i Paper No. 11 / March 2004 Timor-Leste: Independent Review of the Credit Component of the Community Empowerment Project John D. Conroy Summary Findings Timor-Leste: Independent Review of the Credit Component of the Community Empowerment Project This review analyzes experience with the Community revive economic activity. Much of this funding Empowerment Project (CEP) credit component, and would be to recapitalize businesses which have been draws a number of lessons and general principles. destroyed or de-stocked by conflict, for micro- Credit contributed to a revival of rural economic entrepreneurs based on viable proposals evaluated on activities, especially petty trading and the supply of business principles. manufactured goods, helping to re-establish markets and stimulating production and marketing of Severe economic fluctuations in the immediate post- foodstuffs and higher-value crops. While CEP was crisis period will expose borrowers to high risks, not a financial sector project, its use of credit to which further strengthens the case for grants vis-a-vis achieve post-conflict reconstruction goals had loans. These factors also point to the need to phase implications for the revival of financial services. emergency stimulus from grants to credit over time Since the latter was also an important reconstruction considering the need to revive the financial system. goal, CEP had an obligation to avoid compromising The experience of CEP validates the conclusions on the re-establishment of financial services. While a credit reached in the Implementation Completion badly-conducted credit program may succeed in Report (ICR) for SEP I, calling for a transition from reviving business activity, it might do so at an `intermediated grants' to `intermediated credit', with unacceptably high cost in terms of negative impact the former administered by government or quasi- on the credit culture. Whether significant negative government entities and the latter by financial impact has occurred in Timor-Leste as a result of the institutions. poor repayment performance of the CEP and Small Enterprise project (SEP) projects may only become CEP experience offers no new lessons on apparent as microcredit begins to expand throughout microcredit, but reinforces those from experience the country. elsewhere. It does, however, underline some limitations of microcredit in the immediate post- Credit was an add-on feature to CEP, something that conflict crisis setting: could be attached to the project to enhance its ˇ where decision-making is divorced from risk, effectiveness. The amount allocated to credits for credit allocation is inefficient; economic activities was only 13% of total subgrants ˇ subsidized credit tends to undercut credit in the project, but credit is not an accessory that can delivered by institutions striving toward be added with minimal preparation. It requires sustainability; specialist inputs for successful operation, and in any ˇ the absence of any assurance of follow-up credit case its promotion may not be consistent with the is a disincentive to repayment; most immediate needs of post-crisis reconstruction. ˇ failure to recycle credit promptly reduces the As the experience of project implementation showed, multiplier effect on economic activity of a given the credit elements of CEP probably caused amount of loan capital; operational difficulties and diverted staff to a degree ˇ the contamination of non-repayment spreads disproportionate to their weight in total spending. rapidly if immediate and effective action is not taken; Whenever credit is included in a package of ˇ microcredit is not well-adapted to the needs of assistance it must be taken seriously, in the sense that seasonal agriculture; and appropriate targets for repayment should be defined ˇ credit-management groups are organic entities and appropriate measures for collection determined. forged by training and sustained by mutual If this does not seem appropriate, consideration obligation, not arbitrary and administratively- should be given to employing outright grants, rather convenient assemblages. than loans, in order to finance activities necessary to SOCIAL DEVELOPMENT PAPERS Conflict Prevention & Reconstruction Paper No. 11/ March 2004 Timor-Leste: Independent Review of the Credit Component of the Community Empowerment Project John D. Conroy This Working Paper Series disseminates the findings of work in progress to encourage discussion and exchange of ideas on conflict and development issues. Papers in this series are not formal publications of the World Bank. The findings, interpretations, and conclusions are those of the authors and do not necessarily reflect the views of the World Bank Group, its Executive Directors, or the countries they represent. The papers carry the names of the authors and should be cited accordingly. The series is edited by the Conflict Prevention and Reconstruction Unit in the Social Development Department of the Environmentally and Socially Sustainable Development Network of the World Bank. To request copies of the paper or more information on the series, please contact the CPR Unit at cpr@worldbank.org. Papers are also available on the CPR Unit's website: http://www.worldbank.org/conflict under publications. Printed on Recycled Paper Table of Contents Acronyms Foreword I. Introduction ........................................................................................................................................... 1 Project Background ........................................................................................................................... 1 Problems of Choice ........................................................................................................................... 1 The Objectives of CEP Credit ........................................................................................................... 2 CEP Credits in the Broader Economic Policy Context...................................................................... 2 Data Problems in the Analysis of CEP Credit................................................................................... 2 II. Assessment of Objectives and Outputs............................................................................................... 4 Summary Achievement of Objectives............................................................................................... 4 Detailed Discussion of Outputs......................................................................................................... 5 Institutional Development Impact ..................................................................................................... 9 Poverty Impacts............................................................................................................................... 10 Sustainability of Revolving Funds................................................................................................... 11 World Bank Performance................................................................................................................ 11 Government Performance................................................................................................................ 12 III. Judgments ........................................................................................................................................ 13 Major Factors Affecting Implementation........................................................................................ 13 Comparison with Alternative Development Projects ...................................................................... 14 Financing the Revival of Economic Activity after Conflict............................................................ 15 Strengths and Weaknesses............................................................................................................... 17 Major Lessons Learned ................................................................................................................... 18 Specific Issues for Timor-Leste after CEP and SEP ....................................................................... 19 Annexes: 1: Field Interview Data............................................................................................................................... 21 2: Major Factors Affecting Implementation............................................................................................... 26 3: Data Sources and Methods of Investigation........................................................................................... 30 Tables: II.1: Initial CEP Lending by Region............................................................................................................. 6 II.2: CEP Lending and Repayments at end-August, 2003 ($)...................................................................... 6 III.1: CEP Credit Component: Summary Strengths and Weaknesses....................................................... 17 ii Acronyms ADB Asian Development Bank CEP Community Empowerment and Local Governance Project CFO Chief Financial Officer Conselho de Posto Sub-district Development Council Conselho de Suco Village Development Council ICR Implementation Completion Report KDP Kecamatan Development Project MFI Microfinance Institution MFIET Microfinance Institution of East Timor MFWG Microfinance Working Group MIS Management Information System NGO Non-governmental Organization PAD Project Appraisal Document Posto Sub-district PMU Project Management Unit SEP Small Enterprise Project Suco Village UNTAET United Nations Transitional Administration in East Timor UPK Unit Pengelolahan Keuangan (Sub-district Council Finance Unit) iii Foreword This independent review of the credit component of the East Timor Community Empowerment and Local Governance Project (CEP) is part of a broader effort to evaluate and draw lessons from the CEP experience in (now) Timor-Leste. The broader, independent evaluation of CEP, from which this paper is extracted, is being carried out under the direction of Jacqueline Pomeroy, Senior Social Scientist in the East Asia Social Development Unit. The paper was written by John Conroy and benefited from inputs from Jacqueline Pomeroy and Elisabeth Huybens, Timor-Leste Country Manager. The paper was edited by the CPR Unit. Ian Bannon Manager Conflict Prevention and Reconstruction Unit TIMOR-LESTE: INDEPENDENT REVIEW OF THE CREDIT COMPONENT OF THE COMMUNITY EMPOWERMENT PROJECT I. Introduction Project Background The Community Empowerment and Local Governance Project (CEP) was initiated in late 1999 and agreed upon with the United Nations Transitional Administration in East Timor (UNTAET) in response to the emergency situation in East Timor after the referendum and subsequent devastation that affected the country. CEP was funded through three separate Trust Fund agreements. Although there were some adjustments, the overall CEP objective remained "To strengthen local level social capital to build institutions that reduce poverty and support inclusive patterns of growth." A key objective of CEP was to support productive local economic activities through a credit component. This Review evaluates CEP's credit component. The remainder of this section considers the trade-offs that had to be faced initially, the objectives of the credit component, the broader policy context and data problems. The second section assesses objectives and outcomes, institutional and poverty impacts, sustainability and the performance of the Bank and Government. The third section analyzes factors that affected implementation, major lessons and policy recommendations beyond CEP. Field interview data are presented in Annex 1. Major factors affecting implementation are discussed in greater detail in Annex 2, and data sources and methods of investigation in Annex 3. Problems of Choice CEP I was a direct response to an emergency. Indeed it was the first formal project formulated by the international community to address the crisis of late 1999. A recent evaluation of the successes, problems and tradeoffs in the East Timor reconstruction program,1 points to the tradeoff between developing a coherent policy framework and quickly rebuilding infrastructure. "In East Timor it is notable that the sectors that have made more progress in building a coherent policy, sustainable institutions and strong levels of management capacity such as the health sector were less strong in the initial stages in achieving physical reconstruction targets" (p.iii). In education, by contrast, relatively rapid progress occurred with school reconstruction and re-enrollment of children, but progress in terms of the sector policy framework was slower. Similar problems of choice can be seen in various elements of CEP.2 In terms of providing credit, the tradeoff was between a need to inject liquidity into rural areas to stimulate economic activity for immediate relief of distress, on the one hand, and an implicit need for financial sector reconstruction, on the other. Attending to the latter would have required careful thought about the use of credit as a tool and the policy framework for its use. A similar choice had to be made in the World Bank's Small Enterprise Project (SEP), whose credit objectives were parallel to those of CEP, although with a more explicit financial sector orientation and with a financial institution as the lending agency. 1Klaus Rohland and Sarah Cliffe (2002), "The East Timor Reconstruction Program: Successes, Problems and Tradeoffs", CPR Working Paper, No. 2. Washington DC: World Bank. 2The term `CEP credit' is used in this report to denote the bridging set of credit activities spanning three successive projects, CEP I, II and III. Where a particular phase of CEP is discussed this will be made clear, but the term `CEP' will be used to refer to the project as a whole. 