.0 (good and very good) on a scale of 5.0 (excellent) for its capacity to generate employment and income, and promote social inclusion. Of the 19 settlements interviewed, 12 considered the project Excellent and seven viewed it as Very Good or Good. Interviewees also valued the project for its capacity to foster the integration of other public programs, while pointing to several issues requiring solution, listed below in order of priority: Suggestions Reduce bureaucratic requirements and speed up processing Review financing ceilings and increase resources per family Strengthen program dissemination and training of families at all stages Ensure qualified TA services and their commitment and performance Promote greater integration of policies and strengthen partnerships Facilitate access to basic infrastructure (remote and semi-arid areas) Improve the structure and qualifications of STUs Individualize the SAT Loans The BCR makes a number of critical observations from data collected: Opinion was unanimous that the complexity of bureaucratic requirements caused significant processing delays. However, data also showed that the average time taken to reach the contracting stage had been significantly reduced about 50% from 2002 to 2008, due to efforts to simplify procedures, as shown below: No. of Proposals Average Time to Year Received Contracted % Contract (days) ,749 1, States included were: Piauí, Rio Grande do Norte, Paraíba, Minas Gerais and Espírito Santo. 56

67 Revised financing ceilings (on packages) were also a high priority, to permit the purchase of better quality land and permit adequate SIC investments to generate survival activities until they could obtain PRONAF credit (see Lessons). The BCR recommended that this issue be studied further, on the grounds that PRONAF credit should be based on stronger analysis of productive activities to improve their quality, conceptually and operationally. Better information dissemination was seen as fundamental to the quality of entering beneficiaries and organization of groups, as well as expansion of the Project to territories not yet served. Training at all stages from initial group organization through to the launching of productive activities was judged a top priority. Better basic infrastructure, especially access roads and water supply were important to people in more remote, semi-arid areas. Beneficiaries suggested that land loans be individualized with resulting individual responsibility for repayment, alleging that if one family did not pay this reflected on the credit standing of the entire group/association. The quality and quantity of technical assistance services were judged inadequate especially to help farmers launch marketable production. Technical assistance to promote association members ability to live together and to conduct certain kinds of group activities such as management of equipment was also considered important. The BCR sees these as exceedingly important issues. While the participation of CONTAG and its FETAGs and local unions were judged to have made an important contribution to many aspects of project implementation and their participation could be further strengthened, they needed better qualifications and better information in order to increase their efficiency and effectiveness. Some noted that in certain situations, the unions had virtually assumed the activities of the CMDRS and that their decisions were being almost automatically ratified by the other members, as a formality. Economic and Financial Findings: Income results from the analysis of ten cases showed that incomes from agricultural activities were still quite low, averaging R$9, per family/year or a net income of R$6, per family/year. After repayment of the land loan, and where relevant, PRONAF credit, balances remaining averaged R$5,216 per family/year or some R$190.00/month per family agricultural worker. There were some marked exceptions to the average: (a) a producer of manioc, sweet potato and small-animal livestock in Paraíba achieved a gross income of R$28,078 per year, netting R$18,282, equivalent to R$1,523/month after allowing for repayment of loans/credits and setting aside R$1,350 for new on-farm investments; (b) incomes of producers surveyed in Bahia and Espírito Santo showed incomes well above the average due to their inclusion of more lucrative crops such as coffee, cacao, rubber, fruits and vegetables; (c) the highest income was a producer in Rio Grande do Norte (Baraúna) cultivating six hectares of irrigated fruits, who achieved a gross income of R$52,323/year, netting R$35,752 after payment of debt obligations and leaving R$6,000 for new investments. 57

