Report No. 16790 World Bank Support for Small and Medium Industries in Sri Lanka: An Impact Evaluation June 24, 1997 Operations Evaluation Departnent Document of the World Bank Currency Equivalent Currency Unit: Sri Lanka Rupee 1978 Rs 15.61 1987 Rs 29.44 1979 Rs 15.57 1988 Rs 31.81 1980 Rs 16.53 1989 Rs 36.04 1981 Rs 19.25 1990 Rs 40.06 1982 Rs 20.81 1991 Rs 41.37 1983 Rs 23.53 1992 Rs 43.83 1984 Rs 25.44 1993 Rs 48.32 1985 Rs 27.16 1994 Rs 49.42 1986 Rs 28.02 1995 Rs 51.25 Fiscal Year GOSL January 1 to December 31 Director-General, Operations Evaluation : Mr. Robert Picciotto Director, Operations Evaluation Department : Mr. Roger Slade for Ms. Elizabeth McAllister Acting Division Chief : Mr. Ren6 Vandendries Task Manager : Mr. Robert Buckley Abbreviations ADB Asian Development Bank AWDR Average Weighted Deposit Rate AWPR Average Weighted Prime Rate BOC Bank of Ceylon CBOC Commercial Bank of Ceylon CBSL Central Bank of Sri Lanka CITI Clothing Industry Training Institute DEC Development Economics and Chief Economist DFCC Development Finance Corporation of Ceylon DFI Development Finance Institution EDB Export Development Board ESW Economic Sector Work FTZ Free Trade Zone GDP Gross Domestic Product GOSL Government of Sri Lanka HNB Hatton National Bank IDA International Development Association IDB Industrial Development Board IDP Industrial Development Project IFC International Finance Corporation ILO International Labour Organization IMF International Monetary Fund MOFP Ministry of Finance and Planning NDB National Development Bank of Sri Lanka NSB National Savings Bank OED Operations Evaluation Department PB People's Bank PCR Project Completion Report PFDP Private Finance Development Project PFI Participating Financial Institution PME Public Manufacturing Enterprise PPP Purchasing power parity SCB State commercial banks SLFP Sri Lanka Freedom Party SMI Small and Medium Industry UNDP United Nations Development Programme UNP United National Party The World Bank Washington, D.C. 20433 U.S.A. Office of the Director-General Operations Evaluation June 24, 1997 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: World Bank Support for Small and Medium Industries in Sri Lanka: An Impact Evaluation Attached is the Impact Evaluation Report on the World Bank Support for Small and Medium Industries in Sri Lanka, prepared by the Operations Evaluation Department (OED). Between 1979 and 1991 the World Bank provided Sri Lanka with $110 million in financing for small and medium industries (SMIs) through four credits. The Asian Development Bank provided parallel financing of $45 million for the last two projects and in June 1997 approved a third project for $55 million. The last World Bank project was fully committed in FY96. The first of these projects initiated the Bank's financial sector work in Sri Lanka after the country shifted away from being an inward-looking, and public sector-dominated economy. They continued with the deepening of the policy dialogue, and were accompanied by studies and policy-based lending. The credits were designed to help small scale entrepreneurs become better integrated into the economy. They also served as a fundamental component of the Bank's efforts to help liberalize the financial sector. This evaluation provides an empirically based perspective on the Bank's strategy and impact on small and medium industries (SMIs). It makes use of a survey of more than 300 firms, half of whom were beneficiaries, while the other half served as a control group. Structured interviews with participating financial institutions (PFIs) were also undertaken. The analysis considers the effects the credits and other Bank interventions had on SMIs and PFIs. This broader perspective on Bank interventions is part of a greater OED emphasis on a country, rather than a project, being the "unit of analysis." The central finding is that Bank support for SMIs was unquestionably an important, if subordinate, aspect of Sri Lanka's development strategy. Through a longer-term prism, it is clear that the credits played a significant role in developing a more diversified, private sector-oriented economy. They also contributed to the development of a more effective financial infrastructure. Finally, they efficiently helped generate a significant number of jobs, largely for lower-skilled workers. Cumulatively, then, the projects had desirable effects on income distribution, as well as financial and industrial development. Given these results, the projects' outcomes should be seen in a more positive light than that presented by the audit for the first two credits, which rated them as unsatisfactory. Slow, but ultimately realized, policy reforms and subsequent projects addressed the weaknesses of the earlier projects. These projects also had satisfactory outcomes. Indeed, they laid the groundwork for a very fruitful policy discussion that continues today. Their sustainability is likely and their institutional development is strong. ii The major lessons of these projects are three. First, the sharp decline in Bank support for SMI projects in recent years may have been a missed opportunity. From the review of these projects it is clear that if correctly structured, SMI operations can be effective employment programs. The Sri Lankan experience suggests that this kind of result is more likely in economies undergoing substantial adjustment, and which also have high rates of unemployment, such as the reforming socialist economies. Second, the resolution of the most important problems confronting SMIs in Sri Lanka requires broader financial sector reforms, rather than just more credit, as was emphasized in the last project. However, broader international experience suggests that Sri Lanka's slowness in fully realizing these reforms is not unusual. Finally, international experience also suggests that greater fiscal discipline than has been achieved in Sri Lanka may be a necessary condition for financial reforms to be sustainable. Attachment Contents Preface ............................................................................................................................... 5 Executive Summary.......................................................................................................................7 1. Introduction..................................................................................................................... 13 Broader implications of the study..................................................................................... 14 The plan of the paper ........................................................................................................ 15 2. The Initial Situation And Diagnosis.............................................................................. 17 Overview........................................................................................................................... 17 The specifics of the Bank's SMI strategy......................................................................... 21 Evaluation of the diagnosis: W as it correct? .................................................................... 23 3. The Evolving Objectives of Bank Assistance and Their Implementation.................27 Overview...........................................................................................................................27 Objectives and instruments of the SMI strategy............................................................... 27 The consistency of objectives and instruments.......................................................... 27 The evolution of project design........................................................................................30 Policy questions posed by the evolution of SMI support.................................................35 4. The Effects of Bank Support for SM I Firms................................................................ 39 Overview......................................................................................................................... 39 Employment...................................................................................................................... 39 Income distribution ........................................................................................................... 46 Growth ............................................................................................................................. 47 Nonlending service .......................................................................................................... 48 Government policy and SMIs........................................................................................... 51 This report was prepared by Robert Buckley (Task Manager). Major contributors to the study were Dipak Mazumdar and Edgard Rodriguez (Consultants), who also designed the survey instrument, helped structure our approach, and along with Alice Galenson and Manuel Pefialver, made extensive helpful comments; as did a number of staff from the Sri Lanka country team and OEDD2. Other contributors included: Asita de Silva and Oliver Rajakaruna (Consultants). Econsult (Colombo, Sri Lanka) carried out the survey of firms. Norma Namisato provided administrative support. 2 Contents (cont.) 5. The Impact of Bank Support for SMIs on the Financing Patterns............................ 57 Overview........................................................................................................................... 57 Financial market development.......................................................................................... 57 The role of the government in the financial system ......................................................... 62 6. Impact and Sustainability..............................................................................................67 Overview...........................................................................................................................67 Impact and sustainability.................................................................................................. 68 A longer-term perspective on the evaluation of the early SM I credits......................71 Lessons learned.................................................................................................................72 Although episodically effective, policy-focused lending has been very productive...................................................................................................................72 As employment programs, SMIs in Sri Lanka have been relatively cost-effective..............................................................................................................72 Financial sector reform would be significantly more efficient than further SM I suppor .................................................................................................... 73 Bibliography ............................................................................................................................. 75 Annexes I. Methodology and data tabulations on the survey of the small- and medium- scale industry lending scheme.......................................................................................... 83 II. Summary statistics: SM I survey of firms, Sri Lanka, 1996 ............................................. 88 IUl. Projects related to the SM I credits.................................................................................... 89 IV. Chronology of events........................................................................................................90 Boxes 4.1 The return on policy dialogue expenditures.....................................................................49 4.2 Effectiveness of directed credit programs........................................................................ 53 6.1 SM Is and financial fragility.............................................................................................. 70 Figures 2.1 Policy measures and growth trends .................................................................................. 25 3.1 M acroeconomic trends and distortions............................................................................. 32 3.2 Output of small and large firms, 1971-1995 ....................................................................34 3.3a Sri Lanka - Labor force composition, 1971 census.......................................................... 36 3.3b Sri Lanka - Labor force composition, 1993 labor force survey .......................................36 4.1 Employment by firm size - Sri Lankan manufacturing, mid-i 980s-1995 ....................... 40 4.2 Average capital and labor requirements among Sri Lankan firms...................................42 3 Contents (cont.) 4.3 Allocation of SMI credit and industrial value by province .............................................. 54 5.1 Debt ratios by firm size: Sri Lanka vs. higher-income countries.....................................59 5.2 Long-term debt/Total assets ratio..................................................................................... 60 5.3 Long-term debt/Total debt ratio .......................................................................................60 5.4 Short-term debt/Total assets ratio.....................................................................................60 Tables 2.1 Development and manufacturing: Sri Lanka and other countries....................................19 2.2 Share of income, variousyears ......................................................................................... 21 3.1 Components of the four SMI credits ................................................................................ 28 3.2 Borrowing terms of SMI credits.......................................................................................31 4.1 Capital intensity (K/L) of firms, Sri Lanka, 1985 and 1995 ............................................41 4.2 Firms using both overdrafts and trust receipts .................................................................43 4.3 Selected characteristics of complex and basic firms........................................................44 4.4 Problems faced by firms...................................................................................................48 4.5 Financial ratios over time: All S s................................................................................52 5.1 Firms requiring external finance to maintain their current growth rate...........................61 5.2 Interest rates by size, 1995 ............................................................................................... 62 5.3 Collection performance of participating banks ................................................................ 64 5 Preface This study is an Impact Evaluation of four credits, totaling $110 million, that the World Bank provided for small and medium industries (SMIs) in Sri Lanka between 1979 and 1991. The Asian Development Bank (ADB) provided parallel financing of $45 million for the last two projects and in June 1997 approved a third project for $55 million. The last World Bank credit was fully committed in FY96. This study attempts to measure the longer-term effects of Bank support for SMIs. To do this, a survey of more than 300 SIs was carried out by a Sri Lankan firm. Participating financial intermediaries (PFIs) as well as goverinent officials also provided perspective and information on the credits. This study analyzes the survey data and interviews. It also examines the effects of nonlending activities and policy dialogue on SMIs and PFIs. In short, the study brings together detailed empirical evidence on the effects of the credits and policy reforms on beneficiary behavior. For the preparation of this Impact Evaluation, OED missions visited Sri Lanka in November 1995 to identify and arrange for a survey of SMIs, and in February and June 1996 to undertake pilot surveys and discuss the topic with participants in the credits. A draft of the study was sent to the Borrower and participants for comment. Their helpful inputs, in particular those of the National Development Bank of Sri Lanka, the Central Bank of Sri Lanka, the People's Bank, the Bank of Ceylon, and the Institute of Policy Studies, have been taken into account and are greatly appreciated. 7 Executive Summary 1. Between 1979 and 1991 the World Bank provided Sri Lanka with $110 million in financing for small and medium industries (SMIs) through four credits. The Asian Development Bank provided parallel financing of $45 million for the last two projects and in June 1997 approved a third project for $55 million. These credits were accompanied by economic sector work, policy dialogue, and policy-based lending designed to affect the environment in which SMIs functioned. This study attempts to measure the longer-term empirical effects of Bank support for SMIs. Particular emphasis is given to the effects of nonlending activities. This emphasis is taken because in Sri Lanka SMIs stand at the center of policy issues in industry, trade, and financial policies. In such a context, an impact evaluation of Bank support must be expansive enough to consider the synergy between lending and nonlending interventions. The Bank-Government strategy for SMIs 2. The SMI strategy had three major objectives. First, it sought to engage the Government of Sri Lanka (GOSL) in a dialogue about the efficacy of economic policies regarding trade, and the appropriate roles of the public and private sectors. Second, the strategy sought to work with the GOSL to restructure the financial sector from one that serviced a centrally planned industrial economy, which favored large enterprises, into one that responded to the demands placed on it by entrepreneurs. Finally, the SMI credits, like most Bank-supported SMI lending, sought to generate jobs so that Sri Lanka's unemployment problem could be addressed on a cost-effective basis. The presumption was that small firms would be more labor-intensive employers of lower- wage earners than would large firms. The projects 3. Beginning in 1979 the projects provided loans, with an average size of about $10,000, through the banking system to almost 16,000 small- and medium-scale private enterprises. Each of the credits focused on policy reform and institution building, and together with three Industrial Development Projects (IDPs), served as the basis for a Bank-GOSL policy dialogue on the financial sector. Tariff rationalization and reduction, and private sector development were also emphasized in the credits' policy dialogue. These policy issues were also addressed in two adjustment loans in the late 1980s. The last project was fully committed in FY96. Methodology 4. This evaluation provides a longer-term perspective on the Bank's strategy and impact in one sector of a specific country. A survey was undertaken of 304 firms, 149 of them beneficiaries of Bank-supported SMI loans, and 155 nonbeneficiary SMIs. The latter served as a control group. Structured interviews of participating financial intermediaries were also undertaken. The analysis considered the effects of both the credits and nonlending services on SMI and participating financial institution (PFI) behavior. This broader perspective on Bank 8 interventions is part of a greater OED emphasis on the country, rather than a project being the "unit of analysis." Findings 5. The central finding of this study is that Bank support for SMIs was unquestionably an important, if subordinate, aspect of Sri Lanka's development strategy. The credits and the attention given to the policy distortions that impeded the sector played a significant role in developing a more diversified, private sector-oriented economy. They also helped generate a more effective financial infrastructure and a significant number of lower-paying jobs. In many important respects, the SMI strategy was the beginning of a new policy process. It was a process that was to take a less inchoate form as other economic reforms were implemented, and the SMI credits supported the development of this process. 6. Firms and employment. Employment at beneficiaries increased by more than 8 percent a year, more than three times as fast as general employment growth. This growth translates into about 22 additional employees per firm, and the value added of their employees was almost 20 percent higher than employees of nonbeneficiaries. In contrast, employment at nonbeneficiaries increased by less than 6 percent a year. Beneficiaries also increased their labor-intensity by much greater amounts than did firms in the control group. Importantly, our analysis also indicates that unlike the situation in many other countries, the credit went to beneficiaries who were indeed credit constrained, as argued in the Staff Appraisal Reports. As a result, these credits lifted a major impediment to productive investments. Moreover, firm access to credit, both for beneficiaries and nonbeneficiaries, improved substantially over time as financial liberalization took root and Sri Lanka became one of the largest manufacturing producers among low-income countries. 7. The poverty and employment effects of the credits appear to be significant: more lower- wage workers were employed by beneficiaries. In addition, most of the smallest firms became larger, graduating to larger sizes at a rate that exceeds the experience in other countries. Finally, the regional concentration of beneficiaries was more diffused than the existing heavy industrial concentration in the capital city. 8. On the negative side, 40 percent of SMIs assisted went out of business within eight years of borrowing, and they produced largely for domestic markets rather than for export. These firms were often, as a number of Sri Lanka observers have noted, low-skill enterprises. As a result of the high failure rate, Bank estimates of employment effects due to SMI support are overstated. However, the exit rate is similar to those observed in other developed and developing countries. Nor has it resulted in significant losses by the PFIs due to an inability of SMIs to repay their loans. In addition, while many SMIs are low-skill firms, our analysis also indicates that a significant share of them, about 25 percent, undertook more complex transactions. These complex firms performed much more strongly than did the basic firms, and their continued growth is likely to lead to the kinds of specialization that will lead to industrial upgrading, and ultimately higher growth rates. Furthermore, the SMI export performance is not unusual. In fact, on a cross-country basis it appears relatively strong. Finally, while the PFIs uniformly praise the technical assistance provided under the credits, like other studies, we were unable to discern any effect of technical assistance expenditures on firm performance. 9 9. Financial institutions. A primary rationale for the initiation of the SMI credits in Sri Lanka was to help accelerate the creation of a financial system that would provide information on the performance of private firms. When such a system of intermediation exists, entrepreneurs can more easily mobilize resources, as well as garner rewards for undertaking high return investments. However, until such a system is established, financial institutions have less incentive to invest in the capacity needed to provide this service. These institutions will invest much less in capacity development as long as a significant share of the gains from their investments accrue to others. The result will be that financial development will be slowed. Thus, an important question is how did the credits affect PFI profitability and outreach. 10. PFIs were profitably involved with the credits. Bank support contributed to an expansion of funding for the most promising activities, and a general expansion of financial system depth. During the projects implementation, interest rates became market-determined, and varied with the riskiness of borrowers. An effective repayment system was also established under the credits, and ultimately recoveries were at commercial rates. In most cases, the PFIs improved their financial positions: the second tier development finance institution (DFI) involved in the program was privatized, and the PFIs expanded their own lending to SMIs and microenterprises. Indeed, by some measures, it appears that SMIs in Sri Lanka now have as much access to credit as do the SMIs in developed economies. The only qualification to this general improvement in financial sector performance has been the performance of the State Commercial Banks, which continued to suffer from overstaffing and interference by Government. But, even in these cases, substantial progress was made in terms of institutional performance indicators. 