2 The Objectives of CEP Credit The outputs expected of CEP credits were loans and the generation of productive economic activities. Implicit in this was the need to recapitalize microenterprises, in parallel with the labor-intensive reconstruction of community infrastructure under the subgrant program. These objectives were set in the broader context of alleviating poverty (Project Appraisal Document [PAD] I) and the development of district level capacities, specifically in planning development activities (PAD II). Regarding the benchmark for repayment of credit, the Grant Agreement for CEP I (February 2000) stated `more than 80%' as the performance indicator. No method for calculating this measure was specified. The Grant Agreement for CEP II revised the target down to 60%. However, the PAD for CEP I (May 2000) set no specific benchmark. Performance indicators listed for CEP II in the PAD of April 2001 included `amount of credits repaid', but again no specific target was set. By contrast, an explicit target repayment rate was set for SEP I, the first small enterprise project in Timor-Leste. Further, this benchmark rate was defined specifically, to be calculated as a proportion of amounts due at the date of calculation. It is worth noting that none of the documents refers to CEP credits as `microcredit' or uses the term `microfinance'. However, as the discussion in this Review makes clear, the project design was influenced by microfinance practice, specifically as concerns micro-lending. It is therefore appropriate to apply good practice standards of microfinance to reviewing CEP credit operations. CEP Credits in the Broader Economic Policy Context CEP was not itself a financial sector project, but its credit activities had implications for financial sector actors, including embryonic microfinance institutions and the commercial bank involved in SEP I, as well as for other commercial banks contemplating entry to the market in Timor-Leste. This is relevant to key issues such as its impact on the credit culture in rural areas, and the longer-run sustainability of rural financial services. The Timor-Leste National Development Plan3 calls for increasing numbers of private banks and financial institutions to be established, with a `widening rural presence'. It also urges government support for activities of donors and NGOs to develop `micro-savings and credit schemes', again with emphasis on rural outreach. This would assist in monetizing subsistence agriculture, supporting what the Timor-Leste Poverty Assessment described as a `crucial intervention' for raising rural incomes.4 This would be to encourage rural households to produce for the market and to shift into production of higher-value crops. Complementary to this would be the establishment or rehabilitation of periodic markets in rural communities, an initiative which CEP infrastructure subgrants were intended to support. Pioneering the provision of credit in rural areas of the post-conflict economy, and contributing to the creation of a sound credit culture, would support these objectives, as would the successful `graduation' of numbers of borrowers in villages across the nation. Data Problems in the Analysis of CEP Credit Before providing quantitative assessments of CEP credit, it is necessary to consider the nature and reliability of Project data sources. The Unit Pengelolaan Keuangan (UPKs), or Financial Management Units, are located at the subdistrict or posto level of CEP field administration, and are responsible for managing disbursements and collections in a number of villages (suco) located within the territory of the posto. The UPK is appointed by the Conselho de Posto, an elected body responsible for allocating funds for subgrants and credits after considering proposals forwarded from the various conselho de suco within 3Planning Commission (2002), East Timor National Development Plan. Dili. 4World Bank (2003), Timor-Leste, Poverty in a New Nation: Analysis for Action. Dili. 3 the subdistrict. The UPK is supposed to be independent of the Conselho and it reports all expenditures and repayments to the Project Management Unit (PMU). It is thus the foundation of the CEP management information system (MIS). Data sets recording CEP credit were built upon the initial recording of transactions at the UPKs and their entry into the CEP MIS. Data were sent from the UPKs via posto facilitators to district level, then to regional coordinators and finally to the MIS at the PMU in Dili. Good information depends on appropriate procedures, properly applied. For CEP credit, there is reason to question both the procedures and their application in practice. This has implications for the quality of the data reported to, and perhaps especially by, the CEP PMU. For CEP credit, there seems to have been little reliance on aggregated data as a tool for the analysis of Project progress. For example, the MIS had never been used to compile a table showing the geographic distribution of the original CEP loans, nor to track the volume of credit recycled by the CEP. The Mission found it possible to track and understand credit activities on the basis of UPK records, as reported up through the system from posto level. However, lack of uniformity in the data reported by posto, and as consolidated at district and region levels, suggested a lack of coordination and standardization of reporting. These deficiencies reflected poorly on the PMU's design and use of the MIS. The Chief Financial Officer (CFO) had been in place less than a year at the time of the Mission's first visit. Upon his arrival he was no doubt faced with a number of urgent issues. CEP credit was, in dollar terms at least, a relatively minor element in the totality of the Project and other concerns may have appeared more pressing. Capacity-building for staff in the quantitatively larger financial operations of the subgrant program had been neglected and required much attention. Personnel changes within the PMU also contributed to deficiencies in institutional memory. A review of the data-gathering for, and presentation of, credit information resulted in changes to the consolidated monthly credit reporting format from April 2003. These were making their way through the system between the first and second visits of the Mission.5 By the second visit it was apparent that the changes had not achieved the desired improvement. Nor was it obvious that anyone within the PMU was concerned with interpreting the new series of CEP credit data as reported monthly by the MIS, or that the officer in charge received any feedback from management on the reports he prepared. Finally, CEP credit suffered from conceptual confusion over appropriate measures of repayment which had their origin in Project design. This last point is discussed below, where the repayment rate for CEP credits is reported. However, the problems of CEP credit data go beyond the measurement and analysis of repayment. Other significant variables, including gender, could have been analyzed for program management purposes, but were not. These problems are traceable to a lack of strategic vision in the PMU. 5The consultant responsible for reviewing credit operations visited Timor-Leste twice, in June-July and October-November 2003. 4 II. Assessment of Objectives and Outputs Summary Achievement of Objectives According to best-estimate calculations by the Mission, the total value of loans made to end-August 2003, totaled $1.305 million. This consisted of first-round lending of $1.047 million, subsequent, `recycled', loans totaling $0.211 million and `new' loans during CEP III totaling $0.047 million. Recycling of loans was continuing at the time of the second mission visit. The number of loans made was 864 in the first round, perhaps 150-170 recycled loans to end-August 2003, and 22 new loans released during CEP III. This gave a total of perhaps 1,050 loans, with a mean value of around $1,240. Much of this lending permitted the recapitalization of previously existing microenterprises. Total repayments to end-August 2003 were recorded as $0.410 million, equivalent to 31% of the total advanced to that date. Project data do not permit calculation of collections as a proportion of repayments having fallen due to end-August. This would put a better light on collection performance (although it would still be below 40% in all probability). Some rescheduling of loan repayments had commenced from early 2003, although in cases examined by the Mission the results were not promising. If the outlook for SEP I, as reported by its Implementation Completion Report (ICR), was for about 60% final repayment, the final result for CEP lending is likely to be less than 40% at the conclusion of the Project.6 The number of loans made was substantially greater than originally expected. The value of individual loan proposals was typically well below the level set in Project guidelines. The guidelines took the concept of a community group, whose members would work together within the subgrant program for infrastructure, and extended it to group activities in the economic domain. In fact, many persons named as members of the groups to which loans were made were not active participants, and it is accordingly not possible to estimate the number of beneficiaries or their gender. In terms of generating productive economic activities, about 54% of initial loans were applied to establishing or recapitalizing small retail stores or market stalls. Primary industry financing accounted for a third of lending, and the balance, only 13% of the total, was spread between a number of other minor activities. Subsequent lending broadly reflects this initial pattern, with possibly an even greater proportion of loans applied to retailing. Thus the principal output of the Project was support for petty trading and the supply of formal sector, manufactured, goods to a rural economy in which many or most families are responsible for their own subsistence. This was an important intervention, both in the short- term situation of disrupted market supply after the crisis, and in a longer-term sense. Concerning the sustainability of economic activities financed, given that the great majority of loan- recipients were in arrears, the PMU had difficulty estimating the degree to which the businesses concerned were viable and active. The Mission visited a sample of posto and interviewed numerous borrowers, including a number in default. The broad conclusion was that the businesses were generally in better shape than the loan program that financed them. A clear majority seemed still to be operating. The ICR for SEP I came to a similar conclusion. And as with SEP I, it is fair to say that CEP credits contributed to a revival of private sector economic activities in rural areas, although other sources of private, profit-seeking capital also emerged along with the inputs of other agencies, including NGOs. Finally, considered simply as a credit operation, CEP did not meet any reasonable standard of sustainability, with collection rates even in the most well-managed subunits well below those considered 6At present (March 2004), the repayment rate for SEP I has reached 71%, due to persistent efforts by the Government. 5 necessary for sustainable microfinance (even apart from the issues of operating costs and interest income). Detailed Discussion of Outputs Number and type of income-generating activities supported Data are available to describe economic activities financed by the CEP. The initial round of Project lending financed the startup or recapitalization of some 864 micro-businesses, or groups of micro- businesses, in every district of Timor-Leste. Total disbursements amounted to over $1 million for these initial loans at an average of $1,212. This was a much smaller loan size than had been anticipated.7 Recycled loans totaled some $211,000 and financed between 150 and 170 group enterprises. `New' lending during CEP III financed another 22 groups, giving a total of perhaps 1,050 at a mean loan size of around $1,240. The new reporting format introduced from April 2003 was designed to provide a listing of activities financed by recycled loans, but the data were not available at the time of the mission visit. Instead it was necessary to estimate their number. The new data format was supposed to enable the total value of loans to be disaggregated by the gender of group-recipients, but this also had not been done. In round figures, more than half, or about 54%, of initial credit was applied to small retail stores or market stalls (kios). These were primarily concerned with the supply of manufactured goods and processed foods to the rural economy, rather than the distribution of local produce. Their economic significance is potentially considerable and is discussed below. Primary industry financing accounted for a third of lending, with 14% for agriculture, 13% for livestock and 6% for fishing. The balance, only 13% of the total, was spread between a number of other minor activities including transport, restaurants and various crafts. Retail stalls are located in marketplaces or dispersed in villages. Loans for primary industry included threshing and milling machines and hand tractors. These relatively large and indivisible assets were vulnerable to over-investment, against which the Project had no defense since it was not the only source of investment capital in many communities. Also these and some other loans were for seasonal activities whose income streams were ill-adapted to regular repayment schedules over an 18 month period. Examples are cited in the case study material presented in Annex 1. Livestock loans (for cattle, goats and poultry) were an important category and in some coastal areas loans financed nets and other fishing equipment. Sustainability of the businesses supported The PMU had some difficulty in assessing the extent to which the businesses concerned are viable and active. There is provision for posto monitors to report on the condition of businesses under the heading of Physical Targets (Economic Infrastructure), but this report (which clearly derives from the subgrant process for community infrastructure projects rather than any business analysis) is not returned by many or most UPKs. It provides for UPKs to report the progress of microenterprises under one or other of the headings baik (good), cukup (fair), or kurang (inadequate). There does not appear to be any consolidated reporting of this material for the Project as a whole. Annex 1 reports the impressions of the Mission gained from a purposive sample of posto and borrowers. The broad conclusion is that the businesses were generally in better shape than the loan program. The ICR for SEP I, which found that loan default was by no means the same thing as business failure, came to a similar conclusion. 7The original manual for fieldworkers (CEP, Manual Fasilitador Posto, 2000) stated a range of $5,000 to $25,000 for subgrants, to be applied also to economic credits. The revised financial administration manual issued in mid-2002 modified this to $2,500- $5,000. Loans within the original range would have put the CEP credit program in the same league as SEP I and the loan size was clearly unrealistic in light of CEP's objective to contribute to village economic regeneration. In fact, few loans fell within either of these ranges. 