94972 P IEG Working Paper Series ● No. 2015/1 World Bank Lending for Financial Inclusion: Lessons from Reviews of Select Projects Anjali Kumar, Sushma Narain, and Swizen Rubbani © 2015 Independent Evaluation Group, The World Bank Group 1818 H St., NW Washington, DC 20433 IEG: Improving World Bank Group Development Results Through Excellence in Independent Evaluation The Independent Evaluation Group is an independent unit within the World Bank Group; it reports directly to the Bank’s Board of Executive Directors. IEG asse sses what works, and what does not; how a borrower plans to run and maintain a project; and the lasting contribution of the Bank to a country’s overall development. The goals of evaluation are to learn from experience, to provide an objective basis for assessing the results of the Bank’s work, and to provide accountability in the achievement of its objectives. It also improves Bank work by identifying and disseminating the lessons learned from experience and by framing recommendations drawn from evaluation findings. IEG Working Papers are an informal series to disseminate the findings of work in progress to encourage the exchange of ideas about development effectiveness through evaluation. The findings, interpretations, and conclusions expressed here are those of the author(s) and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent, or IEG management. The World Bank cannot guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply on the part of the World Bank any judgment of the legal status of any territory or the endorsement or acceptance of such boundaries. ISBN-10: 1-60244-257-6 ISBN-13: 978-1-60244-257-3 Contact: IEG Communication, Learning and Strategies (IEGCS) e-mail: ieg@worldbank.org Telephone: 202-458-4497 Facsimile: 202-522-3125 http:/www.worldbank.org/ieg ii Acknowledgments The authors of this discussion paper are particularly grateful to the authors of four core Project Performance Assessment Reports reviewed here. Manuel Hinds and Brendan Horton undertook the investigative work and field missions described in this paper, in Mexico, Brazil, Ghana and Mozambique. Valuable comments from peer reviewers are also acknowledged especially Stephen Rasmussen and Kabir Kumar of CGAP, and other commentators, including Stuart Rutherford, formerly Chairman of SafeSave, Bangladesh. The authors also wish to thank the numerous World Bank staff who have reviewed and commented on the project performance assessment reports as well as on the present discussion paper. Finally, thanks are due to all the country officials, researchers and academia who contributed their time and insights to the authors of the PPARs and the authors of the present discussion paper. The views expressed in this paper are those of the authors alone, who also take responsibility for any factual errors. iii Contents Acknowledgments.................................................................................................................... iii 1. Introduction: Financial Inclusion and World Bank Lending................................................ 1 2. Data and Approach ............................................................................................................... 3 2.1 Access to Finance: Loan Commitments and Loan Closures ....................................... 3 2.2 Projects Reviewed: An Introduction and Summary .................................................... 7 Overall Project Objectives ........................................................................................... 7 Macroeconomic and Sector Context ........................................................................... 7 2.3 Key Questions Addressed ......................................................................................... 10 3. Findings: Borrower Institutional Arrangements................................................................. 11 3.1 Large Public Sector Banks ........................................................................................ 12 Mexico: Banrural ...................................................................................................... 12 Brazil: Banco do Nordeste and CrediAmigo ............................................................. 13 3.2 Lines of Credit and Participating Financial Institutions ............................................ 14 Mozambique: On-lending Arrangements .................................................................. 16 Bangladesh: On-Lending through a Government-Established Apex ........................ 18 3.3 Apex Banks and Rural and Community Banks ......................................................... 19 Ghana: Association of Rural Banks’ Apex Bank ...................................................... 19 4. Findings: The Business Environment and the Broader Community .................................. 22 4.1 Market-Based Finance or Government Support? ...................................................... 23 4.2 Other Regulatory Factors .......................................................................................... 24 4.3 Government Commitment to Change........................................................................ 25 5. Findings: Bank Loan Design .............................................................................................. 27 5.1 Adjustment versus Investment Lending? .................................................................. 27 5.2 Other Elements of Successful Loan Design .............................................................. 28 Setting Realistic Objectives and Designing Measures to Achieve Them ................. 28 Mitigating Risks ........................................................................................................ 28 Incorporating Lessons from Past Lending and Analytical Work .............................. 30 Providing Technical Support and Capacity Building with Finance .......................... 30 6. Findings: Outputs and Outcomes ....................................................................................... 32 6.1 Outcomes - Outreach, Impact and Sustainability ...................................................... 32 Project Beneficiaries, Impact on Expanded Inclusion, and Poverty Reduction ........ 32 Regional Considerations ............................................................................................ 36 Gender Considerations .............................................................................................. 37 6.2 Outcomes—Financial Institution Stability and Sustainability .................................. 37 Unintended Consequences for Stability—Multiple Lending and Overindebtedness 40 7. Lessons Learned and the Way Forward ............................................................................. 42 Borrower Institutions and Bank Loan Structure ........................................................ 42 The Policy Environment ............................................................................................ 43 Other Factors in Bank Loan Design .......................................................................... 45 Achieving Broader Goals: From Outputs to Outcomes and Impact .......................... 46 Appendix 1: World Bank Access to Finance Projects ..................................................... 48 iv Tables Table 2.1. World Bank Projects with Access to Finance Content by Sector Codes (Loan Commitments) (2002-13) (Nos. and %) ................................................................................... 3 Table 2.2. World Bank Projects with Access to Finance Thematic Content (2002-13) (Loan Commitments)........................................................................................................................... 4 Table 2.3. World Bank Projects with Access to Finance Sector or Theme Content Exiting the Bank’s Portfolio between FY06 and FY14a ............................................................................. 5 Table 2.4. Characteristics of Projects Reviewed ...................................................................... 6 Table 3.1. Key Indicators of Expansion of the CrediAmigo Program.................................... 15 Table 3.2. Ghana Rural Financial Services Project - Intermediate Outcome Indicators (2001- 11) ........................................................................................................................................... 21 Table 6.1. Financiera Rural Selected Indicators ..................................................................... 33 v 1. Introduction: Financial Inclusion and World Bank Lending 1. The global profile of financial access has grown. Less than a decade ago, in 2006, the United Nations Committee on Building Inclusive Financial Sectors urged central banks and countries to add the goal of universal “financial inclusion” to the two traditional financial system goals of stability and prudential regulation for the safety of depositors’ funds. Articulating its commitment to financial inclusion, the G-20 set up the Global Partnership for Financial Inclusion (December 2010), a platform that also included interested non-G20 countries and relevant stakeholders, to carry forward work on financial inclusion. The Bank, initially represented by the Consultative Group to Assist the Poor (CGAP) and now a partner in its own rights, has been a recognized contributor to this effort and is one of the four implementing agencies for the G20 program. Given increasingly recognized links among access to finance, growth, income smoothing, and poverty reduction, many countries are now adopting the goal of universal financial access.1 Moving beyond the ‘microcredit only’ approach, financial inclusion has become recognized to include a variety of products and services as well as the cost effective provision of such services by a spectrum of financial service providers. Recognizing that 2.5 billion adults worldwide are ‘unbanked’ and that close to 200 million micro to medium enterprises in developing economies lack access to affordable financial services and credit, the World Bank Group has put forward a vision for achieving universal financial access by 2020.2 2. Public interventions to expand access have gained acceptance because of the recognition of financial market imperfections such as informational asymmetries, transaction costs, and contract enforcement costs. These are particularly binding on the poor or on small entrepreneurs who lack collateral, credit histories, and connections. Such constraints make it difficult for poor households or small entrepreneurs to finance investment projects, with potential adverse implications for resource allocation, growth, and poverty alleviation. Credit constraints can limit the entry of new firms and the Schumpeterian process of “creative destruction,” which requires that talented newcomers have access to necessary financial services. Perhaps even more important for poor households is the access to a wider spectrum of financial instruments, including safe savings (for specific needs, for insurance, or to smooth intertemporal consumption), and payment networks for small financial transactions. 3. Based on the case for public interventions to support financial access, the World Bank Group has a long record of such support, extended through a spectrum of instruments: its global convening power, its analytic work, and its financial support through lending and technical assistance.3 This paper examines the effectiveness of the latter - World Bank 1Such emerging association and potential causality are discussed in, for example, Honohan (2004), Burgess and Pande (2005), Clarke, Xu, and Zhou (2006), Beck, Demirgüç-Kunt, and Levine (2007), Copestake, Balhotra, and Johnson (2001) and Hulme and Mosley (1996). 2 World Bank Group: http://www.worldbank.org/en/news/press-release/2013/10/11/universal-financial-access- vital-reducing-poverty-innovation-jim-yong-kim 3The Bank has produced a large amount of analytic work in this area, notably, World Bank (2014), Demirgüç- Kunt, Beck, and Honohan (2009) and Honohan, and Beck (2007). 1 lending as a vehicle for supporting financial inclusion—based on the experience of a select set of projects where in-depth evaluative information is available, especially based on Project Performance Assessment Reviews (PPARs) recently undertaken by the Independent Evaluation Group (IEG) of the World Bank.4 This background paper compliments and contributes to a detailed large scale evaluation of the role of the World Bank Group in financial inclusion. 4. The purpose of the paper is to present a more granular view of such projects through the in-depth focus on a limited number of case studies, with a view to understanding what factors in the design of such lending have helped achieve objectives of expanded access, and what forms of interventions may have been less successful. It examines the nature of Bank lending vehicles, the partnering borrower institutions, the country environments in which its loans were extended, as well as broader elements of good practice that make for loan success. It examines the beneficiaries targeted and results achieved. It aims to draws lessons that suggest what factors could lead to success or failure in Bank operations focused on financial access.5 5. The remainder of the paper is organized as follows: Section 2 briefly describes the set of the Bank’s projects selected for detailed review. Sections 3 to 6 contain the core findings of the review. Section 3 focuses on alternative forms of borrower institutions that have served as vehicles for Bank projects, particularly, public sector banks, apex bank structures that include the private sector, rural banks, nonbanks, non-governmental organizations (NGOs) and microfinance institutions, in terms of the degree to which the Bank has been able to successfully partner with such institutions to expand financial access. It also looks at alternative forms of Bank loan design—policy-based loans, investment loans and lines of credit, Learning and Innovation Loans (LILs), matching grants, technical assistance and combinations thereof, and reviews evidence on the role of loan structure (including partnerships with other donors/lenders) and project success. Section 4 considers the effect of the broader business environment, in terms of financial regulation. Section 5 reviews elements of good practice that have contributed to success in lending that could be applicable to loans with any objective, and examines their application in the present context. Section 6 tries to construct a “bottom line,” reviewing available evidence on outcomes and impact; especially in terms of the ultimate beneficiaries reached. Section 7, the final section, summarizes the main messages emerging from the review and concludes with observations about ways forward. 4 IEG (2014a, 2014b, 2014c, 2014d). These PPAR documents are available to external readers, who are encouraged to read them in tandem with the present chapeau report. The paper also draws upon two IEG Implementation Completion Results Reports (2008a and 2008b) 5 Especially due to the inclusion of information from two Implementation Completion Results Reports, as well as information from external sources, the output is presented in the form of a working paper, rather than a formal “meso product” of evaluation. The paper draws upon information in the form of materials such as supervision reports—Implementation Status and Results Reports—produced by Bank task teams). The availability of information and hence the depth of treatment of all questions is not the same across all projects examined. Although the paper is intended to stand on its own, reference can be made to the underlying project documents and evaluative material as needed. 2 2. Data and Approach 2.1 Access to Finance: Loan Commitments and Loan Closures 6. Looking at all new loan commitments, around a third of all financial sector projects by number and at least a quarter by value have components related to financial access, as broadly defined by the Bank’s sector and thematic code classifications.6 In some years, the proportions have been considerably higher (Table 2.1 and Table 2.2). Table 2.1. World Bank Projects with Access to Finance Content by Sector Codes (Loan Commitments) (2002-13) (Nos. and %) Commitment Total Total amounts of No. of Projects w/ commitment Commitment No. of commitment projects w/A2F projects A2F sector - projects w/ value of A2F Avg. A2F sector finance - projects w/ sector content Fiscal w/A2F content finance components % of Access to sector A2F sector as % of all year sector as % of no. sector (weighted) Finance projects content ($ finance sector content of finance content ($ millions) projects (%) (no.) billions) commitments (no.) projects (%) ($ billions) (%) 1 2 (3=2/1) 2002 38 11 29 5.48 0.81 15 111 26 2003 43 9 21 3.68 0.93 25 463 22 2004 46 16 35 5.55 1.21 22 446 27 2005 46 12 26 4.24 0.92 22 273 27 2006 47 10 21 5.05 0.92 18 238 24 2007 33 12 36 3.67 1.01 28 332 44 2008 31 14 45 4.03 1.12 28 381 32 2009 41 9 22 11.77 2.01 17 869 55 2010 59 18 31 16.59 2.97 18 1,474 45 2011 36 12 33 2.72 1.32 49 193 25 2012 29 12 41 6.05 2.09 35 583 46 2013 28 13 46 5.20 1.74 33 675 46 Total/A 477 148 31 74.04 17.04 23 6,037 35 vg. Source: World Bank data. Note: Access to finance sector percentages of each project, averaged overall access projects for a given year. Includes only IDA/IBRD commitments. A2F = access to finance. 6Sector content in terms of access to finance is defined as including lending under the following sub sector codes: FE - Micro- and SME finance, FH - SME Finance, and FI - Microfinance. Thematic content in this area is defined to include theme code 41 – Micro, Small and Medium Enterprise support. These definitions may omit some projects that support other areas of financial inclusion or include projects with low financial inclusion content, and the universe thus identified should be regarded as an approximation. 3 Table 2.2. World Bank Projects with Access to Finance Thematic Content (2002-13) (Loan Commitments) Criteria Number of projects World Bank Commitments FY02-13 3,623 World Bank Commitments with Access to Finance thematic content 207 at least 25% 94 at least 50% 33 at least 75% 17 Source: World Bank data. 7. As illustrated in Table 2.2, components addressing financial access are typically small elements in larger projects. A large number of these projects have only partial content focused on access to finance, and a number include sectors other than the financial sector— rural finance, for example, has been a major area of focus for lending for financial inclusion. In all, between FY02 and FY13, for example, on a thematic basis, the World Bank approved around 207 projects that had components with themes related to access to finance. However, only 33 had a focus on access related themes of 50 percent or more. Similar results are obtained using a sector code classification. 8. IEG evaluations provide an independent review of each closed and completed Bank project. IEG prepares Implementation Completion Report Reviews (ICRRs) for all projects, that validate the findings of Implementation Completion and Results Reports (ICRs) prepared by project teams, and provide overall outcome ratings on a six point scale. While ICRRs are based on desk reviews, IEG’s PPARs represent in-depth investigations that include field visits to the project, conversations with stakeholders, and, in some cases, beneficiary surveys, as well as other externally available third-party information. They review factors affecting outcomes and provide information on issues of particular importance to the achievement of project objectives. The number of projects with financial access content that have been completed and closed for which such ICRRs and PPARs are available is a necessarily different group from the set of new loan commitments, as they represent projects that in some cases were initiated much earlier, and they will not include projects that are still under execution.7 As shown in Table 2.3, only 30 projects with an access to finance content of at least 35 percent are included in the group of projects that have closed since 2006. Scattered throughout the Bank’s regions, the majority (50 of 125) fell under the Finance and Private Sector Board, but a large number also fell under other sector boards, especially, Agriculture and Rural Development (18 of 125 projects) Almost all loans for support to financial access took the form of investment lending; only 18 were policy-based loans. Almost all (97 of 125) received outcome ratings of Moderately Satisfactory or better. 7IEG’s annual Results and Performance reports show that broadly, on average (depending on the years, sector composition, and so forth) investment loans take six to eight years to be completed. 4 Table 2.3. World Bank Projects with Access to Finance Sector or Theme Content Exiting the Bank’s Portfolio between FY06 and FY14a Number of projects with Access to Finance content 125 Number of projects by sector board Total commitment all Access to Finance Bank projects (sector 12.6 Financial and Private Sector Development 50 and/or theme) (FY06-13)($ billion) Access to Finance commitmentb Agriculture. & Rural Development/ Environment 22 At least 35 but less than 55% 12 Social Protection 15 At least 55 but less than 75% 8 Economic Policy 10 At least 75 and up to 100% 10 Number of projects by region Average percentage of Access to Finance content Africa 31 Average A2F sector content (%) 32.8 East Asia Pacific 16 Average A2F theme content (%) 19.5 Europe and Central Asia 27 Number of projects by IEG outcome rating Latin America and the Caribbean 28 Highly Satisfactory 7 Middle East and North Africa 13 Satisfactory 38 South Asia 10 Moderately Satisfactory 52 Projects by lending instrument Moderately Unsatisfactory 11 Specific Investment Loan 68 Development Policy Loan/Structural Adjustment Policy Unsatisfactory 11 18 Loan Highly Unsatisfactory 3 Adaptable Program Loan 12 N/A 3 Financial Intermediary Loan 12 Number of projects by disconnect Technical Assistance Loan 8 IEG Disconnect (1) 38 Learning & Innovation Loan 5 Sector Investment and Maintenance Loan/ Emergency 7 2 IEG disconnect (2 or more) Recovery Loan Source: World Bank data (August 2014). a In the present table, access to finance is defined as projects having either access to finance sector content or thematic content. b Either thematic or sector content; the higher of the two if both are available. 