11. Policy impediments. Sri Lankan firms now have more access to finance, and particularly long-term finance, than they did 10 years ago. Thus, the financial infrastructure necessary to underwrite, monitor, and fund profitable investments has unquestionably been established in Sri Lanka. The financial sector environment has improved significantly, and it is likely that the Bank's financial sector policy dialogue made a substantial contribution to this improvement. Similarly, trade restrictions have been significantly reduced. Indeed, Sri Lanka's tariff restrictions have fallen from among the world's highest to levels that place it as the most open economy in South Asia. 12. Despite these improvements, SMIs still face high real borrowing costs-of 8 to 10 percent-and are rationed out of access to short-term credit. Both of these problems are largely the result of the failure to complete the financial reform program initiated under the last two SMI credits. Finally, the coinsurance for SMIs, provided by the Central Bank, was an effective exercise in risk-sharing when this program began. It encouraged private sector participation in SMI lending. However, there is no obvious reason why it should be continued. It now causes "good" SMI borrowers to subsidize "less good" ones, and exposes the Central Bank to a risk the private sector can now bear more effectively. 13. Impact evaluation. Cumulatively, the SMI program was a very effective job-creation program. It also had desirable distributional consequences and nicely complemented the emphasis given by the GOSL to encouraging more foreign direct investment. It also had beneficial financial sector effects. The implementation problems that arose in the early projects were largely addressed and corrected in later projects. In this longer-term context, and with the benefit of much more detailed data on beneficiary performance, the impact of Bank support for SMIs in Sri Lanka should be seen in a more positive light than that suggested by the audit of the first two SMI projects. These credits did not have unsatisfactory outcomes. Nor should their 10 sustainability be viewed as uncertain. The subsequent credits helped address the policy constraints that impeded the early projects. They also resulted in a sharp improvement in loan recovery measures from less than 70 percent at the time of the audit to better than 95 percent. Changing the outcome ratings of the early projects to satisfactory increases Sri Lanka's performance on this measure to the fourth highest of the 36 countries which have completed 40 or more projects, and to the highest level of all IDA countries, except China. Lessons learned 14. Policy effectiveness. The policy dialogue financed studies that contributed to a rationalization and lowering of tariff barriers, and better debt recovery mechanisms. Effective performance under the credits also contributed to the privatization of a large DFI and to general financial sector deepening. These changes made significant contributions to overall economic policy as well as to sectoral performance. However, this result was achieved despite the fact that the dialogue was only episodically productive, and indeed, in some instances, appears to have had little value added. For example, detailed studies on the rationalization of the two state commercial banks, which were funded under the credits, have yet to be acted upon. 15. This pattern of policy dialogue success-high overall productivity and low probability of success for many individual items-suggests one explanation for why recent statistical studies of the effects of aid find it has so little impact on policy: success is highly idiosyncratic. That is, even if most aid expenditures in a particular country have little effect on specific policies, this result does not imply that the overall effects are negligible. Conversely, individual projects with satisfactory outcomes can be contemporaneous with a poor overall outcome on country dialogue, and ultimately with development effectiveness. This pattern of success also suggests that the Bank, due to its simultaneous and ongoing operations in so many countries, probably has a comparative advantage in undertaking "investments" in policy dialogue. 16. Bank lending for SMTs. Our results also provide some perspective on the Bank's overall approach to Small and Medium Enterprises. They can do this because the Bank's SMI strategy in Sri Lanka mirrors general concerns that have been raised about Bank support for SMIs. The SMI projects in Sri Lanka began at a time when the Bank had an active SMI program-more than six SMI projects a year were being approved. They ended with a "last" project being approved in FY91, just as Bank SMI lending ceased, at least temporarily. 17. Our results indicate that in Sri Lanka SMIs have been an effective employment program. The efficacy of SMI support for job creation compares favorably with all other government employment programs. This result suggests that in economies undergoing substantial adjustment and which have high rates of unemployment-such as those that have recently undertaken SMI programs, e.g., Albania, Poland, Russia, South Africa and Palestine-SMI projects may well be an effective strategy. Of course, the side effects such programs can have on financial sector development bear careful monitoring. The unintended consequences on the financial system of what is an otherwise effective employment program can be important. 18. SMis as financial institution loans (FILs). The central problems confronting SMIs in Sri Lanka-high real borrowing costs and limited access to short-term funding-cannot be addressed just by additional funding for SMIs. The efficacy of Bank support for SMIs, like the efficacy of all other credits, will continue to be hampered by the high real borrowing costs that the financial structure imposes on all borrowers. These basic financial sector problems will only 11 be addressed by improving the private sector's competitive access to resources. This access is now restricted in order to fund the government's borrowing needs. In short, financial reform is always difficult to complete when there are imbalances in the government's fiscal position of the scale observed in Sri Lanka. Reducing the financial market restrictions that support these large deficits and directed credits would help address the central problems SMIs say they face: the high real cost of credit. 19. In the Appraisal Report for SMI IV, the Bank and the GOSL articulated a strategy to move from sectoral issues to broader financial sector concerns. In basic respects, this strategy remains valid today. Financial reform would do more for SMIs, as well as the economy, than would additional financial sector support for SMIs. However, a review of the international experience in dealing with financial sector crises, particularly in countries under conditions of fiscal stress, indicates that this process takes a considerable amount of time. In this light, the aspirations for the amount of time needed to complete financial reform made in the Appraisal Report were highly ambitious. Furthermore, the international experience also suggests that fiscal adjustment may well be a prerequisite to the achievement of sustainable financial liberalization. 20. It is perhaps also worth noting that in small volatile economies an SMI FIL can often reduce government risk exposure by a greater extent than will a more general FIL. Unlike the situation in developed countries, the credit risk on SMI loans in Sri Lanka appears to have been considerably lower than that on loans to larger firms, as made, for example, under the IDP credits. A similar pattern seems to apply to a number of developing economies. This result suggests that "directed credits" for SIfs, made at prevailing interest rates may also indirectly benefit the banking sector. For example, by providing funding which limits commercial bank exposure to the largest, riskier firms, SMI credits may, in fact, imply less banking sector, and thus ultimately, government risk exposure than do unconstrained FILs. 13 1. Introduction 1.1 Overview. This study is an impact evaluation of Bank support for small and medium industries (SMIs) in Sri Lanka. Between 1979 and 1991 the World Bank provided Sri Lanka with $110 million in financing for SMIs through four credits. These credits accounted for almost 10 percent of the country's total IDA support during this period. In addition, the Asian Development Bank (ADB) provided parallel financing of $45 million for the last two projects, and in June 1997 approved a third project for $55 million. However, much more important than the volume of lending involved is the role small and medium enterprises play in the Sri Lankan economy. In this economy the SMI sector stands at the center of policy issues in three of the key sectors that determine growth-i.e., industry, trade, and finance. Furthermore, SMI lending was also used as a vehicle to address employment concerns. Consequently, in Sri Lanka an evaluation of Bank support for SMIs is a study of the pursuit of growth with equity. Indeed, in many respects, the insights and empirical evidence offered by this kind of analysis are some of the basic building blocks of a constructive dialogue on development strategy. 1.2 The Projects. Beginning in 1979, the projects provided credits through the banking system to almost 16,000 small- and medium- scale private enterprises. The average credit size was about $10,000 per loan. A revolving credit fund was established at a second tier lending institution, the National Development Bank (NDB), which also provided implementation assistance to participating financial institutions (PFIs). Each of the credits focused on policy reform and institutional strengthening in trade and industrial policy. In addition, together with three Industrial Development Projects (IDPs), these credits served as the basis for a Bank- Government of Sri Lanka (GOSL) policy dialogue on the financial sector. The last project was fully committed in FY96. 1.3 Methodology. This evaluation provides detailed empirical evidence, and a longer-term perspective on the Bank's role and strategy with regard to SMIs. Considerable attention is given to the effects that the policy dialogue had on the projects. Such a perspective is part of a greater OED emphasis on the country, rather than a project, being the "unit of analysis." The study adduces empirical evidence on the behavior of beneficiary firms and a control group, firms which did not receive loans.' The performance of PFIs are also evaluated. At the most basic level, the evaluation attempts to answer the following questions: Who benefited from the credits? Was the behavior of beneficiaries any different from that of other firms? Did the credits generate higher productivity or employment as intended? How was the profitability of PFIs affected as a result of their involvement with the credits? Did Bank support contribute to an expansion of the most promising activities? How did the credits affect growth and equity? And finally, how did these credits fit into the Bank-GOSL overall development strategy? 1.4 The evaluation of the nonlending aspects of the credits also addressed a series of questions: Were the projects' outcomes affected by the policy environment? Did the projects affect the policy environment in significant ways? Finally, how do the projects appear with the See Annex I for a description of the survey of firms. 14 benefit of a longer-term perspective? Did they build on the lessons of earlier projects? That is, did later projects address the problems that arose in implementation of the earlier ones? Broader implications of the study 1.5 Bank support for SMI lending in Sri Lanka also provides a unique historical perspective on a number of broader policy issues. Policy effectiveness. The audit of the first two SMI projects and an OED study (1987) of the Bank's relationship with Sri Lanka argue that these projects gave an unusually large emphasis to policy dialogue. This is an emphasis that a recent DEC study, by Burnside and Dollar (1996), suggests was likely to have been fruitless. Did this dialogue affect project outcomes or development measures? And, if so, which aspects of the dialogue seem to have been most productive? By focusing on the policy dialogue embodied in specific projects and surveying beneficiaries, this analysis can consider the direct results of these discussions. It can also show the broader effects on the economy of this important dimension of Bank support. Like Bank analysis of economic sector work (ESW), by Schneider (1996), this study suggests that in Sri Lanka the policy dialogue was very productive, if difficult to measure. * Firm access to credit. This study also augments recent DEC financial sector work that indicates that small- and medium-sized firms in developing countries have been systematically afforded less access to long-term credit.2 Using the comparative data developed by the DEC study, the results provide some sense of how Sri Lankan firms compare with those in other countries. Then, this study can take this analysis one step further: it analyzes how Bank support and government policies affected firm access to credit. * Bank lending for SMIs. Finally, an evaluation of the Bank's SMI strategy in Sri Lanka can provide some perspective on the Bank's overall approach to Small and Medium Enterprises. It can do this because the Bank's SMI strategy in Sri Lanka, in many respects, mirrors both the Bank's financial sector concerns with lending to Small and Medium Enterprises, as well as its concerns with helping to restructure formerly socialist economies. The SMI projects in Sri Lanka began at a time when the Bank had an active SMI program, more than six SMI projects a year were being approved. They ended with a "last" project being approved in FY91, just as Bank SMI lending ceased, at least, temporarily.3 1.6 At the time of the first credit the Bank, along with the UNDP and the ILO, was actively promoting SMIs.4 The SMI sector had become something of an alternative to the large scale industrialization development paradigm that had characterized many development strategies. 2 Demirgdag-Kunt and Maksimovic (1996a and 1996b). See Webster, Riopelle and Chidzero (1995) for a review of Bank support for SMIs and data on Bank lending, and World Bank (1978) for the Bank's initial SMI strategy in Sri Lanka. 4 Little, Mazumdar, and Page (1987) for a review of this experience. 15 During the 1980s the Bank provided almost $4 billion for 80 projects to support this development perspective. By the time of the last project, however, SMI loans were viewed as "directed credits" which were to be avoided by Bank projects.5 More recently, and particularly in reforming socialist economies, the Bank has shown a renewed interest in SMIs. And once again, the policy concerns that initially generated the emphasis on SMI assistance in Sri Lanka are the same ones which have led to the resurgence in SMI projects. Thus, the motivations for the Bank's once more advocating SMIs were also present in the strategy that prompted the initial SMI lending in Sri Lanka. The plan of the paper 1.7 Chapters 2 and 3 review the Bank's initial diagnosis of the situation, and the evolution of the projects over time, respectively. These chapters are similar in spirit to an audit of the SMI projects within the evolving policy environment. For those already familiar with this context or those more interested in the empirical results, these chapters can be skipped. Chapter 4 reviews the empirical evidence on the behavior of firms, and Chapter 5 analyzes the financing patterns of firms and PFIs. A final chapter analyzes the impact and sustainability of the support for SMIs. It evaluates the institutional development of the projects, and considers how the projects affected the overall development strategy. Finally, it presents a number of lessons learned from these projects. These concerns are most explicitly made in Bank Operational Directive 8.30. 17 2. The Initial Situation And Diagnosis Overview 2.1 This chapter reviews the Bank's diagnosis and approach to the sector. The first project is given particular emphasis for two reasons. First, OED analyses have shown that successful identification and appraisal are key elements in ultimate project performance. When well diagnosed and prepared a project's likelihood of having a satisfactory outcome is 14 percent higher than that of the typical Bank project. Moreover, outcome is more than two and one-half times more likely to be satisfactory than is the case when identification and preparation are not adequate. The audit for the first two SMI projects (World Bank 1991c), rated both as having unsatisfactory outcomes, an unusual outcome for a country which at that time had only three other unsatisfactory projects. This audit also raised concerns about SMI III which was under implementation at the time of the audit. Consequently, based on the audit, the effectiveness of the initial diagnosis appears problematical and worth examining. Was this diagnosis the beginning of a flawed relationship? 2.2 Second, one of the concerns of the audit was the emphasis given to policy dialogue in the projects. This is a concern that was recently made more general in a Bank study by Burnside and Dollar (1996). In a study of 56 countries over a 24-year period, they found that aid has no measurable impact on policy. Thus, the emphasis given by the credits to the importance of the policy dialogue also appears to be problematic. This emphasis was established in the initial diagnosis. Hence, by evaluating the initial strategy in light of a longer-term perspective, and with the benefit of more detailed empirical evidence, a richer understanding of the value of the policy dialogue in one country can be inferred. 2.3 The initial situation. These credits were initiated when the economy had just begun to shift from being one of the world's most inward-looking, and inhospitable to the private sector. Until the end of 1977, two years prior to the first SMI credit, the domestic interventions of the government were extensive: restrictions on foreign investment, the nationalization of business undertakings, an overvalued dual exchange rate system, control of interest rates and credit supplies, restrictions on wages and the ability to terminate employment, a publicly-owned manufacturing sector that produced at low capacity behind a high protective wall, and extensive licensing requirements to establish firms. The public sector dominated production and exchange, nationalized banks dominated a financial sector that lent at negative interest rates, and the export earnings of the agricultural sector, which was ultimately nationalized, were heavily taxed to provide one of the world's highest levels of social expenditures. These export taxes also financed 6 See the OED Annual Review ofEvaluation Results 1992. See De Long and Summers (1993) for data on the cost of capital equipment in 81 countries for 1980. Sri Lanka had the highest cost of equipment for all the countries in the sample. They suggest that the discrepancies in this cost arise largely due to tariff and exchange rate policy. The more inward looking the economy the higher the cost. 18 government expenditures on unprofitable large scale public manufacturing enterprises. Finally, price supports and public distribution systems existed for more than 6,000 consumer goods. The private sector was at most tolerated. Furthermore, the Bank's dialogue with the government in power from 1970 to 1977 had broken down.8 2.4 The manufacturing sector in Sri Lanka contributed about 15 percent of GDP and employed about 10 percent of the labor force. About 30 percent of industrial value added was accounted for by agro-industrial processing of tea, rubber, and coconut. Private industry consisted of an organized subsector of some 7,700 firms and an unregistered subsector of more than 20,000 cottage industries. Private industry contributed half of industrial value added, with unregistered firms accounting for two thirds of this. Private sector firms were a fraction of the size of public sector companies. The average number of employees of the largest 1,400 private firms was small, 42, while the typical unregistered firm employed family labor and two or three additional workers. Average fixed capital investment of registered firms was only 15 percent of that in public sector industry. 2.5 As Table 2.1 shows at the beginning of its reform program Sri Lanka was more industrialized than comparator countries. Employment was more heavily concentrated in large enterprises than it was in even the most industrialized economies. At the same time, the share of employment in small and medium enterprises was correspondingly lower. 2.6 Three major subsectors-textiles, food processing, and chemicals-accounted for more than three fourths of value added in manufacturing and more than 60 percent of employment. The public sector dominated production in the first and last of these sectors, and even accounted for almost half of the output of the food processing industry. Industry was highly concentrated in the Colombo area, accounting for almost 90 percent of value added and more than 80 percent of the employment of the organized private sector. By the mid-1970s growing shortages of foreign exchange had begun to cripple many industries, and raw materials shortages were intensified by the monopoly positions of public sector importers or producers. 8 On general macro policies during this time period see Athukorala and Jayasuriya (1994); see Khatkhate (1982) for a description of Sri Lankan financial policies in the 1970s; De Silva (1981) shows that social welfare expenditures in 1970 were equal to almost 15 percent of GDP, a level three times higher than the average figure for low-income countries at that time; and finally, a 1987 OED study details Bank-GOSL relations. It indicates that lending was suspended in the early 1960s following petroleum nationalization. In 1970 new Bank lending was again halted. GOSL policies-particularly the renegotiation of three loans to the previous government-and the nationalization of foreign plantations were issues in the breakdown. 19 Table 2.1: Development and Manufacturing: Sri Lanka and Other Countries Income Group GNP per Urban Manufacturing Employment Employment (1977 US$) capita Population GDP (% of Share" of Share" of (US$) (% of total GDP) Small and Large population) Medium Enterprises Enterprises (100 or more (5-99 workers) workers) 100-50021 302 21.0 13.7 29.2 70.8 Sri Lanka (1980) 280 21.6 17.8 18.0 82.0 501-1,000 744 44.9 17.3 36.7 62.8 1,001-2,0004/ 1331 55.8 25.8 31.2 68.8 Over 2,000s5 6069 73.1 25.7 31.8 68.2 USA (1973) 8750 73.0 24.2 22.5 77.5 1/ Employment shares have been rescaled to exclude Cottage Industries (1-4 workers) which are not covered by Sri Lankan statistics. 2/ India, Tanzania, Kenya, Indonesia, Ghana and the Philippines. 3/ Nigeria, El Salvador, Peru, Colombia, Ecuador and Korea. 4/ Turkey, Taiwan, Panama, Costa Rica, Brazil, Argentina. 5/ Hong Kong, Greece, Italy, Japan, Canada, Norway, Germany, US and Kuwait. Sources: Biggs and Openheim (1986, Table 1.1). Sri Lanka: 1980 Annual Survey ofEstablishments, Department of Census and Statistics, cited by UNIDO; other data from World Bank database. 2.7 In November 1977 Sri Lanka liberalized the economy. Most quantitative restrictions on imports were replaced with tariffs, the currency was sharply devalued, and a floating exchange rate was adopted. Measures to attract foreign investment and foreign banks were implemented. In short, the policies which had prevented a "foreign trade orientation" were precipitously ended. Price controls were eliminated for most goods, and the private sector was encouraged as an engine of growth. The Bank was asked to help foster this process, and responded with a study of the industrial sector in 1978.9 This study spelled out the strategy that the Bank would follow in its implementation of the first two SMI loans in 1979 and 1981, as well as the first of three Industrial Development Projects (IDPs) which began in 1982. 2.8 In large measure the 1978 study repeated many of the recommendations made in a 1953 Bank study of the Sri Lankan economy. The previous study, World Bank (1953), had also recommended that emphasis be given to developing small and medium domestic businesses to complement a strong encouragement to direct foreign investment. However, in the intervening years, the Sri Lankan economy had been transformed. It no longer was a low-income, pre- industrial, agriculturally based economy. By 1978 it had become, as Table 2.1 suggests, an economy that had aggressively marshaled resources, often through nationalization and expropriation, in an attempt to achieve rapid industrialization through import substitution and nationalization. As a result, it no longer had the characteristics of a pre-industrialized, plantation economy. By 1977 Sri Lanka had been transformed into an economy that no longer had a small Issues and Prospects for Industrial Development (1979). 