6 Thus in most of the eight posto visited a clear majority of enterprises were said to be still active, while others were dormant for seasonal reasons. This impression was strengthened by visits to a number of defaulters. Given that almost all businesses financed were household enterprises, only close examination of cashflows would enable an observer to determine the extent to which other sources of income were propping them up. The Mission came across instances where a household member had found wage- employment and the UPK believed this was assisting with the maintenance of payments and/or the continuation of the activity. Also, a number of successful retail enterprises had diversified into other activities (for example livestock rearing) and this had reduced their vulnerability to market conditions. On balance it seemed reasonable to conclude that most of the enterprises financed were still going concerns, and that they were valued by households as part of a strategy of diversifying sources of income. It may be that in some cases, as with those employing depreciating capital equipment, the longer-run viability of the enterprise is dubious. However, access to follow-up credit would make a difference to the sustainability of many of the economic activities in the CEP, and there is certainly a significant residual of economic activity that would benefit from such financing support if it were available. Credits disbursed Although lending commenced in the first half of 2001, CEP produced consolidated lending and repayment data at the national level for the first time only in May 2002. But by that time some recycling had commenced so that it was never exactly clear just how the original loans had been allocated, on a geographical basis. The Mission's estimates (Table II.1) fill that knowledge gap. There had been no single-point-of-time data series produced to chart the growth of lending and repayment, and none of the periodic Project Aide Memoires had been able to cite such data. Table II.1: Initial CEP Lending by Region Region No. of Groups Amount Advanced ($) Eastern 138 156,919 Central 466 553,693 Western 122 170,287 Autonomous 138 166,160 TOTAL 864 1,047,059 At the time of the first credit Mission, an attempt to produce a national level time series of lending and repayment data had commenced (from April 2003). An inspection of the completed returns suggested some conceptual difficulties experienced by the PMU, and/or clerical errors in data entry. The second credit Mission calculated consolidated data for end-August 2003. By this time it was possible to distinguish between total lending, recycled loans, and new lending amounting to $46,549 which had occurred during CEP III. Table II.2 presents the data, compiled from regional- and district-level reports, as a best-estimate on the information available. The Mission recommended to the PMU that it should present monthly reports in this format in the months remaining to completion. Table II.2: CEP Lending and Repayments at end-August, 2003 ($) Total Recycled New Loans Repayment Region Loans Repayments Loans CEP III Rate (%) Central 202,901 56,853 34,982 11,000 28.0 Eastern 682,434 214,710 120,041 8,700 31.5 Western 226,445 66,240 29,309 26,849 29.3 Autonomous 193,032 72,213 26,872 -- 37.4 TOTAL 1,304,032 410,016 211,204 46,549 31.4 Note: Repayment rate is the proportion of repayments to date to total loans. 7 Broadly, by end-August, total lending had grown to some $1.3 million, from the original base of $1.0 million, with repayments to that point of $0.4 million, or around 31%. About $0.2 million of the funds repaid had been recycled. The Project has no crosscheck procedure for reconciling UPK cash balances with reported figures for credit and other expenditures, since cashflow is recorded at posto level in simple double column cashbooks, without disaggregation. Repayment rates The repayment rate of 31% cited in Table II.2, above, is a simple and rather misleading measure, for reasons explained below. On this measure there is little to choose between the Central, Eastern and Western regions, each with rates between 28% and 31%, although there is considerable internal variation at the district and posto levels, as the field studies described in Annex 1 suggest. This diversity may reflect the degree of decentralized decision-making built into Project design, with responsibility for loan approvals residing with conselho, and for collection with UPKs, at the local level. The Autonomous region, comprising two rural posto attached to urban Dili and the enclave of Oecussi, has a better performance, at 37%. There was no lending in urban Dili. The measures of repayment reported by the Project and discussed above are misleading, because they are calculated with total lending to date as the denominator and amount repaid to date as the numerator. Thus the denominator includes installments on loans which have not fallen due at the date of calculation. This is a design flaw traceable to confusion in the CEO I Grant Agreement of 2000, and the PADs for CEP I and II, as described above. An estimate in which the denominator is loans due to date, such as was specified for SEP I, would have been a more appropriate measure, even if still rather limited for management purposes. It describes cumulative collection performance over time. The ICR for SEP I was critical of the Project's failure to measure, in addition, current repayment performance. A review of credit operations annexed to the CEP mid-term review (August 2001) suggested that CEP cumulative repayment rates be calculated in the same way as those of SEP I. This procedure was adopted in a PMU report dated May 2002, but soon after, with the publication of the CEP III Finance Management Manual, the MIS reverted to the former measure. Unfortunately this was retained for the new reporting format adopted from April 2003. Even then, the Project still had a standard form, Kontrol Kredit/Pinjaman (Credit Control) which had provision to record the aging of overdue loans. Some posto monitors continued to provide this information in their monthly returns in mid-2003, but most did not. Making full and regular use of this data source would have provided at least a rudimentary early warning system for credit arrears for the PMU. Revolving funds and second round credits Data reported in Table II.2 suggest that recycled funds amounted to $0.211 million, on initial lending of $1.047 million at end-August 2003. This onlending was equivalent to 20.2% of the initial sum advanced and was financed by collections of slightly more than $0.4 million. Clearly this was a disappointing result if the intention of the revolving funds had been to stimulate economic activity by multiplying the supply of credit in rural areas. From data available prior to the introduction of new reporting procedures in April 2003 it was difficult to identify how many new groups had formed and where they were. The new reporting format did not improve this situation. In most places the relending totals are small. In the first round of field visits the Mission came across only one posto (Liquisa) where recycling had occurred, and two others where cash reserves were sufficient for the conselho to be actively considering it. Not surprisingly, Liquisa, in the Western region, 8 was also a well-performed posto in terms of repayment. In the meantime, an amount of some $0.23 million sat in safes around the country because conselho were not encouraged by the experience of first- round lending to recommence the process. During the second credit Mission, two posto in Covalima (Suai and Fatumean) with good repayment and relending records were visited, to shed light on some better examples of lending in the CEP. Both UPKs had recycled funds actively, with Suai in particular pursuing a policy of relending as soon as funds permitted and favoring well-performed borrowers with second loans. This contrasts with practice in some other regions where second loans have been issued more on an entitlement basis, to new borrowers, rather than on the basis of performance within CEP. Suai had made 15 new loans in three tranches as funds accumulated, compared with 33 in the initial round of lending. This means that some funds were in their fourth round of circulation, which is the closest any posto had come to realizing the CEP vision of a revolving fund to multiply the circulation of capital in the post-conflict economy. Fatumean had made 6 new loans in two tranches, as against 13 in the initial round. In both posto, rates of repayment for the new loans appeared encouraging. Beneficiaries and benefits CEP has no records of beneficiaries other than the names of persons listed as the members of each borrowing group. A priori, one would expect enterprises financed in the social and economic setting of Timor-Leste, with loan sizes averaging around $1,200, to be organized on a household basis. Impressions gained during field visits supported this assumption and the notion of group enterprises proved to be something of a fiction in many, perhaps even a majority of cases, as discussed below. The implication of this is that CEP should not have expected formal employment benefits from such a program of lending. Benefits should be expected instead in terms of income-generation. This was clearly the intention of the PADs for CEP I and II, which spoke in terms of financing productive activities rather than job-creation. As mentioned above, the Timor-Leste Poverty Assessment identified crucial interventions for raising rural incomes, among which encouraging the heads of rural households to shift out of full-time agriculture, to become produce traders, is the most potent. Posto Bazartete is a rural community discussed in Annex 1, which illustrates the potential of this strategy. The most notable feature of the distribution of CEP credit in Bazartete (as in Suai and some other posto visited) appeared to be its importance in supporting petty trading and the supply of formal sector, manufactured, goods to a rural economy in which many or most families are responsible for their own subsistence production. Such traders underpin the operations of local markets in villages and subdistrict centers. Most other market traders are farmers who come to sell small amounts of seasonal produce they have produced themselves, but who also come to buy. The availability of formal sector consumer goods in those marketplaces and in village kios stimulates the production and marketing of subsistence foodstuffs and the production of higher-value crops. This was particularly noticeable in Bazartete, where the most significant loan was made to a group with ten member families, all of whom were stallholders in Bazartete market. Loans ranged from $500 to $2,500 with most toward the lower end of that scale. Reflecting these levels of working capital, merchandise stocked by traders ranged from small collections of sembako (the most basic of household necessities: rice, cooking oil, soap, etc.) to more varied offerings of plastic utensils, brushware, and canned foods and beverages. The marketplace was still only partly reconstructed and the tenure of stallholders had not been secured. Some had erected structures on the sites they occupied, using permanent building materials and making cash investments of hundreds of dollars. All lived within the suco in which the market is located and felt that their residence, occupation of the site and site investments would help to secure their tenure once 9 local authorities were in a position to address the issue. To the extent that they are assisted to maintain themselves in the market while the issue of tenure is resolved, CEP credit may prove significant for these people, in addition to the beneficial impact of their trading on the surrounding agricultural economy. Institutional Development Impact The credit component of CEP is one of three interventions by the international community with implications for financial sector development and the credit culture in Timor-Leste. The others were the Bank's SEP I and the Asian Development Bank's (ADB) Microfinance Institution, both of which were also lending operations. `Credit culture' is best described as the set of expectations within a community as to the behavior appropriate for borrowers. It is the result of a complex set of incentives and disincentives acting upon borrowers. These incentives and disincentives are created by the expectations and practices of lenders, the legal environment in which lending is conducted, and the economic environment for investments financed by credit. The expectations of lenders are important because a low level of lender expectation is conveyed subtly to staff and borrowers in myriad ways. Thus ambiguities in the Grant Agreements and PADs about expected levels of repayment (discussed above) probably acted to de-emphasize repayment as a Project goal. The practices of CEP in regard to record-keeping, collection and follow-up of arrears were also not conducive to high repayment rates and it is probable that this communicated a message to borrowers. Records were not produced in timely fashion, nor in a manner permitting the analysis of trends in repayment. Repayments were collected monthly (rather than weekly as in the case of most microfinance lenders) and the schedule called for first payments to be made after a period of two or more months grace in many cases. Follow-ups do not seem to have been triggered until several payments had been missed, by which time the momentum of default was often unstoppable. For clients the effect must have been to devalue the significance of the contract. Further, the example of deliberate default in at least some proportion of loans cannot have been conducive to institutionalizing a positive credit culture. In terms of legal environment, most borrowing was conducted without any guarantees or offer of collateral. In some cases where informal pledges of land or other property were exacted, no legal right of enforcement existed so that any such pledge was without effect. The economic environment was not particularly conducive to repayment (for example, the necessity to convert loan proceeds to rupiah imposed capital losses on borrowers and no doubt led to resentment, while the deflation of the urban `bubble economy' during 2002 had spread-effects into the rural economy). But it is the business of a lender to maintain credit discipline in the face of economic shocks, even to the point of rescheduling, if necessary. Rescheduling does not seem to have commenced in CEP until contracts had expired, whereas good practice would be to canvass this option at the point where the borrower's difficulties had become obvious. In a number of cases observed by the Mission, rescheduling had been agreed with borrowers but repayment had either not resumed or had rapidly tailed off, once again raising questions about the Project's will to enforce contracts. This catalogue of errors, similar to but more serious than those observed in SEP I, cannot have supported the emergence of a good credit culture in rural Timor-Leste.8 Mitigating this is the fact that the Project was seen as a Government/World Bank operation, so that it remains open for commercial lenders and microfinance institutions to try to convey a different set of signals to the populations previously served by CEP credit, and to achieve a more positive credit culture among them. 8The founder of an NGO microfinance program operating in Maliana and Cailako subdistricts wrote that `In terms of establishing a culture of repayment, I think that the credit operations of CEP have made it more difficult for other microfinance providers'. 