68 Ratios for land utilization were still quite low with only 16% of total land on the 19 properties surveyed ranging from 2.7% in Piauí to 39.6% in Espírito Santo. Use of family labor averaged 34.8%, ranging from 13.4% in Piauí to 49.5% in Rio Grande do Norte, suggesting that a significant proportion of income was still coming from off-farm and other sources. Sustainability issues were evident for settlements in semi-arid areas lacking water. Most production at least on settlements surveyed in these areas was still at a subsistence level with small, low yield herds and inadequate pasture. The issue of the size of individual lots was not considered the real issue since there are many viable small farm production systems - but rather, the location and quality of the land purchased including water availability. The BCR recommended a concentrated effort to qualify multipliers to transmit experiences with small farm production systems to project beneficiaries, including those seeking PRONAF credit; and, to concentrate project actions in areas with concentrations of rural poorest and with better conditions for implementation: Northeast, Northern Minas Gerais and Espírito Santo. To demonstrate the productive potential of settlements, which is obviously dependant on factors that cannot be extrapolated to all of them (e.g., land quality, water availability, technical assistance and credit), the BCR separated out better-performing farms among surveyed beneficiary groups. For example: (a) a producer of manioc, sweet potato and small-animal livestock in Paraíba achieved a gross income of R$28,078 per year, netting R$18,282, equivalent to R$1,523/month after allowing for repayment of the land loan and other credit obligations and setting aside R$1,350 for new on-farm investments; (b) producers surveyed in Bahia and Espírito Santo showed incomes well above the average due to their inclusion of more lucrative crops such as coffee, cacão, rubber, fruits and vegetables; and, (c) the highest income was garnered by a producer in Rio Grande do Norte (Baraúna) cultivating six hectares of irrigated fruits, returning a gross income of R$52,323/year, and net income of R$35,752 after payment of SAT and other debt obligations, and leaving R$6,000 for new investments. C. FECAMP/UNICAMP Studies: 5.4 A brief summary of the evaluation trajectory initiated by the Economics Foundation of the University of Campinas (FECAMP) a decade ago, is useful. The ICR s discussion of the impact evaluation record is predicated on the fact that the core project objective, methodology, mechanisms and target population were essentially the same for both operations and in five Northeast States directly overlapped for almost two years. The ICR also acknowledges, however, that Crédito Fundiário (CF) improved certain design features of Cédula da Terra (CdaT) and thus the FECAMP impact evaluation results of CdaT are viewed as highly indicative of CF, but not necessarily fungible. A comprehensive impact evaluation of CF is underway and more definitive results are expected in It is significant that the CF operation financed the FECAMP 2003 and 2006 studies of CdaT, thereby acknowledging the importance of completing the evaluation cycle of the pilot phase of the program. The timeline, methodologies and major impact/other findings are detailed below. First, FECAMP conducted in 1999 a preliminary socio-economic baseline of beneficiary families, and analyses of land purchased and the governance structure. The study collected information directly from a representative sample of beneficiaries as well as association presidents, gathered data provided by STUs in participating states, and from the Brazilian Institute of Geography and Statistics (IBGE), the Getúlio Vargas Foundation, the National 58

69 Institute for Colonization and Agrarian Reform (INCRA), and the Bank of the Northeast of Brazil. Results were presented in various documents, e.g., Buainain et al (1999a and 1999b), and Souza Filho at al (2001). A second, broader study by FECAMP in 2000 sought to track the socio-economic profile of beneficiaries based on a comparative analysis of the CdaT beneficiary population and beneficiaries on land settlements sponsored by INCRA, and included rural producers who had not benefited from any agrarian reform program. Due to the initial stage of program implementation, both studies dealt with the program s immediate effects on beneficiary wellbeing without analyzing technical aspects of their socio-economic impacts. This study served as the basis for comparing results of the final field survey conducted in 2006/2007. Efforts were made to maintain the maximum correspondence between the two samples, so as to evaluate the performance of the same households over time, permitting a more rigorous analysis of the socioeconomic impact of CdaT. A third FECAMP study done in 2003, also financed by Crédito Fundiário, evaluated the immediate socio-economic impact of CdaT on agricultural production, productivity and incomes of beneficiaries, using as its reference data collected by the Preliminary Evaluation of It also defined a panorama of the characteristics of production of CdaT settlements and beneficiary income generation strategies, which supported the methodological direction of the 2006/2007 study, the final in the CdaT series. Results of this study were reported extensively in the CdaT ICR (Report No , December 2003) and briefly reiterated here. FECAMP s final study in 2006/2007, also financed by Crédito Fundiário, evaluated the program s impact on beneficiaries using the 2000 study as the basis. Complementary research yielding rich results included: (a) a socio-economic profile comparing CdaT beneficiaries with two other groups; (b) special study of beneficiary rotation on settlements; (c) special study of social capital formation (Costa and Romano, 2007); and (d) an econometric analysis of CdaT impacts. The research compared CdaT beneficiaries with a control group of non-beneficiaries, and a group of future beneficiaries waiting to participate (pipeline). This study evaluated comparatively, incomes, assets, food security, education level, health and social capital formation of the three distinct groups in 2006 compared to Methodology: FECAMP exhaustively details the methodology, main elements of which were: (a) utilization of the FECAMP 2000 database; questions to evaluate impact on wellbeing; and, except for the rotation study, no attempt to delineate results by State; (b) use of sub-samples based on property size, and Propensity Score Matching to improve the basis for comparison of settlements and beneficiaries; (c) evaluation of CdaT impact compared with a proxy for the living conditions of the landless pipeline cohort. Propensity Score Matching was used again to select the pipeline sample for comparison with CdaT beneficiaries; and (d) the control sample was conditioned on the sampling procedure used in 2000, also with the added restriction of not including properties beyond a certain size. The study covered 88 municipalities in the Northeast States of Maranhão, Pernambuco, Ceará, Bahia and Minas Gerais, practically repeating the 2000 settlement sample but including additional municipalities to meet selection needs for the control sample. A manual for interviewers ensured harmonized application of the questionnaires and interpretation of questions. 59