20 manufacturing sector, it accounted for 17 percent of GDP, and was characterized by employment in large firms. 2.9 It also had an industrial structure that, as Levy (1993) describes, was more similar to a socialist economy than a low-income, pre-industrialized economy. The large state-owned enterprises set up vertically integrated operating modes that discriminated against the development of small and medium enterprises. Consequently, there was a smaller portion of small- and medium-sized firms (see Lakshman and others 1991). Agricultural employment also followed a bipolar distribution. Employment in the nationalized plantation sector did not follow the usual pattern of contracting in size and developing processing-related small-scale agro- industries. Nor did the agricultural population shift into nonagricultural industries. 2.10 But most of all, Sri Lanka had developed an industrial structure that had not produced growth with equity. The record of the public corporations, nationalized plantations, and constraints on private sector development is one of failure. Over the 1970s the growth in output of the public sector was stagnant while that of the private sector grew at 13 percent per year. In addition, if measured on a purchasing power basis, per capita GDP growth over the 1960-77 period was negative 1.2 percent. By 1977 there was widespread support for a shift in the economy's orientation (see Athukorala and Jayasuriya 1994). A focus on SMI development was a significant, if subordinate, part of that shift. 2.11 The support for SMIs was significant because it was designed to help develop the institutions and mechanisms that would be demanded by a more spontaneous, market-oriented economy. It was designed to assure that when the systematic discrimination that existed against small private businesses ended, these firms could prosper and grow. Nevertheless, it was also a subordinate policy because its success depended in large measure on the enactment of more fundamental reforms of the economy. 2.12 For the economy to prosper reforms were needed to open up the economy to competition and foreign investment. At the same time, if the extensive social welfare system were replaced with a more targeted system of transfers to the poor, SMIs would also be important. If the former policies were enacted, support for SMIs would be an effective instrument to accelerate the rate at which the private sector would develop. If the latter policies were implemented, employment opportunities would be needed to offset the cutbacks in employment at inefficient public sector enterprises. Thus, as a complement to the broader policy changes, the Bank's support for SMIs was designed to enhance and accelerate the gains from these policy shifts. 2.13 The 1977 shift in policy has unquestionably fostered a higher level of growth, as well as a significant increase in manufacturing exports (see Wignaraja 1997). It has also generated considerable controversy about its effects on poverty (see the series of comments in the 1987 World Bank Economic Review). Nevertheless, broad measures of equity have shown considerable improvement. For instance, as shown in Table 2.2, the share of income of lower- income groups in 1990 increased since liberalization, after having fallen during the inward- looking decade. In addition, according to some measures, the earlier period was also one of decline in absolute income levels. Thus, since 1977 the benefits of a growing economy have been widely shared. Not surprisingly, a more market-friendly policy environment has become the accepted policy paradigm. 21 Table 2.2: Share of Income, various years Share of income 1970 1980 1990 Lowest 2 quintiles 11.7 10.1 13.1 Lowest quintile 7.5 5.9 8.9 Source: World Development Report, various years. 2.14 Shortly before the liberalization, the political party, the United National Party (UNP), which initiated SMI lending, had been elected in a landslide, winning 104 of 168 seats. It was explicitly committed to reintegrating Sri Lanka into the world economy, and remained in office for 17 years, finally leaving office in 1994, three years after SMI IV had been approved. However, more important than the constancy of the implementing political party is the change in the terms of the policy discussion. The political party in power in 1997, a coalition led by the Sri Lankan Freedom Party (SLFP), which had been instrumental in the public sector-led industrialization strategy, also discusses industrial policy in market-oriented terms. It supports an open economy led by the private sector. Further, despite the Bank's description of SMI IV as the last such project, the current government remains enthusiastic about providing support to SMIs (see the statement of President Chandrika Bandaranaike Kumaratunga in Business Today, November 1996). 2.15 In short, the shift in political perspective that helped initiate the first SMI loan was a part of a process of rehabilitating the institutions and conventions of a market economy. These institutions had characterized Sri Lanka since the late 1800s when it became one of the first developing countries to reach sustained growth. They had, however, been fundamentally changed, particularly during the 1970s. In what follows we consider how the specific components of the SMI lending and policy dialogue fit into the Bank-Government of Sri Lanka strategy of restoring market institutions. However, before doing so, we first consider the Bank's SMI strategy in detail. The specifics of the Bank's SMI strategy 2.16 The SMI strategy articulated in the Bank study of 1979, and refined in additional studies in 1981 and 1989, had three major objectives. It sought, first of all, to engage the Government of Sri Lanka in a dialogue about the efficacy of economic policies regarding trade, the appropriate roles of the public and private sectors, and the financial system. For SMI projects the immediate focus of this dialogue was on the way these policies affected the functioning of the 90 percent of enterprises that employed 50 or fewer employees. In the Bank's perspective, these policies impeded the performance of the economy in general, as well as the small enterprise sector. Hence, the dialogue was more far-reaching than the specific SMI policy environment. 2.17 Second, the strategy sought to restructure the financial sector from one that serviced a centrally planned industrial economy, structured to favor large enterprises, into one that responded to the demands placed on it by entrepreneurs. In other words, it sought to create a financial system that was able to mobilize resources on a sustainable basis, allocate risks 22 efficiently, and fund high return investments. The SMI component of this strategy was a directed line of credit, so that banks, rather than the government, could underwrite and monitor project performance. It had both positive and defensive elements. 2.18 On the one hand, it was designed to defend against a continuation of lending patterns where banks largely carried out government directives. Prior to liberalization, the financial system had been lending at negative real interest rates to public manufacturing firms with poor recovery. As part of this shift, the credit sought to eliminate the existing advantages to the large, loss-making public enterprises that dominated the economy. In some respects, the SMI loans followed the pattern of earlier Bank support for the private sector. 10 Like the earlier loans, the SMI credits explicitly avoided lending to the large publicly owned manufacturing enterprises. Indeed, in many respects, the emphasis on SMIs was synonymous with a private sector focus. However, with the SMI loans, the Bank also stressed the policy environment within which private sector firms and financial intermediaries operated. 2.19 There was also a more positive objective in the Bank's financial sector strategy. The positive thrust was both specific to the SMI sector, and more generally concerned with how the SMI sector fit into the broader financial policy regime. On the former issue, the strategy sought to provide funding to private sector entrepreneurs. However, it also recognized that in an economy with little of the financial infrastructure of a market economy-a moribund stock market and no bond market, little in the way of company registration outside of the large public sector, years of price and import controls, as well as the threat of expropriation for successful business endeavors-that more than just credits for SMI borrowers were needed. Hence, the broader financial context within which SMI lending took place was also reviewed, and became a focal point of Bank-GOSL policy dialogue. 2.20 The SMI component of the Bank's financial strategy assumed that a lack of project finance could have deleterious effects on both investment efficiency and income distribution. If only those who already had wealth could access credit, then future income would largely accrue to those who already owned it. In short, SMI lending was seen as a way to try to begin to tap the entrepreneurial skills and interests which were outside the public sector, and to fill the gap in resources available to this sector. At the same time, more attention focused on the policies that caused this resource gap. 2.21 Finally, the SMI credits, like most Bank-supported SMI lending, sought to generate jobs so that Sri Lanka's unemployment problem could be addressed on a cost-effective basis. The presumptions were that small firms would be more labor-intensive employers of lower wage earners than would large firms, and that the shift to a private sector-oriented economy would cause public enterprises to shed workers. Hence, support for the investments made by SMIs would have desirable employment and distributional effects. These latter effects were thought to be particularly important in a high unemployment economy that was shifting from an extensive social safety net to one that relied more on prices, and better targeted subsidies. 10 Prior to the 1977 liberalization, the Bank had supported four small loans for private enterprises. See Annex III for more details. The four loans had an average size of $6 million and were exclusively for private firms. These projects financed the activities of the Development Finance Company of Ceylon (DFCC), a private sector firm that was established with the help of a World Bank project in 1956 and which participated in the SMI credits. It was the fourth Development Finance Institution (DFI) that the Bank had helped create. See the early review of this experience in World Bank (1957). 23 Evaluation of the diagnosis: Was it correct? 2.22 In most important respects, the diagnosis that SMI lending would be a significant, if subordinate, component of Sri Lanka's development strategy was appropriate and has been shared strongly by the Sri Lankans. As noted earlier, the old regime's restrictions on trade, finance, and industrial policy had reduced the spontaneous development of both SMIs and FDI which could lay the foundations for a more innovative, productive economy. But, these restrictions had also demonstrated that such constraints on entrepreneurs were ineffectual. Not only had growth fallen, so too had income distribution deteriorated. The policies had resulted in fragmented, top-heavy industrial and agricultural sectors which had few of the linkages to the household sector or the outside world that can lead to sustained growth. 2.23 Thus, lending for industrial development of SMIs was unquestionably an important aspect of the development strategy. Attention to the concerns of the domestic private sector, which had been discriminated against, made good sense. Nevertheless, if the diagnosis was relatively strong, the initial implementation was constrained by circumstances. Implementation can be faulted on a number of levels, both in the broader policy environment, and, less significantly, as regards the project itself. Most of the design issues identified in the Bank's strategy were addressed in the subsequent SMI operations. These design issues will be discussed in the next chapter. Here the focus is on how the policy environment constrained SMI activity. 2.24 Public expenditures. With liberalization donor support increased sharply. In Sri Lanka's agriculturally based economy, most of this support, not surprisingly, took the form of agriculture and infrastructure projects. But, in the event, the donor projects took the form of a massive public investment program for a particular section of the country, the Accelerated Mahaweli Development Program, as well as a large-scale housing and urban rehabilitation program. Not only was the conceptual approach to the infrastructure program problematic-government-led investments of a questionable rate of return to develop a virgin area-so too was its scale unmanageable." Similarly, the agriculture projects did little to develop a thriving, efficient agricultural base. Moreover, bad luck also placed pressures on the fiscal situation. As a result of a very sharp decline in the terms of trade, from 102 in 1977 to 58 in 1980, export performance deteriorated, and so too did export-related government revenues. 2.25 Finance. The first SMI credit was approved in a banking system that had aggressively opened up to competition. For instance, interest rates were liberalized in the year preceding the credit. In the same year, 14 new foreign banks opened in Colombo, and a heavily subsidized rural credit scheme was terminated. It was, nevertheless, still a system in which the government made extensive, and, in some respects, increasing interventions in credit allocation. In 1979 a Medium- and Long-term Credit Fund was revived with an eye towards directing resources into Athukorala and Jayasuriya (1994) show that the estimated rate of return to the Accelerate Mahaweli Scheme was low and that its scale was enormous. Public investment in 1980-83 more than doubled to 17 percent of GDP, largely as a result of these projects, from 8 percent during the 1970-1977 period. Isham and Kaufinann (1995) show that increases in public investments to levels up to 10 percent of GDP have positive effects on projects supported by the Bank. However, increases to levels in excess of 10 percent, as occurred in Sri Lanka, have negative effects on project rates of return. 24 priority areas. And by 1981, a National Credit Plan, modeled on a similar scheme in India, was 12 introduced to monitor the allocation of credit to priority areas. 2.26 Implementation experience. Immediately following the approval of SMI I the share of private manufacturing in GDP declined. This occurred because the government rapidly expanded its infrastructure investments and continued to provide substantial amounts of assistance to the loss-making public enterprises. In fact, transfers to public enterprises were more than twice as high in the 1977-85 period as they had been in 1970-77. Between 1978 and 1982 the Mahaweli scheme alone accounted for 45 percent of public investment, and the deficit during this period rose to over 22 percent of GDP, substantially crowding out private investments, as documented by Athukorala and Rajapatirana (1991). In short, stabilization needs and liberalization conflicted and the government reversed the liberalization process. One way to consider this reversal in the policy environment is shown in Figure 2.1. The figure presents an overall measure of policy performance as well as various aspects of economic performance.13 Over successive four-year periods, the policy performance measure changed from a figure suggesting a very strong overall policy environment in 1978-81 to one lower than the "typical" developing country environment. 2.27 Cumulatively, in this kind of deteriorating macroeconomic environment, it would have been difficult for any SMI project to work very effectively. Nevertheless, as also shown in Figure 2.1, ultimately, if not during SMI I, or even SMI II, many of the reforms discussed by the first industry study were in fact ultimately carried out, and the policy environment continued to improve. For instance, * Sri Lanka has become one of the largest manufacturing exporters among low-income countries. In 1993-94 it exported $2.4 billion in manufactured products, up from $15 million in 1977, and more than any other low-income country except China and India. 14 * By 1995 the private sector had unambiguously become the engine of growth. It accounted for more than 90 percent of manufacturing output while in 1979 it had accounted for less than 40 percent. * The cost of capital equipment for manufacturers has declined from more than double the international level to one where Sri Lanka has become the most open of South 15 Asia economies. This kind of price decrease means that equipment investment can now be financed with only one third of national savings rather than the two thirds it took prior to opening up. 12 Athukorala and Rajapatirana (1991). This policy performance index, developed by Burnside and Dollar (1996), provides a measure of the effects of openness, inflation and budget surplus on the policy environment. The mean for this index is 1.3 with a standard deviation of 1.2. The 1.3 figure represents the predicted growth rate if policy distortions have no effect on growth. 14 Sources: World Bank (1996a) and Wignaraja (1997). On a per capita basis Sri Lanka's manufacturing exports exceed both India and China's by a substantial margin. Its annual average growth of manufacturing exports in constant prices over the 1977-92 period was over 30 percent per year. 15 See Wignaraja (1997) for cross-country comparisons. 25 * The financial sector has broadened and deepened. For instance, measures of financial depth suggest that Sri Lanka's financial sector's development, like measures of its human development, are also unusually high.16 * Finally, sustained economic growth with broad improvements in equity has been realized. Figure 2.1: Policy Measures and Growth Trends 3.00 2.50 % 2.00 1.50 1.00 0.50-- 0.00 1975 1980 1985 1990 1995 Real GNP per capita (1975-1) .... Policy index - Private sector share of manufacturing (%) - Cost of equipment index . ..Credit to the private sector/ODP (1975=1) Note: The cost of equipment index is estimated on the assumption that it is influenced only by the real exchange rate (RER) and the effective rate of protection (ERP) for manufacturing. See De Long and Summers (1993). Sources: IFS database, Central Bank of Sri Lanka annual reports, Athukorala and Rajapatirana (1991), World Bank, Sri Lanka: A Break with the Past (1988), Sri Lanka Private Sector Assessment (1995a). The policy index measure is from Burnside and Dollar (1996). See footnote 13 for more details on its meaning. 16 For instance, in 1994 Sri Lanka has a 64th percentile rating among low- and lower middle-income economies on a purchasing power parity (PPP) basis and a 50th percentile rating on a per capita income basis, reflecting Sri Lanka's strong performance on basic services. In terms of financial development, its performance has been even stronger. It has a 69th percentile ranking among low- and lower middle-income countries, according to holdings of broadly defined monetary aggregates. At the beginning of the reforms, as shown by Khatkhate (1982), Sri Lanka had little financial depth or breadth. 26 2.28 Thus, in many important respects, the implementation of the initial SMI strategy was flawed. Shortly after approval, the government pursued policies which reduced the credit's effectiveness. Nevertheless, this strategy was the beginning of a very different policy process. It was a process that was to take a less inchoate form as other economic reforms were implemented, and as the Bank and GOSL moved to a better understanding of the constraints on SMI activity. In the next chapter the implementation experience that built on this initial diagnosis is reviewed. 27 3. The Evolving Objectives of Bank Assistance and Their Implementation Overview 3.1 This chapter considers the evolution of SMI lending in the Bank-GOSL development strategy. It first delineates the objectives and instruments used to support SMIs. Then, it reviews the evolution of project design and policy dialogue under the four credits. This review provides some context for the policy questions that our empirical analysis focuses on in the next two chapters. Objectives and instruments of the SMI strategy 3.2 The objectives of the SMI strategy called for directly affecting the economy through improving the performance and prospects of SMI firms and the participating financial intermediaries (PFIs) which dealt with them. Ideally, the development of these firms and intermediaries would, in turn, contribute to the development of the financial system, help generate more productive opportunities, and beneficially affect the functioning of the labor market, particularly for lower-skilled workers. But, for these firms and PFIs to function effectively the constraints on their operation needed to be reduced. Thus, besides providing resources to support firms, the credits also sought to engage the GOSL in a dialogue on how to reform the impediments that prevented the spontaneous development of this sector. 3.3 Table 3.1 indicates that $16 million, or almost 15 percent of the $110 million in credits, was targeted on nonlending support of some form, either for the institutional development of PFIs or SMIs, or for broader policy reform issues. Slightly more than 50 percent of this funding was for studies of policy issues, such as a study of effective protection rates, the state commercial banks, and industrial policy. The rest was for technical assistance (TA). The consistency of objectives and instruments 3.4 The SMI lines of credit did target assistance based on the size of firms. However, the firm-size limitation was sufficiently high that only the largest firms were excluded from participating. In addition, the eligibility limitation increased in real terms by more than 50 percent over the four credits. Furthermore, the targets of assistance were not those thought to be necessarily the most remunerative. Rather, they were those discriminated against by the existing economic incentives. Indeed, a premise of the early credits was that policymakers should be agnostic about which sectors of the economy were likely to have the highest returns. Ultimately, the SMI lines of credit were designed to help eliminate the need for such lines of credit. Table 3.1: Components of the Four SMI Credits Project Credit Training/TA to Amount TA to SMIs Amount TA for Policy Reform Amount Selected Loan Agreements Line PFIs ($m) ($m) ($m) (Sm) '_ I SMI I 12 Capacity building 0.59 Support to IDB and sub-sector 2.86 (1) Study on effective protection; (2) Study on 0.55 (1) GOSL to consider modification of policies of NDB and PFIs studies export incentives affecting SMIs; (2) GOSL to provide $1.2 million to upgrade IDB; (3) Timetable for completion of studies SMI II 28 Capacity building 0.33 Support to IDB and EDB and 1.67 None 0 (1) GOSL to review findings of two studies and of NDB and PFIs further sub-sector studies develop action plans; (2) IDB and EDB to develop action plans and personnel development SMI III 15 Capacity building 0.5 (1) Establishment of a Technical 0.5 (1) Operational review of the two NCBs (2) 4 (1) TOR and timetables for various studies; (2) of NDB and PFIs Assistance Facility (in NDB); (2) Upgrade of CBSL supervision capacity; (3) Use of AWDR as market reference; (3) onlending Establish Export Marketing Support export promotion programs; (4) rate to be adjusted with market and be positive in Fund in EDB (ADB funds) Support reforms in trade, industrial policy, and real terms; (4) Conditions for PFI participation efficiency of PMEs required commercial loan recovery rates and capitalization. SMI IV 40 Capacity building 0.5 Further support Technical 0.5 (1) Reorient investment & export promotion 4 of NDB and PFIs Assistance Facility policies; (2) Improve customs & excise policies; (3) Review insurance industry; (4) Review impact of financial reforms on industry; (5) Develop subsector environmental protection policies; (6) Continue review of NCBs TOTALS 95 1,92 5.53 8.55 29 3.5 That is, rather than being seen as yet another directive that was a continuation of the centrally planned approach to the economy, the credits were, in fact, a vehicle designed to help undo this approach. Instead of telling entrepreneurs what to do, their objective was first to encourage entrepreneurs to make decisions and take risks about investments, and then to fund them to do so. Furthermore, by providing these credits through financial intermediaries, which would underwrite and select the projects to be funded, the project would accelerate the development of the kind of financial infrastructure which is a hallmark of economies best able to achieve growth with equity. Additional and complementary projects, as described in Annex III, would also develop this financial system and private sector orientation. The table in the Annex also indicates the periods in which each of the credits took place, and presents a number of evaluative measures of project performance. 