10 Apart from the potentially negative impact on credit culture, there is little evidence of positive institutional development impact. The model of credit allocation decisions being made by elected bodies, with disbursements and repayments administered by officials in a non-commercial environment, was in any case the product of an emergency situation. It is not a model which should continue to operate after normal economic activity has resumed and the regulatory environment for a financial system has been put in place, since it cannot contribute to renewal and expansion of the financial system. The experience did have the merit of demonstrating the existence of effective demand for loans, albeit at the highly concessional interest rate of 10% per annum on the reducing balance. There was also a minority of borrowers who had the beneficial experience of employing borrowed capital productively and successfully servicing a loan. These individuals can be recommended to other lenders. Individuals in UPKs and conselho probably benefited from the experience of decision-making and administration in regard to credit. There were also some isolated cases where UPKs and conselho, as institutions, appear to have performed creditably. These could provide the foundation for some further experiments in decentralized community revolving funds, or the skills acquired can be employed in a different institutional setting. The PMU was relatively isolated from Government. It will be unfortunate if lessons concerning the inability of governmental or quasi-governmental agencies to manage credit are not communicated emphatically to the Government. Poverty Impacts The PAD for CEP I expected outputs in terms of credits issued and generation of economic activities, and set these in a broader context of poverty alleviation in the post-conflict situation. CEP credit did not set out to target the poorest, as some microfinance institutions, including those conducting Grameen Bank replications, have attempted to do in Timor-Leste. A person associated with one such institution commented that "[in] end-2000 and early 2001, there was a lot of expectation in Cailako about the availability of credit from "banku mondial"...But in the event, very few poor women got CEP credit in [Cailako]. In the other subdistricts as well, we experienced very little overlap between the households which qualify under our House Index and Asset test...and the households which received CEP credit". The test referred to, assesses a woman's poverty by the quality of her housing and the assets she possesses. The former East Timor was among the poorest provinces in Indonesia, and suffered severe structural problems of poverty. Other, non-credit elements in CEP, and particularly those dealing with governance, were concerned with laying foundations to address those structural problems. But the immediate consequences of the devastation imposed poverty on an even larger proportion of the population, demonstrating the vulnerability of many above the poverty line. Much of this larger poverty was in principle amenable to relief through immediate economic stimulus. CEP credit had the task of relieving the immediate poverty consequences of the 1999 devastation through regenerating economic activity. This was to be done by injecting liquidity into rural communities in parallel with their engagement in labor-intensive infrastructure works which injected spending power into those same communities. CEP was concerned with poverty alleviation in the face of widespread distress in late 1999 and 2000. But, as mentioned above, CEP credit did not set out to target the poorest (although other Project elements targeting vulnerable groups certainly did). Rather, much CEP credit went to recapitalizing microenterprises damaged or destroyed by violence. While some credit went to assist in the restoration of rural production, more went to restoring the supply of manufactured consumer goods and processed foods in subsistence agricultural communities where production had been disrupted. And, as discussed above, such a strategy also served the longer-term goal of encouraging a more diversified agriculture as well as relieving immediate shortages. 11 CEP has very limited information about Project beneficiaries and their characteristics, as reported above. Nor had it conducted any impact surveys which might suggest poverty alleviation outcomes. However, from the perspective of poverty alleviation, one of the most glaring deficiencies in its reporting is the failure to prepare gender-disaggregated information concerning beneficiaries. This failure is difficult to justify in a project commencing in the year 2000 and which emphasized gender considerations in its design. It prevented the Project from forming any conclusions about the impact of its credit on women, a group particularly vulnerable to poverty. Sustainability of Revolving Funds The issue here is the sustainability of CEP revolving funds as financial mechanisms. The most important consideration is that loan losses are likely to constitute at least 60% of loan funds disbursed to end-August 2003. Even relatively well performing posto, such as Suai and Fatumean in Covalima district, would exhaust their capital after just a few rounds of lending at current levels of performance. This is without consideration of the level of costs incurred by the PMU and the expenses and allowances of UPKs and conselho members. These burden the Project with unit costs of lending much greater than would be borne by an efficient microfinance institution (MFI) handling the same volume of funds. To achieve operational sustainability the same MFI would find it necessary to charge effective interest rates at least two or three times as high as the 10% (reducing balance) rate levied on CEP credit, and even more if an imputed cost of funds were applied to the capital employed.9 If it is clear that CEP revolving funds are not sustainable in a financial sense, then perhaps it is worth asking whether sustainability was really an issue for CEP. Given the emergency situation which gave rise to CEP credit, and the urgent need to restore liquidity and economic activity in rural areas, the primary objective should perhaps have been to get money out quickly to those best able to use it in the `generation of productive economic activities'. Simple grants might have been better options for this immediate purpose. Credit (as opposed to grants) offers the possibility of further multiplying capital through recycling, but in CEP only posto Suai, as discussed above, even approached the CEP credit vision of active recirculation of loan capital. These issues are discussed further below. World Bank Performance In difficult circumstances after the events of late 1999, the Bank contributed to identifying a coherent set of credit activities, one of which was CEP credit. While each of these projects had its own emphases, as a suite they were designed to restart economic activity and re-establish elements of a financial system. To understand choices that were made at that time it is necessary to recall the severity of the crisis, both economic and civil, which the former East Timor was undergoing. A comment quoted in the Introduction concerned the tradeoff involved in developing a coherent policy framework in the face of an imperative for action. World Bank performance has to be judged with proper consciousness of that dilemma. In regard to CEP credit, the Bank developed a mechanism in which lending for economic activities was grafted on to a community-based grant mechanism, making use of the pre-existing framework and procedures of the Bank-funded Kecamatan Development Project (KDP) in Indonesia. As an emergency response the project showed resourcefulness, although perhaps too much reliance was placed on generalist staff for project preparation, where specialist inputs could have been justified. The result was a set of activities in which there were elements of internal inconsistency. 9The concepts of operational and financial sustainability of microfinance institutions are discussed on the website of CGAP (the Consultative Group to Assist the Poorest) at www.cgap.org . 12 One such was that the complex task of establishing local governance structures involved some confusion between goals and instruments. The governance structure was not an instrument for achieving credit goals, rather it was a goal in its own right and therefore not necessarily optimal for the achievement of credit objectives. In addition, a number of operational practices in the credit component fell well short of established good practice. These flaws appear traceable to design `on the run' by generalists during the first year of operation, in 2000. One example was the conceptual problems associated with the definition and targets for repayment in the PADs for CEP I and II. These proved to have implications both for the usefulness of the Project MIS and for the priority given by the PMU to securing repayment. A second instance concerns the `relief' objective of CEP--that of reviving private microenterprise activity by means of credit. This goal was not entirely consistent with the project's implicit financial sector reconstruction objective. Indeed there is an important question whether the introduction of credit was premature in the immediate post-conflict environment. This was the conclusion reached by the Bank's ICR for SEP I and this Review concurs in the case of CEP credit. The Bank Task Team's supervision of CEP I, II, and III is recorded in 11 Aide Memoires between December 1999 and July 2003, which gave varying degrees of attention to credit. Periodic financial management reviews occurred and these also devoted some attention to the administration of credit and repayment. However, prior to the mid-term review in August 2001, which was the occasion of a detailed review of the credit component, no financial sector or microfinance specialist appears to have participated in any IDA mission. The credit review occurred after first round lending had been substantially completed. Hence its focus was primarily on repayment and recycling issues, and on strategies for the eventual transition of CEP borrowers to orthodox sources of microfinance. Even after the mid-term review in August 2001, several Aide Memoires failed to give the credit issue serious attention. There is therefore a question as to the timeliness of Bank supervision inputs to the credit component. Earlier intervention by a microfinance specialist, perhaps late in 2000 when the lack of preparation for credit operations was becoming apparent, could have had beneficial impact on lending operations. Government Performance As the first World Bank project to commence under the Trust Fund arrangements, the project was initially implemented by CAA/Oxfam under contract to UNTAET, and subsequently by a PMU. Government responsibility passed from UNTAET to the Ministry of Internal Administration, and more recently to the Ministry of State Administration (Division of Territorial Administration and Local Government). Working relationships between the Government and the PMU have not been close, and it is not clear how well-informed Government has been about the credit program. The PMU has lacked expertise in financial and, especially, credit management from the beginning of CEP I and this contributed to its failure to design and implement adequate systems and records. Credit activities have in any case had lower priority than the subgrant program, which is substantially larger in terms of expenditure. So far as credit is concerned, senior management of the PMU does not appear to have regarded the MIS as a tool for decision-making nor to have required it to produce coherent program data. The appointment of an international CFO was delayed until December 2000, when the UNTAET- appointed Finance Manager was promoted to the job. This was shortly before lending was to commence but he appears to have given immediate priority to the larger issues of financial management associated with the community subgrants. In the event, a financial management manual for the project with procedures for credit did not appear until mid-2002, after initial lending had been substantially completed. In January 2002 the possibility of appointing the Timorese Deputy CFO to the post was considered and rejected. The World Bank financial management specialist recommended instead the early appointment of an expatriate successor to the CFO to permit a minimum three to four month overlap with the departing 13 CFO. This time would have been used to design improved financial management systems for UPKs, including for credit operations, which the Bank's financial management specialist regarded as an urgent priority. In the event, the appointment was not made until late July 2002. The overlap between the appointee and his predecessor, which might have permitted these objectives to be secured, was not achieved. The Project thus suffered from discontinuities and the second CFO has not been able to apply himself to the credit program in any detail, although unlike his predecessor he has sought to familiarize himself with field operations. Overall, the PMU could fairly be characterized as lacking both leadership and strategic vision in regard to CEP credit. This is apparent from the process of ad hoc and unsupervised tinkering with the reporting system which has not produced information dividends, probably due to lack of supervision and feedback to the officer concerned. Moreover, a perceptible slackening in PMU pace and commitment was noted between the two visits by the credit Mission in the second half of 2003. Without special effort the Project is likely to finish without producing convincing data to document its achievement of CEP credit objectives, just as it has consistently failed to produce such information to guide management decisions throughout the implementation phase. III. Judgments Major Factors Affecting Implementation A number of factors can be identified as having exerted significant and largely negative effects on the implementation of CEP credit. A detailed discussion of these factors is contained in Annex 2, and include: ˇ the adoption, with relatively little modification, of a pre-existing model for service delivery, KDP, which the World Bank had commenced to implement in Indonesia, including the then East Timor, in 1998. Bank evaluations of KDP have shown a range of factors contributing to poor credit performance, all of which find their counterparts in the experience of CEP in Timor-Leste. Even more fundamental is the criticism that while participatory planning and decision-making are appropriate methods for decision-making in relation to public goods such as CEP infrastructure, they are incompatible with sound credit management. ˇ CEP followed KDP in adopting groups as the basic unit of organization for credit. This appears to have been based on a misunderstanding of the group methods employed in classic microcredit operations such as Grameen Bank and ignored the fact that the basic unit of economic activity in the communities served by CEP credit was always going to be the household. Many, perhaps most, of the groups formed for CEP credit did not operate as intended, and a proportion were merely fronts for the operations of individuals. Under these circumstances, the Project often had difficulty securing accountability from the groups which were nominally responsible for loans. ˇ Human capacity deficiencies were the Achilles heel of CEP credit, and were never addressed adequately. The definition of procedures and the training for their implementation seems always to have lagged behind practice. Thus the first financial management manual with a section on credit appeared only in mid-2002, when some three-quarters of initial lending had already been completed. Training for credit operations seems