70 Results: FECAMP 2006/ Beneficiary and household profile: Project beneficiaries showed low educational levels; 45% of heads of household (HH) were illiterate compared to 32% in the control group and 31% of the pipeline (consistent with the ESALQ findings summarized below). Only 2% of project beneficiaries had reached middle or superior level, compared to 15% and 12% respectively for the control and pipeline group. At an average 48 years, project beneficiaries were younger than the control group (average 60 years) but older than the pipeline (average 39 years). The relative youth of the project and pipeline cohorts is viewed as positive, allowing them more time to invest in and consolidate family farms potentially reversing the trend towards a rapidly ageing rural population and population loss. HH were overwhelmingly male in all three cohorts reflecting the prevailing culture; the project established no barriers to women as beneficiaries. Only 14% of project HH were still working off-farm in 2006 compared to 98% pre-project, and predominantly in agricultural activities. This might be interpreted as a sign of on-farm potential or, of the lack of opportunities off-farm for project beneficiaries due to educational limitations/other. In contrast, 29% of the control group HH worked off-farm, often as a complement to farm activities, and predominantly (60%) in urban areas, i.e., more stable and probably better-paid work. Some 35% of the pipeline group worked off-farm. Living Conditions: The vast majority of project beneficiaries were living in houses by 2006, as opposed to huts and other temporary structures (4%), similar levels to the control group; 12% of pipeline families were living in sheds/huts. In terms of water supply, only 34% of project beneficiaries were connected to the public water supply system compared to 28% prior to the project - compared to 52% of the control and 44% of pipeline (up from 31% and 22% respectively). Extension of the public system to project settlements has been slow. Clearly, project beneficiaries still remain more dependent on riskier water sources (reservoirs, springs, rivers etc) than the other two cohorts. However, sanitation access (bathroom/toilet) has risen sharply for project beneficiaries, to 79% from 43% pre-project, compared to 87% for control group and 51% for the pipeline group. Access to electricity showed impressive growth, from 54% at the outset to 95%, on par with the control group by Some 76% of pipeline families had access. The Federal Government s Luz para Todos program has been especially effective in this regard. Health results show that 23% of the project beneficiaries had been affected by parasitic infections and 16% by diarrhea between August 2005 and July 2006, with similar results for the pipeline group. The control group showed much lower incidence of both. Levels of food security and hunger were also surveyed and the situation for CdaT beneficiaries had improved greatly from 42% in 2000 to 66% by 2006; conditions for the control group and pipeline were 70% to 83% and 34% to 66% respectively. Assets: On average in 2006, CdaT beneficiaries surveyed were farming a total of 24 ha (up from about 8 ha in 2000), compared to the control group s 63 ha and pipeline s 5%. Not surprisingly, the total domestic/household assets of the control group significantly exceeded both other groups at R$7,500 compared to R$2,448 and R$2,232 for the project beneficiaries and pipeline respectively. In all cases, there had been a sharp increase in household durable goods ownership since 2000, with the project beneficiaries showing an increase of 159%. In terms of ownership of productive assets, excluding land, all three groups showed low levels of capitalization and technology. Some 30-40% of households - beneficiaries and the control group - had basic on-farm infrastructure endowments but the pipeline group was disadvantaged. In terms of value, there was sharp differentiation in the average value of such assets: whereas the control group had R$8,964, beneficiaries had R$1,161 and pipeline R$586. In 2000, these values were R$2,656, R$232 and R$173 respectively. Animal assets showed equally sharp contrasts: in 2006 the control group owned an average R$11,525 while beneficiaries and the pipeline owned 60