3.6 This agnostic perspective on which types of industries and firms could best produce growth and exports is quite distinct from that proposed by Lall and others (1996) which, by reference to the experience of the "East Asia Miracle," recommends a much more proactive export policy.'7 That study, as well as Wignaraja (1977)'s, argue that the lack of a coherent, proactive industrial strategy in Sri Lanka has led to the development of a fragile, low-skill export base. A base that has few linkages to the domestic economy, and one which effectively "traps" Sri Lanka into producing low-skill exports. We will consider some of the recommendations for SMIs of this proactive industrial policy in the next chapter. Here we focus on the Bank-GOSL strategy. 3.7 If the design of the SMI projects was agnostic about which sectors were the "best bets" in the economy, it was not agnostic about the competitive disadvantages-in both the financial sector and throughout the economy-that prevented small firms and individuals from mobilizing resources. In fact, the early Bank reports on the industrial sector argued that such policy impediments were the central problem confronting SMIs. Thus, it is not surprising that such a central place was given to the policy dialogue in the implementation of these credits. 3.8 The credits also addressed the deeply authoritative nature of the dirigisme model that had prevented those armed only with ideas and initiative from seeking financial partners. In this respect, the SMI credits were targeted on creating greater equity of opportunities as well as growth. Thus, through the SMI credits, the Bank sought to direct resources to a potentially dynamic, and certainly disenfranchised, part of the Sri Lankan economy. By supporting various government bodies which could assist and train firms-IDB, EDB, and CITI-the projects also sought to enhance the technical and managerial skills needed for productive investments. 3.9 Finally, the credits sought to provide these funds through the financial system in such a way that Sri Lankan institutions and conventions would develop the skills and legal infrastructure so that this process would become self-sustaining. However, the projects' SARs are clear that self-sustaining does not mean that there would be continual Bank funding for SMIs. Rather, self-sustaining finance would be achieved when the impediments that prevented financing from flowing freely to SMIs had been eliminated. Thus, a key rationale for Bank support for SMIs was that this was an instrument that could help offset the effects of policy distortions. 17 The Lall and others study provides a lengthy discussion of the support for small and medium enterprises in Korea, Japan and Singapore. 30 The evolution of project design 3.10 Financial institutions. Over the period of the projects SMI lending changed from being a mildly subsidized, directed line of credit lent largely by public banks (77 percent of SMI 1) to market rate lending by mostly private banks (63 percent of SMI IV). SMI I was lent at a fixed interest rate with high delinquencies. It shifted underwriting responsibilities from the Industrial Development Board (IDB) to the PFIs, a major change in orientation. In fact, the real interest rate credit at approval was positive but fixed. Unfortunately, due to an increase in the rate of inflation after loan approval, it quickly became negative and was not adjusted (see World Bank 1991a). Nevertheless, the first credit disbursed more rapidly than expected and a second project was prepared. 3.11 The second credit moved to positive, adjustable interest rates and gave greater emphasis to encouraging exporters, relying particularly on the Export Development Board (EDB). Its disbursement was slowed by the outbreak of civil unrest. The third credit continued the emphasis on encouraging private sector development, but it also shifted towards much greater concern with financial sector development. In particular, in 1988, after it was approved, a new Banking Act was introduced, which required all banks and finance companies to meet the traditional capital adequacy standards. Then, in 1990 further steps were taken to remove legal constraints to debt recovery.'9 Following these changes, the last credit, SM! IV, was a market-determined source of funds provided largely through private banks. The project disbursement slowed, and the role of private banks increased because for a period of the credit the public PFIs did not satisfy performance standards-on delinquency rates, portfolio infection rates, and capital adequacy. Due to their inability to satisfy the standards, they were suspended from participation in the program. The result was that the smaller private banks accounted for a greater share of lending. However, because of their smaller size, the disbursement rate slowed. 3.12 Table 3.2 presents the borrowing terms and dates of approval for each of the four SMI credits. The last loan also changed the basis for determining the lending rate. The new cost of funds index more fully reflected the longer-term nature of the loans, thereby increasing interest rates but reducing the SMIs' interest rate risk exposure. Over the four credits, the interest rate spread to the National Development Bank (NDB) declined from 3 percent on SMI I to 1 percent on SMI IV, and the amount refinanced was reduced from 80 to 70 percent. The last credit was provided through an apex DFI, the NDB, that was empowered by the first SMI loan, and was itself privatized during the last SMI in 1993, a pattern not typical of many Bank-supported DFIs. 18 These measures included: (a) a minimum capital ratio of 4 percent of banks' liabilities, including interbank and contingent liabilities; (b) a minimum liquid asset requirement of 20 percent of liabilities less liabilities to the Central Bank of Sri Lanka (CBSL) and shareholders funds; (c) a limit on lending to single borrowers of 30 percent of capital funds; (d) limiting maximum individual shareholdings to 15 percent; and (e) no lending to directors except with the approval of the Monetary Board. Additionally, revised regulations under the Finance Companies Act require minimum capital for finance companies of SLRs 5 million and a minimum capital-to-deposit ratio of 10 percent. The latter ratio was to be replaced by a 12 percent minimum risk asset-based measure of capital in 1991. The minimum liquid asset requirement has been set at 15 percent of deposits and an additional 6 percent of deposits have to be placed in approved securities-CBSL securities and treasury bills-lodged with CBSL. 19 There was legislative approval of a comprehensive package of 13 debt recovery measures. 31 Table 3.2: Borrowing Terms of SMI Credits Interest Rate Inflation Term Maturity Loan/Value (%) (years) (%) SMI I (FY79) 10 - 12 16 fixed 5-7 80 SMI II (FY81) 18 16.8 variable 10 80 SMI III (FY87) 13 - 16 7.9 variable/fixed 10 75 SMI IV* (FY91) AWDR + 5 20 variable 10 70 * AWDR= the average weighted deposit rate. The inflation rate is the average for the period of the loan. It is from Athukorala and Jayasuriya (1994). The interest rate on SMI III is the rate implied by use of the AWPR -= the average weighted prime rate as the index for funds. It is a shorter term and lower rate than that of the AWDR index used in SMI IV. 3.13 Real interest rates on the SMI lending, as well as on interest rates generally, have been positive since the early 1980s, and significant progress has been realized on strengthening financial institutions. For example, the stock market was unified in 1985, consolidating the two stock exchanges which had not operated effectively, a market for Treasury securities was established in 1988, and the beginnings of bond finance emerged in the 1990s. In addition, over the period of the credits, project finance grew and spread throughout the banking system, and the system improved. But, serious financial sector problems remain. 3.14 The GOSL and the Bank had envisioned a financial reform program that would enable Sri Lanka to move gradually but steadily toward a more competitive market-based system. It sought greater use of indirect controls, and a larger private sector role in the financial system. Increased autonomy, a commercial orientation, and greater management efficiency were the aspirations for the financial institutions that remained under government control. All of these measures, unfortunately, did not materialize, and some of the reforms that would have addressed these problems have had little progress. For example: * The studies of the SCBs funded under the SMI credits recommended the privatization of at least one of them and the restructuring of both; these measures have stalled. The two SCBs still dominate the other 21 banks in the sector (4 private and 17 foreign), accounting for 60 percent of the banking system's total assets. As shown in Figure 3.1, interest rates periodically increase sharply, and real interest rates remain high, generating a demand for more circuitous, inefficient supports for SMIs. * The two largest pension funds, which hold about 15 percent of the country's financial assets, are directed to hold 90 percent of their funds in government securities. Additionally, the National Savings Bank, which holds almost one fourth of all deposits, also channels its resources into low-yielding government securities. Interest income on these deposits is tax exempt. 32 Figure 3.1: Macroeconomic Trends and Distortions 20.0% Trendline: Unemployment rate (2 period mo\Ang 15.0% 10.0% . .. 5.0% / \- ..F Flacal Deficit/GDP - - - - Real Interest Rate Li. . nemiployment Rate Sources: IFS database and Central Bank of Sri Lanka * Finally, relative to the early days of SMI lending, the government's fiscal position has improved. But the improvement is relative to very high levels and, at 10 percent of GDP, the deficit is still multiplicatively higher than Sri Lanka's comparators. Furthermore, in many respects the imbalances in the GOSL's fiscal position have impeded complete financial liberalization. 3.15 Firms. At the beginning of the SMI credits the structure of industrial finance in Sri Lanka was very simple: private industry was self-financed while the vast majority of public corporate investments were financed directly by the government's budget. More than 80 percent of institutional finance was provided by commercial banks in the form of working capital and short-term finance. More than 60 percent of their lending went to public corporations. The two largest government-owned banks had about 90 percent of the bank branches in the country, and the foreign banks lent mainly for trade finance. Little credit was available to small- or medium- sized fims. The only provider of term finance was DFCC, which was the only institutional source of equity capital and the largest long-term lending institution in Sri Lanka. 3.16 A major objective of the financial liberalization program, as well as the industrial policy articulated in the Bank's 1979 study, was to force public enterprises to reduce their reliance on budgetary transfers as a me an ncts The it manet that the 80 percent of funding they had received would have to be provided by the financial sector. The initial objective was not to privatize public corporations, but rather to increase their autonomy and provide financial disciple in vestments they undertook. However, there was little available long-term finance. Large GOSL borrowing needs relied on low interest rate funding from pension and provident funds, and large public firms relied more on the banking sector than they would have if a more developed capital market existed. The result was that little credit was available from the banking sector for SMIs and other private sector borrowers. privatize pblic corpoations, butrather to ncrease thIratnmadpovefincl 33 3.17 In this context, the creation of NDB, as a public institution, to provide term finance to larger public corporations, and the development of the SMI credit lines at banks can be viewed as temporary measures. They were actions taken to assure that smaller private firms could access credit while financial discipline was being developed. Larger firms could ultimately access a bond market, which would be developed later. However, as shown in Annex IV, which presents a chronology of events with a bearing on the SMI credits, many reform measures stalled. 3.18 The financial demands of large deficits, generated first by a massive public investment program, and then, after 1982, by the defense expenditures on social conflict, caused the "temporary" measures to have a longer life. Potential demanders of long-term bonds remain captive public sector investors, and SCBs continue to support GOSL initiatives. Finally, at the same time, the pressure of opening the economy up to foreign competition led to an onslaught of requests for revisions of tariff protections. The result was that more than 200 revisions in tariffs were made in the two years prior to the approval of SMI I in 1979. As a direct result of these pressures, SMI I included a study of the effective protection rates. 3.19 The tariff study showed that the protection rates in force, after so many ad hoc 20 adjustments, were still high and were assigned with little, if any, underlying rationale. As a result of this study tariff restrictions were lowered significantly in 1984 by amounts ranging from 5 to 100 percent. Instead of over 500 tariff categories above 75 percent, the new schedule had only 86. Ultimately, by 1995, the tariff rate reached 35 percent. This rate is still higher than the rate recommended in the study in 1980, but it is also a rate comparable to Sri Lanka's South Asian counterparts. 3.20 The major change in policy perspective over the course of the credits was that private sector development became a key and lasting theme of government policy. As shown in Figure 3.2, following the tariff reforms of the mid-1980s and a series of adjustment loans in the late 1980s, the production levels of small and large enterprises no longer moved countercyclically to each other, as they still do in India, according to Patel and Srivastava (1996). Their output became positively correlated with GDP and the output of large firms, even if the amplitude of smaller firms' output was considerably larger.21 What occurred was that reforms had taken hold. Smaller-scale firms had become integrated into the economy. 20 Selected Issues ofindustrial and Trade Policies in Sri Lanka, Report no. 4795-CE, 1983. Wignaraja (1997) provides a very interesting discussion of the ineffectiveness of the protection afforded by the tariffs vis-a-vis their use in Korea, Taiwan, and Thailand. 21 The correlation of the output of large and small firms was negative in 1971-85 period and positive in 1986-95. 34 Figure 3.2: Output of Small and Large Firms, 1971-1995 40.00 au GDP Growth --- Small Industry Growth -a- Large Industry Growth 30.00 20.00 10.00 at0 0.00 -10.00 -20.00 O~~C OD Do M~O ~ o , o r o Sources: Central Bank of Sri Lanka (several issues). Figures deflated by GDP-at-factor-cost deflator (1982=100) from International Financial Statistics (IFS). 3.21 By the late 1980s, as public sector ownership of manufacturing firms fell into disrepute throughout the world, the objective of GOSL became privatization rather than commercialization of these firms. And once again, the Bank-GOSL dialogue went far beyond concerns with SMI issues, focusing in SMI III, for instance, on the details of the far-reaching privatization program. The program was then enacted with the help of an adjustment credit in 1989. Besides increasing the responsiveness of the economy to economic pressures, these changes have reduced the public presence in the functioning of the economy, and limited the need for government expenditures on loss-making enterprises. 3.22 Problems of course remained, not the least of which was a long-lasting and brutal ethnic conflict, and an associated inability to stabilize and bring interest rates to manageable levels. Real interest rates in excess of 8 percent, as they have been with few exceptions since 1983, make borrowing an expensive proposition. In addition, the political instability of the conflict has resulted in much lower levels of FDI. While FDI did increase significantly, from almost zero in the 1970s to 2.8 percent of gross domestic investment during 1978-92, it never reached very high 35 levels. For instance, Sri Lanka's FDI was less than half of 1 percent of GDP, whereas in Taiwan and Thailand the comparable figures were two to three times larger, respectively. 22 3.23 Nevertheless, the policy environment which had been so hostile to private investment and small and medium enterprises, had certainly become more conducive. The more open and competitive economy also led to more opportunities, even if the total number employed in manufacturing remains still below the number of unemployed, as shown in Figure 3.3. Policy questions posed by the evolution of SMI support 3.24 By the time of approval of SMI IV, concern with the potential conflict in Bank objectives in supporting directed credit lines had become pronounced. The seeming contradiction between relying on lines of credit while pursuing policy dialogue that emphasized financial liberalization, prompted the writing of Operational Directive (OD) 8.30. This directive, which is alluded to in the SAR, lays out guidelines for interpreting the circumstances under which Bank loans can be made through the financial sector, and, as discussed in Chapter 1, effectively ended Bank support for SMIs. 3.25 The OD directs that lending through distorted financial sectors is to be avoided if policy distortions are not being addressed. Subsequent Bank work by Isham and Kaufmann (1995) provides strong empirical support for this perspective. The rate of return on loans made in distorted economic environments is far below that of the return in more open, competitive environments. However, for SMI lending, which in Sri Lanka was designed to offset existing distortions, the application of OD 8.30 becomes a question of determining how much progress has been made on addressing the other distortions. As the SAR for SMI IV says: "IDA's financial intermediation operations in Sri Lanka have become more sectorally oriented, with an increasing emphasis upon the viability of the system and its institutions .... In parallel with the change in focus, the criteria for subproject eligibility have been modified to move the lines of credit from directed credit operations to general capital investment funding operations. In the case of the SMI operations, this approach is reflected in the ... expansion of sectoral eligibility from purely industrial financing to the financing of investment in any productive enterprise. This move was initiated under SMI I and has been gradually strengthened under each successive operation." 3.26 The SAR also makes clear that these credit lines were not to be a substitute for the kinds of financial reforms that will permit the spontaneous, sustainable provision of funding from domestic institutions. The reforms that would accomplish this were clearly identified in the SMI IV Credit Document, and agreed to by the GOSL. In this perspective, because these reforms were not completed, further SMI lending could be seen as, in effect, substituting for the more fundamental and unfinished policy agenda. 3.27 On the other hand, given events in Sri Lanka, the plans in SMI IV may have been too ambitious. For instance, the Bank might agree that, in retrospect, the timetable for the financial liberalization program that the GOSL agreed to at the beginning of SMI IV was too ambitious. Alternatively, if other factors, such as the performance of other sources of growth-e.g., a much lower-than-anticipated level of FDI-had become more important, it may be reasonable to revisit the SMI strategy. 22 See Wignaraja (1997), pp. 97 and 136. 36 Figure 3.3a: Sri Lanka - Labor Force Composition, 1971 Census Other Unemployed 1% Agriculture 40% NAD. f k 7% Sw 11% Transport 4% Wholesale Construction 8% 2% Manufacturing 8% Figure 3.3b: Sri Lanka - Labor Force Composition, 1993 Labor Force Survey Other Unemployed 4% 14% Agriculture N.A.D. 33% Srvices 18% T r I Manufacturing 4% Wholesale Construction 9% 4% Source: ILO Yearbook of Labour Statistics. 37 3.28 However, even if the Bank and the GOSL were to revise their views on the efficacy of further SMI assistance, it may well be that the failure to act on reforms could create a situation that was inhospitable for further SMI support. Recall, for example, the discussion in Chapter 2 about how the policy environment adversely affected the implementation of SMI I. Empirical analysis of firm and PFI behavior can be useful in assessing whether the Sri Lankan conditions are such that the partially liberalized financial sector is able to use support for SMIs effectively. By analyzing the relative productivity and financial structure of firms of various sizes, and the effects SMI credits have on PFI sustainability, we can provide evidence that affords insights into the effectiveness of the GOSL's renewed support for SMIs, and the Bank's views on the need for further financial reform. u i 39 4. The Effects of Bank Support for SMI Firms Overview 4.1 This chapter summarizes the results of our survey work on how Bank credit affected firm behavior in Sri Lanka. The approach taken is to respond to a series of questions or assumptions about SIls that were made or are implicit in Bank and/or GOSL documents. Employment 4.2 As was made clear in Sri Lanka's Industrial Policy of 1989, a central rationale for supporting SMIs was their putative ability to generate jobs and more labor-intensive employment opportunities. In an environment of high unemployment of long duration, it is understandable that job growth is an important objective. Question 1: Did SMIs stimulate employment? 4.3 Yes. Our results indicate that employment growth of SMI beneficiaries exceeded that of nonbeneficiaries by substantial amounts. Over the late 1980s to 1995 period, beneficiary employment grew faster than 8 percent per year versus less than 6 percent for nonbeneficiaries, and a slightly higher than 2 percent growth figure for the labor force. As shown in Figure 4.1, employment among larger SMI firms almost doubled between the late 1980s and 1995, and the average beneficiary expanded employment by 22 to 56 employees. Our results also suggest that larger SMIs hired more workers and they hired them faster. Smaller firms tended to expand employment at lower rates. This relationship between small and medium SMIs appears to be the same whether we talk about the pre-1992 period (1985 to 1992) during a decline in manufacturing output, or during the boom at the beginning of the 1990s (1992 to 1995). Regardless of the size category, beneficiary firms have been able to expand employment at a greater rate than nonbeneficiaries. Indeed, during the boom period the mean and median employment growth rates of beneficiaries was more than twice that of nonbeneficiaries. 23 See annexes I and II for summary statistics on beneficiaries and nonbeneficiaries and a discussion of the sample. In what follows we rely on the definition of small, medium, and large SMIs used by the 1991 OED study of Bank support for SMIs, based on number of employees. Small firms have up to 10 employees; medium up to 100; and large SMIs more than 100. It is a definition that varies across countries and times. Our definition is also consistent with that used by the CBSL. 40 Figure 4.1: Employment by Firm Size - Sri Lankan Manufacturing, mid-1980s - 1995 9000 8000 *Mid-1980s # 1992 7000 8 6000 5000 4000 3000 2000 1000 00 O0012- 1-9 10-29 30-49 50-99 100-199 200+ Firm Size by Number of Workers Source: Econsult (1996). Question 2: How did SMIs increase employment? 4.4 The higher rate of employment growth by SMI beneficiaries stems from two sources: (i) a better performance on output growth; and (ii) an increase in the labor intensity of the firms. Beneficiaries' sales not only grew more rapidly than did nonbeneficiaries, they altered their capital labor ratios in very fundamental ways, as shown in Table 4.1. The table also indicates that capital intensity tends to decline with firm size rather than the other way around as many proponents of SMIs argue. 