always to have had lower priority than training for general financial management, and the staff of UPKs, who form the first line of administration and reporting for credit, were particularly neglected in this regard. However deficiencies in understanding and capacity were not confined to this level. For example, a deficient appreciation of the importance of data collection and analysis for program management appears to extend to the higher levels of the PMU. ˇ Problems of repayment have been damaging for CEP credit operations. External factors such as the instability of exchange rates and the macroeconomy in 2001-02 made life difficult for borrowers, 14 affecting their capacity to service loans and reducing repayment rates. But internal problems, stemming from a lack of technical expertise which led to flaws in credit program design, and exacerbated by the capacity deficiencies of UPKs and staff, have been even more damaging. By comparison with MFIs which work hard to imprint a sense of obligation in their borrowers, CEP has failed to implement good practice procedures to secure repayment rates in line with even the modest expectations of the project design. Comparison with Alternative Development Projects With the joint concerns of financial sector reconstruction and economic revival in mind, the international community prepared three related projects for the immediate post-conflict period.10 Apart from CEP with its credit component aimed at rural borrowers organized in groups, the World Bank prepared SEP, which contained an element of small individual loans for private business, most of them located in or near district-level towns. It provided loan capital of $4.0 million, roughly four times as large as the initial lending of CEP. Like CEP this was aimed at establishing a revolving fund facility, but unlike it, SEP credit was implemented by a financial institution, Bank CGD. It was aimed at `small' as distinct from `micro' enterprise and contained provision for business development services directed to the needs of borrowers. As in the case of CEP credit, risk was divorced from the responsibility for credit decisions. The implementing bank risked no own-capital but stood to profit from commissions. The interest rate on SEP lending was the same as for CEP, 10% per annum on the reducing balance. In the event, disbursement proved slow (though not so slow as in CEP) and the Bank failed to make any second round loans. Loan losses were about 40%, while business development services were delivered in untimely fashion and to only a minority of borrowers. As against this, the project was judged to have had significant benefits in terms of economic regeneration and in the `information dividend' earned by Bank CGD, which was thought to have gained valuable experience of lending in an unfamiliar market. The other related initiative was the ADB Microfinance Development project, which established a regulated financial institution, the Microfinance Institution of East Timor (MFIET). It commenced operations only in May 2002, having had to surmount legislative and regulatory hurdles. It operates under a limited banking license empowering it to accept deposits, make loans (with a minimum 65% of portfolio devoted to microcredit), provide payment and collection services, and to provide current (checking) account services. It is a member of the inter-bank clearing system but may not describe itself as a bank. MFIET has targeted a largely urban microfinance market segment. While it makes loans as large as $5,000, the bulk of its lending is in amounts around $100 or less. Loan quality is good, with portfolio at risk at 2.34% as of March 2003. For micro-lending, loan officers form groups of 4 to 8 people for training and loan administration, which is a costly and time-consuming process. These loans start at $50, have a term of 16 weeks at an interest rate of 1.5% flat per four week period, and weekly repayments. This is equivalent to an annual effective rate of 21.5%. There is another category of loans from $200- 500, catering to the working capital needs of market vendors. It had mobilized deposits of some $0.6 million as against loans of $0.4 million as of March 2003. It intends to offer wholesaling facilities for MFIs to enable them to expand their lending. MFIET was placed to provide banking and payment services to CEP in one of its rural branches and it is unfortunate that this opportunity was not taken up by CEP 10Although it was beyond the scope of the Mission to analyze in detail the performance of other, NGO-sponsored credit operations which started to emerge in mid-2001, it appears that these operations have grown strongly and are reported to have high repayment rates. 15 This detailed account of MFIET enables comparisons between its operations and those of the other two post-conflict credit projects. It differs importantly from both in that it: ˇ attempted to embed its operations in a policy framework, for which the tradeoff was delayed startup; ˇ attempted to create an enduring institution, rather than a `project'; ˇ risks its own capital in making lending decisions; ˇ applies group lending techniques genuinely, as employed in microcredit; ˇ levies an interest rate designed to assure financial sustainability; and ˇ provides deposit services, mobilizing funds that can be recycled into lending as circumstances permit. All of these characteristics distinguish MFIET from the more immediately relief-focused activities of SEP and CEP and illustrate some of the themes touched on in this Review, especially in the next section. These include the tradeoff between speed and institutionalization, the painstaking preparations necessary for sustainable credit, and the difficulty of combining credit with `relief' in the absence of appropriate institutions. Financing the Revival of Economic Activity after Conflict The international community now has considerable experience of financing post-conflict revival of private economic activities. Microfinance has often been employed for this purpose. However, many such initiatives occurred in situations where the pre-conflict economy possessed banking infrastructure and/or operating microfinance institutions. Correspondingly, there were human resources available with necessary skills or the educational background to assimilate them. The rural areas of East Timor were markedly deficient in either of those prerequisites. Hence much of the discussion of how banking and/or microfinance programs might best be revived in post-conflict situations is of limited relevance for Timor- Leste. Also, the discussion usually assumes that post-conflict microfinance will be attempted in environments where population density is sufficient to enable low-cost operation. But Timor-Leste is a country whose relatively low population densities and poor transport and communications would pose difficulties for microfinance under any circumstances. The international literature on post-conflict microfinance is largely concerned with the requirements of non-government actors (MFIs or other financial institutions). It does not assume that government will step in to establish microfinance mechanisms, although it does consider how donors might support international and indigenous NGOs to conduct microfinance. The literature warns against two sources of confusion. First is the confusion between support for relief (in the early stages, when grants may be the appropriate mechanism) and the support for development (as an ongoing activity requiring financial services, among other interventions). Second is confusion arising from the conflation of microfinance with non-economic (political or social) objectives. The decision to recapitalize local rural economies was a correct one, given the economic conditions of late 1999. The question is whether credit, administered according to CEP terms and procedures, was the appropriate mechanism. In this context, the dilemmas of choice faced by the Bank in 1999-00 should be remembered. In the interests of speed, priority was given to providing credit via government channels (since non-government mechanisms did not exist or were too thin on the ground). Unfortunately this decision confused relief (which was needed) with development (which was probably premature). Also, the decision was made to employ community-based mechanisms whose success was itself an end for the project, rather than simply a means for the achievement of infrastructure and credit goals. Thus both sources of confusion described in the previous paragraph were present in the mechanisms of CEP credit. As to the relief/development distinction, microfinance instituted too early may be confused by the population with relief measures such as grants. Thus CEP credit was promised from 2000, and was 16 disbursed together with infrastructure subgrants from early in 2001. There is evidence of confusion in the minds of borrowers as to the distinction between the two, especially since they were administered by the same staff through the same channels. Similarly, CEP credit, along with the subgrant program, was loaded up with an administrative structure designed to assume a widening range of local-level responsibilities in an independent Timor-Leste. CEP credit was to provide a demonstration of the potential of this administrative structure and the governance mechanisms that underlay it. Credit had to be made to work within this structure and with a particular model of governance, rather than having a structure and governance designed for its own needs. Credit was inappropriate as an early intervention, especially when administered via a government or quasi-government entity. The issue of what interest-rate was appropriate is therefore largely irrelevant, although the rate set (10% on the reducing balance) was quite `soft' under the circumstances and could be said to contain a grant element. Absolute grants, coupled with the creation of a supportive environment for the entry of MFIs, and including support for capacity-building of their staff, would have been a better way to handle the recapitalisation of rural microenterprise. Such grants could have been administered by a competitive process in which proposals were judged for feasibility within the governance structure set up by CEP. Meanwhile, MFIs would enter the market according to their own capacities and would select poorer borrowers able to service conventional microfinance loans. Grants in the same size range as the loans disbursed by CEP would have been appropriate, allowing flexibility for larger capital sums for individuals and for smaller amounts when shared between group members. CEP credit did not target the very poor and neither should a grant scheme have done so, given the purpose of the grants, and especially since there were other mechanisms to deal with acute food and shelter issues. It would have been important, psychologically, to establish that grants were simply a phase in the transition from conflict, and to establish a widely publicized `sunset clause' for their availability. This would have established the presumption that credit would resume as the normal source of investment funds, and also minimize the risk that grants would undercut emerging microcredit programs. International experience suggests a role for MFIs to provide savings services in post-conflict situations. This can assist households to cope with volatility in income and food supplies, and eventually permit the revival of household investment. But a functioning banking system, for which MFIs might perform an agency role, is necessary to safeguard the savings of depositors. Managing liquidity to meet surges in demand for cash is another problem, for which CEP staff would have been ill-prepared. In the absence of a legal entitlement to accept savings, and given the absence of banking facilities outside Dili, it would not have been appropriate for CEP to offer savings services. Government policy is now to extend a range of deposit-taking institutions as widely as possible in rural areas. This would be an appropriate course of action, especially if regulated financial institutions enter into partnership or agency relationships with MFIs to extend the reach of safe and liquid deposit services. 17 Strengths and Weaknesses The major strengths and weaknesses of the CEP credit component are summarized in Table III.1. Table III.1: CEP Credit Component: Summary Strengths and Weaknesses Design Implementation Strengths ˇ Goal of reviving economic activity was ˇ Significant outcomes achieved in micro- appropriate to circumstances. enterprise revival/development, quite distinct from issues of repayment. Assisted in stimulating and re-monetizing agricultural production. Some borrowers gained experience of successfully servicing loans with potential to secure credit rating. Demonstrated effective demand for credit albeit at low nominal interest rate. ˇ CEP seized opportunity presented by pre- ˇ Provided initial platform, though without prior existing KDP framework and staff. KDP credit experience. ˇ Allocated genuine responsibility to ˇ Individuals gained transferable administrative community representatives. and political skills. In best cases communities were pro-active in support of project goals. Recycling performance better than SEP I credit. ˇ CEP placed Timorese in positions of ˇ Many PMU members accepted responsibility, responsibility, designed modes of community and grew in capacity despite very limited consultation, negotiation. preparation. Weaknesses ˇ Emergency response not embedded in a policy ˇ Insufficient attention to financial sector framework. implications. ˇ KDP model applied uncritically. ˇ Failure to modify or discard inappropriate KDP elements. ˇ CEP applied a `development' measure (credit) ˇ Many repayment difficulties traceable to in a `relief' situation. premature use of credit. ˇ Local administration and governance were ˇ Sound credit analysis not always basis for designed as project outcomes, rather than as decisions, administrative arrangements not inputs to credit component. designed to facilitate credit. ˇ Credit component was compromised by ˇ Credit instrument not well-adapted to pursue social/political goals. multiple goals. Some role-conflict experienced by individuals performing multiple functions. ˇ Credit decision-making was divorced from ˇ Political and interpersonal factors influenced risk. allocation. Not enough `teeth' or will to enforce payment. ˇ Ambiguity existed concerning repayment ˇ Contributed to poor collection rates, targets. analytical management discouraged. ˇ Credit procedures as designed were not good ˇ Final repayment likely to be only 30-40% of practice. loan capital. Potentially negative impact on credit culture. Negative impact on unsubsidized MFIs. Conselho hesitant to authorize recycling (though performance better than SEP I). ˇ Group methods were misinterpreted in the ˇ Group solidarity benefits not fully realized, design peer pressure mechanisms not effective, some groups manipulated. 