9. This paper illustrates the Bank’s achievements in terms of its lending for access to finance based on the examination of a small group of closed projects that exited the Bank’s portfolio between FY06 and FY13. The sample is defined to include, primarily, four projects with access to finance components on which in-depth reviews have just been completed, in the form of PPARs. These span two very different Bank regions – Latin America (Brazil and Mexico) and Africa (Ghana and Mozambique). For these projects, supplementary information was purposefully collected at the time when the PPAR missions were undertaken, with a view to drawing together lessons on their access to finance elements, based on a set of questions related to financial inclusion.8 The sample also draws on another two projects in Bangladesh, for which in-depth investigations were launched and supplementary information collected, albeit without a field visit.9 Evaluative information is 8 These PPARs were postponed at the request of the country unit. Draft desk PPARs, which include information based on a literature survey, conversations with team members and researchers outside Bangladesh, have, however, been prepared and will be referred to subject to caveats regarding the necessarily partial nature of these two PPARs. 9 These pioneering projects in Bangladesh provide particularly interesting examples of the Bank’s attempts to engage with microfinance. Although field visits were not possible because of country circumstances, the team collected considerable information beyond what was available through IEG’s first-tier ICRRs, through extensive consultations with the project team as well as subject experts. 5 however available from IEG reviews of their ICRs. With the exception of Mozambique, which was a broad private sector development project with just a line of credit for small and medium-size enterprises (SMEs), all the other projects were focused specifically on expanding access to finance for underserved segments of the population or regions of the country. 10. The extent to which the selected projects focus on financial access ranges from 99 percent of access to finance content in the case of the Brazil Northeast Microfinance project to 40 percent in the case of the Ghana Rural Financial Services Project (Table 2.4). Table 2.4. Characteristics of Projects Reviewed Brazil Ghana Mexico Mozambique (Bangladesh) (Bangladesh) Northeast Rural Financial Rural Finance Enterprise Development Second Poverty Financial Services Project name Microfinance Services Project Development Structural Project (PoDE) Alleviation for the Poorest Development Project Adjustment Operation Microfinance Project Approval date 30-May-00 8-Jun-00 16-May-03 27-Jan-00 18-Jan-01 21-Aug-02 Closing date 31-Mar-07 31-Dec-07 19-Dec-09 30-Jun-06 30-Jun-09 31-Dec-07 Loan amount ($ million) 38.5 5.67 505.06 24.96 159.64 5.69 ICRR (September IEG ICRR / PPAR (date) PPAR (April 2014) PPAR (May 2014) PPAR (January 2014) PPAR (March 2014) ICRR (April 2010) 2008) Percentage of Access to Finance content (by 99 40 80 43 96 50 theme and/or sector)a To improve the To promote growth To support the To help broaden the base To diversify To extend financial access of and reduce rural liquidation of the loss of private participation in microcredit services through microenterprises poverty in Ghana making financial Mozambican economic operations to reach innovative delivery throughout the by broadening and institution, Banrural and growth. the urban poor, and mechanisms to a Project objectives Northeast region to deepening financial to put in its place micro-entrepreneurs section of the (summary) sustainable, formal intermediation in Financiera Rural, a and to expand poorest to improve financial services. rural areas. nonbanking outreach to reach a their socio- decentralized institution greater number of economic status the poor. Support to Regulatory support Closure of a loss Line of credit to provide Financing of a Financing of a specialized for the Bank of making public loans to small and revolving fund in an revolving fund in microcredit program Ghana & Ministry agriculture bank and medium-scale apex MFI, to be an apex MFI public Nature of Access to in a public bank, of Finance; support conversion to a non borrowers as well as very used to extend Bank to be used to Finance components including a line of to informal banking decentralized small, first-time bank microcredit to the extend credit microfinance institution. borrowers urban and rural poor. microcredit to the entities. poorest. Bank Loan Type SIL FIL SAL LIL FIL LIL Banco do Nordeste Bank of Ghana Financiera Rural Ministry of Industry, Palli Karma-Sahayak Palli Karma- Overall implementing (Rural Finance Commerce and Tourism Foundation Sahayak agency(ies) Inspection Foundation (PKSF) Division) Implementing Banco do Nordeste Microfinance Unit Financiera Rural Central Bank, DFI (GAPI) PKSF and Partner PKSF and Partner agency(ies) for Access of the Ministry of and PFIs, Organizations Organizations to Finance component Finance Source: Project documents. Note: DFI = Development Finance Institution; FIL = financial intermediary loan; ICRR = Implementation Completion and Results Report Review; LIL = Learning and Innovation Loan; PFI = Participating Financial Institution (commercial bank); PPAR = Project Performance Assessment Review; SAL = structural adjustment loan; SIL = specific investment loan. a Either thematic or sector content; the higher of the two if both are available. 6 2.2 Projects Reviewed: An Introduction and Summary Overall Project Objectives 11. Two of the projects reviewed, the Rural Finance Development Structural Adjustment Operation in Mexico and the Rural Financial Services Project of Ghana had a rural focus, whereas the goal in Brazil’s Northeast Microfinance Development Project was to increase access to finance in the underserved north eastern region of the country. The fourth project, Mozambique’s Enterprise Development Project (PoDE: Projecto para o Desenvolvimento Empresarial) was a broad-based private sector development project, which also included, as one of its goals, an increase in access to finance for underserved SMEs. In Bangladesh, the Second Poverty Alleviation Microfinance Project (Microfinance II) carried forward the gains of its predecessor (Microfinance I) in increasing rural microfinance, but it also targeted other segments such as the urban poor, hardcore poor and micro- entrepreneurs in the so-called “missing middle”—those whose needs had outgrown the small loans offered by microfinance institutions (MFIs), but who lacked access to formal loans.10 These projects were all predicated on then emerging evidence of the causal link among finance, growth, and poverty reduction among the economically active poor. The last project reviewed here is the Financial Services for the Poorest (FSP), also in Bangladesh, an LIL that aimed to further test for sustainable models of microfinance delivery for the poorest or the bottom 10 percent of the population that was not served by mainstream microfinance and which were largely dependent on government safety nets. Macroeconomic and Sector Context 12. Project objectives in each loan reflected the overall country and sector context and were thus largely relevant. In Mexico the project had the overarching objective of closing down and liquidating a large loss-making public bank, Banrural, and setting up a non-bank development finance institution in its place, among other things improving transparency, incentives, and governance and minimizing potential contingent fiscal liabilities. The project was conceived in the context of an overall weak financial system emerging from the Mexican peso “Tequila crisis,” triggered by the sudden devaluation of the Mexican peso in December 1994. In the aftermath of the crisis, and despite some restructuring efforts, development banks lagged far behind private banks in terms of efficiency. Their accumulated losses and poor loan recovery were attributable to unsound loan pricing policies, as well as nonreimbursable services (for example, training, technical assistance) and inefficient internal operations. 13. Similarly, in 2000, when the loan was approved, Brazil had just emerged from a period of hyperinflation and was still under the grip of a serious recession that started in the late 1990s. Average GDP growth rate in 1998-2000 was low but positive, at 1.5 percent in real terms. The country’s overall poverty rate of 17.4 percent was well above the norm for middle-income countries, having increased by 40 percent over the period of hyperinflation.11 10Fora review of the debate on the microcredit impact on poverty in the Bangladesh context, see Khandker and others (2014), Pitt and Khandker. (1996), Li, Squire, and Zou (1998), Khandekar (1998, 2005), and Li, Xu, and Zou (2000). 11 World Bank (1995) 7 There were also wide disparities in the incidence of poverty within the country. The Northeast accounted for only about 30 percent of Brazil’s population (45 million) but more than half of the country’s poor. The poverty rate of the Northeastern states exceeded the national average, ranging from 19.4 percent (in the province of Sergipe) to 50.9 percent (in Piaui). There was also a marked disparity between urban and rural poverty. For Northeast Brazil in 2002, poverty ranged from 34 to 38 percent in urban areas and from 55 to 57 percent in rural areas.12 14. There were also substantial regional differences in terms of financial depth and access to financial services in Brazil. In 2000, there were 12 bank branches per 100,000 people in the South but only 4 per 100,000 in the North and 5 per 100,000 in the Northeast. About 75- 86 percent of the formal credit outstanding in the Northeast Region originated from public banks. The disparity in coverage and apparent lack of incentives for increased coverage by private banks, who found more creditworthy clients in urban centers provided the primary justifications for the government’s continued support of publicly owned banks, which in 2000 accounted for 46 percent of bank branch infrastructure and 55 percent of financial sector assets in Brazil despite generally high default rates and administrative costs. In this context, the project’s objective of improving the access of microenterprises in the Northeast region to sustainable, formal financial services was relevant, although it aimed to do so through the expansion of a single institution, CrediAmigo, which was a pilot microfinance program embedded in a regional public development bank, the Bank of the Northeast of Brazil. 15. The Rural Financial Services Project in Ghana was designed when the country enjoyed steady economic growth and inflation showed a downward trend. However, poverty—particularly rural poverty, at 39.2 percent—continued to be high. And poverty in the rural northern savannah regions was even higher, at three-fifths of the population. Given the predominance of poverty in rural areas, improving the competitiveness and productivity of the rural economy was considered crucial for sustained growth and poverty reduction. Against the backdrop of fragmented and largely informal financial services in rural Ghana, lack of access to finance, coupled with a network of mostly weak rural and community banks, was identified by the Project Appraisal Document as a key constraint for the small farming households that dominated rural economy. When the project was conceived in 1998, total commercial bank credit to the agricultural/rural sector had dropped from 31 percent of all lending in 1983 to less than 10 percent in 1998. Informal sources, such as Ghana’s sususor community-based collective lending vehicles, thus gained greater prominence.13 Community-owned rural banks and microfinance organizations that could potentially fill this void were weak and fragmented and operated with outdated technologies.14 12 Skoufias (2007). 13 Susu groups are savings groups, with group members arranging for collection and payment. The ROSCA (rotating savings and credit association/club) is a modified form of susu group. Under the ROSCA model, members decide on the amount to be contributed by each member at regular intervals. The total collected is given to each member in turn (IFAD 2000. 14 See, for example, Pearce et.al. (DFID, 2004) which states that the rural financial sector in Ghana is dominated by informal service providers that undertake minimum intermediation and often specialize in either deposit-taking or lending. The few who intermediate do so within group-based 8 16. In this context, the Ghana Rural Finance Project was designed to improve the delivery of financial services in rural areas, through a multipronged approach: strengthening of the rural and community bank system and building the capacity of semiformal and informal entities by strengthening their linkages to the rural banks. It also intended to strengthen the thin and overextended oversight of the rural and community banks. Regarding the strengthening of the informal sector, it was expected that linkages with formal rural banks would help bring together such entities in a mutually beneficial framework. Informal final borrowers were considered too fragmented, lacking the education, information, and assets or collateral needed to effectively obtain financial services. Conversely many formal financial institutions in urban and quasi-urban settings lacked the experience and organizational arrangements to link "backwards" into the rural sector. Thus, overall, Ghana’s Rural Finance Project’s objectives were to "promote growth and reduce rural poverty” by broadening and deepening financial intermediation. Improving poor donor coordination was intended to be a parallel objective. 17. The fourth country in which a detailed project investigation had been undertaken, Mozambique, had emerged from a civil war in the previous decade. Following first- generation reforms, it had grown rapidly, at 8 percent per year on average. It had also reduced the poverty headcount index from 69 percent in 1997 to 54 percent in 2003. Despite steady gains, however, Mozambique’s growth was largely driven by foreign aid and foreign investment in megaprojects and was not inclusive. Its megaprojects were concentrated in energy (Cahora gas fields), industry (Mozal Aluminum plant), and mining (Moatize coal mines, Moma Titanium). These foreign-owned, export-focused, “enclave” capital intensive projects were not able to develop linkages with the broader economy, so that there was limited “trickle down” to local small entrepreneurs.15 18. In this context the Bank’s Enterprise Development Project (PoDE) intended to broaden the base of private sector participation in economic growth by boosting the competitiveness of Mozambican private firms; enhancing access to term finance; and building the capacity of selected government and nongovernment agencies. It aimed to enhance access to term finance through lines of credit, together with other components focused on strengthening the business and business organizations, and through direct support to specific sectors to improve their competitiveness. 19. In Bangladesh, despite modest economic growth attained in the 1990s, half the population still lived below the poverty line. The challenge for the government was to spread the benefits of growth among the poor and reduce poverty. Bangladesh adopted a two–fold poverty reduction strategy: a macroeconomic strategy to accelerate economic growth and improve prospects for employment coupled with limited direct poverty-focused interventions supplemented by NGO activities to alleviate poverty and to foster human development. systems with little or no access to external resources. Consequently, the informal sub-sector lacks the capacity to meet the requirements of the rural population. 15 Although they represent 62 percent of manufacturing, “megaprojects” only employed 3 percent of its labor force and 2 percent of urban private sector employment (IMF 2011). The report also commented on the increasingly concentrated pattern of Mozambique’s growth take-off, focused on the large foreign projects, and also pointed to weaknesses in the business environment that prevented diversification. 9 20. Unusually, compared to other countries, microfinance had gained national prominence as a solution to the expansion of financial access, as demonstrated by the pioneering success of Grameen Bank. Early World Bank studies using Grameen Bank data also demonstrated a causal link between micro credit and poverty reduction.16 The Government set up an apex funding and capacity building institution, Palli Karma-Sahayak Foundation (PKSF) as a vehicle for lending funds on concessional terms to NGOs and microfinance institutions (MFIs), which then on-lent the funds to the poor. PKSF in its initial stages suffered from funding constraints. Thus at the Government’s request the IDA- supported first microfinance project for the rural poor (Microfinance I) compensated for a decline in donor funding and became a major funding source of PKSF, accounting for 90 percent of its funding sources. The phase II project attempted to accompany the expansion in outreach with diversification to reach the urban poor; the hard-core poor, and micro- entrepreneurs, in addition to providing technical assistance to PKSF, the apex microcredit “wholesaler,” and establishing a legal, regulatory and supervisory framework for NGOs and MFIs. 21. Mainstream microfinance, however, did not reach the bottom 10 percent of the population. To deepen outreach to this segment, IDA supported LIL projects aimed to test models for delivery to the poorest, and to see whether, in addition to safety nets, microfinance combined with skills enhancement could sustainably increase the welfare of the poorest. 2.3 Key Questions Addressed 22. In the following sections, the paper refers to the projects reviewed to explore the following themes:  What borrower institutions has the Bank successfully been able to partner with, as vehicles for expanding financial access (public banks, private apex banks, nonbank financial institutions, microfinance)?  What Bank loan design features are best suited to expanding financial access (loan type, analytical underpinning, technical assistance, matching grants)?  To what extent do factors in the environment of the country, outside the parameters of the project, affect its likely outcome (overall economic management, the financial sector policy environment and financial sector infrastructure)?  Has the Bank/IDA been able to work with other donors/partners in financial access projects?  Who are the ultimate beneficiaries of these projects been? Has the Bank been able to reach out to the most underserved? 16 Pitt and Khandker (1998) 10 3. Findings: Borrower Institutional Arrangements 23. Borrower arrangements in each project were dictated by the circumstances and the objectives pursued. Although there is clearly no universal institutional arrangement that can work in all circumstances, to achieve different project objectives, lessons of good and poor practice may be drawn from the form in which each project was structured and how it worked with local financial and regulatory institutions to achieve its objectives. The projects used different institutional vehicles, ranging from public banks; development finance institutions; and apex bank structures that worked in tandem with private commercial banks, rural community banks and in one project, an apex microfinance institution, to implement their objectives. The following section examines the extent to which successful outcomes for inclusion can be achieved with different borrower arrangements. Borrower Institutional Arrangements and Bank Loan Structure: Key Findings In these examples, different institutional vehicles have been used for World Bank lending for financial inclusion:  Public sector banks and development finance institutions  Lines of credit  Apex bank structures, including an apex microfinance institution Public Banks:  Are prone to problems of inefficiency and recurrent losses.  Expanded financial outreach and limited losses have been achieved but through transformation to a development agency model.  Moving from first tier to second tier lending can be difficult and may take years to achieve.  Such agencies can suffer from shifts in the political climate.  Ring-fencing “independent” microfinance programs in such banks is difficult and vulnerable to future recapture by the parent institution.  Tensions about the extent to which public or private sector principles apply persist. Lines of Credit:  Such operations may increase the supply of funds but do not increase sustainable access to finance if other institutional, legal and regulatory constraints to financial sector development are not addressed.  Lines of credit require assessments of not just demand but also factors that could dampen demand, such as high interest rates, the availability of non-loan income to the banks, e.g., low-risk government securities.  On the supply side there can also be factors that may limit commercial banks’ appetite to lend to small and medium enterprises, even if they have the funds—for example, limited information on borrowers’ creditworthiness, poor laws on collateral, or the availability of other opportunities for investing, such as government paper, with potentially lower risk adjusted returns.  Loan design features may also be a consideration—for example minimum or maximum size, earmarking or preferential schemes—for example, for first time borrowers, domestic borrowers.  Credit line operations that are able to disburse often display elements of subsidy shared with the borrower, even if not initially designed with such features. Credit lines accompanied by programs of technical assistance or training can be effective. Apex Bank Arrangements:  Apex arrangements have been the typical approach for direct support to rural finance institutions, smaller microfinance or informal financial institutions—with only partial success. 