71 R$2,755 and R$796 respectively. In 2000, animal assets of the control group were also much higher. Notably, from 2000 to 2006, the control group had more than doubled the value of animal assets. In summary, total average gross household assets for beneficiaries in 2006 were R$9,529 compared to R$4,631 in 2000 and the pipeline R$6,124. Assets of the control group were about five times that of beneficiaries at R$48,715, up from R$32,286 in Animal assets predominated for both beneficiaries and the control, whereas for the pipeline, urban assets were more important. Indebtedness of the three groups in 2006 averaged R$2,373 for CdaT beneficiaries, R$3,561 for the control group and R$256 for the pipeline. Income: Total monetary and non-monetary income in the period from August 2005 to July 2006 was R$20,323 for the control group, whereas beneficiaries and the pipeline averaged R$7,901 and R$6,505 respectively. The sharp differentiation of the control group is accounted for by far more developed agro-livestock activities; greater income from other programs and social security payments (due to the much higher average age of the control group); and, income from other properties (rent and sales) and other commercial activities. While the income differential between the CdaT beneficiaries and the pipeline was not dramatic, the difference was accounted for primarily by agro-livestock activities and the valuation of production for consumption. CdaT beneficiaries were able to produce an income of R$4,471 from these sources compared to R$2,370 for the pipeline. FECAMP attributes this difference directly to the CdaT cohort s access to land. Income from off-farm work by the pipeline accounted for 30% of total gross income whereas for the CdaT group it was a mere 13% (compared to 36% pre-project). Both the CdaT cohort and the control group by 2006 had managed to treble their incomes. For the CdaT group, this meant pre-project income had risen from R$2,472 in 2000 to R$7,901 in Importantly, the average value of collective production on CdaT settlements reached barely 3% of the average income of CdaT beneficiary households, indicating that the collective strategy has not worked well (or not been adequately supported). Importantly, the econometric analysis of impacts based on income variables shows that the CdaT cohort did not advance significantly more than the control group. However, income of the CdaT group showed an evolution in income from production far superior to income from paid labor, indicating that there was a real effort to re-direct activities towards agriculture and reduce temporary labor, the latter a characteristic of CdaT beneficiaries pre-project. Importantly, the program has been an access door to productive activities, to some extent controlled by local conditions and by the social capital characteristics of entrants. The program is able to transform settlers into rural producers and this is shown by net income gains, but it also indicates that they are starting to encounter a new kind of risk, substituting that of not finding work to the typical risks of rural production. Technology: Barely 8% of CdaT beneficiaries had irrigation on their individual properties and only 3% had irrigation on the collective farming area of the settlement. The control and pipeline groups were about the same. The implication is that the many CdaT beneficiaries located on land in the semi-arid Northeast are likely to encounter productive obstacles. Further, the purchase of certified seeds remains low all three groups were still mostly acquiring seeds and seedlings from own production or neighbors indicating that productivity and quality are likely to be affected. Fertilizer use was also low in all three groups. Manual planting and harvesting still predominated in all three groups, but 34% of the CdaT and control groups were using tractors for land preparation. 61

72 Access to Credit: As with technology usage, access to credit was also found to be modest in all three cohorts. An important reason cited to researchers was the high level of delinquency on previous PRONAF credit which impeded all groups access to a new round of credit applications. The main source overall however, remains PRONAF (PRONAF-A for CdaT and CF beneficiaries). Technical Assistance: The study surveyed frequency and quality. About 60% of CdaT beneficiaries had not received technical assistance in the period from August 2005 and July The situation for the control group was much worse. Among CdaT beneficiaries, 21% had monthly access (compared to the very low 4% for control group, and pipeline group 2%). Even so, 67% of CdaT beneficiaries rated TA good quality, unlike the control group, 26% of which rated TA as poor and 21% as average. Rotation of Beneficiaries on Settlements: The study sought to quantify and explain the rotation of beneficiary families and to assess its implications. The research was conducted almost exclusively in the semi-arid Northeast region and the methodology is described in the report. The region has traditionally been one of heavy out-migration and a strong process of urbanization. Both internal and external causes were studied and rotation was not automatically viewed as a negative or positive phenomenon. Looking at the period from 1997 to 2006, the study found: (a) average size of groups settled was 33 families but there was a statistical bias towards groups with less than 30; (b) the average number of families leaving the settlement was 11.5, and of substituting was slightly lower at 9.6; (c) overall, the numbers leaving were about 34.6% of the numbers which had initially settled, fairly similar to the 27% ratio encountered on INCRA settlements. In about 24% of settlements, the attrition rate exceeded 50%, making the settlement unstable. The potential for substituting families who leave is not necessarily negative: while it can affect income and the group s capacity to repay its land loan, it can also signify renovation of the group. The study also found that the imminent need to re-pay the land loan did not provoke more intense out-movement of families. Looked at by State surveyed however, there were indications of problematic official/other support for settlements in some states. States with exit rates exceeding 40% were Pernambuco, Bahia and Minas Gerais (50%). Maranhão and Ceará were significantly lower. Of interest, the latter states showed rates of substitution lower than exits meaning inter alia that affected settlements found equilibrium at somewhat fewer families. The study concluded that there was a link between a State s CdaT governance structure and the percentages of exits and substitutions. Table 5.1.1: Percentage of Leavers and Entrants Year Leavers vs. Beneficiaries at the Start % Entrants vs. Beneficiaries at the Start % Rotation Balance % Total: The study found (a) in the period from many beneficiaries perceived that life on the settlement was not what they had anticipated, perhaps due to selection problems or due to opportunism about 4% of families who exited did so because the initial food/other support ended. Importantly, a key reason was the intense, collective nature of initial on-property activities financed by the SIC: fencing, electricity, housing improvements and, the cessation of the initial 62