18 ˇ MIS design was ad hoc, and designed `on the ˇ MIS not used to inform management on run'. achievement of goals, nor to support operational decision-making. ˇ Capacity building was not integral. ˇ Capacity deficiencies were pervasive and affected performance negatively. Major Lessons Learned In terms of applying lessons from the experience of CEP credit to analogous situations in future, a number of more general principles may be drawn from the experience of this Project. These include that, while CEP was not a financial sector project, its use of credit to achieve goals for post-conflict reconstruction had implications for the revival of financial services in Timor-Leste. Since the latter was also an important goal of reconstruction, CEP had an obligation to avoid compromising the reconstruction of financial services. While a badly-conducted credit program may succeed in reviving business activity, it might do so at an unacceptably high cost in terms of negative impact on the credit culture. Whether significant negative impact has occurred in Timor-Leste as a result of the poor repayment performance of the CEP and SEP projects may only become apparent as MFIs and orthodox financial institutions extend their credit operations throughout the country in future. Credit was employed in CEP as an add-on feature, something that could be attached to the project to enhance its effectiveness. The amount allocated to credits for economic activities was only 13% of total subgrant expenditure in the project. But credit is not an accessory that can be added to a package with minimal preparation. It requires specialist inputs for successful operation, and in any case its promotion may not be consistent with the most immediate needs of post-crisis reconstruction. As the experience of project implementation showed, the credit elements of CEP probably caused operational difficulties and diverted staff and UPK energies to a degree disproportionate in relation to their weight in total spending. More broadly, it is generally not advisable to have one institution deal both with public infrastructure programs and credit programs, given that they require markedly different procedures and skills. Whenever credit is included in a package of assistance it must be taken seriously, in the sense that appropriate targets for repayment should be defined and appropriate measures for collection determined. If this does not seem appropriate, consideration should be given to employing outright grants, rather than loans, in order to finance activities deemed necessary for the revival of economic activity. Much of this funding would be applied to recapitalizing businesses which have been destroyed or de-stocked by conflict. Such grants should be given to micro-entrepreneurs on the basis of viable proposals evaluated on business principles. This should be done in the interest of economic revival, even though immediate benefits may accrue to private individuals rather than to the public. Moreover, even a grant scheme for the revival of economic activity would probably best be implemented through an institution separate from one specializing on infrastructure rehabilitation, as the appropriate procedures and skills of facilitators would likely remain very different. Related to the point above is the likelihood of severe economic fluctuations in the immediate post-crisis period (as for example in exchange rates and the level of economic activity in a `bubble' economy), which are likely to expose borrowers to unacceptably high risks and may produce outcomes contrary to those intended. This further strengthens the case for grants vis-a-vis loans. These factors point to the need to phase emergency stimulus for economic activity from grants to credit over some period of time. This time-period would need to be related to the revival of economic activity and the financial system, including commercial banks and MFIs. In this respect the experience of CEP further validates the conclusions on credit reached in the World Bank's ICR for SEP I. These called for a 19 transition from `intermediated grants' to `intermediated credit' over time, with the former administered by government or quasi-government entities and the latter by financial institutions. The experience of credit delivered via CEP offers no new lessons for good practice in microcredit. Rather it reinforces a number of lessons well-understood from experience elsewhere. It does, however, underline some limitations of microcredit in the immediate aftermath of crisis which bear upon the concerns of donors and the World Bank, including: ˇ where decision-making is divorced from risk, credit allocation is inefficient; ˇ subsidized credit tends to undercut credit delivered by institutions striving toward sustainability; ˇ the absence of any assurance of follow-up credit is a disincentive to repayment; ˇ failure to recycle credit promptly reduces the multiplier effect on economic activity of a given amount of loan capital; ˇ the contamination of non-repayment spreads rapidly if immediate and effective action is not taken; ˇ microcredit is not well-adapted to the needs of seasonal agriculture; and ˇ credit-management groups are organic entities forged by training and sustained by mutual obligation, not arbitrary and administratively-convenient assemblages. Specific Issues for Timor-Leste after CEP and SEP The Government's policy focus is rather different from the more-wide-ranging interests of donors and the World Bank. Nonetheless, the lessons outlined above, relating to timing the commencement of microfinance and good practice in its implementation, are relevant to Timor-Leste's future concerns. After the immediate crisis it now faces a stage of transition and rebuilding with the objective of achieving sustained growth. In the language used above, it has passed from `relief' to `development' and now has the task, outlined in its National Development Plan, of extending financial services sustainably and as widely as possible throughout the countryside. Regulated financial institutions will be slow to extend their operations outside major urban areas, although the possibility that Bank Rakyat Indonesia may reopen its Unit banking system in Timor-Leste will offer some scope for expansion. The conditions agreed for its return to Timor-Leste should be such as to assure services are provided in at least some locations outside urban Dili, with outreach into adjacent rural areas. MFIs as a group are still embryonic and with very limited outreach. Nonetheless the establishment of a conducive policy and regulatory environment will be helpful for their growth, as will the preparedness of regulated financial institutions to enter into mutually-beneficial partnership or agency arrangements with MFIs. The MFIET wholesale lending facility for MFIs has potential for such arrangements although it is capital-constrained and its expansion will be slow. Other institutions may wish to partner with MFIs as deposit-collection agencies and thus provide a service in more remote communities that would be valued more highly than credit by the majority of households. The PMU should also investigate the possibility of securing ongoing benefits from the CEP credit experience for deserving individuals. Where genuine microenterprise development has occurred as a result of CEP, a proper continuation strategy would be to attempt to link such enterprises with other sources of credit to assist in making them sustainable. There will be difficulties for such people in making the transition to commercial credit, since they have not been accustomed to paying sustainable interest rates, while MFIs and other microfinance providers will also have some reservations about such clients. Overcoming these difficulties would require a re-orientation of the borrower to a new set of circumstances. 20 Where well-performing CEP loan clients live within the catchment areas of institutions providing microfinance services, CEP should take steps to introduce them to those institutions. This should be done in close consultation with the institutions concerned. CEP should also attempt, through its facilitators and other staff, to prepare CEP loan clients for initial interviews with microfinance institutions, by helping them to review their credit needs and cashflows. CEP should consider issuing certificates to borrowers who have discharged their loans. All qualified clients, wherever they live, should receive documentary evidence of loans taken and repayments made. This would serve as a credit reference for the individuals concerned. CEP will avoid discrediting the system if it sets strict criteria for eligibility. MFIs and financial institutions will in any case apply their own criteria to assessing such applicants. The Government could consider extending this service to well- credentialed SEP clients, which would require access to the project records of Bank CGD. CEP should consult MFIs, the Microfinance Working Group (MFWG) and commercial banks about the form of documentation acceptable to them. It will be unfortunate, but unavoidable, if CEP is unable to introduce well-performed clients to alternative sources of finance simply because there is no institution within reach. Under these circumstances, CEP must do as much as possible to document the good credit record of a client for future reference when financial services do become accessible. CEP could also assist MFIs and other financial institutions by issuing a negative list of defaulters. Members of the MFWG are already communicating such information among themselves and with the MFIET. The numbers of clients in CEP and SEP are not large, so consideration could be given to including their records in a credit database under discussion between commercial banks and the Banking and Payments Authority. This would go some way toward repairing any damage done to the rural credit culture, by associating default with the denial of future credit. If the intention to prepare a `black list' is publicized by CEP, this may have some positive effects on collection for the time remaining for the project. 21 Annex 1: Field Interview Data The Mission made field visits in June-July and October-November 2003. This involved inspecting records at the posto level to gain insight into the CEP loan portfolio. However, limited time available for fieldwork and the necessarily non-random nature of data collection posed problems for attempts to generalize. On the first occasion the Mission spent four days in the field, visiting six posto in three regions. On the second occasion the Mission spent two days visiting two more remote posto in the southern district of Covalima (in Central region). This district had been suffering from localized flooding and road closures during the first visit and it was not feasible to visit the area then. This would have been a serious omission since Covalima has relatively large lending and a high degree of recycling. Summary details of the sample are given in Table 1.1. A total of eight posto (of 62) in three regions (of four) were visited, representing a purposive sample of posto responsible for 19% of total credit and 47% of second round or recycled credit as at August 2003.11 Table 1.1: Posto Visited and Credit Program Data Total Repay- Repay- Original New Loans Paid Groups w/o Groups in Region/ Loans ments ment rate12 Groups Groups Off Arrears Arrears Posto ($) ($) (%) (No.) (No.)13 (No.) (No.) (No.) Western Bobonaro 18,500 6,732 36.4 11 1 - 10 Maliana 50,056 12,223 24.4 35 - 1 - 34 Balibo 8,614 734 8.5 4 - - - 4 Central Suai 68,228 30,878 45.3 33 15 9 9 30 Fatumean 31,839 20,602 64.7 13 6 5 6 8 Liquisa 31,705 17,470 55.1 13 5 4 5 9 Bazartete 18,639 6,216 33.3 5 - - 5 - Autonomous Metinaro 19,558 5,279 27.0 17 - - - 17 TOTAL 247,139 100,134 40.5 131 26 20 25 112 The Mission visited the distrito of Liquisa, and two posto, Liquisa and Bazartete, within the district. In the case of Posto Liquisa, 13 credit groups were formed during 2001, taking the cumulative total of lending to $22,205. By the last four months of 2002, the UPK was able to fund five new groups by recirculating money repaid by the first round of borrowers, bringing the total lent to $31,705. The interest rate for these new loan contracts was set at 15% per annum on the decreasing balance, an increase from 10% in the first round.14 In terms of repayment, around $15,100 (a little less than 50% of the amount advanced) had been repaid to end-May, according to verbal advice from the CEP Coordinator for Liquisa District. However a written return completed by the CEP Monitor for the district recorded repayments of $17,470 (around 55%). In summary, Posto Liquisa had formed 18 groups for credit under the CEP by May 2003. The number of groups decreased by four after October, with the discharge of those groups' fully-paid loans. Thus the groups current at end-June comprised nine original and five new groups, making a total of 14. The remaining nine original groups were overdue to varying degrees. Their repayments were, respectively, 2, 11Data for the June-July visit are drawn from May reports. Data for the October visit are extracted from data for that month obtained at the posts before they were received at the PMU. The aggregation of these data is therefore subject to error and the percentages cited, as well as the totals given in Table 1.1, are subject to this reservation. 12Repayments to date as a proportion of total loans extended. Note that relending, as in Suai, Fatumean and Liquisa, acts to depress this measure. 13Indicates recycling of loan fund has commenced. 14The review of the credit program conducted in August 2001 had recommended recycled credit should bear a rate of 1.5% per month. Liquisa seems to have been one of the few posto to adopt this course. 