11 3.1 Large Public Sector Banks 24. To what extent was the Bank able to partner with public sector banks to achieve the expansion of financial access? Among the projects reviewed, two worked with large public sector banks, Banrural in Mexico and the Banco do Nordeste in Brazil. Both countries had big loss-making public sector banks that suffered from legacies of extreme inefficiencies, poor supervision, and, to some extent, misplaced attempts to expand financial access through special government programs. In both cases, the Bank’s projects were successful in achieving the expansion of financial inclusion to rural communities in Mexico and to the poor northeast of Brazil. Yet, in Mexico, the government chose to shut down its loss-making rural bank, replacing it with a non-bank (non-deposit-taking) entity, Financiera Rural, financed directly by the government and under its direct oversight. The project was successful in stopping the losses at Banrural and the contingent fiscal liabilities of the government, and expanding financial access. 25. In Brazil, the project worked through a special program for microfinance within one of the large public banks, separated in large measure from the parent bank.17 The CrediAmigo program supported by the Bank established a quasi-autonomous program, situated within a large and loss-making federal bank, the Banco do Nordeste do Brasil (BNB). The project illustrates the remarkable extent to which such a program managed to maintain largely sound lending practices, and on this basis, to expand outreach to the underserved segments. Yet the program was not able to sever its links with the public bank and eventually benefitted indirectly from privileged access to funding. Private microfinance entities pointed out the difficulties of entry in the context of the dominance of the public program. Mexico: Banrural 26. For several decades Banrural had provided subsidized credit for agriculture, thus arguably reaching out to an underserved rural community. However, the institution also suffered from extreme inefficiencies, including excessive administrative costs and low collection efforts, leading to very large accumulated losses. The institution had been able to remain liquid in spite of its huge losses by attracting deposits at unsustainably high interest rates. Its losses represented a fiscal drain on the government. Yet the government recognized the need to provide access to financial markets to rural groups not served by commercial banks, especially in the form of prudent lending. Thus it decided to create Financiera Rural. This development agency, which would not receive deposits or issue any debt, was to be totally financed with an endowment created by the government, and it was required to maintain the value of its capital endowment in real terms, report periodically to Congress, lend only to low- and middle-income rural producers, and move, over time, fully to second- tier lending through banks and non-bank financial institutions. Its capacity to provide 17During this period, the government chose to restructure and recapitalize its loss-making federal banks and to introduce a new level of transparency in its operations, recognizing and absorbing past losses. Going forward the federal government assumed the risks and costs of the various federal credit programs managed by these institutions, reflecting transparently in its budget any subsidies provided through these banks. It was against this backdrop that the Bank supported the CrediAmigo program (Brazil Second Programmatic Financial Sector Adjustment Loan (PSECAL 2, 2003). 12 subsidies was limited by Congress, making those that they did provide more targeted and transparent. 27. The government was guided by the 2001 joint International Monetary Fund (IMF) and World Bank, Financial Sector Assessment Program (FSAP) that recommended Banrural’s closing, which pointed out that development banks suffer from an inherent tension between their social policy mandate (that is, to foster access to financial services for households and firms that are costlier and riskier to serve) and the objective of avoiding losses in their bank-like activities. It was expected that the development agency model would overcome this tension by setting a strict limit to the funds that the public sector would be putting at risk and forcing, through the second-tier structure, the participation of third parties in credit decisions—preferably private institutions that would put their own capital at risk. The government decided to act on this advice in the case of rural credit. The Bank team prepared a financial model that simulated the situation of the public sector agricultural bank, Banrural, that showed that no “good bank” could be rescued from its remains and that it was better to close it. In collaboration with the Inter-American Development Bank, the Bank worked in the preparation of the pro forma financial statements and projection model used to assess whether Financiera Rural could meet the “zero net-income” test by the end of the project. 28. Thus overall, the Banrural/Financiera Rural project illustrates a case where the Bank was able to support the closure of a public sector bank, which had grown to reflect the traditional problems of such institutions. The Bank was also able to provide for substantial continuity in its services for the financially underserved in a more sustainable way, through a public development agency, and work toward this in a gradual manner, in keeping with the government’s concerns. As the government had articulated its constraints in converting the Tier 1 structure into a Tier 2 structure, the team agreed to a phased approach toward this end. Brazil: Banco do Nordeste and CrediAmigo 29. At the time of preparation of the Brazil Northeast Microfinance Development Project, and up to its approval in the year 2000, the aim was to isolate the CrediAmigo program unit from the parent bank and eventually to spin it off.18 While CrediAmigo successfully expanded outreach to the financially underserved Northeast of Brazil, it was not eventually spun off from its parent institution. The Bank supervision team tried to ensure the soundness of operations at CrediAmigo, and its functional independence and autonomy, despite its absorption into Banco do Nordeste and the access to BNB funding reflected in eventual redundancy and partial cancellation of the Bank loan. There were also advantages in remaining within Banco do Nordeste for CrediAmigo—access to management information systems, ATM machines, treasury services, routine personnel management, security and insurance—that would otherwise be costly for an independent CrediAmigo. However, the 18It later became increasingly clear to the Bank that the parent institution was beset by problems similar to those experienced by Banrural; in a separate and subsequent operation, the Bank supported the government’s reform program for the public federal banks, including the Banco do Nordeste, which attempted to introduce standard norms for capital adequacy and loan loss provisioning, and greater fiscal transparency (World Bank 2002). 13 advantages of staying on within CrediAmigo were well known even at the design stage. CrediAmigo’s independence was considered important and central to its access to market funding and long-term sustainability, both by the Bank and by Banco do Nordeste senior management.19 30. The government introduced significant policy changes in 2003, during project implementation. These provided for expanded availability of funding for microfinance programs within the public banks. There were clear advantages for Banco do Nordeste of keeping CrediAmigo within its operations. The government supported expanded funding for microfinance through the earmarking of a portion of deposits for microfinance, with the alternative of transferring them to the Central Bank as unremunerated reserves. As many banks did not have the infrastructure or the experience to undertake microfinance lending, the expertise acquired by CrediAmigo in this regard provided Banco do Nordeste with an opportunity to gain deposits (remunerated at potentially low rates) from such institutions, to undertake microfinance lending on their behalf. CrediAmigo did not become a demonstration of sustainable microfinance that encouraged private participation. Moreover, the relatively expensive line of credit given by the Bank to CrediAmigo became less attractive and was no longer needed, leading to its partial cancellation. Thus, overall the Bank project’s implementing arrangement, though successful in increasing access for the underserved, eventually did benefit from state support and did not demonstrate, as it set out to do, the effectiveness of market-based finance. 31. The project team’s confidence in the design, the preparation, and the policy maker’s commitment was such that they did not include the risk regarding “continuity in policy framework” that was a prerequisite for the success of the intended demonstration effect project. Better risk management could have been reflected in the design and alternatives worked out in consultation with Banco do Nordeste. 3.2 Lines of Credit and Participating Financial Institutions 32. The CrediAmigo program also illustrates an attempt to expand inclusion through a line of credit – in this case, offered directly by the CrediAmigo program to final borrowers. The program was clearly successful in achieving outreach, as illustrated in Table 3.1. However, the program also illustrated the extent to which the greater benefits of expansion of a sustainable model of microfinance, though the “crowding in” of new private sector lenders, did not occur. According to the World Bank/IMF financial sector assessment of Brazil, in 2012 BNB remained by far the largest microcredit provider in the country, with a market share of over 70 percent, more than a million active clients in the CrediAmigo Program and another 650,000 in a parallel program it established, AgroAmigo. Three state-owned banks continue to be the main source for financial products and services for low-income consumers. The crowding out impact of the government’s programs appears to dominate any potential crowding in from the success of the CrediAmigo example.20 19Project Appraisal Document, Brazil Northeast Microfinance Development Project 20A review (in Agencia Brasil 2013): Microcrédito é recorde no país, mas atende apenas parte dos pequenos empreendedores. O Globo 2013, Dec 22) cites that the recent increase in microfinance can be almost 14 Table 3.1. Key Indicators of Expansion of the CrediAmigo Program Target 1999 2000 2003 2004 2005 2006 Indicator 1. Extent of 150,000 outreach: Number of 37,976 57,967 138,497 162,868 195,378 235,729 additional active clients Indicator 2: Breath of outreach: number of No target n.a. 358 1172 1165 1200 1420 municipalities in NE Brazil with CrediAmigo clients Intermediate Indicator 1: Depth of outreach: No target 17,613,634 32,816,435 91,516,126 111,541,109 141,317,421 177,621,238 Outstanding loan portfolio (Reais) Intermediate Indicator 2: Average loan size over 65 27 25 20 18 19 19 Northeast Brazil, per capita GNP (%) Source: Brazil Northeast Microfinance Development Project, 2014. Note: n.a. = not available. 33. In the case of Brazil, the line of credit offered was a “retail” line, with CrediAmigo disbursing directly to end borrowers as a first-tier lender. CrediAmigo did not move toward a second-tier model. Was the Bank more successful in achieving sustained expansion of lending to the underserved in cases where it set up arrangement with apex banks that were then intended to work as second tier lenders to a wider number of other financial institutions? The Mozambique Rural Financial Services Project provides one example. entirely attributed to the entry of two additional public sector banks, Banco do Brasil and the Caixa Economica Federal. Researchers in Brazil point out that despite recent growth; microfinance in Brazil still reaches only a fraction of its potential clients, pending more universal participation by private sector banks. According to Lauro Gonzalez, coordinator of the Center for Microfinance Studies in the School of Business Administration at São Paulo’s Fundação Getulio Vargas, there needs to be investment and innovation at private banks, that need to develop appropriate technologies and products for the public—but private bank participation is effectively inhibited by subsidies offered through the Crescer Program. 15 Mozambique: On-lending Arrangements 34. In Mozambique, the Bank/IDA offered a line of credit for SMEs, with the Central Bank as the apex borrower. IDA funds were made available to the government at zero percent and passed on to the Bank of Mozambique for on-lending to participating financial institutions, at the six-month deposit rate. Thus the Bank of Mozambique, as trustee, received the mark-up between IDA funding provided to the government and the lending rate to participating financial institutions (PFIs). It was intended that participating financial institutions would on-lend, in turn, at market terms, to final borrowers. Several commercial banks had initially indicated their interest in increasing longer term lending if they had an appropriate longer-term funding base. Initially, nine banks signed up, but disbursements were very slow until the fourth year, when addition of a noncommercial development finance institution among the participating financial institutions and relaxation of certain loan design features increased disbursement. 35. The project illustrates a number of reasons why commercial banks could lose interest in a Bank-sponsored line of credit, despite its rare offer of long-term funds. Some of these factors related to the broader environment. On the macroeconomic front, a first factor was the high market interest rate structure (up to 35 percent in the year 1999), which limited the interest of potential borrowers. Second, the financial sector environment was not conducive to expanded market-based intermediation. One of the large participating commercial banks became insolvent as the IDA operation was getting under way, and the policy response was to grant such potentially failing banks quasi-exclusivity in the purchase of t-bills at rates sufficiently high to allow them to recapitalize. As the increased issue of government t-bills implied that banks could place their money in relatively well-remunerated and safe assets, the attractiveness of lending to potentially risky clients was reduced. There were other weaknesses in the financial system that made it difficult for lines of credit to succeed, for example, the credit disciplinary environment was weak, with little information on credit risk, which made lending to small borrowers especially risky—and unattractive compared to holding government paper. 36. Third, there were also factors related to the specific details of the design of the line of credit by the Bank in its loan to Mozambique that slowed disbursement. Subproject eligibility criteria for beneficiary firms required them to be “domestic enterprises” or 51 percent locally owned, and this restricted the pool of eligible enterprises. A low loan ceiling was specified for first time borrowers, at $15,000, and both eligibility and collateral requirements were severe. One part of the line of credit took the form of a special facility for small borrowers, but eligibility required that borrowers had to have (i) no history of previous borrowing from banks; (ii) minimum equity of 10 percent of the amount of the proposed project; and (iii) maximum borrower total assets of $40,000. Eligibility requirements for a second, traditional, part of the line of credit included the conditions that prospective borrowers have (i) no outstanding loans in default; (ii) minimum equity requirements of 25 percent of the subproject; and (iii) a maximum $500,000 in total assets. Clearly, such design aspects of a credit line, that fragment and limit the potential client base, can be restrictive. 16 37. Disbursement finally accelerated thanks to first, the expanded role of a public development finance institution among participating financial institutions: GAPI.21 Two features of GAPI’s involvement were important in this regard: first, unlike the commercial banks, GAPI received project funds with some subsidy from the Central Bank, alleged to be in the order of 2.5 to 4.5 percent. GAPI passed on a portion of the subsidy to its clients who thus paid lower rates on their loans than those charged by the commercial banks. A second feature of GAPI’s involvement was the assistance it was able to offer, together with the loan, to assist small customers with loan processing. Changes in the design elements of the loan also helped the later accelerated disbursement: the raising of the loan limit for first-time borrowers from $15,000 to $ 40,000 and the expansion of eligible borrowers to include foreign-owned enterprises in addition to Mozambican enterprises. Nonetheless, Bank funds allocated to the credit line were reduced by almost 50 percent.22 38. Other international lenders used significantly different pricing arrangements for their lines of credit to Mozambique, with different foreign exchange risk arrangements. The European Investment Bank also offered Mozambique two lines of credit (1999 and 2003) in foreign exchange at a nominal interest rate of 1 percent to participating financial institutions; these on-lent in domestic currency at interbank rate plus a risk margin of their choice, it seems typically with roughly the same margins as applied to the IDA funds. In this case, the participating financial institutions took the foreign exchange risk. Overall, the IDA line of credit in this project was marginally successful at best in expanding access of SMEs to medium-term finance.23 21Sociedadede Promoção de Pequenas e Medias Empresas—Company for the Promotion of Small and Small and Medium Enterprises. 22 Was the assumption of foreign exchange risk a reason for slow disbursement? In the case of this operation, the commercial banks took no foreign exchange risk, as this was assumed by the government and local commercial banks received money in local currency. The government, in fact, also took no foreign exchange risk vis-à-vis repayment of IDA funds, since the IDA funds were written off in the context of the Multilateral Debt Relief Initiative. Similar arguments apply to loans denominated in foreign exchange. In such cases, the cost of funds was significantly lower to the PFI, because interest rates in foreign exchange charged by the central bank were significantly lower than advances made in domestic currency. However, in this case, the subborrower took the foreign exchange risk. Available evidence suggests however that subborrowers were able to manage this without difficulty to the extent that the repayment rates of all loans was in excess of 95 percent. This was likely because on-lending in foreign exchange was encouraged only for exporters and others who had foreign currency revenue streams which could provide a hedge against such risk. Overall loan repayment rates as noted above were in excess of 95 percent. 23 Although it is difficult to trace attribution of success or failure to this specific project, there is general evidence that indicates that access to credit, especially for SMEs, remained low. The FSAP Update (2009), for example, showed that the private sector credit-to-GDP ratio at 12.8 percent in 2009 was particularly low, up only marginally since the 2003 FSAP. The outreach of the banking sector had expanded but access to financial services overall remained low and fragmented. Though large corporations have relatively good access to credit, SMEs do not. Likewise, the World Bank Group’s Investment Climate Assessment (2009) reported that the percentage of firms accessing formal finance (13 percent in 2009) was not only low but had in fact declined from previous surveys (29 percent in 2002 and 26 percent in 2005). State-owned enterprises, which dominate the enterprise sector, account for 80 percent of total nongovernment credit. And access to finance was the second most often cited 17 Bangladesh: On-Lending through a Government-Established Apex 39. In Bangladesh, the government set up an apex funding and capacity building institution, PKSF, as a vehicle for on-lending funds, on concessional terms, to (NGOs and MFIs. PKSF was the implementing agency for both projects reviewed—the Second Microfinance Project (MFII) and the LIL FSP project. Although this review focuses on the second phase of the project, it is noteworthy that PKSF successfully disbursed and implemented a previous microfinance loan, focused on the rural poor. 40. The second microfinance loan gave PKSF a much wider role, including components for the hardcore poor and for microenterprises. PKSF was not able to deliver on these components. Funds for these components were discontinued and reallocated to rural microfinance, even though the Project Appraisal Document had set a careful limit on rural microfinance, to “not over-flood” rural microfinance. Similar difficulties were encountered by the second Bangladesh project under review: funds earmarked for learning and innovation were converted to capacity building activities for recipients of the government’s “Ultra Poor Program,” launched while the loan was still under implementation. 41. Some of the largest participating microfinance institutions under the PKSF (BRAC, ASA, and PROSHIKA), as well as some medium-sized ones, reduced their dependence on PKSF for funding and began borrowing funds from local and international financial markets at competitive rates.24 This in fact was a stated goal of the MFII project, which sought to encourage market-based funding for its participating microfinance organizations, through guarantees for loans to these institutions as the project appraisal document had proposed. However, the large ones among these left PKSF because of the capping of service charges by PKSF. Thus, the objective of achieving market-based funding was only indirectly achieved, and through mechanisms other than the project. Moreover, the Bangladesh MFI growth, was spearheaded by mandatory deposits and internal retained earnings (Rutherford, 2012). Smaller and medium-sized microfinance institutions as well as PKSF itself, however, continued to depend on soft loans. Although nongovernment MFIs have been an important part of the financial inclusion agenda in Bangladesh, Bank support for the two microfinance projects likely contributed to this growth but was not central to it. There have since been obstacle to business operations by Mozambican firms; this was confirmed as a key obstacle through empirical analysis linking access to finance indicators to firm-level productivity in Mozambique. The empirical analysis of this ICA, taking into consideration demand factors, found a highly credit constrained environment: more than two-thirds of the sample was credit constrained (World Bank 2009a). Although the project increased access to finance only marginally, the other components of this private sector development project, not reviewed in this working paper, did relatively better. 