73 R$130.0/month household payment; (b) was a period of insecurity due to changing policies and CF incentives, and the demonstrated difficulty of generating income sufficient to pay the land loan; and (c) a mature period from 2005 onwards, when community leaders realized that the non-payment of land loans had implications. Also, SIC payments had been received by families and exhausted, without necessarily having yet accessed other credit such as PRONAF. Pressures on beneficiaries to pay their land loans, once the grace period had elapsed, caused internal conflicts and the exit of about 8% of families unable to adapt to the rules Conclusions: The study concludes that there is a natural routine of replacement but no deliberate strategy on the part of CF settlement associations to reduce the number of beneficiaries. Further, rotation can be associated with the sale of lots and the CF did not prevent that from happening. Sales in the first phase ( ) averaged R$1,257, rising to around R$3,000 in the second phase ( ) indicating that the average value of CF lands sold was rising. Further, a relatively high percentage of families leaving their lots had paid four or more installments on their land loans. Importantly, the research showed that neither the price paid for the land nor land quality was a significant factor driving attrition. Key factors driving attrition were as shown in the following table: Table 5.1.2: Reasons for Leaving a Settlement Reasons Percentage Lack of adaptation to settlement rules 23 Various family problems 20 Lack of resources 19 Precarious infrastructure conditions 10 Attraction to better opportunities elsewhere 10 Lack of adaptation to life on a settlement 9 Frustrated expectations of an improved life 6 Natural resources deficiencies (water supply etc) 3 Total: 100 Lack of adaptation to far settlement life explains some 32% of attrition, in large part due to beneficiary selection issues and corrected by the entry of substitute candidates. The study notes that inequitable division of the property into lots, non-payment of land installments explain divergence from the group. Low mobilization capacity, i.e. moving the group on specific issues, is another factor explaining attrition. Of significance, almost half of the impacts of families leaving the settlements were considered favorable to the settlement, by those interviewed. About 57% of all families leaving a settlement moved to a town or village nearby. (a) In terms of collective work, the study notes that research shows that the imposition in rural projects of collective models of organization and forms of crop cultivation are inconsistent with farmers traditions and generate conflict. (b) In regard to repayment of the land loan, conflict can also be generated by families inability to confront the need to pay and leaving the settlement becomes the easiest solution. (c) Precarious infrastructure conditions are an important driver of families leaving including lack of schools, poor roads, lack of transport and electricity, and bad housing conditions. (d) Schools/teachers are almost the first demand made as families start to settle and frequently, the reason for leaving is to ensure children get an education. 63

74 (e) Equally great importance is attached to roads not only to market produce but to access schools and medical help. (f) Some settlements/families still wait significant periods for public services and feel abandoned, in spite of the SIC payments. An important issue is the tension surrounding possible non-payment of the land loan: (a) delinquency compromises the credit rating of the person and impedes their requesting other credit, especially at term; and (b) destabilizes the relationship within the group because repayment is collective and the non-payment by one means the others have to cover the deficiency. When this happens, the groups generally pressure the delinquent to sell their lot or leave the settlement. Pressure to find better opportunities in the rural space constituted a high 37.6% of all leavers. Conclusions: Rotation within settlements indicates the challenges faced in selecting candidates for settlement, as well as the challenges of repaying the land loan. Substitution demonstrates beneficiaries preoccupation with sustainability and the commitment to pay the land loan contracted. The upshot of this process is that there was an adjustment, whether to achieve a better use of available resources to improve the beneficiary team, or to avoid further aggravating a settlement s situation. Rotation does not necessarily negatively affect the life and sustainability of a settlement. Research on rotation suggests that CdaT beneficiaries show strong identification with the land they have acquired and this justifies the positive results in the evaluation of price paid for and quality of, this land. Results: FECAMP FECAMP (2003), also financed by CF, evaluated the immediate socio-economic impact of the Program on agricultural production, productivity and incomes of beneficiaries, using as its reference data collected by the Preliminary Evaluation of It also defined a panorama of the characteristics of production of CdaT settlements and beneficiary income generation strategies, which supported the methodological direction of the 2006/2007 study, the final in the series. Main conclusions are reiterated selectively below: 44 Data was collected from a sample of 108 land acquisition projects and as many associations, and 313 beneficiaries surveyed in August 2003 and compared to data collected from the same sample projects in 1999, just after implementation started. An additional comparison was made to the results of the 2000 study (see above). In the period from 1998 to 2003, beneficiaries nominal income rose by an average 180%, from $2,037 (typical beneficiary family) to R$5,777, equivalent to a monthly per capita income of R$122, significantly above the poverty line of R$70 then in use by the Federal Government. Correcting the nominal income values for the increase in agricultural labor salary rates in the period, a good proxy for inflation in rural areas since it indicates the opportunity cost of labor, real income increased by 75% between 1998 and In terms of minimum salaries, average family income in 1999 was equivalent to 0.7 minimum salaries, while in 2003 it was equivalent to 2.3 minimum salaries of R$200, or R$500 (applicable up to April 2003). This represented a substantial increase especially considering that the PNAD (National Household Survey) surveys 44 See Implementation Completion Report No of December 2, 2003 for a broader presentation on the FECAMP 2003 findings. 64