41 Table 4.1: Capital Intensity (K/L) of Firms, Sri Lanka, 1985 and 1995 No. Firms K/L (1985) KIL (1995) NDB Beneficiaries Large 23 0.74 0.19 Medium 20 0.59 0.25 Small 85 0.47 0.40 Non-Beneficiaries Large 45 0.15 0.15 Medium 26 0.20 0.18 Small 64 0.26 0.18 TOTAL 263 Note: 263 firms had data on fixed assets and total employment for 1985 and 1995 (out of a sample of 295 firms). Source: 1996 Econsult survey. 4.5 The relationship between input intensity and firm size is presented in another way in Figure 4.2. This figure shows that in Sri Lanka, as in most other countries, the capital-labor combinations of smaller firms are farther away from the origin than are those of larger firms. This location means that smaller firms tend to use more of both inputs, and accordingly, are less efficient than are larger SMIs. The figure also indicates that median-sized beneficiaries employ both more capital and labor for a given output level than do nonbeneficiaries, suggesting lower 24 efficiency in input usage. Thus, the smallest SMIs, both beneficiaries and nonbeneficiaries, are the least efficient of the SMIs. More disturbingly, based on Figure 4.2, the efficiency of medium- sized beneficiaries appears to be below that of nonbeneficiaries. However, statistical tests could not reject the hypothesis that the mean total factor productivity of beneficiaries and nonbeneficiaries were identical. Question 3: What kind ofjobs were created? 4.6 Mostly lower-skilled jobs were created, a result that has been a cause for concern by some. For example, Wignaraja (1997) argues that this kind of concentration is a problem. He says that if only simple, low-skill jobs are generated, then industrial upgrading and, ultimately, development will not take place. His conclusion is that more attention should be given to creating more technologically advanced jobs. Others, such as Little, Page and Mazumdar (1987), citing the example of India, have argued that entrepreneurial and PFI decisionmaking, rather than firm characteristics, should guide funding and investment decisions. Similarly, a Bank study on industrialization in Korea, India and Indonesia (1991d) warns of the high incidence of "government failure" in attempts to "pick firms" and force the pace by which countries move up the ladder of complexity. 24 The figure also indicates that large SMI beneficiaries employ both more capital and labor. However, the difference between beneficiaries and nonbeneficiaries is not statistically significant. 42 Figure 4.2: Average Capital and Labor Requirements Among Sri Lankan Firms 6.00 Fixed Assets Small-NDB unit of 5.00 Value Added (KIVA) 4.00 Small-N) m-NDB Medium - NDB a 3.00 0 Large - NDB 2.00 Medium - Non-NDB / 0 Large - Non-NDB 1.00 0.00 I 10.00 11.00 12.00 13.00 14.00 15.00 16.00 17.00 18.00 19.00 Labor per unit of Value Added (LVA) Note: 263 firms had data on fixed assets and total employment for 1985 and 1995 (out of a sample of 295 firms). Source: 1996 Econsult survey. 4.7 To provide some empirical content to this discussion, we applied a number of definitions of firm complexity to the firms in our sample. We found that most of the firms are what might be termed basic, noncomplex firms, regardless of which definition is used. For example, whether we use the definition of Bradburd and Levy (1995), Wignaraja (1997), or that implied by Biggs and Srivastava (1996), 73 percent to 81 percent of our observations, can be characterized as noncomplex firms.25 However, we also found that the creation of more complex firms has not been limited. Indeed, we found that a group of high-growth complex firms has spontaneously emerged without special targeting. 4.8 For instance, based on Biggs and Srivastava (1996) analysis a group of "transactionally" complex firms can be defined. These firms: (i) rely on overdraft facilities from banks, and (ii) make use of interfirm contracting, which we measure by trust receipts rather than accounts 25 Bradburd and Levy (1995) identify six characteristics by which the complexity of firms' operations can be characterized. They conclude that noncomplex firms produce food products, furniture and woodworking, garments and textiles, i.e., 73 percent of the firms in our sample. 43 receivable.26 According to this perspective, firms that engage in both practices have an ongoing banking relationship which allows them to respond to shifts in demands or deliveries. They also have developed relationsh s with clients or input providers that imply that they undertake more sophisticated transactions. 4.9 Table 4.2 shows the share of firms in our sample in each size category that are complex according to this last definition. It also shows the percentage of their sales financed with trade credit. By this definition, 23 percent of the firms, equally split between beneficiaries and nonbeneficiaries, are "complex." The amount of trade credit relative to sales for these firms appears to be higher than that obtained in the five African countries considered by Biggs and Srivastava (1996), but the pattern of complexity increasing with firm size is quite similar.2 Only 11 percent of the smallest SMIs are complex, while almost half the larger ones are. Only 26 percent of the basic firms produce anything for export, while 38 percent of complex firms do. In addition, the wages paid to production workers by the complex firms are more than 40 percent higher than those provided by the basic firms. Furthermore, even though production workers at complex firms are paid substantially more, the ratio of their administrative workers' wages to that of production workers increases to 1.74 vs. 1.55 for simple firms. Table 4.2: Firms Using Both Overdrafts and Trust Receipts Frequency Small (1-49) Medium (50-99) Large (100+) Total Share of firms (%) 11 18 49 23 Trade credit as% 9 11 15 13 of sales Source: Econsult survey. 4.10 As far as the types of jobs created, it appears that Wignaraja (1997) and Lall and his co- authors (1996) are right. Regardless of which definition is used, the jobs created by SMIs in Sri Lanka have largely been those of relatively basic producers who follow simple contracting and financial practices. Moreover, if we use Wignaraja's definition of complexity, we find that these firms invest more in fixed assets than in new equipment, and behave just like the basic firms as far as employment, sales, and financial patterns. Thus, at first glance, it also appears that most 26 A trust receipt is typically used to finance imports or exports. For example, a customer can get a loan from a bank against a letter of credit sent by his buyer. Our sample data did not permit us to separately measure "accounts receivable" as did Biggs and Srivastava. 27 In contrast, defining the complexity of Sri Lankan firms by reference to the U.S. wage structure, as does Wignaraja (1997), is a problematic perspective for SMIs in Sri Lanka in a number of regards. First, it defines a firm as a high-skill one only if its workers undertake the functions of the highest skilled workers in one of the world's most technologically advanced societies. While this perspective may be appropriate for an analysis of exports, such as he undertook, it is less so for SMIs. In Sri Lanka, SMIs produce largely nonexported goods; these are not products which compete with the goods produced elsewhere. Second, this definition emphasizes that skills are dictated only by production behavior. As Lall (1992) shows, production is only one of a number of skills which are important for industrial upgrading. The ability to make investment and financial decisions, and the ability to interact with other firms on an ongoing basis can be equally important skills. 28 The simple average for a firms in Cameroon, Ghana, Kenya, Tanzania, and Zimbabwe is S percent. 44 SMI firms do not appear to be vehicles through which technological advances and much higher rates of growth can be diffused throughout the society. However, this view changes if we use a broader notion of complexity. 4.11 Table 4.3 presents data on the behavior of firms that are complex according to the Biggs and Srivastava approach. The data indicate that if a "transactional" definition of complexity is used, there is indeed a group of firms generating higher rates of growth and employment, as well as more sophisticated transactions. While this group of firms accounts for a small share of firms, 23 percent, it is, nevertheless, a much larger share than that estimated by reference to a "job skill" definition of complexity, i.e., about 9 percent. Moreover, it is a group that is emerging quite spontaneously, without special targeting. Table 4-3: Selected Characteristics of Complex and Basic Firms Complex Basic 1985-92 1992-95 1985-92 1992-95 Employment growth 9.6 8.9 4.6 2.0 Sales growth 25.4 7.7 3.4 6.5 Share of sales exported 29.6 19.4 Share of firms that export 38 26 Source: Econsult survey. 4.12 In the end, we are left with the conclusion that most of the SMI jobs created were relatively low skilled. Certainly the compressed wages of production workers by firm size is consistent with this result. Production workers in the largest SMIs received only 7 percent more in wages than those employed by the smallest SMIs, and 4 percent more than those employed by medium-size firms. But, as is discussed further under question 5 below, there is nothing inherently "wrong" with this result. 4.13 Indeed, the opposite may well be the case. We have shown that there is a significant share of complex SMIs which are indeed more dynamic and generating higher rates of growth. Consistent with this result is the fact that real wage growth of those employed by SMIs-i.e., unskilled workers in the unorganized private sector-was higher over the 1979-93 period (it grew at 2.9 percent) than that of any other employment category (i.e., agriculture, 1.7 percent; wages of those covered by wage boards, less than 1.0 percent; and government employment, 1.7 percent). In addition, we show below that there is no evidence that technical assistance has played a role in the apparently spontaneous development of these firms. In sum, most SMI employment has not been high tech and export-oriented. It has taken place, however, in a sector 29 Computed from Table 1.8 in the Poverty Assessment (1995). 45 of the labor market that appears to be very resilient, and it has taken place in a way that suggests industrial complexity is developing without special government assistance.30 Question 4: How durable are the jobs? 4.14 Many of the jobs do not last very long. As Winter (1995) argues, small firms are not only disproportionate creators of new jobs, they are also disproportionate destroyers of jobs. Our survey indicates that within eight years of receiving a loan 40 percent of the beneficiaries no longer exist.31 While we do not know whether these firms failed or just could not be located, this exit rate is consistent with the failure rate for SMIs in developed and developing economies. Weston and Brigham (1978), for example, estimate that the small business failure rate in the US is between one third and one half of all firms. Winter (1995) cites similar evidence showing that over a decade the exit rates for U.S. manufacturing firms exceed 60 percent. Similarly, Behrman and Deolalikar (1989) find that in Indonesia over 1975-85 only 45 percent of firms with more than 20 employees, a less vulnerable group than our sample, survive. Finally, Roberts and Tybout (1996) show that average exit rates for U.S. firms over a five-year interval were between 41 and 50 percent. They also show that the average annual firm exit rate was higher in Colombia and Chile than it was in Sri Lanka, while the rate in Morocco was similar to our estimate. As Roberts and Tybout (1996) emphasize, exit rates by themselves are not a cause for concern. They reflect the "turnover" rate of an industry, and high levels can be very healthy as long as the corresponding entry rates are higher than the exit rates. Our data, unfortunately, do not permit such exit-entry comparisons. 4.15 However, high exit rates do affect estimates of job creation. The high exit rate for SMIs indicates that the typical analyses of SMIs that estimate the number of jobs generated often 32 overstate their effect. If the high exit rate of firms reflects a high failure rate, the employment does not last. Consequently, caution must be applied to employment estimates. For example, the first three complete SMI projects in Sri Lanka were estimated to have created 67,680 jobs between 1979 and 1993. Considering that between the 1971 Census and the 1993 Labor Survey the total number of employed Sri Lankans increased by 350,788. SMI support would have generated almost 20 percent of the increase in employment. However, when firms close down, the jobs that were created are lost or "destroyed." Hence, the estimate must correspondingly be reduced. 30 Lail and others (1996) provide a list of ineffective tax and regulatory restrictions that impede the development of firms, the reduction or elimination of which would, no doubt, contribute to firm growth and development. 31 Our survey data reveal an exit rate for beneficiaries. Unfortunately, we have no comparable information on nonbeneficiaries. This pattern reflects a sampling characteristic in which only firms that survived were included in the census of manufacturing of nonbeneficiaries. As a result, none of these firms exited as did our beneficiaries. 32 The OED study (1991a) discusses other serious problems with approaches to the measurement of employment effects that rely on simple partial perspectives ofjob creation; Winter (1995) also emphasizes these second order effects. We do not focus on them here. Rather, our objective is to emphasize that measures ofjob creation should account for firm exit. 46 Income distribution Question 5: Did SMI support address poverty? 4.16 Yes. The creation of jobs among unskilled workers lifts people from poverty and improves income distribution. Further, production employees in SMI beneficiaries, in general, received wages 22 percent lower than did similar employees working for nonbeneficiaries. However, recall that SMI beneficiaries also have had: (i) a more rapid employment expansion; (ii) a sharp relative increase in labor intensity; and (iii) higher average value added per worker. In this context, it appears very likely that the lower wages of SMI beneficiary production workers reflect the employment of lower-skilled workers. If so, the effects of the SMI support on poverty reduction were very strong. Such results would be consistent with the findings of Fei, Ranis and Kuo (1979) for SMEs in Taiwan (cited in Levy 1994). Question 6: Which type of SMIs had the greatest poverty impact? 4.17 The larger SMIs had the most beneficial effects on poverty reduction due to their greater employment of low-skill workers and women. While small firms employ proportionally more casual workers-39 percent of all production workers are casuals in small firms, compared to 30 percent and 22 percent for medium and large firms, respectively-large firms employ more casual workers in absolute terms. Indeed, large firms employed more than double the number of casual workers than small firms in our sample-and they paid them 12 percent more than did the small SMIs. Consequently, the advantage of smaller SMIs indirectly contributing to poverty reduction is relatively lower. 4.18 On the other hand, had the objective been to promote total employment of women in manufacturing for, say, social development, small firms would still not have been a good instrument: women represent less than a fifth of administrative workers in small firms versus more than half in large firms (the difference between production workers is narrower: small firms have a third of female production workers while larger firms have more than half). Again, the larger firms also paid administrative workers 70 percent more than small SMIs. 4.19 Finally, among the direct beneficiaries, female entrepreneurs are underrepresented. Although these two groups do not necessarily correspond to the poorest ranks of Sri Lankan society, they represent disadvantaged groups that the World Bank has tried to assist. First, self- employed Sri Lankan women represent 18 percent of all self-employed according to the 1993 Labor Force Survey (ILO, 1994), but only 1 percent of the NDB beneficiaries in this sample (and 5 percent among nonbeneficiaries). These figures do not necessarily mean that there is discrimination against women because the results could reflect lower credit requirements, instead of a refusal to their need of credit. 4.20 Perhaps the main lesson of these results is that poverty alleviation is best addressed by not limiting credit to the smallest firms, as is often done in SMI programs, such as Ecuador's Value added per worker was higher for small beneficiaries and medium nonbeneficiaries. It was equal for large firms. The average over all firms was 20 percent higher for beneficiaries. 47 program. Larger SMI firms are not only more productive than smaller ones, they also employ far more poor people. Growth 4.21 SMs and entrepreneurs. According to a 1985 OECD Report and a number of studies of developed economies (for example, see Birch [1987]), in recent years SMIs have reemerged as dynamic factors in some industrialized economies, such as the US and Italy. In addition, they have also been of continuing importance in generating growth, in economies such as Japan, Thailand, Korea, and Taiwan. As a result, a number of studies have suggested that an emphasis on SMIs should be part of an industrial growth strategy for Sri Lanka. Question 7: Did Bank support for SMIs provide a breeding groundfor growth? 4.22 Yes. In contrast to the results for other countries, micro SMIs in Sri Lanka, i.e., those with less than 10 employees, often grew into larger firms. Over 56 percent of the beneficiary microfirms, which had employed less than 10 workers in 1987 or 1988, employed more than 10 in 1995. Eighteen percent of them grew to employ between 30 and 49 workers and I firm, 2 percent, employed more than 50 workers. In contrast, Grosh and Somolekae (1996) report that only 2.5 percent of microfirms in Botswana ever grew beyond 10 employees. They also report similar low graduation rates for microfirms in Nigeria, Kenya, Malawi, Swaziland, and Zimbabwe. 4.23 Liedholm and Mead (1992)'s findings were somewhat more optimistic. They found that only 20 percent of microfirms in a number of African countries ever graduate. Biggs and Srivastava (1996) report higher graduation rates for microfirms in Ghana (41 percent) and Kenya (50 percent). However, they also show that this graduation rate declines sharply when the period of observation is shortened to a 5 to 6-year period to 20 and 10 percent, respectively. In Sri 34 Lanka, graduation rates over the shorter period still exceed 40 percent. Finally, while not strictly comparable with our data, Kim and Nugent (1994) find that only 30 percent of SMEs in Korea with fewer than 50 employees grew to a larger size. The Korean firms were not only larger but also older. Hence, if only 30 percent of these firms "graduate" in size, we expect a lower rate for smaller SMEs. 4.24 In addition, in an indirect sense, the credits also supported entrepreneurs and growth in a number of important, intangible ways. Lower tariff barriers, more competitive access to credit, less licensing control of production, and fewer restrictions on employment, all create better opportunities for profit and entrepreneurs. The fact that nonbeneficiaries graduated at only slightly lower rates, rates that still exceeded those of microfirms in most other countries, indicates that the economic environment was relatively supportive. Because our data do not include firms that failed our graduation rates are higher than those surveys that allow for failures. Nonetheless, even if the failure rates for micro SMIs were twice those of SMIs generally, the graduation rates would still exceed 25 percent, still a high rate. 48 Table 4.4: Problems Faced by Firms Percent offirms identifying the problems Beneficiaries Nonbeneficiaries Most important Cost of financing 34 21 Tax level and regulations 27 26 Economic policy uncertainty 17 17 Availability/Access to finance 13 16 Least important Import/Export duties 59 51 Marketing support from government 54 46 Source: 1996 Econsult survey. 4.25 Further, in the SMI entrepreneurs' opinions many of the policy problem areas had become unimportant. Table 4.4, for example, shows the most and least important problems faced by SMIs. It shows that tariffs and duties have become one of SMIs "least important problems," while the cost of financing, as opposed to access to finance, is one of, if not, the most important of their concerns. The second most important problem was the level of taxation and government regulations. Uncertain economic policies ranked third and access to finance came in fourth. Nonlending services Question 8: Was the policy dialogue effective? 4.26 Yes, in a number of respects. For example, from the replies of the entrepreneurs in our sample, it appears that SMIs in Sri Lanka are no longer rationed out of access to credit. Financial market conditions have clearly improved. Indeed, extending the broad index of the policy environment from Figure 2.1 to the present period indicates that the current policy environment is the most favorable it has ever been. 4.27 The fact that these policy changes occurred contemporaneously with the SMI credits does not, of course, imply that the changes occurred directly as a result of the credits, their covenants, or the studies funded by the credits. Rather, the reforms were the result of GOSL policy decisions; and these decisions were based on a wide range of inputs, including the policy dialogue. Nevertheless, as far as implementing the GOSL decisions, the credits provided resources for studies that helped define and articulate the specifics of these reforms. As the 1987 OED study puts it while "difficult to document and impossible to quantify ... 30 years of interaction between Bank staff and Sri Lankans officials.. . . Influence was mutual and not unidirectional, but it certainly was significant in the view of Sri Lankan officials" (p. 15). 49 4.28 On the other hand, the recommendations of these studies were often not adopted, or adopted very slowly. For example, despite the credits' concern with tariff reduction, for some sectors these reductions have been very slow in being realized. In a number of the more protected industries-such as basic metals, apparel and leather, and textiles-throughout much of the period, SMI credits provided support to industries that operated behind higher tariff walls. Nevertheless, tariff reductions and rationalizations did occur, sometimes with notably positive effects. For instance, consider the effective tariff reductions enacted in 1984 immediately after the SMI-sponsored study. One way to do this is to use the framework developed by De Long and Summers (1993). In Box 4.1 we show how this framework can help consider the value of some of the technical work undertaken as part of the policy dialogue. The box suggests that, because of the success of one technical study, the rate of return on the expenditures for all nonlending services provided under SMI I and H exceeds 35 percent. This is obviously a very simplified perspective and it suggests a very high rate of return. However, it is not only consistent with other analysis of Bank sector work, it is conservative relative to the rates of return that have been estimated for Bank ESW by Schneider (1996). Box 4.1: The Return on Policy Dialogue Expenditures De Long and Summers show that each extra percent of GDP devoted to equipment investment is associated with a .3 percentage point increase in the annual GDP per worker growth rate. In Sri Lanka private sector investment in plant and equipment is heavily dependent on imports according to Athukorala and Jayasuriya (1994). Over the decade following the 1984 tariff reduction this investment was 6 percent of GDP, slightly higher than in the 1978-84 period. They emphasize that the main source of differences in the cost of capital equipment is tariff and exchange rate policy. The 1984 tariff reductions reduced the price of imported equipment by more than 30 percent. This price reduction is equivalent to increasing investment in plant and equipment by 1.8 percent of GDP (the 6 percent of GDP in such investments times the 30 percent price reduction). Furthermore, regardless of the price reduction, the firms in our survey indicated that the most important change in firm operation was due to the increase in the quality of equipment which, due to the tariff reductions, they could now import more easily. According to De Long and Summer, this kind of increase in newer, more productive equipment in turn, would yield a half percent increase in the annual growth rate of per worker outcome, or .2 percent point increase in the per capita income growth rate.35 An increase in growth of .2 percentage points per year in per capita income growth is an enormous welfare-enhancing measure. Of course the movement to the lower tariffs realized a decade later would have been even more productive. But, to get a sense of how important the changes implemented were consider the present value of the gain realized. If the .2 gain in GDP growth was maintained and discounted at a 4 percent real rate, in 1985 dollars, it would be worth 5 percent of GDP, or about $220 million dollars. If this payout is compared to the $6 million in technical assistance from SMI I and II, or the approximately $8 million of costs, inclusive of Bank staff costs, over the 8 years preceding the reforms, the rate of return on the costs of carrying on the policy dialogue exceeds 35 percent per year. Even if the $40 million in credits are also included as part of the costs, as grants, the rate of return still exceeds 20 percent per year. The labor force is 40 percent of the population. Hence, the overall growth rate in income is correspondingly reduced. 