24According to a 2011 BRAC study, “…given their maturity, the large MFIs in Bangladesh have substantial cumulative client (member) deposits and internal sources of financing to fund their credit activities. This is in complete contrast to MFIs in India, where there has been a substantial focus on debt financing, and other countries, such as those in Africa where donor funds still play a significant role in microfinance” (Sinha 2011). 18 concerns, in the case of the second microfinance loan, that the large availability of funds may have spurred multiple lending.25 42. In both projects, the government on-lent to PKSF at highly concessionary interest rates with the understanding that participating microfinance institutions would have the flexibility to charge market rates.26 IDA agreed to preferential support for small and medium- size participants. However, beyond 2010, the project envisaged funding from market financial instruments. 43. Overall, the line of credit targets with respect to rural and urban microfinance were successfully met. However, other important components, specifically for microentrepreneurs, were cancelled, as were the funds reallocated to rural microfinance. A major advantage of PKSF, cited in the project’s ICR, was its access to concessional government and donor funds. PKSF’s capacity to run microfinance operations using funds from the formal financial markets at market rates is yet to be proven.27 And PKSF was not able to implement the Hardcore Poor component under MFII because of implementation challenges. By contrast, BRAC (an independent large microfinance institution that later left the Bank-sponsored loan) successfully and independently implemented its own graduation program for the poor and documented its implementation and impact. The program is being globally scaled up by CGAP in nine countries and is currently being evaluated by the MIT Poverty Lab.28 44. Although disbursement rates for the traditional loan components were high, this was not the case for innovative or newer segments of borrowers. The Learning and Innovation project did not define who the “poorest” were. This in turn slowed beneficiary identification and arguably led to selection bias in favor of “better-off clients.” In terms of the development of the financial system, it appears that organizations participating with PKSF, especially the smaller ones, did not graduate to market-based sources of funding. Larger ones did, but as pointed out earlier PKSF does not appear to have been instrumental in this process. 3.3 Apex Banks and Rural and Community Banks Ghana: Association of Rural Banks’ Apex Bank 45. In Ghana the Rural Financial Services Project was designed to build sustainable financial institutions to increase growth and reduce rural poverty. It illustrates attempts by the 25 Rutherford, email communication 2012 26 One percent (repayable over 20 years, including a 5-year grace period) for the Microfinance II project. The foreign exchange risk was also borne by the government. Such concessionary terms and conditions were agreed to because the project objective coincided with the government's poverty alleviation objective. In the second operation, the government expected to on-lend $3 million equivalent from the Credit proceeds to PKSF at an interest rate of 1.25 percent. 27 The ICR team suggested that PKSF could chalk out a long-term plan for mobilizing funds directly from the market at commercial rates. It could also consider securitizing its portfolio to mobilize funds from the market, following the successful pioneering of BRAC’s securitization ini tiatives (by Citibank in 2006). 28 Information regarding the graduation program is available at http://www.poverty- action.org/ultrapoor/about. 19 World Bank to intervene at different levels and through different institutions. One component was intended to support both rural community banks and informal finance providers, through a “linkage” program, (inspired by the Indian Self-Help-Group Bank linkage) whereby informal nongovernment microfinance entities in Ghana (including savings and credit bodies such as susu collectors), were to be linked to the rural and community banks, for greater outreach. 46. The linkage program was not successful. In Ghana, the rural and community banks and informal finance providers turned out to be competing for the same market, and synergies did not develop. Although inspired by the successful Self Help Group program in India, no pilot was conducted to test for replicability before scaling up. Although training was provided to select rural community banks on the use of microfinance techniques to help them increase their poverty outreach, the program was poorly implemented and training funds were exhausted. The Ministry of Finance, expressing its commitment, took on the role of implementing agency for the linkage program as well as for creating a consultative forum for microfinance policy development to reduce subsidies and promote sustainable microfinance. However, the “policy dialogue” role of the microfinance forum was limited by its presence in the Ministry of Finance.29 47. The loan also illustrates an attempt by the Bank to establish and support an Apex institution for rural banks, the Association of Rural Banks (ARB) Apex Bank. This component was under the oversight of the Rural Finance Inspection Division of the Department of Supervision at the Central Bank. The component included the preparation and adoption of new banking regulations, policies and institutional arrangements to allow the ARB Apex Bank to provide oversight to the rural community banks. 48. The Apex Bank’s mandate for oversight in practice, however, fell short of proper supervision, mainly because the Bank of Ghana did not formally delegate such inspection and supervision functions to the ARB Apex Bank. And other regulatory and policy reforms were either delayed or not implemented. Nonetheless, even with partial delegation, the Apex Bank began to carry out a number of oversight functions such as those relating to capital adequacy ratios of rural and community banks and maintenance of liquidity reserves. The Apex Bank also played a role in terms of providing clearance services, specie circulation and training. A number of improvements in the rural financial sector were thus brought about by the ARB Apex.30 49. Overall therefore, the project could not fully support the outreach of the rural community banks to the poorer regions or to the informal financial entities in Ghana. The project did play a significant role in the preparation of the regulatory framework for the 29As noted by another donor agency: “This Forum was driven by MoFEP but did not have a strong representative from the microfinance sector (e.g. an apex organization) as co-chair. This is a limitation because a strong counterpart is important for policy dialogue: the Government cannot be expected to engage in policy dialogue with itself” (IFAD 2012). 30Such as reductions in time required for check clearing from two weeks to five days, the approval of advances to salaried workers from 14 days to 3 days, and reductions in over-the-counter withdrawal time on average from 20 minutes to less than 9 minutes. 20 ARB Apex Bank, supporting its establishment, but did not have the leverage to ensure that its objective of full delegated supervision would be realized. The ARB Apex Bank was functionally useful, even if it fell short of providing the level of oversight that may originally have been envisaged. 50. The overall outreach of the rural commercial banks increased during the project period (Table 3.2). The outreach data, however, should be interpreted with the caveat that it is not clear how many of these “new” clients were actually new or how much of this increase can be attributed to the project. Table 3.2. Ghana Rural Financial Services Project - Intermediate Outcome Indicators (2001-11) Intermediate outcome Baseline Target Actual Actual Target Actual Actual Actual indicators 2001 2007 2007 2011 2001-07 2001007 2008-11 2008-11 Values Annual percentage change Number of RCB clients 1269.0 3.900 3261 4570 20.5 17.0 8.8 13.1 (depositors plus borrowers -- Number of depositors 1129.0 na 2671 3766 na 15.4 9.0 12.2 (thousands) Number of borrowers 139.0 na 590 804 27.2 8.0 18.4 (thousands) Deposits in rural banking 38.0 300.0 293.2 910.4 41.1 40.5 32.7 37.3 system - millions of new cedis Loans - millions of new cedis 14.5 200.0 172.1 471.7 336.6 51.0 40.0 4137.0 none none Short Term Investments 24.7 105.2 336.6 27.3 47.3 29.9 specified specified Loan to Deposit ratio for RCBs 38.1 66.7 58.7 51.8 Source: IEG. Note: In the ICR, the target for new loans was set as 200 for 2007 and the loan deposit ratio was set to 50 percent. These numbers are inconsistent, so the PPAR has reset the loan to deposit ratio to 66.7 (that is, 200/300). 21 4. Findings: The Business Environment and the Broader Community 51. Perhaps more than specific partner institutions within a country, the success of the projects reviewed was affected by the policy and regulatory environment of each country and the financial infrastructure available. The projects illustrate the tensions that can arise between expanding access and outreach, on the one hand, and ensuring its sustainable provision at market based terms on the other. They also illustrate frequent and recurring tendencies to support programs for expanded inclusion through some level of subvention. Finally, they demonstrate that success at the level of individual projects is hard to achieve in the absence of supporting elements in the financial environment. Effects of the Policy Environment: Key Findings  In terms of Bank lending instruments, although the instrument of choice is typically an investment loan, policy-based loans may have a larger role to play given their usefulness in addressing all- important issues in the policy environment.  Strong borrower commitment is essential for the success of an adjustment operation involving policy and institutional reforms. Well-managed institutions can reflect financial unsustainability from the point of view of the economy as a whole if the enabling environment reflects distortions in financial markets.  There can be tensions between the achievement of expanded microfinance outreach and the achievement of financial sustainability. It may be tempting in some political environments to expand outreach at the expense of sustainability.  When subsidies are political and entrenched, they are hard to displace and it is not enough for a project team alone to conduct dialogue on this theme; it requires higher level policy dialogue support between senior Bank staff and senior country officials.  Complementary financial infrastructure is also important—clear creditors’ rights, effective laws for the use of collateral, and credit registries.  The Bank can help build lasting reforms by limiting the scope of the projects to what can be accomplished within a certain political and economic environment, even if not fully market oriented. 22 4.1 Market-Based Finance or Government Support? 52. The projects were all impacted by prevailing levels of financial market distortions in their broader environments. In Mexico the financial reform-oriented government recognized the non-sustainability, as well as the potential threat to stability, arising from the lending practices of Banrural. Banrural’s inefficient lending practices were masked by its ability to continue to raise deposits, which were remunerated at unsustainably high interest rates. Its unsound funding practices were combined with poor loan recovery rates on the lending side. Because of its public sector ownership, its accumulated losses implied open-ended subsidization in the form of a buildup of contingent fiscal liabilities for the government. At the same time, the government recognized the need to continue to provide credit to the rural sector and to find a pragmatic solution that set limits on support actually provided by the government, for the provision of such credit. 53. The Financiera Rural solution demonstrated the Bank’s ability to help engineer a solution to a failing public bank that contained subsidies and maintained efficient access; provided that the policy environment was conducive to such a solution. However, it also demonstrated the limits of achieving such a solution within the parameters of a public bank, and the potential fragility of such a solution in the absence of strong political commitment. In 2013, nonperforming loans crept beyond the benchmark “safe” limit of 4 percent, and changes in financial sector policy called for development banks to play a more proactive role in providing (directed) credit at lower interest rates. 54. In Brazil, the CrediAmigo program initially demonstrated the feasibility of setting up a specialized program within a public sector bank to bring good practices of market-based microfinance to the poor. It also demonstrated that subsidies for financial inclusion can in principle be made transparent through the use of a subsidy dependence index. As in the case of Financiera Rural, the program was successful in the expansion of outreach to the financially underserved target population. 55. However, policy changes including earmarked reserves for microfinance lending and interest rate caps introduced in 2003 effectively lowered the cost of funds to public banks for microfinance lending. Further subsidies for microfinance lending were introduced under the Crescer program in 2011.31 CrediAmigo’s planned spin-off as an independent institution, to facilitate its access to market funds and long-term sustainability, was abandoned, and CrediAmigo remained a program within Banco do Nordeste, which continued to benefit from special government programs, with some evidence of crowding out of sound microfinance.32 31 Under the CRESCER program (October 2011), public financial institutions were able to reduce the rates of interest charged for microfinance solidary loans, from 1.2 percent per month (15.4 percent annually) to 0.64 per month (8 percent annually). This boosted CrediAmigo’s strength in the microcredit market. However, it puts the interest rate well below the costs for private banks. At the end of 2013, the average interest rate on BNB’s microfinance lending, at 5 percent per annum, was considerably below prevailing market rates of over 40 percent. 32 See IEG (2014a, Annex C). In an interview, the president of Associação Brasileira de Entidades Operadoras de Microcredito e Microfinanças–the Brazilian Microfinance Association—pointed out that Brazilian private sector microfinance continues to suffer from scarce and expensive funding, despite 23 56. Ghana had had a long history of special credit schemes prior to the World Bank project, usually at subsidized rates with extremely poor recovery rates. Such programs then challenged the ability of the rural commercial banks and rural microfinance institutions to mobilize savings and to collect from borrowers, whose willingness to repay was typically low when loans were known to come from government at subsidized rates (Obeng 2008). Although the project intended to address such issues through policy dialogue, the efforts were not successful. However, government-subsidized funding were channeled through a specialized agency, the Microfinance and Small Loans Center (MASLOC), directly under the authority of the Presidency. MASLOC was established in 2006, one year prior to project closing. 57. Mozambique illustrates an example of a line of credit which was initially intended to be onlent at market terms, but faced severe difficulties and limited offtake, until the point where some element of subsidy via a government development bank promoted its offtake. While nine banks initially signed up as Participating Financial Institutions, more than 85 percent of loan disbursements were eventually undertaken by just two financial institutions. The primary one among these, GAPI, accounted for more than half of total loan amounts disbursed. GAPI received funds at relatively favorable terms from the Central Bank, negotiating a preferential rate equal to 50 percent of the t-bill rate. The result was an implicit interest rate subsidy to GAPI of between 2.5 and 4.75 percent, depending on the year, which GAPI said it partially passed forward to its subborrowers. Broader difficulties in the environment that impeded offtake implied that a subsidy element crept back into the implementation of a loan initially designed to be self-sustaining. 4.2 Other Regulatory Factors 58. In Ghana, the Bank project was successful in supporting certain regulatory reforms that were important for the increased success of the project, even if not successful in addressing the subsidy issue. These included reform of regulations pertaining to the size of reserve requirements of the rural and community banks, which were key to enabling the rural and community banks to increase the amount of their loans; and a law for the Association of Rural Banks’ Apex Bank that allowed the Apex Bank to conduct oversight of the rural commercial banks on behalf of the Bank of Ghana. The latter was completed at the time of the project closing in 2007. Implementation was only partly successful and delegation was partial. The project also illustrates the slow pace of regulatory change. The amended regulations pertaining to non-bank financial intermediaries became effective only in 2008 and the new Bank of Ghana regulations for microfinance institutions were adopted in 2011 (both after project closure). 59. In Mozambique, by contrast, the Bank did not take into account the series of constraints in the broader environment in the provision of medium-term credit to SMEs even though lessons from several earlier projects had brought home these messages. It was not the supply of credit alone that was constrained. Demand for credit was also limited by the high level of real interest rates (over 30 percent). Borrowing was costly to enterprises, relative to sound operational performance data. And statistics prepared by the Brazilian Micro and Small Business Support Service note the decline in private microfinance credit programs. 24 internally generated funds. High interest rates partly reflected a lack of competition in highly concentrated financial markets, but also low credit bureau coverage, which implied that lenders found it difficult to gauge the potential risks of different borrowers, and maintained high rates as a precaution. Other limitations in the environment that made banks reluctant to lend were poor contract enforcement and a limited framework for secured transactions (i.e., for collateralized lending). Moreover, in a highly liquid economy where banks had the opportunity to hold government securities (a safe asset), there was limited appetite to hold risky assets in the form of loans. 60. The second microfinance loan in Bangladesh reflected, and attempted to include in the project design, the broader enabling environment issues facing the sector. In particular, these related to the lack of regulation and supervision of NGO microfinance institutions that had accumulated large amounts of deposit. There was also a lack of credit information system and increasing evidence of growing multiple lending and limited industry performance standards. Setting up the Microfinance Regulatory Authority in 2006, could be regarded as a possible achievement, although the Authority was able to accomplish little beyond registering microfinance institutions—earning it the epithet “Microfinance Registration Authority.”33 61. The limited ability of the project to accomplish changes in the policy environment partly reflected the unique dynamics of Bangladesh microfinance, driven by large non- government microfinance institutions that had no incentive to be brought under the regulatory framework. With the departure of the participating microfinance institutions (BRAC, ASA) from the Bank-financed PKSF funding source, its ability to implement some of the regulatory and other reforms envisaged in the project such as industry performance standards or an industry wide credit information system diminished.34 Concerns regarding risks associated with unregulated MFIs with large deposits have grown (IMF, 2010a). 4.3 Government Commitment to Change 62. Whatever the institutional arrangements, a key factor underlying successful implementation of financial inclusion objectives was the government’s commitment to bring about the required changes, and the adoption of policies conducive to the realization of such objectives. 63. In the Mexico operation the government showed a high degree of stakeholder commitment and ownership to the project and met all implementation-related conditions in a 33 Rutherford, 2012. 34According to Rutherford, “Because they feared intrusive bureaucracy, ASA (at least, probably many others) preferred until recently to leave things as they were—that is, to do without a license to mobilize deposits in return for keeping clear of government interference. BRAC remained curiously uninterested in savings until recently. All that did eventually change in recent years when ASA and others saw the advantages that Grameen was reaping (in its GBII version) from its savings license, but that all post-dates Microfinance II. Had things been otherwise —had ASA and BRAC pushed hard early on for the right to take savings—it would probably have happened, irrespective of what the World Bank/PKSF were doing” (communication, 2012). 