75 from 1999 to 2001 indicate that there was no positive change in average total family income for the rural population with similar characteristics to CdaT beneficiaries. Analysis of the composition of household income is helpful in explaining not only the situation of CdaT beneficiaries in 2003 but also the dynamics of the settlement projects. Clearly, a significant shift occurred towards agricultural production and away from off-farm employment, transfers and pensions. In 1998, income from agricultural production was only 13% of total household income; by 2000 it was about 45% and by 2003, about 65%. Off-farm income decreased from 45% to 7% of total income in the period. This change resulted from the massive upswing in agricultural production: 360% between 1998 and 2000 and 240% between 2000 and Beneficiaries were dedicating more time/labor to on-farm work, evolving towards greater independence from off-farm sources, the latter being the mainstay of similar poor farmers in these same regions. Total monetary income indicative of market insertion was R$3,947, 68% of total income or equivalent to R$324 per months, equivalent to 1.6 minimum salaries (of the period). Of this, R$3,158 was derived from agricultural production on-farm, again indicating an evolution away from the subsistence-level poor, small farmer. In many cases, just the monetary income put families above the poverty line, a notable gain in the Northeast. Settlement projects were found to create employment for an average 3.9 individuals per beneficiary family. Given that the sample used in 2003 could statistically represent only the universe of projects implemented from 1998 on, extrapolating these employment estimates to the universe of project beneficiaries was not possible. However, since the projects implemented after 1998 were in general more successful in initiating productive activities at an earlier stage, a conservative estimate could be made that for the 15,267 beneficiary families of CdaT, about 60,000 year-round occupations were being created by

76 Annex 6. Stakeholder Workshop Report and Results N/A 66

77 Annex 7. Summary of Borrower's ICR and/or Comments on Draft ICR A. Borrower Completion Report: Executive Summary Unofficial Translation 1. Objectives: The Final Report of the Land Credit and Poverty Reduction Project (Loan Agreement IBRD 7037/BR) had the following objectives: Present a general vision of the CF/CPR and its implementation since its initiation until the of 2008 Systematically record the opinions of the principal actors involved in the Project regarding its importance and opportunities and about the main challenges and potentialities detected during its execution. Organize suggestions for changes designed to improve the project and suggest measures of both a general and specific nature to be implemented by the MDA/SRA during the next ongoing stage of the project. 2. Methodology: The Report achieves its objectives through the following methodological processes: Survey of the physical and financial targets achieved by the Project from its initiation until closing at end-december 31, 2008, including an analysis of the results below, emphasizing the following aspects: (a) index of the achievement of targets both in general and by State, by rural territory and by component; (b) principal indicators of project performance (families by project, cost per family, average cost of the SAT (land purchase subprojects) and of the SIC (on-farm investment subprojects). Interviews conducted in the States of Piaui, Rio Grande do Norte, Paraiba, Bahia and Espirito Santo, with State Technical Units, members of State Sustainable Rural Development Councils, technical assistance providers and members of Municipal Rural Development Councils. These interviews utilized questionnaires and sought to capture: (a) satisfaction levels and the importance of the Project; (b) the main positive and negative aspects of the project; and (c) suggestions for improvements to be introduced to the project. Case studies conducted with four settlements (groups of beneficiaries) in each of the abovementioned States, covering a total of 20 cases. Analysis of all this information permitted general conclusions about the functioning of the project and proposed recommendations for improvement, applicable mainly to the continuation of the project in its new phase utilizing national resources only. 3. Results Obtained by the Project: By State: As observed, performance varied between the States with high indices in the States of Piaui and Maranhao, while Bahia, Ceara and Pernambuco showed performance significantly lower than expected. In these latter states, unsatisfactory results demonstrate the need to achieve greater commitment of the State Governments and better structure within State Technical Units, while in Pernambuco the difficulty of obtaining land, especially in the Zona da Mata and Agreste must be taken into consideration. By Component: About 91% of total resources were applied to the SATs and SICs which can be considered a very positive result revealing the high level of direct reach to the endbeneficiaries of the project. 67