50 4.29 This kind of perspective on the possible high rate of return to specific components of the policy dialogue suggests just how important the policy dialogue can be. If acted upon, and targeted on important distortions, it can generate growth opportunities. On the other hand, if the advice is weak or not acted upon, it can, as suggested by Lall and others (1996), impede both SMIs and growth. Thus, the emphasis given to policy dialogue with a committed borrower seems appropriate, even if the gains are difficult to measure and are not always realized easily. Question 9: Was technical assistance (TA) effective? 4.30 Partially. We were unable to discern any impact of the provision of technical assistance on sales or employment growth. Nor does it appear that complex firms (of any definition) relied more on TA. Less than 10 percent of beneficiaries had participated in some sort of short-term training (such as courses, lectures, demonstrations, workshops, seminars, exhibitions, etc.). Only 10 firms in the entire sample of 304 indicated that they changed their production activities in any significant way. 4.31 These results, as to the lack of measurable impact on firm behavior, are consistent with those of Wignaraja (1997), Lall and others (1996), and Athukorala and Rajapatirana (1991). While the many criticisms of the institutional mechanisms through which this assistance was given seem reasonable, it is worth remembering how difficult it is to measure such effects on firm behavior. Because of this difficulty, it is equally difficult to infer whether the failings of the TA expenditures were due to institutional failings by the TA providers, or the existence of simultaneous and offsetting effects. Consequently, on empirical grounds recommendations to greatly expand these expenditures, as in Wignaraja and Lall and others, should be treated with considerable caution. 4.32 Ultimately, the clearest effects of TA was on what might be termed "learning by doing." All the PFIs indicated that their ability to underwrite, monitor and evaluate business lending improved significantly as a result of their involvement with the SMI credits. Similarly, as Box 4.2 suggests, increased firm access to imported capital goods carried with it a form of embodied technical knowledge. All firms in our survey agreed that the two most important changes in their production behavior was with respect to the quality of the product they made, and their capital equipment. These results are consistent with the findings of a number of recent Bank studies of SMEs in a number of countries, summarized in Levy and others (1994). This study says that central government institutions-as opposed to decentralized organizations or the private sector-are generally not the most effective providers of export marketing or technology support.36 4.33 To summarize, the effects of nonlending services are difficult to measure. The audit of the first two SMIs criticized the performance of the operation of the Lapidary School, the Jewelry Training School and the Mushroom Development Center. Such ineffective uses of funds are an important concern, and one that has been mirrored in the assessment of public institutions in other countries (see, for instance, Berry and Levy [1994] for Indonesia, and Itoh and Urata [1994] for Japan). 36 The studies summarized were undertaken in Colombia, Indonesia, Japan, and Korea. See Levy and others (1994) for the citations of the various country studies. A World Bank study of small scale industry in India reached a very similar conclusion. 51 4.34 However, in Sri Lanka, the weaknesses of the support for public providers of TA is less significant in light of the apparent benefits associated with the discussions on trade and financial reform that were simultaneously pursued by the policy dialogue. From this broader developmental perspective the total nonlending services should be seen more positively. The high possible "payoff" on individual components of nonlending services suggests that such expenditures should be viewed as being similar to the investments of a venture fund in potentially profitable endeavors. Like venture fund investments many of the "investments" in nonlending services will fail. However, as long as a few strategic ones succeed, as they appear to in Sri Lanka, the value of the total nonlending services portfolio can be very high. Government policy and SMIs 4.35 SM7s are credit constrained. "Significant . .. potential exists among small and medium industries . . . A potentially major constraint to [this] expansion in the medium term is the availability of bank credit" (Prospects for Industrial Development, 1979). Thus: Question 10: In the competition for funds, have SMIs been disadvantaged relative to larger firms? 4.36 Yes, and while still important, disadvantages in access to credit have declined significantly over time. As detailed in the next chapter, credit has played an important role in allowing SMIs, and particularly beneficiaries, to grow at much higher levels than they would have in the absence of such credit. In addition, as shown in Table 4.5, the financial ratios of all SMIs had, by 1992, improved substantially relative to those of the late-1980s. In the next chapter these ratios will be discussed further, for now it suffices to note that secular improvement in access to finance and corporate structure appears to have occurred for all borrowers, not just SMI beneficiaries. There has been a secular increase in debt issuance and in the share of long- term debt. There has also been an increase in the share of investments in nonfixed assets as well as a much more intensive use of fixed assets. For those firms that we characterized as being "transactionally" complex, these im rovements, described in the last column of Table 4.5, are to even higher levels of performance. 4.37 The effects that the composition of the improvement in access to credit might have on firm performance can be discussed in only the most general terms. The theoretical literature on how firm debt maturity structure affects firm performance is inconclusive (see Caprio and Demirgiig-Kunt 1997). As a result, it is not possible to argue whether long- or short-term maturities would be the preferred form of finance. Nevertheless, from a simple comparison of the performance of complex firms relative to other firms and the financial structures of both types of firms, as described in columns 3 and 4 of Table 4.5, it appears that it is shorter-term debt that is most important. More concretely, as we show in the next chapter, we can draw conclusions about the overall effect of credit availability on firm performance: availability improves it. Firms defined as complex according to production skills, i.e., according to Wignaraja's definition, do not show such improvements relative to those of basic firms. In fact, they show lower turnover, a greater share of fixed assets in total assets and debt ratios more similar to those of all firms in 1985. 52 Table 4.5: Financial Ratios Over Time: All SMIs 1985 1992 1995 1995 Complex firms Long-term debt/Total assets 0.06 0.12 0.10 0.10 Short-term debt/Total assets 0.12 0.18 0.19 0.29 Total debt/Total assets 0.19 0.3 0.29 0.39 Long-term/Short-term debt 0.29 0.38 0.32 0.26 Fixed assets/Total assets 0.81 0.75 0.73 0.66 Sales/Fixed assets 1.93 2.58 3.21 3.48 Source: 1996 Econsult survey. 4.38 The cost to government of supporting SMIs. Even in a deregulated, market-oriented economy many feel that small industry needs special help from government to overcome some of its initial handicaps. The question is, what form should it take? A fairly standard approach, and one adopted by the SMI credits, is to provide this assistance through credit, technical advice, organized along the lines of an extension service; training programs for managers and skilled workers; and the provision of industrial estates that provide sites with infrastructure and regulatory advantages, such as the Free Trade Zone outside Colombo. The idea behind this kind of package is to help inexperienced entrepreneurs over their early hurdles, introduce them to regular marketing channels, and eventually make them self-reliant. Thus: Question 11: Was it expensive to help SMIs? 4.39 No. The interest rates paid by SMI beneficiaries and nonbeneficiaries tend to fall with firm size. In addition, nonbeneficiaries-who are not covered by the CBSL coinsurance program for SMIs, and which will be discussed in Chapter 5-pay higher interest rates. These represent risk-adjusted real interest rates of 5 percent to 7 percent. This is a figure that indicates complete cost recovery even if administrative and monitoring costs were moderately high. As confirmed by interviews with the PFIs and an ADB project completion report (1994), SMI lending to both beneficiaries and nonbeneficiaries has been profitable. By relying on banks, rather than government agencies, to contact the numerous and varied individual entrepreneurs, the credits relied on the self-interest of PFIs to select and monitor performance. Box 4.2 compares these credits with other credit lines now in operation in Sri Lanka. 53 Box 4.2: Effectiveness of Directed Credit Programs In a recent article Stiglitz and Uy (1996) present six characteristics of effective directed credit programs that can be inferred from the experience of what is known as the East Asian Miracle. It is instructive to use these characteristics to compare the Bank-supported SMI credits with two other directed credit policies now in operation in Sri Lanka. In addition, because the other programs are subsidized, it is also appropriate to compare them on the effectiveness of their subsidy targeting. Bank-SMIs Janasaviya New GOSL Trust Credit SMI Program Program 1. Moderate or no subsidy yes no no 2. Institutions monitor performance effectively yes no yes 3. Ability to change rapidly yes no ? 4. Directed towards private firms yes yes yes 5. Limited amounts yes yes yes 6. Credible DFI yes no yes 7. Targeting yes ? no The Bank-supported SMI credits have all the characteristics of the effective directed credits used in East Asia. At real interest rates of 5-7 percent, and with the administrative and credit risk cost estimates of the next chapter, there is no subsidy to borrowers; recovery is at commercial rates; interest rates and terms can be varied by the lender during the course of a loan; private firms use the credit and because they have to repay and the funds are not subsidized, they only take the credit if they can use it. Thus, the credit is targeted on high return going-concerns. The newly announced GOSL SMI program, in contrast, would make credit available to SMIs at interest rates as low as 10 percent nominal rates. The Janasaviya Trust Credit program is targeted on poverty alleviation rather than employment generation. Like the new GOSL SMI program, it also provides subsidized interest rates for eligible borrowers. It violates four of the six characteristics of effective credit programs. In addition, while the targeting, in this case on the poor, is clearly better, the many sources of inefficiency in such transfer devices suggest that credit programs are ineffective poverty instruments. Hence, the JTP is rated as questionable on this criterion. Finally, the central difference between the Bank SMI program and recently announced GOSL program is that the latter provides an interest subsidy. This feature causes the new program to violate two of the six characteristics of effective credit programs. It also violates the targeting criterion because it is not targeted on the poor or high return SMIs. 4.40 Locational effects. Regional industrial dispersal was one of the prime objectives of the GOSL at the time of SMI I, and remains one today as demonstrated by Two Hundred Garment Factory Programme which encourages production outside Colombo and Gampaha. It was also one of the measures of performance considered by OED (World Bank 199 1a) in its evaluation of SMIs in a number of countries. 54 Question 12: Did Bank support for SMIs promote regional decentralization as intended by the Appraisal reports? 4.41 Yes, mildly. As shown in Figure 4.3 the overwhelming share of SMI lending patterns followed those of the general spatial allocation of industry. Among the nine provinces of Sri Lanka, most of the industrial activity concentrates in the Western Province (where Colombo is located) where almost 80 percent of all the industrial value added of the country is produced. In this sense, it is understandable that a high proportion of SMI credit goes to this province, but the share relative to industrial value added is smaller. In terms of total credit, between 50 and 60 percent has gone to the Western province. On the one hand, the Northern and Eastern provinces have a combined industrial value added of 6.6 percent, but they received less than 1 percent of the SMI funds; although, they did receive 9.9 percent of SMI credit under SMI I. Their share no doubt dropped because of the conflict. On the other hand, other provinces receive a higher-than- proportional share of SMI (with respect to their industrial value added): Southern and Northwestern provinces had more than 10 percent of the total SMI credit, respectively, although they produce less than 5 percent of the industrial value added in the country. These figures suggest that SMI has expanded beyond the Western province so that some regional decentralization took place. Figure 4.3: Allocation of SMI Credit and Industrial Value by Province Northern & Eastern mindustrial Value Added (1990) Sabaragarnuwa CISM I I,II WSMI II1 Uva ESMI I North Central North Western Southern Central Western 0 1 20 30 40 50 60 70 80 Source: NDB (SMI distribution by districts) and 1991 Annual Industrial Survey. 4.42 However, the aspiration that assistance to small-scale industry should be used to promote regional decentralization needs considerable qualification. It is true that many small firms are located in rural areas and small towns. These firms tend to be traditional enterprises that are unlikely to be integrated in the process of industrialization. While subsidies would keep such firms going, they would only do so at a high cost to efficiency and long-term growth. The 55 subsidies would have to be very high and they would be targeted on rewarding inefficiency. Modem small firms, in contrast, tend to cluster in cities, perhaps even more densely than do larger firms, to take advantage of, as well as generate, spillover effects. 4.43 In other words, while large firms tend to rely on economies of scale to lower production costs, small firms depend more on economies of agglomeration generated by proximity to firms engaged in similar or complementary activities. The SMIs in our survey tended to operate in 38 exactly this way, that is, under constant return to scale. With this kind of production technology small producers will need access to intermediate material inputs. They will prefer to be close to transport and infrastructure facilities. If they are located in cities, they can benefit from being near urban growth centers where there already are intermediate inputs and skilled workers. Thus, the aspiration that SMIs will disperse industry away from cities is problematic. A considerable amount of Bank work has shown that providing locational incentives for firms is usually expensive and ineffective (see, for instance, Lee 1989). 4.44 To sum up, SMI support produced more employment, increased firm labor intensity and resulted in firms hiring production workers who earned 22 percent less. Nevertheless, the average value added per worker of beneficiaries was almost 20 percent larger than that of nonbeneficiaries. Almost certainly this support made a significant contribution to the relatively strong wage growth among lower paid workers. On the other hand, it does not appear that SMIs are making a significant contribution to export growth. In addition, regardless of how firm complexity is defined, most SMIs in Sri Lanka are basic rather than complex firms. Nevertheless, some types of complex firms are growing much more rapidly and thereby creating opportunities for greater specialization and ultimately industrial upgrading. Even the pattern of production, in which 75 percent of sales are on domestic markets rather than for export, does not appear to be unusual. Murphy, Shleifer and Vishny (1989) show that the Sri Lankan share of exports by SMIs is similar to that for all industry in other comparable countries. Thus, because large firms export more than do SMIs, Sri Lanka appears relatively strong on this score as well. 38 Preliminary analysis of our survey show interesting results about Sri Lankan firms. First, the growth in labor and capital explains growth in value added in an equation that resembles a Cobb-Douglas production function. It has constant returns to scale, i.e., the coefficients for labor and capital add approximately to one, so that proportional increases in labor and capital produce the same increase on average in the value added. This is consistent with the expected results for smaller firms. Second, the factor shares are also consistent with expectations: the growth in employment increases value added between 0.7 percent to 1.1 percent, and increases in fixed assets (investment in new machinery, including buildings and equipment) increase value added by 0.24 percent to 0.34 percent when the increase comes from NDB beneficiaries. Murphy and others (1989) cite the example of industrial output in Korea, the paragon of an open export-oriented economy. Between 1955 and 1973 domestic demand accounted for 53 percent of the growth of industrial output. This figure is, of course, significantly lower than our estimates for Sri Lanka. However, our estimates are, again, just for SMIs and the Korean figure is for all industry. 57 5. The Impact of Bank Support for SMIs on the Financing Patterns Overview 5.1 This chapter considers the effects that the Bank support had on the financial institutions that participated in the SMI projects and the financial practices of SMIs. Structured interviews with PFIs that disbursed more than 90 percent of the credit, as well as the NDB, were used to elicit some of this information. Like the last chapter, the discussion focuses on a series of questions about the effects of SMI operations. The questions address two broad themes: (i) financial market development; and (ii) the role of the government in the financial system. Financial market development 5.2 The concept of the credits was to build up the capability of financial intermediaries so that they could onlend IDA funds to creditworthy borrowers. In effect, the PFIs, acting as the agents of the GOSL, were retailers of the Bank's funds. However, at the time of SMI I the PFIs had limited ability to perform the underwriting and monitoring tasks performed by intermediaries in well-functioning financial systems. Instead, as discussed in Chapter 2, they had largely funded projects identified by GOSL. Hence, the first objective was to provide training and resources to the PFIs so that they would not have to await the development of a funding base. With Bank funds they could provide immediate support of what was thought likely to be high demand by high return investments. The liberalization of the financial system that began before the first project would eventually bring resources to the institutions on an ongoing basis. The Bank's first resources were to help catalyze this area of lending. 5.3 SMIs and financial market development. Financial systems play a key role in development, and their development is highly correlated with higher levels of growth. Thus: Question 1: Did the support for SMIs serve to enhance and complete financial markets? 5.4 Yes, but not because of the volume of SMI lending. The amounts of funds lent by the Bank, even with the cofinancing of the ADB, were not sufficient to change financial market practices. They were simply too small an amount. In total, they never equaled more than 5 percent of the assets of the commercial banks involved in the program. However, as discussed in Chapter 3, by working with the GOSL on a financial policy dialogue that led to a deepening and broadening of the financial sector, the Bank helped sustain a momentum towards maintaining a more flexible, buoyant financial sector. In addition to growth in credit to the private sector, as detailed in para. 2.27, the M2/GDP and other measures of depth all experienced increases at much higher rates than those experienced by comparator countries. 5.5 Cumulatively, figures 5.1 through 5.4 suggest that the access to credit of small and medium firms in Sri Lanka had by 1995 reached levels almost comparable to those of other higher income developing countries. While this result is interesting, it is important to stress that the data for the other countries should be taken as suggestive rather than strictly comparable to 58 our data. 4o The objective of presenting this data is not to make a precise comparison between Sri Lankan firms and those in other countries. Rather, it is to draw inferences about the evolving maturity structure of the credit used by SMIs, and to check the internal consistency of our observations. 5.6 Figure 5.1 presents ratios of Sri Lanka's indebtedness by firm size relative to those of an aggregate measure for 11 developing and 19 developed countries reported by Demirgiig-Kunt and Maksimovic (1 996a). The three columns on the right-hand side describe the behavior of Sri Lankan firms; the three on the left, the averages for the other countries. As expected, this kind of comparison shows that Sri Lankan firms are underserved relative to those in the other countries. However, even with our qualifications about data comparability, one surprise worth noting is that Sri Lankan SMIs perform relatively well on the share of long-term debt. In some respects, the comparison suggests a maturity structure more like that of a developed economy rather than SMIs in a low-income country. 5.7 Figures 5.2 and 5.3 show that in Sri Lanka long-term indebtedness is somewhat higher than the median of other developing countries, and lower than that of developed economies. They also show that these characteristics are more similar to those of the corporate financing patterns of firms in developed economies. In fact, on access to long-term debt Sri Lanka ranks higher than 9 of the other 11 developing countries. However, Figure 5.4 indicates that Sri Lanka's relatively high ranking in Figure 5.3 is due more to the shortage of short-term lending to private firms than it is to the availability of long-term debt. Thus, even though the last chapter suggested: first, that firm access to short-term credit had improved over time; and second, that the availability of short-term credit may be more important than long-term credit, it should be borne in mind that relative to other countries this improvement has been to a low level. 