25 timely fashion. The government was committed to limit financial losses in the weak development banks and in the case of Banrural, willing to allow its liquidation and replacement with a market driven institution development agency model. The government also opted for a gradual move over time to a Tier II lending institution and to not disrupting the flow of finds to agriculture which was a mainstay of the Mexican economy. 64. In Brazil, government policies toward the implementation of microfinance objectives changed considerably over the life of the project. These impacted the level of commitment to project design. Initially intended to support the expansion of market-based microfinance, successive changes in policy some years after the project was initiated went in the direction of increasing directed credit for microfinance support. As a result, although CrediAmigo increased access to finance for a large number of beneficiaries, the project did not scale up market-driven microfinance. And CrediAmigo did not spin off from its parent public sector bank to become an independent market-driven entity as initially intended. 65. In Ghana, the government was committed in principle to support the project’s objectives, but implementation in practice wavered. There was a 21-month delay between approval and effectiveness, which impacted project performance and donor coordination. And the microfinance draft guidelines prepared in 2006 were not issued till 2011—and their revision began shortly thereafter. Moreover, the government continued to support subsidies through MASLOC, which impacted the development of sustainable rural finance. 66. In Mozambique, the government was highly engaged in the design of the project and committed to its success. The private sector development unit in the Ministry of Commerce was committed to reforming the policy environment throughout the project and took a lead role in advocating for a better enabling environment for the private sector. But its restrictions on eligibility to the line of credit for small firms (to those that were owned at least 51 percent by Mozambican nationals) were shortsighted and conflicted with the project’s objectives. And even when they agreed to change this position at the midterm review, the authorities vacillated and delayed the signature of the modified Credit Agreement for 10 more months, so the change was not effective until the last year of the project, with an adverse impact on implementation. The government’s delays in disbursing counterpart funds also compromised the efficiency of a key component of the project. 67. In Bangladesh, the government was committed to the development goals of the projects reviewed—using microcredit as an instrument to enhance incomes and livelihoods of the poor. It also supported the project from funds through the national budget and absorbed the foreign exchange risk. However, in the case of the Learning and Innovation Loan, the government initiated its own “scaling up” project, before the completion of the project, which reduced the potential learning and lessons that could have been derived from the experimental approach of the project. This was a missed opportunity. 26 5. Findings: Bank Loan Design Other Elements of Successful Bank Loan Design: Key Findings  Project development objectives that are too general are hard to effectively monitor and compromise the evaluation of results.  Sound analytical underpinnings are a large element in good project design.  Flexibility in supervision to adapt to changing scenarios is essential for success.  It is difficult to transpose sometimes idiosyncratic institutional arrangements for microfinance services from one country to another and pilot and stakeholder consultations to ensure their suitability for the local context would be very helpful.  The Bank has found it difficult to disentangle rural and regional objectives and to adapt the choice of interventions and the loan design to match the specific objective. 5.1 Adjustment versus Investment Lending? 68. An early question that arises in the design of Bank operations is whether the achievement of project objectives is better suited to adjustment or investment lending. Whereas economic policy intervention is usually conducted through the former, the bulk of Bank lending (around 60 percent) takes the form of investment projects. Lending that aims to support access to finance has taken the form of both adjustment and investment, as reflected in the different borrower institutional arrangements discussed in Section 3. One of the projects reviewed here (Mexico) took the form of a structural adjustment loan, while the rest were investment loans. Two of the PPARs reviewed (Brazil and Mozambique), as well as the two projects in Bangladesh, used lines of credit. One of the Bangladesh operations took the form of a ‘learning and innovation’ loan, intended to test alternative models of credit delivery to the poorest. One project, in Mozambique, offered a matching grant facility. 69. This review finds, broadly, that no single lending instrument is uniquely suited to promoting financial inclusion; each loan type has additional design features that outweigh this factor. However, the review also finds that policy-based lending, relatively little used in this cluster, as well as in the overall set of financial inclusion projects undertaken by the Bank (Appendix 1) can be very effective in certain circumstances, and that “getting the policy environment right” is vital even for the successful outcome of a line of credit. Insufficient attention to the policy framework was a core factor that impeded the successful functioning of the lines of credit in the operations included in this review. The multi-tranche policy-based adjustment loan in Mexico was well suited to bringing about the structural changes that the government sought to achieve and to foster phased institutional strengthening. 27 70. In contrast, as discussed in the previous section, the lines of credit encountered difficulties as a number of other structural elements of the environment needed to be in place for their success. However, investment loans that include capacity building components are more attractive to borrowers and may indeed be essential for first time borrowers. 71. Apart from specific loan structures and features that helped to support the goals of financial inclusion, there were several “good practice” design features of lending operations that emerge from this review that helped achieve overall project objectives. The following section reviews some of these features of loan design that consistently emerged as important. 5.2 Other Elements of Successful Loan Design Setting Realistic Objectives and Designing Measures to Achieve Them 72. Overambitious project objectives are difficult to achieve. As the Mexican project example shows, a focused and realistic objective in keeping with the size and scope of the project is important for a successful operation. By contrast, in Ghana, the universal and broadly stated goals of the project—poverty reduction and growth—were overambitious for a small project, focused essentially on capacity building in rural financial institutions. Also in Ghana, although the professed objective of the project was to support the rural poor, there was no strategy, within the design of the loan, to increase outreach in the lagging northern regions/poorer sections or for agriculture, which was the primary employment-generating activity in rural Ghana. 73. Though there were admitted shortcomings with respect to the design of the line of credit in Mozambique, one good feature was that for first time borrowers in Mozambique also had a matching grant component to incentivize timely repayments. And to support the development of an active business and technical training market, a matching grant of 50 percent was made available to qualified trainers to provide business and technical training services to local firms. The project’s outsourcing of the management of the matching grant to a firm with solid expertise in the field also contributed to its relative success. 74. Simplicity of design was also a virtue as the Mexico project’s clarity of focus showed. By contrast in Mozambique the institutional arrangements for technical assistance were complex, and the capacity-building component included too many partners, straining already weak capacity in the newly created coordination unit. Similarly in Bangladesh MFII had too many components—regulatory reform, setting up a credit bureau, microenterprise loan for the missing middle, loan to hardcore poor, and rural and urban microfinance. Many of the components had to be dropped. Mitigating Risks 75. The projects also demonstrate the importance of recognizing potential risks. The successful project in Mexico explicitly took account of a series of risks that could have derailed the operation. First, there was the risk that political pressure could lead to too rapid an expansion of credit by Financiera Rural. To limit this risk, the Board of the Financiera Rural limited credit expansion in the first year of operation to no more than 10 percent over the credit given by its predecessor, Banrural. Second, there was the risk that with the closure 28 of Banrural, credit to lower-income rural producers would fall. The project determined that Financiera Rural would lend primarily to low and middle-income rural producers. Higher- income clients (who were among the worst repayers of Banrural) would have to borrow from commercial banks or other financial intermediaries. The average size of Financiera Rural’s own operations has remained roughly constant in real terms. 76. Measures were also taken to ensure that selected financial intermediaries had adequate experience and good governance through eligibility criteria and technical assistance. Finally and perhaps most significantly, the project explicitly recognized the risk that government policies of providing interest rate subsidies could limit Financiera Rural’s ability to charge rates that guaranteed its financial viability and could deter the entry of private intermediaries. To mitigate this risk, the authorities committed to interest rates congruent with the objective of maintaining the capital endowment of Financiera Rural, and moreover, adjusted the interest rates of other government programs in the rural sector to avoid interest rate arbitrage. These measures were intended to avoid the creation of market distortions and allow entry of competitive private lenders. Nevertheless, years after project closure, this risk did materialize, as Financiera Rural was been forced to compete with cheaper sources of funds from the federal and state governments. Financial sector policies announced in July 2013 call for development banks to play a more proactive role in providing (directed) credit at lower interest rates. This illustrates, once again, the frequent tensions that can arise between expanded financial inclusion and financial sustainability. 77. In Brazil, project design recognized the “moderate” risk of resistance to the proposed spin-off of CrediAmigo from Banco do Nordeste into a legally and financially independent financial entity. However, the Bank’s strategy to address this risk was not effective, as the spin-off was not implemented. Although this may have been partly due to the changed policy environment, the spin-off was not fully clarified at the design stage as it was dependent on the outcome of the review of the regulatory framework by the central bank and the approval of the Banco do Nordeste board. The design also failed to clearly lay out whether CrediAmigo would be a “partial or completely independent entity.” Thus, although the changed political environment after 2003 may have stalled the proposed spin-off, the ambiguities at the design stage were equally reflected in the cancellation of the spin-off. In this regard, the project team may have been too optimistic, especially given the strongly entrenched roles of the major public sector banks in Brazil. 29 Incorporating Lessons from Past Lending and Analytical Work 78. In Mexico, the level of preparation and analytical underpinnings of the project were based on wide-ranging stakeholder consultations. Project design benefited greatly from proactive dialogue and exchanges of views with the Ministry of Finance. The contribution of the 2001 FSAP had a decisive impact on the course taken for development bank reform in Mexico and the selection of the business model chosen for Financiera Rural in particular. Moreover, the Bank participated in early discussions with the government on what to do with Banrural. The Bank prepared a financial model that simulated the situation of Banrural that showed that no "good bank" rescue could be carved out and that it was better to close it. Work done by the Bank over the years in support of Mexico's financial reform and modernization was of great help. This financial sector work continued, for example, with the 2006 FSAP mission and the First Programmatic Finance and Growth Development Development Policy Loan approved at the same time. 79. In Brazil, the project team had invested in loan preparation and worked closely with CGAP and other specialists in microfinance (ACCION International); they based the project on lessons learned from the pilot phase at Banco do Nordeste. The team had also undertaken m work in terms of mapping out the microfinance landscape in Brazil for the first time, benchmarking Brazil with regional comparators, and setting parameters for sustainability. The project team also correctly assessed the implementing agency’s commitment and capacity. These factors contributed to the relevance of project design at the start of the project. 80. In contrast, as discussed, in Mozambique, lessons from earlier, unsuccessful credit line projects were not incorporated, resulting in poor uptake and restructuring. Providing Technical Support and Capacity Building with Finance 81. Technical assistance to borrowers and financial intermediaries can be crucial to increase the demand for credit funds. In Ghana, the “institutional support to the Ministry of Finance and Bank of Ghana” component, provided training to the Banking Supervision Department staff to improve their ability to oversee the microfinance sector. There is some evidence of improved supervision of the Apex Bank and associated rural and community banks.35 The training provided to the staff enhanced their skills in supervision and examination, the production of more timely reports, and the adoption of remedial measures to address distressed rural and community banks.36 This remained a positive achievement of the project. 35 Capital adequacy improved, together with rates of supervision, eventually achieving a 100 percent inspection rate. 36 During the March 2012 PPAR mission, the Bank of Ghana reported that the good performance with regard to their supervision of rural and community banks has been maintained since the closure of the project thanks in part to the electronic reporting and analysis of primary rural and community bank returns by the ARB Apex Bank. This has even permitted the Bank of Ghana to envisage the introduction of selective on-site inspection 30 82. In Mozambique, the credit line picked up only after the inclusion of GAPI, a development finance institution that provided both training and loans to its enterprise clients, relative to other participating financial institutions. This feature was subsequently embedded in a follow-up project in Mozambique that IFC undertook, which not only gave loans but also provided support through its advisory service capacity building of both financial intermediaries and borrowers.37 It would appear that IFC’s approach of working directly with an interested financial intermediary and combining financial support with technical assistance to the bank concerned was successful. 83. In Bangladesh, capacity building of the poor borrowers was a planned output under an LIL (FSP). However as previously pointed out the launch of government’s ultra poor program effectively replaced the FSP loan. The remaining funds under the loan’s revolving funds were transferred to the capacity building component under the ultra-poor program. The ICR lists many outputs under this component; however the outcome—that is, numbers of the poorest who were able to get employment or were able to set up an income generating activity following the trainings—was not tracked and remains unknown. 37IFC’s subordinated loan to the Banco Comercial e de Investimento was for $8.5 million for SMEs, which was successful in fully disbursing. The loan was also accompanied by a technical assistance loan from IFC for credit approval processes and marketing for SMEs, improved risk management and internal controls, and branch compliance. 31 6. Findings: Outputs and Outcomes 6.1 Outcomes - Outreach, Impact and Sustainability Outputs, Outcomes and Impact on Inclusion: Key Findings  There is mixed evidence regarding the extent to which these projects were able to achieve intended outcomes, in terms of expanded financial inclusion among intended beneficiaries  In some cases, there is evidence of outreach to target segments, but it is less clear that the objectives have been achieved in a sustainable way.  In some other cases, while target outcomes of poverty reduction appear to have improved, difficulties of attribution are significant, given the relatively small role these projects played.  Defining beneficiary selection criteria is important to avoid selection bias.  These Bank projects have focused mainly on credit and do not reflect the fact that the poor- like the rich- need access to a broad array of financial products and services.  In addition to a focus on a wider array of products, a broader use of different instruments could also have been pursued, including new technologies and institutional models such as mobile or correspondent banking.  Potential beneficiaries could also have been served by other means, including government transfers and social protection. Project Beneficiaries, Impact on Expanded Inclusion, and Poverty Reduction 84. As the preceding discussion has illustrated, there is mixed evidence, overall, as to the extent to which these projects achieved intended outcomes, in terms of expanded financial inclusion among intended beneficiaries (whether underserved segments of the population of individuals or enterprises, or poorer regions of the country), or their final impact, in terms of poverty reduction. In some cases, there is evidence of outreach to target segments, but it is less clear that the objectives have been achieved in a sustainable way, from the point of view of the financial system as a whole; this illustrates the dilemma that can arise in trying to achieve expanded financial inclusion in market-based financial systems. In some other cases, although target outcomes of poverty reduction appear to have improved, difficulties of attribution to these specific projects are significant, given the relatively small role they played. 85. In Mexico, though not the stated objective of the Mexico project, the smooth and well-executed transition of Banrural to Financiera Rural ensured that the flow of funds for rural finance was not disrupted, and lending to the rural population increased in volume and 32 efficiency. Financiera Rural intermediated significantly larger credit volumes than Banrural and at a lower cost. In 2001-02, Banrural provided an average of $800 million in loans with a staff of 3,459, whereas Financiera Rural provided $1.19 billion in loans in 2005 with a much smaller staff of 1,061. By 2012 its loans had increased to $1.4 billion equivalent, while the staff had increased to 1,200. Thus, Financiera Rural multiplied the ratio of loans per person by 4.8 from 2001-02 to 2005. The ratio went down to 4.1 in 2011 as a result of the increased support by the promoters to upgrade the abilities of the potential borrowers.38 The improvement is much higher (on the order of eight times) if the comparison is made in terms of collected loans (net of nonperforming loans). 86. Table 6.1 traces financial performance indicators for the project in Mexico that confirm the strength of Financiera Rural as a financial agency. In particular, they show that efficiency has been improving over time as lending by employee has increased, turning Financiera Rural into a profitable financial intermediary. Table 6.1. Financiera Rural Selected Indicators Banrural 2003 2004 2005 2011 2001-02 I. SOLVENCY INDICATORS Loan-arrears 2.10% 2.90% 2.50% 4.7% Loan-loss provisions coverage 850.40% 298.70% 269.10% 129.0% Capital at risk (arrears/equity) 0.30% 1.20% 1.40% 2.75% II. PROFITABILITY Return on equity -0.30% -1.10% 2.20% 3.5% Return on assets -0.20% -1.10% 2.10% 3.5% III. EFFICIENCY Adm. Expenses / Loan Portfolio 11.80% 9.30% 6.80% 6.92% Operating Costs / Average Loan Portfolio 8.18% 8.96% Loan Portfolio / Number of Employees 231,281 248,000 642,300 960,500 948,276 Source: ICR and Financiera Rural. 38 Inits role of development agency, Financiera Rural carries out promotional activities funded with budgetary allocations from the federal government. Delivery is outsourced to experts in different fields and clearly accounted for in the financial statements. 33 87. In the case of Brazil, two ex post evaluations of the CrediAmigo program are available, one conducted by the Bank and another by the Fundaçao Getulio Vargas, a prestigious Brazilian think tank. 39, 40 These studies both show that access to credit increased rapidly in the target region of the Northeast, faster than in the rest of the country, and poverty also declined in this region. Regarding the impact on poverty, the think tank study found that people who borrowed from CrediAmigo for five years were between 36 and 41 percent more likely to leave poverty than those who had just started to borrow from the institution. Although less than a robust impact evaluation, the results suggest that CrediAmigo helped fuel the certainly expanded access to credit in the Northeast, which exceeded planned targets. An increase of 25 percent was recorded at the end of 2013 compared to 2012.41 The IEG mission visited some of CrediAmigo customers in favelas (poor neighborhoods) and saw how the projects changed their lives by financing income earning and livelihood-expanding activities. However, these people represented only a very small percentage of the poorest of the poor. 88. However, although access to formal finance through the program increased substantially, it is less evident that the expansion was sustainable, from the point of view of market-based finance. After the initial years, CrediAmigo’s parent institution, Banco do Nordeste de Brasil, benefitted from the government’s special programs for microfinance support through public banks, that included a series of measures including, on the funding side, earmarked deposits for microfinance lending, and on the lending side, interest rates caps below market rates. There is some evidence that such government support may have been at the expense of crowding out private participation. 