78 Activities in the Rural Territories : In the Northeast, Minas Gerais and Espirito Santo, the SDT/MDA, through its PRONAT and Citizenship Territories were active in 82 Rural Territories and the project supported proposals in 78 of these territories (95.1%). The number of municipalities with proposals contracted was 534, which corresponds to 23% of total municipalities in these territories. Families attended were 24,590, equivalent to 61.9% of total beneficiaries in these States. Twenty three of these territories are included in the Citizenship Territories Program. Within these, some 14,648 families were benefited, corresponding to six percent of total beneficiaries in the territories and 36.9% of those benefited in the States. Considering that the rural population of the Territories corresponded to 53.7% of the rural population of the respective states and that in the Citizenship Territories this percentage is 31.6%, it can be concluded that the level of attention of the project in rural territories is reflecting the real priority of these geographic spaces, demonstrating compatibility with the strategic rules and priorities for sustainable rural development and for poverty alleviation. 3. Performance Indicators: The average of 20 families per group is adequate for achieving efficient organization along with a minimum scale of production which justifies and gives viability to the processing of products, of mechanization, the production of inputs (mainly seeds and seedlings), of transport, storage and other. The area of 20.7 ha per family, despite opinions which consider it insufficient, can also be considered adequate since it exceeds the average size of family farms in the Northeast which is about 12.9 ha/farm and not much lower that the average area of INCRA settlements which in the Northeast is about 29 ha. The average cost for land (SAT), as well as the cost per hectare of areas acquired showed a significant decline in 2005 and 2006, to levels of about 57.5% of the values of 2002 to From 2007, a small increase can be observed with values rising to about 68.3% of the average of the first three years. This fact is due as much to the increased efficiency of beneficiaries in negotiating land prices, as to the incentives for achieving prices below the values generally achieved in each municipality or even, to generate more resources for complementary investments (SIC) so as to offer more favorable conditions for the settlement of families and facilitation of productive activities. These reduced values for land acquisition translated into the simultaneous increase in resources for complementary investments which increased from an average R$7,195 per family in the first three years to R$11,247 in the years 2007 and 2008, an increase of 56.3%. These notable achievements were the result of constant improvement of the project s operational strategy especially in the strengthening of Social Management, which minimized external influences on the land purchase process, increasing the self-determination of beneficiary groups. 4. Opinions of interviewees: The State Technical Units, members of State and Municipal Rural Development Councils (and especially civil society) technical assistance providers and beneficiary groups, were in agreement on the importance and opportunity provided by the project 68

79 and its effective capacity to combat rural poverty. This confirmation is based on the opinions synthesized in the following table which obtained average outcomes of 3.2 to 4.0 (good and very good) of a total 5.0 (excellent). Score Opinions of Interviewees STUs ATER CMDRs Total 1. Capacity to generate jobs and income Social inclusion and stronger citizenship Potential for growth and better quality of life Capacity to pay the land loans obtained Strengthen worker organization =weak; 2.0 = regular; 3.0 = good; 4.0 = very good; 5.0 = excellent These same groups of institutions or persons suggested diverse measures for improving the project among which the following were the most important: Reduce bureaucratic requirements and speed up the process; Review the financing limits and increase the resources per family Strengthen the dissemination of the program and the capacity of families at all stages. Qualify the ATER (technical assistance and rural extension providers) and seek greater commitment and performance Promote greater integration of policies and strengthen partnerships Facilitate access to basic infrastructure Improve the structure and qualifications of the STUs Individualize the SAT loans General analysis of the information presented in the table and in the Borrower Completion Report reveals that problems can occur in generating a satisfactory income in projects such as those in Rio Grande do Norte, Paraiba and Piaui which are located in the semi-arid zone, and in obtaining irrigation capacity. Up till now, the productive systems used are still centered on subsistence cropping with low monetary returns and with livestock of low productivity, with very small herds and with insufficient pasture. These systems are very vulnerable to drought. The low utilization of the family labor available suggests that the income needed to maintain families is still coming from work off-farm, government income transfer programs and pensions. This assertion is backed up by the evaluations of the University of Sao Paulo which showed that non-agricultural income in 2005 was around 69.8% of total income. A strong tendency is also observed to maintain productive systems in the majority of cases centered on livestock activities of low profitability, using native pasture (caatinga) which, under normal livestock management, would not offer adequate forage. The suggestion for increasing the size of properties does not have much substance, since the average of about 20 ha per family is well above the average of family properties in the Northeast which is 12.9 ha, and 86.7% of these establishments have an area exceeding 20 ha. It merits mentioning that of a total of 773 settlement projects of INCRA in rural territories of the Sao Francisco Basin, in the States of Minas Gerais, Bahia, Sergipe, Alagoas, Pernambuco, Rio Grande do Norte and Paraiba have an area averaging 29.0 ha per family, a bit above the average obtained by Credito Fundiario. 69