5.8 The immediate question is, why do Sri Lankan firms have such relatively limited access to short-term debt? One answer is the uncompetitiveness of the deposit market. Because of the very large role played in deposit mobilization by the NSB and the two SCBs, and the lack of competitiveness in this market, these funds can be used to channel funds to GOSL initiatives rather than private firms.41 If this answer is accurate, the source of Sri Lanka's relatively weak performance on firm access to short-term credit is the government crowding them out through its control of depository institutions. The firms in each of our categories were defined by number of employees and the firms in the Demirgtg-Kunt and Maksimovic studies by asset size. Each firm size in our categorization scheme is smaller than in their studies. Smaller firms, in general, perform less effectively than larger ones. As a result, because Sri Lankan firms are smaller, they should perform less well than firms in Demirgttg-Kunt and Maksimovic studies. On the other hand, there is another bias in the opposite direction. Our estimate of the debt ratios for a typical Sri Lankan firm is based on aggregating beneficiaries and nonbeneficiaries. By definition, 100 percent of the SMIs with access to Bank credits have credit, while only 70 percent of nonbeneficiaries do. Consequently, the estimates from our sample are higher than would be the case if our sample represented the universe of Sri Lankan firms. Nevertheless, the general overall comparability and consistency of trends is comforting as to the quality of our data. 41 A commonly used measure of market concentration is the Heyerdahl Index-the sum of squares of a firm's market share. For financial institutions an industry is considered concentrated when this measure exceeds 10 percent. The Heyerdahl measure for just the three depositories noted exceeds 20 percent. Hence, there is extreme concentration in Sri Lanka's deposit market. 59 Figure 5.1: Debt Ratios by Firm Size: Sri Lanka vs. Higher-Income Countries NLTDITA 0 STDITA 0.8 0LTD/TD 0.5 AFM BTDITA 0.4 0.3 0.2 TDITA LTD/TD 0.1 STDfrA 0 Small MeimLTDfTA High Meim Large Sml Income Sdal Medium Lag Lanka LTD/TA: Long-term debt/Total assets STD/TA: Short-term debt/Total assets LTD/TD: Long-term debt/Total debt TD/TA: Total debt/Total assets Source: 1996 Econsult survey. 60 Figure 5.2: Long-term Debt/Total Assets Ratio 0.3 0.2 0.25 01 k JI Figure 5.3: Long-term Debt/Total Debt Ratio 0.4 0.4 0.0 0.20 0.2 0.10 0.1 0.00 II1 Figure 5.4: Short-term Debt/Total Assets Ratio 0.6 0.0 0.4 0.3 0.2 0.1 0 Source: 1996 Econsult survey 61 5.9 SMIs and resource allocation. One of the central benefits of financial systems is their ability to allocate savings to high return investments. Thus: Question 2: Did the PFIs use the Bank's resources to support relatively more productive entrepreneurs? 5.10 Yes. Our data provide evidence that a high proportion of Sri Lankan firms, and an even higher share of beneficiaries, could sustain their current levels of sales only because of their access to credit. This finding is important because in many developing countries additional credit has little or no effect on output. For example, Demirgdig-Kunt and Maksimovic (1996b) show that in the 1980s two thirds of the firms in Brazil and South Africa could have sustained their growth levels without access to external finance.42 In addition, in the US, according to Cole and Wolken (1996), on the order of half of small firms finance production without any credit. In these environments, finance is not the constraint on firm growth. In contrast, in Japan, Thailand, and Korea two thirds of firms require external financing to realize their growth rates. 5.11 The beneficiary firms in our sample have a pattern much more similar to firms in East Asian countries. More than 70 percent of them required external finance to achieve the growth rates realized. Table 5.1 presents the share of beneficiaries and nonbeneficiaries whose growth exceeded that predicted by exclusive reliance on internal sources of finance. It indicates that the growth achieved by small firms, in particular, was much more reliant on external finance than that achieved by nonbeneficiaries. More medium-sized nonbeneficiaries achieved higher growth rates, but overall more than 70 percent of beneficiaries' growth rates were greater than the level that could be generated by relying only on internal funds. In addition, while SMI beneficiaries were generally less profitable than nonbeneficiaries, smaller beneficiaries were substantially more profitable, as well as more likely to be growing more rapidly than were small nonbeneficiaries. Table 5.1: Firms Requiring External Finance to Maintain their Current Growth Rate Percent offirms requiring finance * Firm size Share needing finance Beneficiary Large (100) 0.80 Medium (50-99) 0.59 Small (1-49) 0.71 Total 0.71 Nonbeneficiary Large (100) 0.75 Medium (50-99) 0.69 Small (1-49) 0.60 Total 0.67 * Computations based on a model of firm's financial position described in Demirgl9-Kunt and Maksimovic (1996a). Source: 1996 Econsult Survey. 42 We use the model in Demirgaq-Kunt and Maksimovic (1966b) to estimate the proportion of beneficiaries and nonbeneficiaries that could maintain their current growth rates without borrowing funds. Their model is a simple financial planning perspective that provides an estimate of the rate of firm growth that could be sustained by relying strictly on internally generated funds. Firms that grew at faster rates than that which could be supported by internal funding required external finance to achieve their growth levels. 62 5.12 Thus, from this evidence it is understandable why a private bank, like Hatton National Bank, has established a microenterprise program out of its own funds: this is a line of business that has a high return as well as high demand. In this respect, our data suggest that the SMI program may well have had positive demonstration effects. Further, as we showed earlier, in Sri Lanka over time most smaller firms have grown into larger ones with more employees. Consequently, by growing with their clients this kind of lending can help establish longer-term relationships with firms which are also increasing in size. Question 3: Did the Bank support for SMIs help mobilize savings for the most promising activities? 5.13 Yes. Because the smallest SMIs have less resources available to sustain shocks or liquidity problems, they are inherently greater credit risks than are the largest SMIs. Table 5.2 presents the interest rates charged to the largest and smallest size categories of beneficiaries and nonbeneficiaries. It indicates that interest rates tend to decrease with firm size as one would expect. In addition, if the CBSL coinsurance fee of 1 percent, which is required of all beneficiaries and is discussed below in para. 5.23, is subtracted from the return on beneficiary loans, lenders earn an average of 2.3 percent more on nonbeneficiaries' loans. These results suggest that lenders charge interest rates in such a way that larger and less risky SMIs pay lower interest rates, and higher expected risks pay higher rates. Table 5.2: Interest Rates by Size, 1995 (Means of interest rates) Beneficiary Nonbeneficiary 0(0) O/o) The means of interest rate: Large (100+) 17.03 18.44 Small (1-49) 18.09 19.34 Source: 1996 Econsult survey. 5.14 The table indicates that PFIs are discriminating between different risk categories, and rewarding more profitable firms with a lower, but still substantially positive, interest rate. Further, unlike the situation in a number of countries where the Bank also supported SMIs, even the largest firms in our sample remained eligible for SMI support. Indeed, larger firms accounted for 25 percent of the value of loans. As a result, our data do not suggest that the SMI credits have so far misallocated credit away from promising investments as they have in Ecuador for example, according to Jaramillo, Schiantarelli and Weiss (1993). The role of the government in the financial system Question 4: Do SMIs receive subsidized credit? 5.15 No. In the context of Sri Lanka's high real borrowing costs-over 8 percent in 1995- the modest interest rate reduction received by SMI beneficiaries relative to nonbeneficiaries should not be seen as a subsidy. Sri Lanka's high borrowing costs are directly attributable to policies which subsidize the performance of the SCBs. However, the cost of this subsidy is now borne in an implicit tax. This tax takes the form of the higher interest rates paid by those who 63 need to borrow funds, as do most firms in Sri Lanka. In this perspective, the interest rate reduction simply shifts some of the implicit tax for the subsidy to SCBs away from some of the SMIs who borrow to undertake investments. Thus, by lowering implicit taxes on productive expenditures SMIs' mild interest rate reduction furthers the credit's ability to target resources on promising activities. 5.16 Finally, Sri Lanka, like most Asian countries, has not developed a bond market, and in the late 1980s experienced severe problems in the development of nonbank institutions that would service this form of intermediation (see World Bank 1991b). Instead, it has relied on the NDB to fund banks to intermediate between longer term, largely external savers, and the demanders of funds. NDB was effectively capitalized by the interest rate transfers on the SMI credits. Its autonomy was also assured as a covenanted condition of a number of the credits. Finally, it has received, as well as disbursed, technical assistance through the credits that enhanced its technical strengths. Its privatization in 1993 stands in sharp contrast to that of many DFIs supported by the Bank. A 1989 Bank study, for example, found that 18 DFIs supported by the Bank had arrears in excess of 50 percent. 5.17 A number of studies have shown that the "signals" provided by a credible DFI, such as Japan's Development Bank, can have significant effects on the ability of promising firms to mobilize resources. In Sri Lanka, these indirect effects on investment efficiency are certainly muted by the distortions affecting institutional long-term savers. Nevertheless, the assistance that the SMI lines provided to the establishment of an effective DFI can be very important, and NDB's effective record in a difficult environment is a substantial accomplishment. 5.18 SMIs and risk allocation. Sri Lanka is an economy subject to more-than-average business risks. Its GDP growth rate, for example, is more volatile than 75 percent of non-African developing economies and all developed economies. Further, the prereform policies, by essentially making the government the equity partner in a very specific industrialization strategy, exacerbated this sensitivity. Policies which either allocate risks to those who are not well placed to bear them, or which prohibit risk-mitigating strategies can be very costly. Thus: Question 5: Did Bank support for SMIs contribute to a more efficient allocation of risks? 5.19 Yes, in two ways: (i) as a result of the credits default risks are now borne largely by entrepreneurs and private banks; and (ii) the pricing of the SMI credits more accurately reflects the underlying cost of funds. However, serious risk misallocations still persist. The improvements 5.20 Credit risk. Table 5.3 shows measures of loan recovery for the PFIs and the four SMI credits. It shows that the high initial arrears rates on SMI I and II, exceeding 30 percent in the late 1980s, as well as the weak recovery performance of the SCBs have been addressed. All participating banks have commercial level recovery rates for all four SMI lines.43 Clearly the loan recovery measures enacted in 1990, and the eligibility requirements for participation in SMI The modest decline in collection performance over the various lines appears to be due more to the seasoning of the loans rather than performance variation over the different credits. In addition, the collection rates on SMI IV may be temporarily depressed due to utility outages that occurred in 1996. 64 IV were very effective. However, it should be noted that the CBSL did capitalize the insurance fund that coinsures SMI borrowers, and it is not clear how this funding is accounted for in budget documents. 5.21 Interest rate risks. Under SMI IV, there was a shift from relying on the short-term average weighted prime rate (AWPR) to the average weighted deposit rate (AWDR), as an index for setting SMI borrowing rates. This shift reduced the interest rate risk exposure of the participating SMIs. The latter rate is a longer-term interest rate and so it is both higher and varies less than does the former rate. Table 5.3: Collection Performance of Participating Banks* Participating Bank SM II SMT III SMIIV Aggregate (%) (%) (%) (%) Bank of Ceylon 99.3 96.9 93.0 96.7 Peoples Bank 98.8 95.4 95.4 96.2 DFCC 98.0 95.4 87.3 93.2 Commercial Bank of Ceylon 99.9 98.0 90.9 94.9 Hatton National Bank 100.0 99.6 97.3 98.3 Sampath Bank 93.1 89.7 91.7 Seylan Bank 91.5 82.6 83.3 RRDB Kurunegala 94.7 89.4 90.6 RRDB Kegalle 96.0 96.3 RRDB Kandy 91.0 91.1 All Banks (weighted average) 98.9 96.0 91.5 94.9 (*) Cumulative collection ratio as of 6/30/95 Sources: National Development Bank of Sri Lanka and Central Bank of Sri Lanka. These data have been subjected to audits that verified PFI eligibility for participation in the SMI program. 5.22 On the other hand, the highly concentrated deposit market still directs more resources to public projects and penalizes entrepreneurs through higher interest rates. It also continues to impose significant risks on the public sector, a risk-bearing for which this sector has shown little aptitude. The SCBs have been recapitalized by the GOSL, but the history of public sector banks and DFIs in developing countries is, with a few notable exceptions, such as NDB, not an optimistic one. When this concentrated public ownership is combined with still high, though declining, level of public ownership of traditional agricultural exports, it is clear that significant misallocations of risk remain. 5.23 Finally, one of the ways that commercial repayments rates have been achieved on the SMI loans is through a cross-subsidy across SMI beneficiaries. The cross-subsidy occurs because the CBSL runs what is now an actuarially-sound coinsurance program for which it charges SMI beneficiaries a fee of 1 percent. All beneficiaries must pay the fee, and in return CBSL bears a 65 portion of the credit risk due to lack of repayment. Larger loans are entitled to coverage of a smaller portion of the debt. When the program was initiated, and banks were just beginning to establish underwriting and monitoring norms, the partial risk mitigation afforded by this program was very helpful in encouraging lenders to participate. These fees have been used to pay for a portion of the default rate implied by the high exit rates of SMIs, and cumulatively, according to the CBSL, the funds accumulated by these fees have more than covered all such costs. 5.24 As a result of this program, SMIs, who are "good borrowers" and who nevertheless have to buy the insurance, have helped pay for the borrowings of less effective SMIs. Given the access to credit noted in Figures 5.2-5.4, and implied by the reduced reliance on informal credit markets, there is no obvious reason why this program should continue. It causes good borrowers to have to pay more, and subjects the GOSL to a business risk that the private sector can now easily absorb. Question 6: Did Bank support for SMIs improve PFIs profitability? 5.25 Yes. The margins charged on SMI lending, on the order of 6 percent, were more than twice as large as those needed to cover roughly the 2.0 percent cost of the default risks of the loans.45 Because of the refinancing of SMI loans by NDB the PFIs did not bear interest rate risk. As a result, profitability required that their administrative costs could be covered with a 3.5-4.0 percent spread. All the banks involved agreed that these margins were more than sufficient to cover these costs. Further, a number of the banks expanded their SMI and microenterprise lending from their own portfolio, as our data on firm access to credit make clear. This pattern is one of the clearest forms of evidence of profitability. Indeed, while it is encouraging that banking lending for SMIs is profitable and hence sustainable, given the high spreads earned one can, in fact, question whether it is "too profitable." Finally, the privatization of NDB shows that despite reductions in the margins for onlending to 1 percent wholesaling SMI finds has also been profitable and sustainable. 5.26 To sum up, PFIs profitability undertook SMI lending and expanded their involvement in significant ways. The access to credit of nonbeneficiaries has improved significantly. The interest rate adjustments across beneficiaries and nonbeneficiaries by size suggests that PFIs are systematically pricing risks. This result, along with the finding that most firms needed the credit to sustain their current growth rates, implies that the credits are providing support for the most promising activities. Finally, the government's involvement in risk-bearing has declined as a 44 Our results indicate that less than 3 percent of SMI borrowing comes from informal lenders such as those Pettah money market in Colombo. This already small share is down from that of a 1986 Central Bank survey cited in Athukorala and Rajapatirana (1991). It is also lower than the share of informal borrowings in Cameroon, Kenya, and Tanzania, as can be inferred from Tables 2.1 and 2.4 in Biggs and Srivastava (1996). 45 The credit risk is equal to the present value of the probability of nonrepayment times the collateral value minus salvage value. For SMIs, salvage value on defaulting loans has been 50 percent (CBSL). The CBSL shares losses on all SMI loans with the originating PFI on a sliding scale of 60 to 90 percent of those losses, with smaller loans receiving higher coverage. The weighted average size of insurance coverage is almost 70 percent. Thus, the average loss per default is on the order of 20 percent (70 - 50). The CBSL charges an insurance premium of 1 percent per year on outstanding balances for the duration of the loans guaranteed. The fees collected have been more than sufficient to cover losses. Hence, if the PFI is subject to the residual risk, i.e., 30 percent of loan amount, it follows that a premium of 2.0 percent would be adequate. 66 result of the credits and financial sector dialogue. Nevertheless, additional ways to shift more risks to the private sector can still be realized. 67 6. Impact and Sustainability Overview 6.1 The ultimate policy objective of the credits was to help achieve economic growth with equity on a sustained basis. Over the past 20 years, a key aspect of the Sri Lankan-Bank strategy to accomplish this goal has been to help diversify the economy's structure so that the country's sensitivity to macroeconomic shocks could be reduced. Diversification is an important objective for two reasons. First, because of its product mix and size Sri Lanka is one of the world's more volatile economies. It is, as a result, an economy that is particularly sensitive to economic shocks. Second, policy reform is one of the key ways to achieve diversification because, as Athukorala and Jayasuriya (1994) show, it was government policies in the first place that had exacerbated the economy's lack of diversification. Through the centralization of ownership and decision-making during the 1970s the economy had become considerably less diversified. Consequently, part of the strategy was to reverse the policies that had increased the risks associated with the economy's already volatile environment. Finally, at the same time that diversification was pursued, particular concern has been given to maintaining Sri Lanka's impressive performance on human development.46 6.2 Sri Lanka's complete removal of exchange rate restrictions on international transactions, as highlighted by its acceptance in 1994 of Article VIII obligations of the IMF, is a landmark of progress toward the liberalization of the economy. Serious problems, however, remain with unfinished financial sector reforms, and the rationalization of employment policy. Furthermore, despite the measures to open up the economy, Sri Lanka's share of world trade in 1990 was virtually unchanged from 1980, and it was almost half the share of 1970. Increases in exports of non-traditional products, such as those produced by SMIs, increased dramatically, but, unfortunately, not enough to expand Sri Lanka's share of trade. 6.3 The SMI program fit into this broader diversification strategy in three ways. * First, it provided financial support and technical assistance to a wide range of private sector producers. This support helped decentralize investment decisions. * Second, it attempted to reallocate risks from the public sector to others, in the private sector, with greater comparative advantage in such risk-bearing. * Finally, it attempted to generate employment in an economy that has had high secular rates of unemployment for the past 20 years, and very high unemployment rates for those between 20 and 30 years of age. 46 Besides its well-known advances in literacy and other human development measures, the World Development Report 1996 identifies Sri Lanka as having the second lowest Gini coefficient of all countries. This coefficient is a frequently used measure of income distribution. The lower the figure, the more equal is the distribution.' 68 6.4 These objectives provide measurable indicators against which program impact can be evaluated. For example, Did SMI support contribute to a more diversified, sustainable productive base? Did the SMI credits result in the private sector bearing more of the risks that the government had in the past borne? And, did the credits efficiently generate employment opportunities? Impact and sustainability 6.5 Firms. As a result of the credits, the corporate financing of SMIs in Sri Lanka became better balanced and sustained at a higher level than it would have been in the absence of the credits. Firms hired more workers than they would have and they operate in a policy environment which is now much more conducive to contributing the highest value added. While the amount of additional employment generated was significantly less than earlier estimates, the cost of providing this employment is a fraction of that of other employment programs, such as the overstaffing of SCBs, or the public employment programs. Further, most of the jobs generated by the credits were of the lower paid sort, the type that poorer, less-skilled workers would fill. 6.6 It is true that a sizable share of this employment is in basic-skill industries, however that term is measured. Nonetheless, there does appear to be a significant share of more complex firms. These firms are, in fact, generating higher sales and employment, and the kinds of sophisticated transactions that can, in turn, generate greater specialization and technological upgrading. Furthermore, Sri Lanka's wage structure relative to Asian countries, suggests that Sri Lanka should be able to expand its export orientation. 6.7 Thus, the impact of the credits on firms' activities and employment was satisfactory, and the institutional development of firms has been substantial. It is highly likely that these results will be sustainable and that Sri Lanka will continue to experience an increasingly diversified productive base. Firms now operate in more competitive markets in which prices reflect resource costs and borrowings must be repaid. Furthermore, despite the continuing high deficits, which impede firms' access to credit, the policy environment, while still fragile, is much improved. 6.8 In terms of measurable indicators, over 1984 to 1992, Sri Lanka experienced the sharpest decrease in export concentration of all low-income countries.47 Thus, the economy is far more diversified than it was when SMI lending began. Nevertheless, due to the sensitivity of FDI to the ongoing hostilities, and the sharply declining trend in FDI since 1993, it will, perforce, remain not only fragile but increasingly reliant on domestic investments such as those undertaken by SMIs.48 6.9 Participating financial institutions. The PFIs became better monitors and evaluators of the financial needs of SMIs. Further, they have demonstrated a willingness to use their own resources to exploit what our evidence indicates is both a profitable and credit-demanding sector of the economy. In Sri Lanka finance was indeed a constraint on investment. The credits improved the operations of PFIs, provided the resources so that a DFI could be privatized, and World Development Report 1996. 48 Net FDI in 1995 was 11 percent of the 1993 level and net portfolio investment turned negative. (Source: CBSL.) 69 credit has been provided on risk-adjusted terms. Finally, entrepreneurs and PFIs became more directly responsible for bearing the business and interest rate risks associated with such lending, even if the public ownership of the two SCBs means that the government is still a large residual risk-bearer. 