89. In Ghana, overall, there is evidence of increased financial outreach by rural and community banks, in terms of growth in the number of depositors, the value of deposits, loans, and advances, including microfinance and the extent of non-credit services. However, this increase cannot all be attributed to the project, as falling interest rates on bank-held Treasury bills during this period obliged the banks to increase lending.42 Rural and community banks also remained institutionally weak and had large numbers of nonperforming loans, with implications for the sustainability of expanded outreach.43 39 See Skoufias, Leite, and Haimovich (2009). Although the study was careful to point out the difficulties of assessing the impact of increased access to credit (in general, or through CrediAmigo), it sought to explain the impacts of increased supply of credit on credit utilization and thus on firm performance. 40 Neri and others (2008), and Neri et.al.(2010) . 41 Agencia Brasil (2013). 42 Nair and Fissah (2010). 43 One study of a sample of rural and community banks suggests that nonperforming loans were indeed an issue at in rural and community bank performance. The average portfolio at risk > 30 days was 16 percent and at risk > 365 days was 3.5 percent (Nair and Fissah 2010). A separate sample analysis of 24 rural and community banks by GHAMFIN (2005) also shows that rural and community banks have higher ratios of nonperforming loans than do savings and loans and financial NGOs in Ghana. An audit of the MCC Agricultural Credit Program also reported that rural and community banks had not been repaying their loans to the Bank of Ghana on time, because their sub-borrowers were not paying them in a timely manner, and also some rural and community banks were making 34 Meanwhile, subsidized and directed lending by government continued through MASLOC. A 2013 study on subsidies found that though the dependency is on the decline, the microfinance sector will continue to be subsidy dependent in the long term.44 90. In Mozambique, indirect poverty reduction effects were expected to result from employment creation in assisted enterprises. There is some evidence to show that the government development finance intermediary that eventually undertook most of the disbursement of the Bank’s line of credit, GAPI, worked with various parts of the value chain to improve small borrower’s production quality and bankability. This potentially had income, employment, and a poverty reduction impact, although definitive poverty reduction assessments are not available.45 91. The Bangladesh LIL (FSP) was designed to test specific models to achieve outreach. However, the achievements of the project were not fully tested, as the government started its own Ultra-Poor Program. The design of the project in fact did not define who the poorest were or how beneficiaries were selected. No baseline survey was conducted. Although funds for the project were eventually utilized for capacity building of the poorest (as the government’s Ultra-Poor Program did not have a capacity development component), there was no evaluation of the impact of these capacity development programs. 92. The ICR of the Bangladesh second microfinance operation (MFII) attempted to collect some evidence of the welfare impact of microcredit programs compared to nonparticipant groups with similar characteristics. Of the 446 families that were sampled food expenditures increased from Taka 60 in 1998 to Taka 202 in 2009 in current terms—in real terms, an increase of 68 percent. For microcredit loans, the average return on 56 microcredit projects was 39.5 percent. Although some of the data are mixed, 93 percent of participants in the microcredit program reported no food shortages, and only 78 percent of nonparticipants reported no food shortages. In a separate study concerning rickshaw pullers participating in the program, income increased by 13 percent relative to counterparts not participating. 93. These samples are tiny relative to the millions of people affected by the project, but the results coincide with a number of observations over time from other sources, on the impact of microcredit, that seem to corroborate this finding. Using long panel survey data spanning 20 years Khandekar and Samad (2014), for example, studied the impact of microcredit on household welfare. The study found that microcredit programs continued to benefit the poor by raising household welfare. Moreover the beneficial effects remained higher for female borrowers. The Bangladesh Household Income and Expenditure Survey more loans that they should have, given their capital. The Bank of Ghana had not established appropriate prudential limits in this regard. 44Aveh, Krah, and Dadzie (2013). 45See Simonetti Roberto, Wuyts Marc and Wuyts-Fivawo Anna (2007). Overall, though, the poverty headcount in Mozambique has stagnated at 55 percent since 2003, a matter of concern for Mozambique and its development partners. 35 2010 data show that in addition to poverty reduction, microfinance outreach significantly increased in rural and urban Bangladesh.46 94. At a more macro level, the Financial System Stability Assessment (IMF, 2010) report also showed that MFI loans accounted for 3 percent of GDP.47 And a 2011 microfinance review found that microfinance accounted for 7.33 percent of the country’s entire financial system.48 The phenomenal growth of microfinance institutions in Bangladesh (representing 30 million members with over $2 billion of annual disbursement) was also noted in Khandekar and Samad (2014). The project (MFII) likely contributed to it. However, given that the major MFIs who dominated the sector in terms of outreach left the MFII program and largely grew with their own retained earnings and deposits, the impact of MFII on the observed growth was likely limited. Planned outreach to newer segments of the population was not reached under the project as these components were dropped. 95. Multiple lending (in terms of the average number of accounts per borrower) is also an issue in Bangladesh. Leading MFIs in Bangladesh have adjusted their outreach numbers downward to account for such multiple lending.49 The claim of the project regarding new borrowers reached needs further examination. 96. It may be noted that Bangladesh's overall poverty reduction program was, however, highly successful, and the incidence of poverty declined from 48.9 percent in 2000 to 40 percent in 2005; the incidence of extreme poverty also declined, from 34 percent of the population in 2000 to 25 percent of the population in 2005. Further declines were noted in 2010.50 However, attribution—as in the other projects—to the project is a challenge. Moreover, the Bangladesh economy was growing rapidly, so the big reduction in the number of people living below the poverty line could also be attributed to this rapid economic growth and a host of other factors. Regional Considerations 97. In Mexico, the strategic focus of Financiera Rural successfully helped increase access in rural areas, broadly defined. Within this broad mandate, the projects had variable success in terms of their ability to target and then plan for region specific groups where the need for higher inclusion, and poverty reduction, were identified. More precisely, targeted region- specific considerations were factored into the Brazilian Northeast Project, where the goal was 46 World Bank (2011). 47 IMF (2010 a) 48 Sinha (2011). 49 During 2009, a number of leading MFIs attempted to rationalize their borrower accounts, which reduced the number of accounts from 29.8 million in 2008 to just 27.0 million in 2009. Experts, however, suggest that this may still be an overestimate of number of clients by 15 percent (Sinha 2011). They point out that Bangladesh is a saturated market and new clients may not have been new. There is a possibility that the same clients were given multiple loans and shown as separate clients in outreach numbers. 50 According to the Bank’s Preliminary Report on Household Income and Expenditure Survey (2011). Based on the upper poverty line, the incidence of poverty is estimated at 31.5 percent at the national level, 35.2 percent in rural areas, and 21.3 percent in urban areas. In 2005, these rates were 40.0 percent, 43.8 percent and 28.4 percent, respectively. 36 to increase financial outreach in the less served northeastern region. Following the project, the outreach in the northeastern region increased, albeit, encouraged by government policies that supported the development of a microfinance program at a public bank, possibly discouraging the development of private microfinance. 98. By contrast, the Rural Finance Project in Ghana did not include any strategy to increase outreach in the poorer and underserved northern region, even though rural poverty reduction was an objective of the project. In Ghana, regional considerations were not a central consideration for the project, when designed. Rather the emphasis was on the systemic improvement of rural microfinance institutions. 99. It was not possible to examine to what extent there have been higher rates of growth of depositors, deposits, and loans in the north over the past decade, in the absence of baseline data. The project did not have a strategy to increase outreach in the poorer northern rural regions or the less served agriculture sector. Evidence shows that the outreach for both remained low, and rural and particularly northern poverty remains a challenge.51 In Mozambique, even though there were no specific regional considerations with respect to the line of credit, nonetheless GAPI—the largest lender of the participating PFIs that was later included among the PFIs—made 40 percent of its loans to SMEs outside of Maputo/Maputo province. Those accounted for about $1 million, that is, 43 percent of the portfolio financed by the IDA line of credit. Gender Considerations 100. There were no gender considerations specified in any of the projects, and neither Brazil nor Mexico had any gender differentiated considerations or noted results. However, in Ghana, after the midterm review, a training program to increase financial literacy for women, with targets for financial inclusion, were added and achieved. It was expected that the project would have a significant impact on the number of rural poor women obtaining access to credit. Estimated results in this area seem to have been quite promising. By the end of the project in 2007 the target for outreach to women was nearly met (44,700 versus a target of 60,000) and exceeded by the 2010 (70,000). In Mozambique, though gender was not a consideration, 10 of 52 borrowers from the credit line were women-owned and -operated businesses, representing 20 percent of total borrowers. In Bangladesh, though the project did not include gender considerations, the majority (90 percent) of the loans were given to women. 6.2 Outcomes—Financial Institution Stability and Sustainability 101. Financiera Rural in Mexico evolved into a solid financial institution, with accounts that transparently show the true costs of its operations. As a financial agency it maintained robust financial performance. Its efficiency and profitability improved over time. Returns on 51Poverty,though lower than a decade ago, continued to be broader and deeper in rural areas than urban areas: the headcount poverty rate among rural households was 50 percent against 19 percent for urban households (with the national average being 40 percent). In the Northern, Upper East, and Upper West regions, poverty rates were 52 percent, 70 percent and 88 percent respectively (based on Ghana Statistical Service, April 2007, Patterns and Trends of Poverty in Ghana 1991-2006.) 37 capital as well as lending by employee increased, and it became a profitable financial intermediary, as well as a more effective channel for supplying financial resources to rural producers. Financiera Rural still has significant space to grow, as a large portion of its equity endowment is invested in negotiable fixed-income securities at lower interest rates than its loan portfolio. The new institutional arrangement introduced by the federal government to support rural financing under the project had a substantial positive impact on the use of fiscal resources, arresting the “hemorrhaging” of contingent and realized fiscal liabilities, successfully contained with the closure of Banrural, and reversing the moral hazard caused by Banrural’s poor lending. 102. Financiera Rural also accumulated substantial capital in terms of knowledge of the rural sector and a philosophy of development. The operation also had an important institutional development impact through the promotional and advisory activities carried out by Financiera Rural to help small rural financial entities such as credit unions and financial cooperatives to become rural financial intermediaries. It is noteworthy that during the 2008- 10 recession the institution remained robust. Nevertheless, there is little room for complacency. In 2013, nonperforming loans (4.7 percent in June 2013) crept beyond the benchmark “safe” limit of 4 percent. Moreover, the financial sector policy announced by the government in July 2013 calls for development banks to play a more proactive role in providing (directed) credit at lower interest rates, a move that could potentially limit their market-based lending. 103. In Brazil, CrediAmigo expanded access to finance for microenterprises and solidified its own operational and financial management. Other players, notably a handful of public banks, later also entered the arena; access to formal microfinance in Brazil has grown. However, as noted, policy changes, including earmarked reserves for microfinance lending and implied government subsidies, interest rate caps introduced in 2003, and further changes under the Crescer program in 2011, have made it difficult to achieve the provision of sustainable, market-based microfinance. It appears likely that there has been some crowding out of sound microfinance as a result of the new programs. CrediAmigo’s planned spin-off as an independent institution, to facilitate its access to market funds and long-term sustainability, was abandoned, and CrediAmigo remained a program within Banco do Nordeste, which in turn continues to benefit from special government programs. At the end of 2013, the average interest rate on BNB’s microfinance lending—5 percent per year—was considerably below prevailing market rates of over 40 percent. 104. In Ghana, rural and community banks saw a 40 percent growth of total assets in 2012, with 700 branches and a large customer base. However, together with strong asset growth, the nonperforming loan ratio for rural and community banks reached 20 percent in 2012, pointing to persistent weaknesses in the credit process.52 The ARB Apex Bank 52Of 136 rural and community banks, 5 had not met the required minimum capital requirements by 2012, and 16 had a capital adequacy ratio below the required minimum of 10 percent. Concerned that the nonperforming loans would erode the already small capital base of the rural and community banks, the Bank of Ghana proposed to double the banks’ minimum capital (from Cedis 150,000 to Cedis 300,000). However, the Bank of Ghana’s previous (2008) increase in the minimum capital (from Cedis 50,000 to Cedis 150,000) was yet to be implemented in some rural banks. 38 continues to provide important services to its member banks (liquidity management, mobilization of funds, and training); however, its supervision functions remain limited and incomplete, and enforcement of its recommendations has been slow.53 Thus, two rural and community banks that the ARB Apex had recommended for closure to the Bank of Ghana in 2010 were not resolved in 2012. 105. With effective interest rates that could go above 100 percent per annum, micro- lending activities also raise concerns of over-indebtedness, especially as rural and community banks do not report to credit bureaus (IMF 2013). Although the Ghana Micro Finance Institutions Network created a “Reference center,” costs of access to information are perceived as too high by microfinance institutions (IMF 2013). Overall, the Bank of Ghana remains concerned with the quality of the portfolios of rural and community banks which show poor risk management practices and information asymmetry between lenders and borrowers, which results in high interest charges and precipitates delinquency. 106. Overall, there remain deficiencies in the soundness of rural and community banks, despite ongoing efforts to address them. Additionally, subsidized and directed lending by government continues through MASLOC. As discussed, subsidy dependence in the rural finance sector will likely continue in the long term. 107. In Mozambique, the landscape with respect to credit for small and medium enterprises has changed, but subsequent to the Bank operation reviewed. IFC extended two loans to a commercial bank (BCI) with the objective of expanding access to medium-term credit for SMEs, which successfully extended financing to the SME segment, including women entrepreneurs. IFC has proposed to replicate this approach with other banks, including supporting the establishment of new greenfield banks to spur competition in the sector, and potentially to add local currency lending. IDA continued to provide training support to firms through subsequent private sector development projects (Competitiveness and Private Sector Development Project/Projeto de Apoio a Competitivade e do Sector Privado) approved in 2009, and a Technical and Vocational Education Training project. Policy and institutional reforms that the Enterprise Development Project (PoDE) helped realize are likely to remain in place. The government appears committed to improving the business environment and the Investment Promotion Center reported that government funding has remained secure since PoDE’s closure. On the institutional front, the Investment Promotion Center continues to function, and the industrial zone in Belualane continues to grow. Mozambique managed the recent economic and financial crises well, and according to the most recent Financial Sector Stability Assessment, the financial sector appears to be in much better health than 10 years ago.54 108. In Bangladesh, PKSF’s role in the sector has been limited, as some of the top MFIs left its fold during the project. PKSF continues to rely on government funds and has not yet developed any plans to seek market funds, which may impact its sustainability. The big 53 The ARB Apex is tasked with some supervision tasks, such as collecting data from all rural and community banks on a quarterly basis and providing recommendations to the Bank of Ghana (which is the supervisory authority that conducts onsite inspections of Rural and Community Banks). 54IMF (2010b). 39 participating organizations are not looking for funds from PKSF and have in fact grown on their own retained earnings or deposits (in the case of Grameen Bank, which alone has the authorization to collect deposits). Some of the medium-sized MFIs have likely benefited from improvements in financial as well as managerial capacity as participants in the Microfinance II project. However, most remain dependent on subsidized sources of funding. Other potential risks are overlapping loans and resulting risk of overindebtedness among clients. There have been attempts among the MFIs to rationalize by staking out areas of individual territory, closing or merging branches, and reducing client numbers. Experiments with loan rescheduling are ongoing—Grameen is the leader in this. Big MFIs are trying, with some success, to increase the share of their loan portfolios held by larger-scale enterprises, lending to individuals, at the expense of the traditional group-based smaller general-purpose loan (Rutherford 2011).55 Unintended Consequences for Stability—Multiple Lending and Overindebtedness 109. At least in two projects—Ghana and Bangladesh—there appears to be a potential for lending by different microfinance institutions to the same borrower; hence, there is potential over-indebtedness. An empirical study of borrowers in Accra finds that about 30 percent of the borrowers interviewed fulfilled the criteria for over-indebtedness although not at crisis proportions (Schicks 2011). However, other studies cite growing over-indebtedness.56 The authorities’ push for mandatory credit bureau participation for rural and community banks was intended in part to address the issue of potential multiple lending. 110. Likewise in Bangladesh there are varying estimates of multiple lending ranging from 20 percent to as high as 50 percent, yet there is little overt over indebtedness, and relatively few reported abusive practices by MFIs. According to an expert commentator, “No-one really knows about the extent of multiple borrowing, but certainly very many clients had loans from more than one provider and in some cases the burden of repayment became too heavy and the loans went into default. However, there are very few cases these days of abusive collection practices in Bangladesh.”57 A study on multiple borrowing using data 55Rutherford (CGAP, 2011). 56 IMF (2013). 