80 These observations suggest that the question of low profitability and the probable increase/surge in delinquencies is due more to the production models adopted in these areas than the size of the properties. Today there are various alternative productive systems with satisfactory returns which could be installed and operated in relatively small areas such as: poultry production, fish farming, simplified water capture using simplified technologies, the PAIS system which is being disseminated by the Bank of Brazil Foundation, beekeeping, flowers, and other alternatives. In this sense, more concentrated actions are recommended, directed to training multiplier agents who could pass these technologies/systems on to project beneficiaries, including for credit subprojects submitted to PRONAF for financing. 5. Conclusions and Recommendations: Conclusions: The information and data explained in the Borrower Completion Report and its annexes permit the following conclusions: The Credito Fundiario Project, in its first stage, showed satisfactory performance, having achieved its physical and financial targets adjusted to realities and having reached its objectives, especially in regard to combating rural poverty. The performance promoted the recognition of the relevance and timeliness of the project, initially between the actors participating directly in its execution and between their direct beneficiaries and subsequently in the governments (Federal, State and Municipal) and in the civil society entities. This recognition is resulting in the consolidation of the project as an important public policy for sustainable rural development and more specifically, for poverty reduction, which materialized in the favorable decision to continue the project and in the passage of Federal legislation institutionalizing the continuation of the project. Some gaps which are still negatively affecting the efficiency and effectiveness of the project have been identified and solutions are gradually being implemented. However, some corrective measures need special attention and these are explained in the following recommendations. Recommendations: The main recommendations, apart from general conclusions and suggestions presented in the Report, are the following: Continue the effort to expand Crédito Fundiário, while concentrating actions in areas with greater concentrations of poverty and with better conditions for putting project components into operation: Northeast, Northern Minas Gerais and Espírito Santo; Continue efforts to improve the project strategy and adjusting its operational rules/norms, giving special attention to measures which have been agreed between the State and Civil Society as a result of stronger Social Management. 70

81 Dedicate special effort to resolving certain challenges of greater importance already recognized by all actors involved in the project, especially those linked to: Better quality and expanded quantity of technical assistance services for beneficiaries, in all the various phases of the project; Simplify procedures, speeding up the steps for contracting financing and execution of the complementary on-farm investments (SICs); With increased quality, facilitate and speed up access of beneficiaries to credit (e.g., PRONAF-A) for productive activities; Intensify project dissemination and promote its expansion to areas not yet attended or scarcely attended; Give special attention to increasing the qualifications of beneficiaries, from the organization of groups to the consolidation of communities and productive processes; Seek to establish mechanisms which facilitate adequate linkages between the project and other rural development public policies, especially those linked to development, production, marketing of products, education, health, social assistance and basic infrastructure. These linkages are needed at all levels (Federal, State and Municipal); and, Intensify the relationships and negotiations with State Governments, seeking greater commitment from them and greater support for the project, especially in the case of those states which are showing results below expectations. 7. Complementary information: The main report presents, in the form of annexes, the following complementary information: Questionnaires/schedule of interviews Tables with data obtained Contracts effected by states and municipalities Contracts effected by Rural Territories (of PRONAT and Citizenship) Panel of Management Indicators (as at 12/31/2008) Brasilia/DF, June 2009 Raimundo da Costa Sobrinho Director NTU/SRA/MDA 71

82 B. Borrower s comments to Draft ICR 72

83 73

84 Annex 8. Comments of Co-financiers and Other Partners/Stakeholders N/A 74

85 Annex 9. List of Supporting Documents - Implementation Status Reports (ISR) - Supervision Aide Memoires - Audit Reports, Financial Management Supervision reports and Procurement Post- Review reports. - Six-monthly panel progress reports by MDA - Avaliação Preliminar do Cédula da Terra. Relatório Técnico, 1999ª and 1999b, Buinain, Silveira, Souza Filho, Magalhães - Estudo de Avaliação de Impactos do Programa Cédula da Terra, 2001: Relatório Sintese. State University of Campinas/Institute of Economics and Ministry of Agrarian Development, November Estudo de Avaliação de Impactos do Programa Cedula da Terra, 2003, Relatório Final, State University of Campinas/Institute of Economics, and Ministry of Agrarian Development, Brazil, Estudos de Reordenamento Agrario: Avaliacao de Impacto do Projeto Cedula da Terra, Jose Maria Ferreira Jardim da Silveira coord., FECAMP/UNICAMP O Impacto do Projeto Cedula da Terra sobre Capital Social, Alberto C. G. Costa and Claudia Romano, special study in Estudos de Reordenamento Agrario, FECAMP/UNICAMP, Rural Poverty Reduction in Brazil: Towards an Integrated Strategy, Poverty Reduction and Economic Management (PREM), Latin America and Caribbean Region, Diagnóstico dos Projetos do Crédito Fundiário e Combate á Pobreza Rural (2003), USP/ESALQ (Escola Superior de Agricultura Luiz de Queiroz), pub. as Crédito Fundiário e Qualidade de Vida no Campo, G. Sparovek, Estudos NEAD/MDA - Diagnóstico Qualitativo Crédito Fundiário (2003 e 2005), FEALQ (Fundação de Estudos Agrários Luiz de Queiroz)/University of São Paulo. - Country Case Study: Reducing Rural Poverty through Access to Land, March 2004, L. Coirolo and A. Roumani, prepared for Shanghai Poverty Conference - Scaling up Poverty Reduction Borrower Completion Report: Relatório Final de Execução do Projeto de Crédito Fundiário e Combate a Pobreza Rural: Acordo de Empréstimo entre o Governo Brasileiro e o Banco Mundial, 7037-BR, MDA/NEAD,

86 MAP No. IBRD