22 3, 5 (two cases), 8 (two cases), 9, 10 and 13 months in arrears, on loan contracts 18 months in duration. This level of delinquency suggests a quite high level of portfolio risk. As visits to other posto were to make clear, this was an all-too-typical pattern of arrears. Only five of the nine groups in arrears were still active and the UPK intended to reschedule the loans of these five. Among the five new groups formed in posto Liquisa since October 2002 some early signs of payment difficulty were apparent by April. Three were up to date, but two had each missed one payment. As noted above, crude calculations of the cumulative repayment rate15 for all loans made by the Liquisa posto showed a figure somewhere in the range 48-55%. By the time of the second Mission, returns for end- August showed repayment of 54.6%, suggesting a slowdown in payment. No new groups had been formed. In the case of Posto Bazartete, total lending of $18,639 had been made to five groups by June 2003. The UPK in Bazartete commenced lending only toward the end of 2002, having postponed credit in favor of applying Project subgrants to infrastructure. Repayments reported verbally by the UPK, up to the day of the Mission's visit, totaled $6,216 with each of the groups said still to be `on-time' (lancar). A monitoring report records the five groups as each having paid four on-time loan installments to end-April, or 33% of total loans. In terms of conventional measures of portfolio quality, this would be recorded as 100% on-time payment with 0% portfolio at risk, although the small numbers involved and the short period since commencement scarcely justify formal analysis. On the face of it, these new groups appeared to be making a promising start. The UPK was also preparing another four new groups for borrowing, to be funded from cash reserves. In addition to its broader impact on the agricultural economy in Bazartete, the significance of CEP credit seemed likely to be in enabling families, some of whom at least are in very modest circumstances, to build assets and improve their lives. One such was a young man who had never worked before commencing to trade. Another was an employee of the Indonesian IDT program until 1999. He had been able to establish himself in the market in 2000 with some savings and had quadrupled the value of his stock since then. A $500 loan from CEP had assisted him in this process. Another, older, man appeared unfit for heavy labor, but had held a place in the pasar since 2001 and was repaying a loan of $500. The strong personality of the Ketua Kelompok (Group Leader) seemed very important for the dynamics of this group. And unlike many others observed by the Mission, these members did seem to function as a group, a process facilitated by their all being located in the same workplace. After this initial visit it was tempting to think that the deferral of lending in Bazartete would result in superior repayment performance, perhaps because of extra time for governance processes at the local level to become better institutionalized. Certainly the Ketua (Chair) of the UPK seemed to think this would be the case. But by the time of the second visit, when end-August data were available, the picture was less encouraging. Total payments had increased slightly, from 33% of loans outstanding to 39%, but arrears had begun to appear and despite earlier plans no new groups had been formed. A still less encouraging picture emerged from a visit to Posto Metinaro, in the autonomous Dili district. This is a dry and sparsely-populated area of limited agricultural potential located on the main coastal road to the east about 45 minutes out of Dili. Fishing is the major economic activity. From data collected at the UPK in Metinaro it appears the posto received credits for 17 groups, located in three suco, commencing from April 2001. The terms of all loan contracts had expired, although none had been paid off, and to end-May repayments totaled only $5,279 on total borrowing of $19,558 (or 27%). Most group payments tailed off during 2001 and scarcely any installments had been paid during 2002. Seven of the 15It is a very crude calculation, because, as explained above in the text, repayments received to date should be related to payments having fallen due to date, rather than to the total of the loans outstanding. 23 17 groups had been persuaded to sign an undertaking to resume repayment under a rescheduling arrangement. While all were at various levels of delinquency, the groups fell into two categories. Thirteen had made no payments since 2001 (with an average of 3.15 payments per group). The remaining four groups were signatories to the rescheduling agreement and had recommenced paying from March 2003, due to pressure from the District Coordinator who had threatened to defer subgrants due under CEP III. His taking this position posed a conflict with the CEP imperative to complete disbursement of the third cycle of subgrants by end-2003. One of the borrowers who had recommenced paying had been appointed Camat, or Head of District Administration, in the meantime. In Posto Bobonaro, in the mountains of the Western region, 11 groups had borrowed $18,500 with a repayment rate of about 36% to date. Disbursement had occurred in three stages (50%, then 40%, finally 10%) apparently under the mistaken impression that the procedure for infrastructure grants also applied to lending. The review of credit operations conducted for the mid-term review (August 2001) indicated that this practice occurred in a number of places. However, many successful microcredit programs make the immediate expenditure of the loan for the agreed purpose a condition of lending, so that this staged disbursement cannot have helped financed activities to succeed, and may indeed have encouraged the misapplication of funds. One loan had been paid off and the group had requested further credit, but no further loans had been made (and in any case under normal procedures new borrowers would be given preference for recycled funds). All other loans were past their 18 month contract dates and in substantial arrears. Various misadventures were cited as reasons for non-payment although 9 of the 11 groups were said to be active still. The largest loan was for $5,000, to a group which had restocked a kios and started a retail outlet for cement. This group consisted effectively of two people who had jointly repaid less than $1,700. Both businesses were visited, and the one borrower who was available for interview claimed to own three homes and farmland. In discussion with members of the UPK and the Conselho de Posto the CEP credit program was compared unfavorably with the activities of a Grameen Bank replication being conducted within the district. The latter was said to be very vigilant in collections (`they arrive right on time every week') and to have more effective leverage on recalcitrant clients. In a separate discussion, a senior official of the NGO concerned told the consultant that the activities of CEP had somewhat increased the difficulty of their achieving high repayment rates and that few of the CEP borrowers would qualify for credit under the NGO's poverty targeting criteria. In Posto Maliana 35 groups had taken loans totaling slightly more than $50,000. Just one group (in fact, a family kios) had discharged its loan, while all others were in arrears and total repayment was at just below 25%. The pattern of payments suggested that perhaps half of all groups had made some attempt to repay, but there were just as many who had made three or fewer payments before abandoning the task. With cash reserves exceeding $12,000, the Conselho de Posto is considering whether to recommence lending Some duplication of projects was evident in Maliana. For example the conselho had approved loans for four threshing machines in one suco. The loss of agricultural machinery had been extreme during the events of late 1999 and it was thought that the machines would be hired out to other villages. But (as has also been noted in the ICR for SEP I) the returns from investment in agricultural machinery were depressed when villagers obtained equipment from other sources, including NGOs and international agencies. Small retail activities were another source of duplication. There were 9 kios financed by CEP in villages in Maliana posto, but many others had also been established and the one borrower with a successful kios had diversified into a range of activities including poultry and livestock, making it less vulnerable to competition in retailing. In the Mission's assessment of the broader economic significance 24 of activities financed by the CEP, small retail kios should be credited with having an important role to play in raising incomes in the agricultural sector. But this is not to deny the problems caused in situations where a `copycat' syndrome leads to excessive investment in such assets. In Posto Balibo, only four loans had been made and none had yielded more than one installment. All were for the finance of agricultural equipment, said still to be in operation in three cases. These four groups were scattered at some distance around the town of Balibo where the posto and UPK were located. The mission visited two borrowers. One was a Chefe de Suco, who was interviewed outside his spacious and renovated house. He said that the group had disbanded within a month and the equipment had been returned to the supplier at considerable loss. With the proceeds he had bought a generator and refrigerator and was making ice-blocks for sale on his own account. The Chefe had signed a letter some months previously, pledging to discharge the debt within 12 months, but when interviewed said he would pay only if the UPK would come to him and then only a small amount per month. It was difficult to imagine the young UPK and Conselho officials we met in Balibo confronting this man. The second borrower appeared to live in more modest circumstances and had also signed a new pledge. She too said the UPK would have to make the journey to collect. The machinery was seen and was in working order although said not to be in frequent use. In both Balibo and Metinaro the mission encountered individuals who had signed letters recommitting to payment as part of efforts by the PMU to reschedule loans in arrears. The Mission's field experiences suggest results are likely to be mixed. The October series of visits took in two posto, Suai and Fatumean, with comparatively good records of repayment and recycling. Both are subject to monitoring by CEP staff for Covalima district and their credit programs have many features in common. In general, the standard of record-keeping in Covalima district is above the average for CEP credit as a whole, for example in tracking successive rounds of credit rather than lumping new loans in with the original lending as is done elsewhere. It appears relevant also that both Suai and Fatumean are located close to the Indonesian border and have good access to supplies of relatively cheap manufactured goods and processed foods, while the continued presence of UN forces in the border regions has maintained some injections of funds into local economies. This has advantaged numbers of CEP-financed kios in these areas and resulted in better than average repayment performance. Thus in Suai, 9 of the original 33 groups had discharged their loans at the time of the visit, and in Fatumean the numbers were 5 of 13. In Suai, marked contrasts in repayment appeared between different suco. In Labarai, 5 of 6 loans had been discharged on time, and the sixth was near to closure, with total repayment at 93%. In another suco all 9 loans were substantially in arrears and average repayment was 31%. In Labarai, 5 of the 6 loans had been for retail kios, whereas in the other village trading ventures in vegetables, livestock and fish appeared to have been less successful. By no means all defaulting borrowers had failed in business, however. Members of the posto UPK felt that negative attitudes from this village were beginning to have wider influence and to affect repayment elsewhere. In discussion, it was difficult to pin down reasons for discrepancies in performance, although the quality and commitment of UPKs and councils appeared likely to play a role. The UPK office in posto Suai was particularly impressive, with systematic records and a wall-display of graphics suggesting good Project governance and an analytical approach to credit. However this office was responsible for both the well- and poorly-performing villages, suggesting that differences in outcomes are due to local factors. The nature of activities financed appeared also to influence repayment rates. In both posto it was noticeable that activities with an element of seasonality did less well than `cashflow' operations such as kios. This is consistent with the experience of microfinance institutions everywhere and although the 25 CEP manual (October 2003) states that special arrangements should be made for repayment in the case of enterprises subject to seasonality, it appears that CEP has not been able to do so. Both posto UPKs had recycled funds actively, with Suai in particular pursuing a policy of relending as soon as funds permitted and favoring well-performed borrowers with second loans. This appears to contrast with practice in some other regions where second loans have been issued more on an entitlement basis, to new borrowers, rather than on the basis of performance within CEP. Suai had made 15 new loans in three tranches as funds accumulated. This means that some funds were in their fourth round of circulation, which is the closest any posto had come to realizing the CEP vision of a revolving fund to multiply the circulation of capital in the post-conflict economy. Fatumean had made 6 new loans, in two tranches. Repayment rates for these new loans appeared encouraging in both posto. 26 Annex 2: Major Factors Affecting Implementation Adoption of the Indonesian KDP Model The events of late 1999 created an emergency which the CEP was intended to address. The Kecamatan Development Program (KDP) a project of the Government of Indonesia funded by the Bank, was seen as providing the tools for the job, since East Timor had been included in the KDP from its start in 1998. Planning and training activities had commenced in 42 of the 62 sub-districts (now posto) with Timorese staff in place, and at various stages of preparation. Moreover, the KDP project in East Timor was not dependent upon the vanished Indonesian administrative structure. So it seemed a particularly apt platform for a program aimed at repairing destruction and establishing appropriate community-level institutions for reconstruction. KDP was thus the model for CEP, and 35 former KDP staff were among the first Timorese appointments to CEP I. The credit elements of KDP were directly incorporated into the CEP project design. Since KDP credit activities in Indonesia have been extensively reviewed, it is appropriate for this Review to consider what lessons those evaluations held for the CEP in Timor-Leste.16 In the absence of civil strife, KDP credit activities in Indonesia had one considerable advantage not enjoyed by CEP. That was access to the services of Bank Rakyat Indonesia, its branches and its village units. Nonetheless, on-time repayment rates for years 1 and 2 of KDP were only 45%. A number of reasons were given, all familiar to a greater or less extent in the context of CEP in Timor-Leste. They included: business failure due either to poor planning or economic conditions, borrower perceptions that loans were in fact non-repayable government grants, village elite interference in credit decisions and poor targeting, insufficient technical training and support for credit management, and lack of loan collection, supervision and enforcement. However, an even more fundamental cause of failure was suggested: while participatory planning and decision-making are appropriate methods for allocation of funding for public goods, such as the subgrants for infrastructure, they are incompatible with sound credit management. The final report of KDP concluded that in its second phase it would `need to reconcile how to adapt a more sustainable micro- credit system while maintaining KDP's key principles of participatory local decision-making and local management of funds'. In the context of Timor-Leste, credit allocation recommendations were made initially in the Conselho de Suco, then forwarded to the Conselho de Posto for ratification. These also are popularly elected bodies and there is no reason to think their workings were any different in essentials from those of the equivalent Indonesian bodies. This is not to say that the allocations made in CEP were invariably `political' (although instances of blatant elite capture are not hard to find). But a political process, in which decision-making is divorced from risk-bearing, is a less efficient way of funding successful enterprises and achieving high repayment rates than lending based on sound credit analysis. Other findings concerning KDP which appear to have been mirrored in the CEP include the following: ˇ The project did not develop a comprehensive manual for revolving fund and credit management. ˇ Financial procedures were structured around the subgrant process, rather than adapted to the needs of credit. ˇ The group mode of organization chosen for the project had many problems. It was assumed that some sort of group liability existed, but how this could be assured in practice was unclear. Experience showed many instances of groups being formed only to qualify for loans and hence lacking any common bonds that would work to assure repayment. 