57 Rutherford outlines the following reasons that Bangladesh is different from, say, Andhra Pradesh in India, which recently suffered a serious microfinance over-indebtedness crisis: First, Bangladesh clients always had some savings lodged at their lenders. This gave security for both parties. When a loan got into trouble, staff nudged the client to pay back until the outstanding loan was small enough to be netted off against savings. This meant low losses of funds for the providers and low losses of dignity for the clients. India didn’t have this highly important savings cushion – ironically, precisely because India had strong regulation forbidding unlicensed saving mobilization. Second, microfinance in India grew much faster than in Bangladesh – it compressed 20 years of Bangladesh history into 2 or 3 years. Third, clients and staff didn’t have long enough to learn -as-you-go. Fourth, India got a tsunami of profit-seeking funds – a huge wave that came all at once. Bangladesh, in contrast, grew more steadily and never had big chunks of profit-seeking funding. Fifth, though Bangladeshi microcredit was quasi-commercial (with actors like ASA being very aggressive), everyone in the industry knew that microcredit was about relieving poverty and wasn’t just another sector of commercial financial business. This tempered behavior (email communication 2012). 40 from Grameen Bank II found that “quite a few loans were repeat loans from the same MFI (and some in that subclass were loan types that deliberately allow 'topping up' such as at Grameen—71 of the loans were top-ups of this kind)” (Rutherford, 2006).58 Furthermore, Khandekar and Samad (2014) also note that while “multiple program membership is growing with competition from microfinance institutions, but that this has rather helped raise assets and net worth more than it has contributed to indebtedness.” Overall observers point out that there was quite a lot of multiple borrowing, varying in intensity between households, but the levels of debt held probably remained quite moderate relative to incomes.59 This refers to the practice of allowing a loan to be “topped up” to its original value. In the case of 58 Grameen bank this was done after 26 weeks of repayment. 59 Rutherford, email communication, 2012. 41 7. Lessons Learned and the Way Forward 111. There are a number of indicative lessons that emerge from these assessments, on specific institutional and policy environments and on loan design features that can be conducive to the goal of expanding financial inclusion through Bank-financed projects. In addition, as for all successful project lending, straightforward features of good practice are once again underscored, such as the need for strong and sustained political commitment, a supportive policy and institutional environment, solid preparatory work and appropriate design, and careful follow-up. These findings are based on a small albeit intensively examined sample of projects, however, and a wider sample would need to be examined to validate the robustness of these findings.60 Borrower Institutions and Bank Loan Structure 112. There are instances where the World Bank has been successful in finding solutions that protect the goals of expanded financial outreach of public sector banks, but limit the need for frequent and costly recapitalizations by the government. The development agency model, as demonstrated by Mexico’s Financiera Rural, can be a good option for transforming a state-owned financial institution, when full privatization is not feasible and there is a need to expand credit to underserved groups in a cost-efficient manner. Such a development agency, properly run, is funded out of initial capital endowments or targeted budgetary appropriations, not from deposits or other liabilities, and it has appropriate government oversight. It can be capable of promoting financial access for underserved sectors and market development through market-friendly instruments that do not distort market prices, such as matching grants and subsidies that are financed by the government’s budget; thus it need not discourage private sector activities. Nevertheless, moving from first- tier to second-tier lending can be difficult and may take years to achieve. In addition, such agencies are not immune from shifts in the political climate. 113. Establishing a ring-fenced mechanism of a broadly independent microfinance institution within an existing public sector bank has been attempted by the Bank, however this leaves it open to influence by the prevailing political climate. As the CrediAmigo example shows, in an environment where public sector financial intermediation or some segments of it may be subsidized by the state, it may be difficult to isolate the ‘independent’ microfinance program. Even if ring-fenced today, unless spun off, the MFI remains vulnerable to future recapture by the parent institution. Tensions about the extent to which public or private sector principles apply persist. 114. Credit line operations may increase the supply of funds but do not increase sustainable access to finance if other institutional, legal and regulatory constraints to financial sector development are not addressed. This was the case with the Mozambique loan. Lines of credit require assessments of not just overall demand but also factors that could dampen demand, such as high interest rates or the availability of nonloan income to the 60The forthcoming IEG evaluation on financial inclusion will provide relevant insights, although from a different analytic perspective. 42 banks such as low-risk government securities. On the supply side there can also be factors that may limit commercial banks’ appetite to lend to small and medium enterprises, even if they have the funds—for example, limited information on borrowers’ creditworthiness, poor laws on collateral, or the availability of other opportunities for investing, such as government paper, with potentially lower risk adjusted returns. Loan design features may also be a consideration—for example, minimum or maximum size, earmarking or preferential schemes, such as for first time borrowers, domestic borrowers, etc. This broad array of factors suggest reasons why credit line operations that are able to disburse often display elements of subsidy shared with the borrower, even if not initially designed with such features. 115. Credit lines accompanied by programs of technical assistance or training are often more effective vehicles of expanded inclusion. Technical assistance to borrowers and financial intermediaries can be crucial to increase the demand for credit funds. In Mozambique, one reason for the success of GAPI in helping disburse the line of credit was that it provided training in addition to loans for its small enterprise clients. And in Mozambique, following the Bank project, IFC launched a successful line of credit which not only gave loans but also provided support to both financial intermediaries as well as borrowers through its advisory services. 116. Matching grant programs can potentially be an effective way of providing incentives to small firms that are clients of expanded access. To be successful such programs have to be well managed. Outsourcing to a well-established institution with the right expertise helps. A successful matching grant program for business services requires a combination of international expertise, local knowledge, autonomy in administration, and adaptability to respond to market demand. 117. Vehicles for direct support to rural finance institutions, or myriad smaller microfinance or informal financial institutions, are limited and apex arrangements have been the typical approach, with mixed success. In the case of Ghana, for example, the Bank was able to help establish the regulatory guidelines for rural banks through support to the ARB Apex bank, but it had only partial success in establishing delegated supervision by the Central Bank to the Apex Bank. The project design did not promote a closer association between the Central Bank/banking supervisor and the Bank supported apex institutions. The Policy Environment 118. National economic policies are as important as institutional aspects for the achievement of sustainable and inclusive financial services. The operations of well- managed institutions can reflect financial unsustainability if the enabling environment reflects distortions in financial markets, as was the case in Ghana and Brazil. Development of an enabling framework for MFIs was rightly a key initial objective of Microfinance II in Bangladesh and did not develop under the leadership of PKSF. Later the central bank took over implementation and eventually established the Microfinance Regulatory Authority. However, progress in establishing the framework was slow and microfinance institutions lacked access to the credit registry or payment system, for example. 43 119. There can be tensions between the achievement of expanded microfinance outreach and the achievement of financial sustainability. It may be tempting in some political environments to expand outreach at the expense of sustainability. In Brazil, the Bank’s decision to remain engaged was anchored on real successes achieved in terms of outreach, which illustrates dilemmas about outreach versus sustainability. Yet expansion of outreach to a desired clientele is not necessarily an indicator of sustainability. 120. Subsidies can be taken into account in project design through a Subsidy Dependence Index, provided it is well defined—but it has proven difficult to reflect subsidies in funding. This is illustrated both by the Credi Amigo (Brazil) and Ghana examples. Greater transparency was achieved in Mexico, with government commitment and policy support reflected in the proviso that Financiera Rural’s capacity to provide subsidies would be limited by Congress, and also, that those subsidies that would be provided would be more targeted and transparent. Financiera Rural‘s budgetary funding established strong incentives to manage it prudently and efficiently and revealed the extent of the government’s commitment to support rural credit.61 121. When subsidies are political and entrenched, they are hard to displace and it is insufficient for a project team alone to conduct dialogue on this theme; it requires higher-level policy dialogue between senior Bank staff and senior country officials. Despite regular policy dialogue, the project team in Ghana, on its own, was not successful in reducing subsidies that were offered under parallel government programs. 122. Complementary regulatory and financial institution infrastructure is also important for the effective functioning of financial markets, including the availability of loans for financial inclusion. These include factors such as clear creditors’ rights, effective laws for the use of collateral, and credit registries. All these elements are needed to increase the risk-taking appetite of potential lenders, as illustrated by the case in Mozambique. 123. The Bank can help build lasting reforms by limiting the scope of the projects to what can be accomplished within a certain political and economic environment, even if not fully market oriented. This was demonstrated in Mexico, where the Bank recognized that some element of subsidization was inevitable for support to rural borrowers, but it designed a solution that moved the burden directly to the government, and away from the financial institutions, at the same time limiting their scope and increasing their transparency. 61The initial capital endowment thus comprised the good assets of Banrural plus a one-time budgetary allocation for lending and initial set-up costs. The new institution was expected to preserve this capital. This meant that to sustain its lending operations, Financiera Rural would have to adequately collect on the loans, or ask for additional budgetary appropriations to restore its initial capital endowment in real terms. This created strong incentives for its management to maintain sound lending policies. Its developmental activities (subsidies, guarantees, technical assistance, an so forth) to develop rural financial markets would be funded directly through the budget. Under this system Financiera Rural was expected to have total transparency and accountability to Congress in the use of its resources. Even more important, not being able to take deposits made the operations of the Financiera Rural transparent, as it could not use deposits to get liquidity while making losses, as Banrural had done for many years. The government’s continued support to sustainability was key to the success of the strategy. 44 The Bank also recognized that moving from first-tier lender to second-tier lender status was not instantly possible, and a phased program toward this objective was adopted. In Brazil too, the Bank attempted to set up a sound and sheltered lending platform within a public sector bank with less sound operations. In such cases the Bank would need to recognize nonetheless that the environment is fragile and may depart further from market-based principles in the future. 124. In terms of Bank lending instruments, although the instrument of choice is typically an investment loan, policy based loans may have a larger role to play given their usefulness in addressing all-important issues in the policy environment. The use of such a policy-based loan was appropriate for the Mexican operation since it provided the necessary flexibility to formulate the right set and sequence of policy actions. The multitranche structure (or, in the present environment, a programmatic structure) was attuned with the need to follow up closely on the implementation of the business model selected for Financiera Rural. In this example, the choice of instrument was reinforced by the government’s need for general budgetary support given contingent fiscal liabilities because of ailing public sector banks, rather than investment funding for Financiera Rural. 125. Strong borrower commitment is particularly essential for the success of an adjustment operation involving policy and institutional reforms. Borrower commitment is also important for project-specific investment loans, not only to ensure that there are no obstacles to implementation, but also to ensure that underlying policy and environment- related factors that may not have been an explicit element of project design, are aligned with project objectives. Other Factors in Bank Loan Design 126. Project development objectives that are too general cannot be effectively monitored. Such objectives generally compromise the evaluation of results. This was the case for the Ghana Rural Financial Services project. Narrowing the objectives to sharply defined actions has the beneficial effect of keeping the operation simple and focused. In the case of Financiera Rural, the sharp focus on institutional building and financial discipline was one of the operations’ best aspects. 127. Sound analytical underpinnings are a large element in good project design. Part of the success of the Mexico project can be attributed to long-standing analytical work done by the Bank on Mexico’s financial sector, including rural finance. Additionally, the financial sector assessment undertaken jointly by the Bank and the IMF in 2001 had a decisive impact on the government’s approach to the development bank reform in Mexico and the selection of the business model for Financiera Rural. 128. Flexibility in supervision to adapt to changing scenarios is essential for success, both when changes are marginal and when they are fundamental to the point of rendering the attainment of the operation objectives impossible. In the former case, flexibility must be manifest in the willingness to sacrifice details for the sake of achieving the ultimate objectives of the operation. In the latter case, this may require courageous decisions from the project team, to the point of significantly restructuring the operation, should it be 45 necessary to remain true to original objectives, as observed in the Brazil CrediAmigo project, in the Mozambique Enterprise Development loan (PoDE) for SME finance, and in the Bangladesh Learning and Innovation project FSP. 129. It is difficult to transpose sometimes idiosyncratic institutional arrangements for microfinance services from one country to another and pilot and stakeholder consultations to ensure their suitability for the local context would be very helpful. The project in Ghana illustrated the difficulties of transposing the idea of promoting associations between informal financial service providers, such as the susus, and the rural banks. While this may have worked in India within the context of Self Help Groups, in Ghana, the rural banks and the informal groups were effectively competing with no obvious gains from association. Because of implementation challenges, the component had to be revised at mid- term review to narrow its focus to a smaller number of rural and community banks and to support them in developing business plans to help them form linkages as well as to rationalize the type of training, group selection criteria etc. This was costly and inefficient. 130. The Bank has found it difficult to disentangle rural and regional objectives; intervention choices need to match the objective to loan design. In the case of Ghana, the project was expected to increase outreach to agriculture and the lagging rural regions, but the design did not include any strategies for doing so. Not surprisingly, outreach to both agriculture and the lagging regions remained low. In this regard, the clear regional focus of CrediAmigo of Brazil, which aimed to support the Northeast regions (with no rural or SME objective, per se), was more successful. In Mozambique, the Bank tracked indicators on a regional basis, and looked at the extent to which the project was successful in outreach outside Maputo, with a view to expanding first–time borrowers, although this was not an element of project design. Achieving Broader Goals: From Outputs to Outcomes and Impact 131. Defining beneficiary selection criteria is important to avoid selection bias. There is a possibility of selection bias because of the strong incentive of the implementing agency to include beneficiaries who are more likely to succeed. The beneficiary selection process needs to be strictly monitored. This was not incorporated in the Bangladesh’s Learning and Innovation project, although demonstrating that microfinance works for the ultra poor was a core component. This, however, is a matter of debate. The poorest are arguably better served through grants (safety nets, training, and so forth) than through credit alone, which may further overburden them in debt. Eventually, these alternative modalities were not adequately explored through the loan. The project included support/safety nets and training components in addition to credit for the poorest. However, it was not clear how many were able to graduate to mainstream microfinance or develop skills that enabled them to have an income generating activity on a sustainable basis. The outreach criterion alone—especially in the absence of a definition of the poor or a clear criterion for targeting—does not substantiate the claim that the poorest can be graduated to mainstream microfinance or sustainably be lifted out of poverty. 132. These Bank projects have not reflected the fact that the poor—like the rich— need access to a broad array of financial products and services. Much of the focus in the 46 projects examined here has been largely on access to credit and a wider focus is needed. Credit was clearly the focus of the financial services to be provided in all the projects under consideration, although to varying degrees, some also provided training to borrowers (in Mexico and Mozambique, for example). None of the projects reviewed included the development of products in the areas of deposits, remittances, money transfers, microinsurance, microleasing, or other services. Yet it is known that, for example, a lack of easily accessible payment transfer channels contribute to a large share of informal remittances in Bangladesh (in the order of 30-50 percent).62 Facilitating the participation of microfinance and financial cooperatives in the remittance and money transfer market could also be an avenue for increasing financial access in remote districts. By linking to formal channels, clients could benefit from other financial services. Similarly a safe place for voluntary savings (and in the case of women often confidential savings too) is an urgent need of the poor for consumption smoothing. But in Bangladesh for example, other than the Grameen bank, MFIs, as NGOs, are not allowed to accept deposits. Increasing access of the poor to these products does not appear to have been explored. 133. In addition to a focus on a wider array of products, a broader use of different instruments and delivery channels could also have been pursued. Arguably, Ghana’s hard-to-reach northern and poorer regions could have benefit from models of agency/correspondent banking, as in Brazil initially and now increasingly explored in other countries.63 In Bangladesh, the selection of participating microfinance institutions did not appear to take into account the unserved regions the north and southwest of the country. Introducing new technologies and institutional models such as mobile or correspondent banking could help not only reduce costs and increase access to financial services but also address the geographic skew in outreach observed in all the countries reviewed. Finally, it does not appear that the projects considered (apart from Financiera Rural) looked at alternative channels of financial inclusion of the poor such as government transfers, social protection, or salary payments through branchless channels such as payment cards. These may be interesting areas for Bank projects focused on financial inclusion, apart from the traditional bank-focused intermediation. 62Bangladesh Human Resource Development Centre study “Remittance Contribution by the Bangladeshi Migrant Workers,” http://www.dhakatribune.com/economy/2013/sep/15/30-40- remittances-coming-through-informal-channels. The article also quotes the 2006 World Bank Global Economic Prospect Report, which claimed that 54 percent of remittances came through informal channels. 63 Kumar and others (2006), World Bank, 2009b 47 Appendix 1: World Bank Access to Finance Projects Exit FY 2006-13 (% Access to Finance by Sector) A2F A2F ICR IEG Commitment Exit Regio Sector Sector Theme Lending # Proj ID Project Name Country n Board Outcome Outcom Content Content Instrument Amount FY Rating e Rating (US$ M) % % ACCESS to 1 2012 P108140 FINANCE for Turkey ECA FPD S S 100 100 FIL 250.0 SMEs - AF2 ACCESS to 2 2012 P112949 FINANCE for Turkey ECA FPD S S 100 100 FIL 200.0 SMEs - AF BD: Microfinance 3 2009 P105742 II-AF Bangladesh SAR FPD S S 100 0 FIL 15.0 AF: Microfinance 4 2012 P104301 Project Afghanistan SAR FPD MS U 100 0 FIL 30.0 BR NE 5 2007 P050776 Microfinance Brazil LCR ARD S S 99 40 SIL 50.0 Development MX (AF) 6 2011 P106682 Savings&RurFina Mexico LCR ARD MS MS 98 0 SIL 21.0 nce SAGARPA 7 2009 P059143 Microfinance II Bangladesh SAR FPD S S 96 0 FIL 151.0 8 2011 P096285 CN-MSE Finance China EAP FPD S S 95 67 FIL 100.0 MG-Microfinance 9 2011 P104883 Add Fin SIL Madagascar AFR FPD MS MS 94 29 SIL 5.0 (FY07) 10 2011 P052186 MG-Microfinance Madagascar AFR FPD MS MS 94 25 APL 16.4 MX Southeast 11 2006 P060577 Reg'l Mexico LCR FPD S S 80 100 LIL 5.0 Development LIL MX Rural 12 2006 P074655 Finance Develop Mexico LCR FPD HS HS 80 29 SAD 505.1 Struct Adj Loan TR Access to Fin 13 2012 P082822 for SMEs Turkey ECA FPD S S 70 67 FIL 180.2 14 2007 P057187 TZ-FIDP II Tanzania AFR FPD S S 70 28 SIL 27.5 PK: Second Poverty 15 2011 P082977 Alleviation Fund Pakistan SAR FPD S MS 70 11 FIL 238.0 Proj BR-RGN Rural 16 2011 P066170 Poverty Brazil LCR ARD MS MS 60 34 SIL 22.5 Reduction 17 2012 P083082 MSME Nigeria AFR FPD MU MS 60 25 SIL 32.0 BJ-Private Sector 18 2008 P039882 (FY00) Benin AFR FPD U U 55 23 SIL 30.4 MN-Private Sector 19 2011 P088992 Development Mongolia EAP FPD S MS 50 50 FIL 10.6 Credit II MX Savings & 20 2011 P070108 Credit Sector Mexico LCR ARD S MS 45 25 TAL 64.6 Strengthening AGRIC DEVT & 21 2006 P035813 CREDIT Azerbaijan ECA ARD S S 43 22 APL 30.0 Enterprise Dev Mozambiqu 22 2006 P049874 (FY00) e AFR FPD MS MS 43 20 SIL 26.0 48 A2F A2F ICR IEG Commitment Exit Regio Sector Sector Theme Lending # Proj ID Project Name Country n Board Outcome Outcom Content Content Instrument Amount FY Rating e Rating (US$ M) % % EG East Delta Egypt, Arab 23 2010 P049166 Ag. Serv. Republic of MNA ARD MU MS 41 0 SIL 15.0 Financial 24 2008 P074731 Services for the Bangladesh SAR FPD HS MS 40 50 LIL 5.0 Poorest BR- Amapa 25 2011 P076924 Sustainable Brazil LCR ENV MU HU 40 22 SIL 4.8 Communities RENEW 26 2010 P072480 ENERGY Turkey ECA EMT HS HS 40 0 SIL 202.0 27 2010 P082801 EFIL 2 Turkey ECA FPD HS HS 35 40 FIL 303.1 MSME 28 2012 P085007 Competitiveness Kenya AFR FPD MS MS 35 25 SIL 22.0 GH-Rural Fin 29 2008 P069465 Srvcs SIL (FY00) Ghana AFR FPD S MS 32 40 FIL 5.1 30 2010 P093568 EFIL 3 Turkey ECA FPD HS HS 30 40 FIL 305.0 MN - Sustainable 31 2008 P067770 Livelihoods Mongolia EAP ARD S S 27 0 APL 18.7 FARM PRIV 32 2006 P049718 SUPPORT Tajikistan ECA ARD S MS 26 0 SIL 20.0 GT 33 2009 P055084 COMPETITIVEN Guatemala LCR FPD MS MS 25 22 SIL 20.3 ESS PROJECT NI 34 2006 P070016 Competitiveness Nicaragua LCR FPD S S 24 25 LIL 5.0 LIL BR RGS LAND 35 2006 P043868 MGT/POVERTY Brazil LCR ARD S S 24 0 SIL 100.0 MA-Social 36 2008 P073531 Development Morocco MNA SP S MU 21 0 SIL 5.0 Agency RURAL FIN (APL 37 2008 P056891 #1) Romania ECA ARD S S 21 0 APL 80.0 SOC SECT DEV 38 2009 P008783 (SSD) Romania ECA SP S MS 20 29 SIL 50.0 MA-Sustainable 39 2012 P117201 Access to Morocco MNA FPD S S 20 15 DPL 200.0 Finance DPL ID - BENGKULU 40 2006 P040061 REGIONAL Indonesia EAP ARD S S 20 0 SIL 20.5 DEVELOPMENT HN Regional Dev 41 2009 P081172 in the Copan Honduras LCR SDV S MS 20 0 SIL 12.0 Valley IN: AP Rural 42 2012 P104724 Poverty Drought India SAR ARD S S 20 0 SIL 65.0 Adaptation Central CAR: EMGRG III 43 2012 P120534 DPO IV (FY11) African AFR EP MS MU 19 0 DPL 8.8 Republic Financial Sector 44 2012 P094704 Cap Bldg. Project Ethiopia AFR FPD MS MS 17 14 SIL 15.0 45 2007 P056337 MINE CLOSURE Romania ECA EMT HS S 16 0 SIL 44.5 BD: SIPP- 46 2011 P110376 Cyclone-07 Add Bangladesh SAR ARD MS MU 16 0 SIL 50.0 Financing III 49 A2F A2F ICR IEG Commitment Exit Regio Sector Sector Theme Lending # Proj ID Project Name Country n Board Outcome Outcom Content Content Instrument Amount FY Rating e Rating (US$ M) % % EG-SOCIAL Egypt, Arab 47 2006 P052705 FUND III Republic of MNA SP S MS 15 25 SIL 50.0 RUR ENT & 48 2011 P087011 AGRIC DEVT Armenia ECA ARD S S 15 14 SIL 20.0 Programmatic 49 2006 P083949 Reform for Peru LCR FPD S S 15 0 PSL 100.0 Growth II SN-Soc Dev 50 2006 P041566 Fund APL (FY01) Senegal AFR SP S MS 15 0 APL 30.0 VN - Rural 51 2010 P072601 Finance II Project Vietnam EAP ARD S S 15 0 FIL 200.0 NE-Financial Sec 52 2011 P074316 (FY04) Niger AFR FPD MU U 15 0 TAL 14.8 MEXICO: KNOWLEDGE 53 2006 P044531 AND Mexico LCR ED S MS 13 25 SIL 300.0 INNOVATION PE 3rd Prog 54 2013 P106720 Fiscal Mgmt DPL Peru LCR EP S S 13 0 DPL 150.0 MX GENDER 55 2006 P066938 (LIL) Mexico LCR GE HS S 12 0 LIL 3.1 PY NTL RES 56 2006 P007918 MGMT I Paraguay LCR ARD S S 12 0 SIL 50.0 ER-Emerg 57 2009 P073604 Demob & Reint Eritrea AFR SP MS MS 12 0 ERL 60.0 ERL (FY02) MX Strength. 58 2012 P112264 Business Env for Mexico LCR FPD S S 12 0 DPL 751.9 EcoGrowth RW-Compet & 59 2012 P106978 Ent Dev Add Fin Rwanda AFR FPD MS MS 12 0 SIL 6.0 SIL (FY08) 60 2007 P008860 POV ALLV 2 Tajikistan ECA SP S S 11 0 SIL 13.8 EG - SOHAG Egypt, Arab 61 2008 P040858 Rural Republic of MNA ARD MS MS 11 0 SIL 25.0 Development ML-Finance Sec 62 2008 P001748 Dev (FY00) - Mali AFR FPD MS MS 10 25 SIL 21.0 (PDSF) NI Offgrid Rural 63 2012 P073246 Electrification Nicaragua LCR EMT S MS 10 25 SIL 12.0 (PERZA) MX Finance and 64 2006 P097159 Growth DPL I Mexico LCR FPD S S 10 20 DPL 501.3 65 2006 P074072 TZ-PRSC 1 (first) Tanzania AFR FPD S S 10 0 PRC 132.0 Emergency 66 2007 P084329 National Afghanistan SAR ARD S MS 10 0 SIL 95.0 Solidarity Project Emerg National 67 2007 P094735 Solidarity - Afghanistan SAR ARD MS MS 10 0 SIL 28.0 Supplemental MW-MASAF APL 68 2008 P075911 3 (FY03) Malawi AFR SP S MS 10 0 APL 60.0 BD - Water 69 2011 P086661 Supply Program Bangladesh SAR WAT MS MS 10 0 SIL 40.0 Project 50 A2F A2F ICR IEG Commitment Exit Regio Sector Sector Theme Lending # Proj ID Project Name Country n Board Outcome Outcom Content Content Instrument Amount FY Rating e Rating (US$ M) % % CN-Enterprise 70 2006 P060270 Reform China EAP FPD U HU 9 29 LIL 5.0 Indonesia Sixth 71 2010 P113638 Development Indonesia EAP EP S S 9 0 DPL 750.0 Policy Loan BJ-Natl CDD SIL 72 2012 P081484 (FY05) Benin AFR SP S S 9 0 SIL 50.0 Forestry 73 2006 P046952 Development In China EAP ARD S S 7 33 SIL 200.0 Poor Areas RY-Second Yemen, 74 2006 P068830 Social Fund for Republic of MNA SP HS S 7 22 SIL 75.0 Dev 75 2009 P110191 Fifth DPL Indonesia EAP EP S S 7 7 DPL 750.0 TN-Integration and 76 2011 P095388 Competitiveness Tunisia MNA EP MS MU 6 16 DPL 250.0 DPL TN-Export 77 2013 P106274 Development II Tunisia MNA FPD MS MS 5 20 SIL 6.0 (adtl financing RY-SOCIAL FUND FOR Yemen, 78 2010 P082498 DEVELOPMENT Republic of MNA SP S S 5 14 SIL 60.0 III HN Trade 79 2011 P070038 Facilitatio & Honduras LCR FPD MS MU 5 13 SIL 28.1 Productivity Enha 80 2009 P074602 BI-ERC (FY03) Burundi AFR PS S MS 5 0 SAL 54.0 Integrated 81 2011 P098583 Human Devt II Maldives SAR SP MU MU 5 0 SIL 5.7 IN: Karnataka 82 2012 P102328 Tanks India SAR ARD S S 5 0 SIL 64.0 (Supplement) Poverty 83 2009 P081968 Alleviation Fund Nepal SAR ARD S MS 4 0 SIL 15.0 NP PAF 84 2009 P080819 Supplemental Nepal SAR ARD MS MS 4 0 LIL 25.0 PH-RURAL 85 2008 P057598 FINANCE III Philippines EAP ARD S S 3 33 SIL 150.0 Social Fund for Yemen, 86 2010 P102181 Development 3 Republic of MNA SP S S 2 13 SIL 15.0 (supplem.) PE SECOND 87 2007 P044601 RURAL ROADS Peru LCR TR HS S 2 0 SIM 50.0 PROJECT MG-Priv Sec Dev 88 2011 P072160 2 (FY02) Madagascar AFR FPD MU MU 1 20 SIL 23.8 Venezuela, VE CARACAS Republica 89 2006 P040174 SLUM Bolivariana LCR UD MS MS 1 0 SIL 60.7 UPGRADE de BJ-Community 90 2012 P121104 Dev't Add. Benin AFR SP S S 1 0 SIL 12.0 Financing MG-Transp Sec 91 2006 P052208 Reform & Rehab Madagascar AFR TR S S 0 50 APL 65.0 (FY00) 51 A2F A2F ICR IEG Commitment Exit Regio Sector Sector Theme Lending # Proj ID Project Name Country n Board Outcome Outcom Content Content Instrument Amount FY Rating e Rating (US$ M) % % ENTERPRISE 92 2007 P008416 REHABILITATIO Georgia ECA FPD S MS 0 50 TAL 15.0 N PROJECT Karnataka 93 2008 P070421 Highways India SAR TR S S 0 40 SIL 360.0 ACC TO FIN 94 2010 P076553 SERVS (APL 1) Ukraine ECA FPD U U 0 40 APL 150.0 PRIV SEC DEV 95 2006 P054966 (APL #1) Ukraine ECA FPD HU HU 0 33 APL 30.0 96 2006 P076543 FIEF LIL Armenia ECA EP S HS 0 33 LIL 1.0 ENT 97 2007 P044852 INCUBATOR LIL Armenia ECA FPD S S 0 29 LIL 5.0 CM-Pub/Priv 98 2008 P065927 Partnership & Cameroon AFR FPD MU U 0 29 TAL 20.9 Growth (FY00) CV Growth & 99 2010 P103329 Competitiveness Cabo Verde AFR FPD S S 0 29 SIL 3.0 Suppl.Credit HARD COAL 100 2007 P066512 SOC Poland ECA EMT MS MS 0 25 SIL 200.0 MITIGATION TN-Agricultural 101 2009 P005750 Support Services Tunisia MNA ARD MS MS 0 25 SIL 21.3 PE Trade Facil. 102 2009 P077788 and Prod. Peru LCR FPD S S 0 25 TAL 20.0 Improv. T. A. RUR ENT 103 2008 P046043 SUPPORT Uzbekistan ECA ARD S S 0 24 SIL 36.1 BI-Pub Wrks & 104 2009 P095024 Emply Crtn Burundi AFR UD S S 0 24 SIL 30.6 AdditSIL (FY06) INDUSTRIAL 105 2006 P009073 TECH Turkey ECA FPD HS MS 0 23 SIL 155.0 AR POLLUTION 106 2006 P006050 MGT. Argentina LCR ENV S MS 0 22 SIL 18.0 MW-Road 107 2006 P001666 Maintenance & Malawi AFR TR MS MS 0 20 SIL 30.0 Rehab (FY99) 108 2008 P074408 SRMP Turkey ECA SP S S 0 20 SAD 500.0 Kyrgyz 109 2009 P073973 VIP Republic ECA ARD HS HS 0 20 SIL 15.1 110 2010 P079978 ENERGY EFF Croatia ECA EMT MS MS 0 20 SIL 5.0 RY-PORT CITIES Yemen, 111 2010 P065111 DEVELOPMENT Republic of MNA UD MS MS 0 20 APL 23.4 PROGRAM CO Business 112 2006 P094301 Product and Colombia LCR FPD S MS 0 17 DPL 250.0 Efficiency DPL I ID Second 113 2006 P096594 Development Indonesia EAP EP S S 0 17 DPL 400.0 Policy Loan ENERGY SECT 114 2008 P088865 TA 3 Kosovo ECA EMT S S 0 17 TAL 2.5 115 2010 P094167 PSSP 2 Turkey ECA SP MS MS 0 17 SIL 465.4 52 A2F A2F ICR IEG Commitment Exit Regio Sector Sector Theme Lending # Proj ID Project Name Country n Board Outcome Outcom Content Content Instrument Amount FY Rating e Rating (US$ M) % % GH-Small Towns 116 2010 P084015 Water Sply & Ghana AFR WAT S S 0 17 APL 26.0 Sanit (FY05) Enterprise 117 2011 P081969 Growth & Bank Bangladesh SAR FPD MS MS 0 17 SIL 250.0 Modernization HR Science & 118 2011 P080258 Technology Croatia ECA FPD S S 0 17 SIL 40.0 VN-ROAD 119 2012 P059663 NETWORK Vietnam EAP TR MU MU 0 17 SIL 225.3 IMPROVEMT PG- (Supplement) Papua New 120 2012 P102792 Road Maint. and Guinea EAP TR S S 0 17 SIL 37.3 Rehab Third Road 121 2006 P037294 Rehabilitation & Bangladesh SAR TR S MS 0 14 SIL 273.0 Maintenance 122 2007 P068808 SDF 2 (APL #2) Romania ECA SP HS MS 0 14 APL 20.0 MU-Development 123 2008 P101570 Policy DPL Mauritius AFR EP HS HS 0 14 DPL 30.0 (FY07) LB - Community 124 2008 P071113 Development Lebanon MNA SP MS U 0 14 SIL 20.0 Project 125 2008 P039929 SIF Georgia ECA SP S MS 0 14 FIL 20.0 EC (APL2) 126 2009 P095555 PRAGUAS II Ecuador LCR WAT HU HU 0 14 APL 48.0 First 127 2010 P092663 Development Indonesia EAP EP S S 0 14 DPL 300.0 Policy Loan RURAL 128 2010 P066199 ENVIRONMENT Azerbaijan ECA ENV U U 0 14 SIL 8.0 SOC & ECON 129 2011 P076730 REC Croatia ECA SDV MS MS 0 14 SIL 45.7 LA-Fin. Lao Management People's 130 2011 P077620 Capacity Building Democratic EAP PS MS MS 0 14 SIL 8.5 Cr. Republic GA-Loc 131 2012 P082812 Infrastructure Gabon AFR UD MS MS 0 14 SIL 25.0 Dev SIL (FY06) RW-Compet & 132 2012 P057295 Enterprise Dev Rwanda AFR FPD MS MS 0 14 SIL 40.8 (FY01) RY-Southern Yemen, 133 2006 P005902 GOV AGRI Republic of MNA ARD U U 0 13 SIL 24.7 PRIVATIZA MX 134 2007 P098299 Competitiveness Mexico LCR FPD MS MU 0 13 DPL 300.8 DPL 135 2007 P057952 SIF 2 Armenia ECA SP S MS 0 13 SIL 20.0 AZ Fin Servs 136 2010 P081616 Devt Azerbaijan ECA FPD S MS 0 13 SIL 12.3 MX-(APL1) 137 2011 P089865 Innov. for Mexico LCR FPD S S 0 13 APL 250.0 Competitiveness 53 A2F A2F ICR IEG Commitment Exit Regio Sector Sector Theme Lending # Proj ID Project Name Country n Board Outcome Outcom Content Content Instrument Amount FY Rating e Rating (US$ M) % % PG-ROAD Papua New 138 2012 P004397 MAINT. & Guinea EAP TR S S 0 13 SIL 40.0 REHAB JO-Recovery 139 2012 P117023 Under Global Jordan MNA EP MS U 0 12 DPL 300.0 Uncertainty DPL PPAF-II Suppl 140 2011 P105076 Social Pakistan SAR ARD MS MS 0 11 SIL 75.0 Mobilization RY Private Yemen, 141 2012 P122414 Sector Growth & Republic of MNA EP U U 0 11 DPL 70.0 Social Protec Note: Projects with a disconnect between ICR outcome and IEG rating are highlighted. Includes only IDA/IBRD projects whose ICRs have been reviewed & rated by IEG. Key: U-Unsatisfactory, MS-Moderately Satisfactory, S-Satisfactory, HS-Highly Satisfactory, MU- Moderately, Unsatisfactory, HU-Highly Unsatisfactory, Sector / Theme: GD-Gender & Development, PSG-Public Sector Governance, EM- Energy & Mining, SD -Social Development, UD - Urban Development, ENV-Environment, TRANS-Transport 54 References Agencia Brasil 2013. Microcrédito é recorde no Demirguc-Kunt, A., and L. Klapper. 2011, and país, mas atende apenas parte dos pequenos forthcoming, April 2015. Measuring Financial empreendedores. O Globo 2013 Dec 22 Inclusion: The Global Findex Database. Brookings Papers on Economic Activity. Armendariz de Aghion, Beatriz and Jonathan Washington D.C. Morduch, 2000, Microfinance beyond group lending. Economics of Transition 8, 401-420. Demirgüç-Kunt, A., Beck, T. & Honohan, P., 2008. "Finance for All?: Policies and Pitfalls in Aveh, Krah and Dadzie. 2013. An Evaluation Expanding Access," The World Bank, of Sustainability and Subsidy Dependence of Washington, DC. Microfinance Institutions in Ghana. CS Canada International Business Management Galor, O. & Zeira, J., 1993. Income Distribution Vol. 6, No. 1, 2013, pp. 55-63 and Macroeconomics. Review of Economic Studies, Wiley Blackwell, vol. 60(1), pages 35- Barr, Michael S., Anjali Kumar, and Robert E. 52, January. Litan, 2007 (Ed.) Building Inclusive Financial Systems: A Framework for Financial Access. Ghana Microfinance Intuitions Network Brookings Institution Press. Washington DC (GHAMFIN, 2005). “Microfinance Poverty Outreach and Performance Assessment: A Beck, T., Demirguc-Kunt, A., & Levine, R., study of Rural Microfinance Institutions and 2004. "Finance, Inequality, and Poverty: Cross- Government Programmes in Ghana,” PO Box Country Evidence," Policy Research Working AH 1392, Achimota, Accra, Ghana. Paper Series 3338, The World Bank. Honohan, P. 2004. "Financial Development, Beck, T., A. Demirgüç-Kunt, and R. Levine Growth, and Poverty: How Close are the (2007). "Finance, Inequality, and the Poor." Links?," Policy Research Working Paper Series Journal of Economic Growth 12 (1): 27-49. 3203, The World Bank. Belaisch, Agnes. 2003. “Do Brazilian Banks Compete?” IMF Working Paper No. 03/113, Honohan, Patrick, and Thorsten Beck. 2007. Washington, DC. Making Finance Work for Africa. Washington, Burgess, Robin, and Rohini Pande. 2005. “Can DC: World Bank. Rural Banks Reduce Poverty? Evidence from the Indian Social Banking Experiment.” Hulme, D. and Mosley, P. 1996 Finance American Economic Review 95(3): 780–95. Against Poverty: Effective Institutions for Lending to Small Farmers and Micro- Clarke, George R. G., L. Colin Xu, and Heng- enterprises in Developing Countries, London: fu Zou. 2006. “Finance and Inequality: What Routledge. Do the Data Tell Us?” Southern Economic Journal 72(3): 578–96. IEG, 2008a. “Implementation Completion Report Review: Bangladesh Financial Services Claessens, S. & Tzioumis, K. 2006. "Ownership for the Poorest (Credit No. C36810; Project ID and Financing Structures of Listed and Large P074731), World Bank. Non-listed Corporations," Corporate Governance: An International Review, Wiley IEG, 2008b. “Implementation Completion Blackwell, vol. 14(4), pages 266-276, 07 Report Review: Bangladesh Second Poverty Alleviation Microfinance Project (Credit No. Copestake, J., Balhotra, S. and Johnson, S. 2001 C3457; Project ID P059143), World Bank. 'Assessing the impact of microcredit: A Zambian case study', Journal of Development IEG, 2014a. “Project Performance Assessment Studies 37(4): 81-100. Report: Brazil Northeast Microfinance Development Project (IBRD-45540; Project ID P050776), World Bank. 55 IEG, 2014b. “Project Performance Assessment Discussion Papers 4366, C.E.P.R. Discussion Report: Ghana Rural Financial Services Project Papers. Land Administration Project (Credit No. Kumar, Anjali, Ajai Nair, Adam Parsons and C3374, C3374A; Project ID P069465), World Eduardo Urdapilleta (2006)”Expanding Bank Bank. Outreach through Retail Partnerships - IEG, 2014c. “Project Performance Assessment Correspondent Banking in Brazil” Report: Mexico Rural Finance Development Levine, R. 2005. "Law, Endowments, and Structural Adjustment Loan (Loan No. IBRD - Property Rights," NBER Working Papers 73670; Project ID P074655), World Bank. 11502, National Bureau of Economic IEG, 2014d. “Project Performance Assessment Research, Inc. Report: Mozambique Enterprise Development Li, Hongyi, Lyn Squire, and Heng-fu Zou, Project (Credit No. C3317, C3317A; Project ID 1998. “Explaining International and P049874), World Bank. Intertemporal Variations in Income IFAD, 2012. “Republic of Ghana: Rural Inequality.” Economic Journal 108(446): 26–43. Finances Services Project. Project Performance Li, Hongyi, Lixin Colin Xu, and Heng-fu Zou. Assessment” International Fund for 2000. “Corruption, Income Distribution, and Agricultural Development., Growth.” Economics and Politics 12(2): 155–82. IMF, 2010a. “Bangladesh: Financial System Stability Assessment” IMF Country Report Millennium Challenge Corporation (MCC). No. 10/38. 2009. “Audit of the Agricultural Credit Program in Ghana.” Audit Report No. M-000- IMF, 2010b. “Republic of Mozambique: 09-005-P Financial Sector Assessment program— Financial System Stability Assessment”. Nair, Ajai and Fissah Azeb. 2010. “Rural Washington, DC. Banking – The Case of Rural and Community IMF, 2011 “Republic of Mozambique: Staff Banks in Ghana.” World Bank, Washington, Report for the 2011 Article IV Consultation, DC No. 11/149, Washington, DC. Neri, Marcelo Cortes. Microcrédito, O Mistério IMF,2013. “GHANA Article IV Consultations” Nordestino E O Grameen Brasileiro, Perfil and IMF Country Report Number 13/187, Performance dos Clientes doCrediAmigo. Rio de Washington, DC. Janeiro: Banco do Nordeste, CrediAmigo, FGV and FGV editora, 2008. Khandker, Shahidur R. 1998. Fighting Poverty with Microcredit : Experience in Bangladesh, New York, NY: Oxford University Press Neri, Marcelo et. al. 2010. Brazilian Microcredit Takes Off. Article published in Khandker, Shahidur R. 2005. “Microfinance Brazilinfocus.com. 30 November. and Poverty: Evidence Using Panel Data from Bangladesh,” World Bank Economic Review, Vol. Obeng, Samuel. 2008. “Rural Banking in 19 (2): 263-286. Ghana: It’s Impact on Rural Farmers.” A case Khandker, Shahidur R.; Samad, Hussain A. Study of Abokobi Rural Area. Samven’s 2014. Dynamic effects of microcredit in Research and Consultancy Unit, Ghana Bangladesh. Policy Research working paper; Pearce, Douglas, Junior Davis, Gideon No. WPS 6821. Washington, DC: World Bank Onumah and Ruth Butterworth, (DFID, 2004). Group. “Making rural finance count for the poor”, Klapper, L., Laeven, L. & Rajan, R G, 2004. Department for International Development "Business Environment and Firm Entry: (DFID), UK, London Evidence from International Data," CEPR Pitt, Mark M. and Shahidur R. Khandker. 1996. “Household and Intrahousehold Impact 56 of the Grameen Bank and Similar Targeted Skoufias, Emmanuel. Philippe Leite and Credit Programs in Bangladesh.” World Bank Fransisco Haimovich 2009. ‘Expanding Access Discussion Papers No. 320. Washington D.C. to Microfinance: Evidence from the CrediAmigo Natural Experiment in the NE of Pitt, Mark and Shahidur Kandker 1998, “The Brazil.’ World Bank, October. Impact of Group-Based Credit Programs on Poor Households in Bangladesh: Does the Simonetti Roberto, Marc Wuyts, and Anna Gender of Participants Matter?” Journal of Wuyts-Fivawo. 2007. “Banking on Rural Political Economy, October. Innovation for Poverty Reduction: A Case Study of Value Chain Lending in Rajan, R. & Zingales, L., 2003. "The Emergence Mozambique.” European Journal of of Strong Property Rights: Speculation from Development Research, 19: 136-155. History," NBER Working Papers 9478, National Bureau of Economic Research, Inc. United Nations, 2006. “Building Inclusive Financial Sector for Robinson Marguerite. 2001.The Microfinance Development”bluebook.uncdf.org Revolution: Sustainable finance for the poor World Bank. 1995. “Brazil: A Poverty Washington, DC: The World Bank. ISBN 0- Assessment.” Human Resources Operations 8213-4524-9. Division, Country Department 1, Report No Rutherford, Stuart, 1999. The Poor and Their 14323-BR. Washington DC. Money An essay about financial services for World Bank, 2002 “Brazil Second poor people. Institute for Development Policy Programmatic Financial Sector Adjustment and Management. University of Manchester. Loan. Program Document”, May 2002 January. World Bank, 2006. Finance Research Web Rutherford, Stuart 2006. Grameen II The First Feature Article. “Better Data Needed for Five Years (2001-2005): Grounded View of Policy Research on Access to Financial Grameen’s New Initiative, MicroSave, Services.” Bangladesh. World Bank, 2009a. Mozambique Investment Rutherford, Stuart (CGAP, 2011). Two Climate Assessment, World Bank, RPED. Cautionary tales from Bangladesh: Washington, DC. http://www.cgap.org/blog/two-cautionary- tales-bangladesh. World Bank. 2009b, Banking the Poor: Measuring Banking Access in 54 Economies. Washington, DC.. Schicks, Jessica. 2011. “Over -Indebtedness of Micro-borrowers in Ghana – An Empirical World Bank, 2010. Bangladesh - Household Study from a Customer Protection income and expenditure survey: key findings Perspective, Center for European Research in and results 2010. Washington, DC Microfinance.” Working Paper No 15, World Bank. 2011.“Tackling Poverty in Bruxelles, November 2011. Northern Ghana.” Report Number 53991 -GH. Sinha, Sanjay 2011. “Bangladesh Microfinance World Bank. 2012. Global Financial Review,” BRAC, Bangladesh. Development Report : Rethinking the Role of Skoufias, Emmanuel. 2007. “Brazil: Measuring the State in Finance. Washington, DC Poverty Using Household Consumption.” World Bank 2014. Global Financial Report 36358-BR, World Bank, Washington, Development Report 2014: Financial inclusion. DC. 57