16For this purpose the Review draws upon two documents. World Bank (2001), Review of the KDP Microcredit Approach, by Detlev Holloh. Jakarta (September, 2001); and Republic of Indonesia, Ministry of Home affairs (2002), Kecamatan Development Program Phase 1: Final Report, 1998-2002.Jakarta. 27 The Group Method of Organization As with the KDP, CEP grants and credits were extended to communities on proposals submitted in the name of the community (or of groups in the case of credits). Funds for infrastructure grants were allocated for the creation of public goods, and it is logical to think that neighbors would willingly band together to work on a project for their common benefit. But this is not always a reasonable assumption in the case of credit extended to groups for economic activities, and during the Mission it became apparent from field visits and other evidence that the concept of a credit group had limited basis in reality. Rather, the household is the normal organizational unit for livelihoods. Under these circumstances it is not surprising that the proceeds of many group loans were divided between group members to finance their individual income-generating activities, and that genuine group economic activities were apparently rare. Stories also abound of groups having been formed by one or two dominant individuals for no other reason than to qualify for loans. The Mission encountered instances of microenterprise activities conducted by groups, but also of groups effectively controlled by individuals, with the latter category apparently more numerous. What is surprising is that the Project design envisioned any other outcome. The confusion may have arisen from two sources. First, credits were a secondary consideration in CEP I and it may have been administratively convenient to process them as an extension of the community subgrants, which were organized on a group basis. Second was a misunderstanding of the group loan method seen in classic microcredit operations such as Grameen Bank. In fact, the Grameen group is an organizational unit, not an economic one. The group is the locus of training and loan-payment collection, and its members' solidarity is underpinned by their common experience of training, their joint savings activities and their mutual guarantees of one another's repayments. But their economic activities are individualistic, not communal. Low repayment rates experienced for CEP credits cannot reasonably be attributed to the break-up of loans for individual activities, since this does not, per se, erode group solidarity so long as group members have received adequate preparation and are bound together by mutual obligation. But neither of these forces for coherence appeared to be present in credit groups formed under CEP. There was no training provided for groups nor any arrangement for cross-guarantees of repayment. One UPK with a good record of credit management informed the Mission that dealing with groups was more difficult when some members were not prospering. The effect was to create dissension within the group and to retard repayment by the group as a whole. Human Capacity Issues The credit activities of CEP were a minor component in the Project in dollar terms. Training activities were directed primarily to the primary and more pressing functions of CEP, especially the administration of subgrants for infrastructure. The Project lacked a CFO until almost the end of 2000 and the absence of detailed Project manuals was a source of complaint. A manual for facilitators appeared late in 2000 around the time that credit disbursement began, but did not discuss credit procedures. UPKs were established in preparation for the second disbursement cycle, during which the bulk of CEP credit was advanced. The key officials for their orientation and training were the DPAs (District Project Accountants) who were put in place after the first cycle, but who did not (as mentioned above) have the benefit of a financial or credit procedures manual and were themselves not always well-prepared for the task. As early as December 2000, an audit report had recommended that as a matter of priority a formalized manual should be prepared, but the first comprehensive financial management manual was 28 published only in mid-2002.17 This included a section on credit, but it appeared after about 75% of the initial amount available for loans had already been disbursed. In the meantime, the preparation for credit management received by the Treasurers of UPKs was inadequate. By mid-2001, when about three- quarters of credits had already been disbursed, the CEP mid-term review noted that `many of the [UPKs] did not receive the full two weeks training that should have been provided'. A report on credit activities annexed to the mid-term review recommended that UPK treasurers in posto and suco with large loan portfolios should receive training urgently in financial management and bookkeeping to improve the management of revolving funds. Inadequate financial procedures at UPKs were documented in subsequent reports by the Bank's financial management specialist and by the Project external audit. In January 2002 the financial management specialist spoke of `ad hoc information' and concluded that: `district and village councils [the conselho] have taken on banking functions for which they have virtually no capacities'. Difficulties he observed, which were leading to inaccurate credit records prepared by UPKs, included complicated manual calculations of principle and interest due, unclear identification of total repayments, lack of clarity in identifying arrears, and confusion as to the amount available for recycling as loans. The auditor's report at June 30, 2002, noted skill deficiencies of the treasurers of UPKs as a `major weakness' exacerbated by changes of personnel and requiring urgent attention. It recommended further attention to capacity-building for UPK officials and increased frequency and quality of CEP district and finance staff visits to UPKs. A further report by the financial management specialist in June 2003 noted that `...data collected by district offices from UPKs is not always timely nor systematic...has not been adequately scrutinized and analyzed to indicate reliable repayment rates, age analysis of loans outstanding nor total amounts still at risk'. These difficulties were still evident at the time of the Mission's second visit, late in 2003. Following the appointment of a new CFO in July 2002, a number of improvements were attempted in reporting from the posto level, intended to give a clearer picture of CEP lending. These included a master list of borrower groups, with names of members, the amount borrowed and the nature of the activity. Another report classified groups by gender, and by activity and amount borrowed. The Mission saw evidence of these reports being prepared in certain posto, but inspection of returns received by the PMU suggested that many were not doing so. This suggests continuing lack of capacity among many field staff, coupled with failures of supervision. Consequently, the opportunity to tabulate important variables, such as distribution of loans and activities by gender, was lost. CEP had also been working toward the production of a new aggregated data set for loans and repayments, as mentioned above. However, the second Mission late in 2003 concluded that the new reporting format had yielded no information dividends and recommended an alternative form of aggregated reporting for the remaining months of the Project. These failures reflected a continuing weakness in Project capacity to report, record and analyze credit data. This appears to have stemmed from the haste with which CEP was implemented and the relatively low priority accorded to capacity building for credit operations over the life of the Project. Problems of Repayment A number of factors which seem likely to have contributed to poor repayment performance were noted from CEP documents, and some of the situations described were also observed during field visits by the Mission. They included factors external to the Project, including: ˇ Currency exchange losses suffered by borrowers during 2001 before the transition to `dollarisation' was complete. Loans were paid in dollars but the Indonesian Rupee remained the 17CEP III Finance Management Manual, June 2002. 29 working currency in villages at the time. Borrowers incurred losses of from 10 to 12%, according to the mid-term review. ˇ Adverse economic events, including the downturn in national income due to the rundown in numbers of international staff during 2002. Also, the revival of privately funded investment in some sectors, and/or the activities of grant-funded enterprises, both of which reduced the returns to investment by CEP borrowers. Factors internal to the Project included: ˇ Initial misunderstandings such as staged disbursement of credit and failure to understand the distinction between grants and credits. These led to resentments which could have impacted on repayment. ˇ Poorly-conceived loan proposals, compounded by inability of UPKs to cope with an unexpectedly large number of very small proposals. ˇ Ambiguity over the question of whether timely repayment would result in further credit. When this issue was clarified (in the negative), the incentive for repayment was weakened among some borrowers who had hoped for the renewal of credit. ˇ Elite capture, entitlement attitudes and intentional delinquency, evident from the status of some individuals concerned and their minimal efforts to repay (and leading in at least one case to a moratorium on repayment by other borrowers). ˇ Inability of UPKs and PMU to monitor repayment, lack of procedures for timely follow-up of defaulters. ˇ Lack of sanctions and procedures to enforce repayment, including in cases where rescheduling occurred, perhaps compounded by the placement of UPK and Conselho outside traditional power structures, and the relative youth of many UPK staff. The result could be interpreted as a disjuncture between CEP governance and longer-established power structures. MFIs operating as financial institutions work hard to `imprint' a sense of obligation in their borrowers. Despite charging much higher interest rates good MFIs typically obtain substantially better repayment rates than are seen in non-financial credit disbursement schemes such as KDP and CEP. Unlike the latter, they tend to start small and grow organically. 30 Annex 3: Data Sources and Methods of Investigation Sources of information for this CEP credit review included files and manuals of the PMU, original reports from posto, district and regional level, obtained in the field and from the PMU, and the consolidated reporting of program performance data prepared by the PMU as part of the Project MIS. This was supplemented by extensive discussions with PMU and field staff, including the Deputy Director of the PMU, CFO, National and Regional Coordinators, and the officer in charge of MIS credit records. The Director of the PMU was not available during either of the Mission visits.18 Information concerning credit program operations was supplemented by a total of 6 field days, involving visits to 8 posto (of 62) in 3 (of 4) regions, accompanied by PMU staff and involving interviews with field staff, and members of UPKs and conselho and borrowers, as well as the observation of economic activities. From donor sources, the Mission consulted the periodic Aide Memoires of World Bank supervision missions, the reports of joint donor supervision missions conducted in November 2000 and November 2002, and a review of credit activities conducted in conjunction with the mid-term review (August 2001). Reports of periodic monitoring visits by a Bank financial management specialist were also consulted. Original project documentation included the TFET Grant Agreement of 2000, and the first and second PADs for CEP, dated 2000 and 2001 and the Completion Report of the initial contractor, dated December 2000. The Mission also had access to the Implementation Completion Report for the Small Enterprise Project (SEP I). A review of the Indonesian KDP credit Program, commissioned by the Bank and dated September 2001 was consulted, as well as the joint donor report, Timor Leste: Poverty in a New Nation (May 2003). The Macroeconomics Background Paper prepared by the Joint Assessment Mission in 1999 was helpful in understanding the genesis of CEP credit. The National Development Plan of Timor-Leste (May 2002) gave insight into the Government's thinking about rural financial sector development. Discussions with members of the Timor-Leste Microfinance Working Group and with management of the Microfinance Institute of East Timor were also helpful in this regard. 18The consultant responsible for reviewing credit elements of CEP visited Timor-Leste on two occasions, in June-July and October-November 2003. CONFLICT PREVENTION AND RECONSTRUCTION UNIT--WORKING PAPER SERIES No. Title Author Date 1 Children, Education and War: Marc Sommers June 2002 Reaching Education For All (EFA) Objectives in Countries Affected by Conflict 2 The East Timor Reconstruction Klaus Rohland November 2002 Program:Successes, Problems and Tradeoffs Sarah Cliffe 3 Humanitarian Assistance, Reconstruction Alastair J. McKechnie March 2003 and Development in Afghanistan: A Practitioner's View 4 Central America: Education Reform in a José Marques April 2003 Post-Conflict Setting, Opportunities and Ian Bannon Challenges 5 Using Case Studies to Expand the Theory of Nicholas Sambanis May 2003 Civil War 6 Financing and Aid Management Salvatore Schiavo-Campo May 2003 Arrangements in Post-Conflict Situations 7 Community-driven Reconstruction as an Sarah Cliffe August 2003 Instrument in War-to-Peace Transitions Scott Guggenheim Markus Kostner 8 Gender Equality and Civil Wars Mary Caprioli September 2003 9 Do Participatory Development Projects Help Patrick Barron September 2003 Villagers Manage Local Conflicts? Rachael Diprose David Madden Claire Q. Smith Michael Woolcock 10 Landmine Clearance Projects: Task Jacques Buré November 2003 Manager's Guide Pierre Pont 11 Timor-Leste: Independent Review of the John D. Conroy March 2004 Credit Component of the Community Empowerment Project 2 Also published as Social Development Paper No. 58