6.10 The increased private role in SMI intermediation is also likely to have beneficial effects on financial sector development. In general, financial depth and breadth increased during the credits, and in some ways, these improvements were at least indirectly related to the policies enacted under the SMI credits. The improvement in debt recovery under SMI III, in particular, seems to have had quite significant effects of the sustainability of the PFIs. In sum, the impact of the credits on PFIs was satisfactory, and it is likely that the results will be sustainable, despite the problems that remain with the SCBs. The institutional development of a number of the PFIs, besides NDB, has been substantial. 6.11 Nonlending services. The GOSL's main gains from the nonlending components of the credits were those provided by the policy dialogue's possible contributions to a much less distorted, more competitive economy. Neither we nor other analyses can adduce evidence that the other nonlending component of the credits-technical assistance for firms-was productive. 6.12 Unquestionably, the putative gains from the policy dialogue were often slow to be realized, and, as has been stressed, in some cases serious distortions remain. For example, as has been repeatedly stressed, the central cost of the operation of the SCBs is higher real interest rates throughout the financial sector. Their cost of intermediation is more than twice those of their commercial competitors (IMF 1996), and because of their tax advantages and concentration, these institutions set prices in both the deposit market and the financial system. If the inefficient operation of these institutions is responsible for just half the 6 percent interest rate margins on borrowings through the banking sector, then the costs of this subsidy are on the order of 1.0 percent of GDP, an amount only slightly lower than the annual revenue generated by the corporate income tax. This is a very expensive distortion, and one that heavily taxes productive investments. 6.13 Moreover, these implicit taxes are borne by investors in an expensive, nontransparent way that undermines financial development, contributes to the pressures for various ad hoc tax relief measures, and is almost certainly regressive. The fact that the share of deposits held by SCBs has declined slightly over time does not imply that these costs have fallen. The regulatory structure of the banking system is such that the competitors do not offer the lower interest rate margins that their more efficient production could afford. Hence, the costs of this distortion are a function of the size of the banking sector, and not the share of deposits held by SCBs. 6.14 On the other hand, the policies taken to improve collections, better match borrowing and lending maturities, and fulfill prudential capital adequacy requirements have reduced the GOSL's contingent liabilities on behalf of the banking system. (See Box 6.1 for a discussion of another way SMIs may have contributed to strengthening the government's position.) Similarly, while the reductions in the trade barriers will reduce tariff revenues, they will also, ultimately, contribute to the development of a more sustainable, faster-growing and more diversified manufacturing sector. The ultimate result should contribute to a more stable and buoyant source of tax revenues at lower tax rates. 70 Box 6.1: SMIs and Financial Fragility Recent analysis by Caprio and Klingebiel (1996) shows that over 80 countries have experienced banking sectors crises since the late 1970s. In many cases the losses realized were huge, totaling more than 20 percent of GDP. These crises occurred in both developed and developing economies. However, one frequent correlate of severe banking sector problems has been a highly concentrated export sector, such as Sri Lanka's. When countries with concentrated exports experience term of trade shocks-as did Sri Lanka, which between 1989-1993 experienced one of the largest shocks of the countries reviewed by Caprio and Klingebiel-crises typically result. The inability of borrowers to repay due to the price shock is transmitted through the banking system to contingent liabilities being realized by the government. In addition to the ubiquity of crises and their increased frequency in countries with export concentration, the most frequent common feature of crisis situations has been the slowness and recalcitrance with which governments have dealt with the problems. While macroeconomic circumstances were the primary precipitating factor in a crisis occurring, incomplete and slow policy responses were the cause of most of the costs of these crises. Thus, Sri Lanka's partial and incomplete measures in dealing with the insolvencies of the State commercial banks, despite studies that clearly laid out the nature of the problems, is fairly typical: slow and consequently more costly. One of the central ways Caprio and Klingebiel propose for averting banking crises in the future, particularly for smaller economies, is to let banks which are diversified across economies bear the risks that might be associated with, for instance, an individual country's export concentration. Alternatively, direct foreign investment would also reduce the risk exposure of domestic banks. Finally, if these measures are not followed, banks in such economies require higher capital requirements than do banks in less concentrated economies if prudential soundness is to be achieved. How do SMI loans in Sri Lanka fit into the question of the financial fragility of the Sri Lankan banking system? They can do so because banks in Sri Lanka's volatile economy are not diversified across countries. They are largely concentrated on domestic loans. In such a context, the relative performance of various forms of domestic lending is of increasing importance. Surprisingly, SMIs score very well on a relative basis. The collection ratio on SMI loans has been multiplicatively better than that on the larger loans made to bigger firms under IDP credits. The arrears rate on loans made under IDP III, a financial sector loan with unrestricted firm size eligibility requirements, was over 15 percent in 1995. For the SMI credits the comparable figure is 5 percent. This type of risk pattern almost never occurs in developed economies. In developed countries the default risk of the largest firms is very low, and the default risk of the smaller firms is so high that half of them do not issue debt. While these figures on performance under SMI and IDP credits do not directly translate into a proportional reduction in credit risk exposure, as might be implied by the arrears rates, they are, nevertheless, very suggestive when the differences are so large. Thus, SMI lending by domestic banks with size limits that eliminate the largest firms-for example, most IDP subprojects were also of sufficiently small size to be eligible to be financed through SMI credits-would reduce the contingent liabilities borne by the GOSL on behalf of a domestically restricted banking system. Of course, reducing the government's ownership shares in the SCBs would be a much more direct approach, and would have a much more significant effect. 71 6.15 In sum, the impact of the credits on the GOSL's financial position has been satisfactory. However, given the distortions that remain in the financial sector the sustainability of these gains is uncertain. 6.16 Technical assistance. Recent reviews of the technical assistance (TA) institutions that support exports reach conclusions similar to those of the audit for SMI I and II. That is, these expenditures do not seem to have been effectively used. Nevertheless, in principle, TA can be a very important instrument for helping SMIs hurdle the barriers that prevent their becoming effective, productive firms. Clearly, the experience of Japan, Korea, and Singapore, reviewed by Lall and others (1966), suggests that when correctly structured TA can be very productive. Levy and others (1994) suggest that an important aspect of the structuring of this assistance is reliance on private sector firms. At the very least, then, the lack of any apparent effects of the TA on firms seems to have been a missed opportunity. A longer-term perspective on the evaluation of the early SMI credits 6.17 The audit of the first two SMIs was critical of: (i) the low recovery rates on loans; (ii) that the projects emphasized the policy dialogue; (iii) the fact that the projects were concentrated in urban areas with limited impact on promoting new technological processes or rural industrialization; and (iv) that NDB did not have systematic information on subproject performance sufficient to calculate the economic rate of return on various subprojects. This audit downgraded the initial evaluation of SMI I from satisfactory to unsatisfactory, and rated SMI II as unsatisfactory. 6.18 A longer-term, more detailed perspective indicates that these criticisms no longer apply. The central problem identified was the low recovery rate-it was on the order of 60 percent in 1989. This was a serious problem. However, it has been addressed. As the eligibility for SMI IV shows, this rate was improved to over 95 percent by 1995. Criticism (ii) also appears weaker with the benefit of a longer-term perspective. The policy dialogue appears to have been highly productive on a number of fronts. Finally, criticisms (iii) and (iv) are problematic on conceptual grounds. The aspiration of changing the regional location of industries cannot generally be done by credit instruments. Nevertheless, the credits did result in a more geographically spread allocation of credit than would be predicted by regional industrial production, even if the technological advances were minimal. Similarly, the concern that NDB should know the economic rate of return on subprojects rather than the financial rate of return misunderstands the role of second tier intermediaries. NDB's privatization indicates that it has carried out its financial concerns effectively. 6.19 In sum, from this broader perspective, the evaluations of SMI I and II, should be seen in a more positive light. Through the prism of a longer-term perspective, the projects were not the unsatisfactory ones described in the audit. Nor should their sustainability be viewed as being uncertain any longer. Finally, institutional development has been substantial. Changing their ratings to satisfactory increases the percent of Sri Lankan projects with a satisfactory rating to 89 percent. This is the fourth highest rating among all borrowers which have had at least 40 projects, and next to China, the best performance of all IDA borrowers. 72 Lessons learned Although episodically effective, policy-focused lending has been very productive 6.20 We estimated that the rate of return on the study of tariffs implied a very high, if difficult to measure, rate of return on the nonlending expenditures of SMI I and II. Besides tariff reforms, the policy dialogue in SMI III and IV also contributed in a number of ways to financial deepening. Such deepening can generate similar growth effects as did the tariff reductions. In addition, the audit of Sri Lanka's two adjustment loans (World Bank 1996b) indicates that the increases in financial depth that had been achieved in the years between liberalization and the adjustment program in the late 1980s contributed significantly to the success of the adjustment credits. Hence, the indirect effects of the policy reforms have been highly productive. 6.21 Of course, this is not to say that all the reforms took place simply because a dialogue was taking place. Nor is it to say that all of the objectives of the policy dialogue were achieved, or even that most of the expenditures on them were useful. Sri Lanka, like most countries that have experienced financial crises, has been slow to complete the financial restructurings identified in SMI IV. Nevertheless, at least three or four of the issues focused on by the policy dialogue have had clear and obvious productive outcomes. Moreover, the apparently productive outputs of the dialogue were not broad recommendations or covenants. Rather, they were studies that provided technical details on how a particular reform could be enacted. In this sense, even if most of the expenditures appear to have had little value, the high "pay-offs" on a limited number of successes can more than compensate for the lack of return on most policy dialogue expenditures. In this respect, the Bank's support for policy reform is much like the investments made by a venture capitalist in risky endeavors: many may not bear fruit, but the expected return can nevertheless be very high. 6.22 When expenditures on policy dialogue are viewed in this light, two conclusions emerge: First, efforts to examine the econometric relationship between measures of aid and policy reform across countries, such as those of Burnside and Dollar's (1996), are unlikely to show much evidence of a link. Such an exercise is much like examining expenditures on oil-well exploration. Perhaps only one in ten wells will succeed, but for profitable firms, success more than pays for the failures. The question is not whether a dollar of aid leads to policy reform, but rather is there a positive rate of return on a "portfolio" of expenditures. Second, if most policy dialogue expenditures are not likely to bear fruit even if the expected return is high, then such expenditures should be undertaken by an institution that is broadly diversified across countries. No other institution matches the World Bank's comparative advantage in this respect. As employment programs, SMIs in Sri Lanka have been relatively cost-effective 6.23 According to a recent study (GuimarAes 1991), Sri Lanka is a society in which "if jobs are not created, the cohesion of the social fabric . .. may be in jeopardy ... especially taking into account the violent convulsions . .. in the recent past. Job creation must be treated as a major, if not the major, objective of any development strategy." In this context, efficient job generation is important, and as we have shown, the SMI support has helped effectively generate jobs. 6.24 Furthermore, when SMIs are compared to three largest ongoing employment programs: (i) the overstaffing of SCBs; (ii) the recurrent public hirings; and (iii) the termination of 73 Employment Act,49 their relative efficiency is strong. All of these other employment programs carry large, nontransparent costs, and none of them are targeted on poorer workers or employment in firms that must repay market rate debt. SMI credits, in contrast, are: targeted on productive workers, as well as workers who are also relatively, if not absolutely, poor. Moreover, the only costs of the SMI credits are attributable to the modest reduction in the high real interest rates that occur in the Sri Lankan financial system. The only other obvious source of employment creation of comparable efficiency is FDI. This is a source that is beyond direct government control, as well as one that has been declining sharply over recent years. 6.25 Thus, in many economies SMIs have much to recommend them. For instance, in high unemployment economies, and particularly those undergoing a shift away from large scale public manufacturing and public-owned banks, or in economies with tentative FDI, such as some of the reforming socialist economies, SMI "directed credits" can be effective. From the Sri Lankan experience, the Bank's post-FY91 revival of support for SMIs in countries such as Albania, Poland, and South Africa is an effective shift in orientation. However, these benefits must be weighed carefully against the possible costs to financial sector development. Financial sector reform would be sign ficantly more efficient than further SMI support 6.26 Our analysis of the financing of SMIs indicates that they confront two central financial problems: the high cost of funds and lack of access to short term rather than long term funds. Both of these problems stem from the structure of the SCBs. Additional SMI support, by itself, would not directly address either problem. Thus, as regards financial sector development, the, perspective stressed on the Appraisal Report for SMI IV still seems appropriate today. Financial reform is the key way to support SMIs as well as improve the financial sector. 6.27 However, when the Bank's aspirations for financial reform are viewed in light of the experiences of other countries undergoing financial distress (see Box 6.1), they appear ambitious. Few countries have reformed so extensively in the short-time period called for in the Appraisal Report, and even fewer have done so under the unstable fiscal conditions experienced by Sri Lanka. Furthermore, SMI lending appears to have helped diversify commercial banks' risks to a greater extent than did credits that allowed the largest firms to borrow from commercial banks. Moreover, this is a pattern that does not appear to be unique to Sri Lanka. Consequently, unconstrained FILs may often impose greater contingent liabilities on governments than would SMI credits. In sum, in many instances, Bank-support for SMIs may imply a greater reduction in the government's role in the financial system than does unconstrained support through FILs. 49 Although the last of these is not explicitly a funded program, its constraints on employers' ability to discharge workers makes it, in effect, into an employment program that is paid for by firms. See Prywes (1995). 75 Bibliography Asian Development Bank. (1994). Project Completion Report of the Small and Medium Industries Project in Sri Lanka. Athukorala, Premachandra and Sarath Rajapatirana. (1991). "The Domestic Financial Market and the Trade Liberalization Outcome. The Evidence from Sri Lanka." World Bank Policy Research, and External Affairs Working Paper no. WPS 590. Washington, D.C.: World Bank. Athukorala, Premachandra and Sisira Jayasuriya. (1994). "Macroeconomic Policies, Crises, and Growth in Sri Lanka, 1969-90." World Bank Comparative Macroeconomic Studies. Washington, D.C.: World Bank. Bank of Ceylon. Annual Report for various years. Colombo, Sri Lanka. Behrman, J.R. and A.B. Deolalikar. (1989)."... of the fittest?: duration of survival of manufacturing establishments in a developing country." Journal ofIndustrial Economics 38: 215-226. Berry, Albert and Brian Levy. (1994). "Indonesia's Small and Medium-Size Exporters and Their Support Systems." Policy Research Working Paper no. 1402. Washington, D.C.: World Bank. Bhalla, Surjit S. and Paul Glewwe. (1986). "Growth and Equity in Developing Countries: A Reinterpretation of the Sri Lankan Experience." World Bank Economic Review vol. 1, no. 1. Washington, D.C.: World Bank. Biggs, Tyler and Pradeep Srivastava. (1996). "Structural Aspects of Manufacturing in Sub- Saharan Africa. Findings from a Seven Country Enterprise Survey." World Bank Discussion Paper no. 346. Washington, D.C.: World Bank. Birch, D.L. (1987). Job Creation in America. New York: Free Press. Bradburd, Ralph and Brian Levy. (1995). "Zimbabwe's New Entrepreneurs: An Emerging Success Story?" Draft paper. Brock, Philip L. (1994). "Economic Policy Reform, Government Debt Guarantees, and Financial Bailouts." Policy Research Working Paper no. 1369. Washington, D.C.: World Bank. Bruton, Henry J. (1992). Sri Lanka and Malaysia. The Political Economy of Poverty, Equity, and Growth Series. New York, NY: Oxford UP. Burnside, Craig and David Dollar. (1996). "Aid, Policies and Growth." Draft. World Bank Policy Research Department. Washington, D.C.: World Bank. 76 Bibliography (cont.) Caprio, Gerard, Jr. and Daniela Klingebiel. (1996). "Bank Insolvencies. Cross-country Experience." World Bank Policy Research Working Paper no. 1620. Washington, D.C.: World Bank. Caprio, Gerard, Jr. and Asli DemirgUig-Kunt. (1997). "The Role of Long-Term Finance. Theory and Evidence." Policy Research Working Paper no. 1746. Washington, D.C.: World Bank. Central Bank of Sri Lanka. (1995). Annual Report 1994. Colombo, Sri Lanka: Central Bank. Chenery, H., S. Robinson, and M. Syrquin. (1986). Industrialization and Growth: A Comparative Study. New York, NY: Oxford UP. Cole, Rebel and John D. Wolken. (1996). "Bank and Nonbank Competition for Small Business Credit: Evidence from the 1987 and 1993 National Surveys of Small Business Finances." Federal Reserve Bulletin. Washington, D.C. (November). De Long, J. Bradford and Lawrence H. Summers. (1993). "How strongly do developing economies benefit from equipment investment?" Journal ofMonetary Economics. Vol. 32, no. 3, pp. 395-415. De Silva, K.M. (1981). A History ofSri Lanka. Delhi: Oxford UP. Demirgdig-Kunt, Ash and Vojislav Maksimovic. (1996a). "Term Finance: Does It Matter? Financial Constraints, Uses of Funds and Firm Growth: An International Comparison." Preliminary Draft. Paper presented to the World Bank Conference on June 14, 1996. ___. (1 996b). "Term Finance: Does It Matter? Institutions, Financial Markets and Firm Debt Maturity." Preliminary Draft. Paper presented to the World Bank Conference on June 14, 1996. Development Finance Corporation of Ceylon. Annual Report 1995-96. Colombo, Sri Lanka Dunham, David and Saman Kelegama. (1995). "Economic Reform and Governance: the second wave of liberalisation in Sri Lanka 1989-93." Research Studies Governance Series no. 2. Institute of Policy Studies. Colombo, Sri Lanka. Grosh, Barbara and Gloria Somolekae. (1996). "Mighty Oaks from Little Acorns: Can Microenterprise Serve as the Seedbed of Industrialization?" World Development, vol. 24, no. 12, pp. 1879-1890 GuimarAes, J.P.C. (1991). 'Industrialization in Sri Lanka: What can we learn from the NICs?" Research Studies Industrialisation Series no. 1. Institute of Policy Studies. Colombo, Sri Lanka. Hatton National Bank. (1996). Annual Report and Accounts 1995. " Colombo, Sri Lanka 77 Bibliography (cont.) Hulugalle, Sriyani. (1994). "Privatization Program of Public Sector Commercial Enterprises. Sri Lankan Experience." Draft Report. IMF Institute. (1996). Financial Programming and Policy: The Case ofSri Lanka. Washington, D.C.: IMF. Isham, Jonathan and Daniel Kaufmann. (1995). "The Forgotten Rationale for Policy Reform. The Productivity of Investment Projects." Policy Research Working Paper no. 1549. Washington, D.C.: World Bank. Itoh, Motoshige and Shujiro Urata. (1994). "Small and Medium-Size Enterprise Support Policies in Japan." Policy Research Working Paper no. 1403. Washington, D.C.: World Bank. Jaramillo, Fidel, Fabio Schiantarelli, and Andrew Weiss. (1993). "The Effect of Fifnancial Liberalization on the Allocation of Credit. Panel Data Evidence for Ecuador." World Bank Policy Research Working Papers no. 1092. Washington, D.C.: World Bank. Khatkhate, Deena. (1982). "Anatomy of Financial Retardation in a Less Developed Country: the Case of Sri Lanka, 1951-1976." World Development, vol. 10, no. 9. UK: Pergamon Press. . (1988). "Assessing the Impact of Interest Rates in Less Developed Countries." World Development, vol. 16,, no. 5. UK: Pergamon Press. Kim, Linsu and Jeffrey B. Nugent. (1994). "The Republic of Korea's Small and Medium-Size Enterprises and Their Support Systems." Policy Research Working Paper no. 1404. Washington, D.C.: World Bank. King, Robert G. and Ross Levine. (1993). "Finance, entrepreneurship, and growth: Theory and evidence." Journal ofMonetary Economics, vol. 32, no. 3, pp. 513-542. Lakshman, W.D., and others. (1991). "Changes in the Industrial Structure and the Role of Small and Medium Industries in Developing Countries: The Case of Sri Lanka." Institute of Developing Economies, Tokyo. Lall, Sanjaya. (1992). "Technological Capabilities and Industrialization" World Development, vol. 20, no. 2, pp. 165-186. UK: Pergamon Press. Lall, Sanjaya, and others. (1996). "Building Sir Lankan Competitiveness: A Strategy for Manufactured Export Growth." Report for the Government of Sri Lanka National Development Council. First draft report, volume I. Lee, Kyu Sik. (1989). The Location ofJobs in a Developing Metropolis: Patterns of Growth in Bogotd and Cali, Colombia. New York: Oxford UP. Levy, Brian (1993). "Obstacles to Developing Indigenous Small and Medium Enterprises: An Empirical Assessment." World Bank Economic Review, vol. 7, no. 1, pp. 65-83. 78 Bibliography (cont.) Levy, Brian, and others. (1994). "Technical and Marketing Support Systems for Successful Small and Medium-Size Enterprises in Four Countries." Policy Research Working Paper no. 1400, Washington, D.C.: World Bank. Liedholm, Carl and Donald Mead. (1992). "The Structure and Growth of Microenterprises in Southern and Eastern Africa: Evidence from Recent Surveys." Draft Paper (Department of Economics, Michigan State University, East Lansing, MI), January. Little, Ian M.D., Dipak Mazumdar, and John M. Page, Jr. (1987). Small Manufacturing Enterprises. A Comparative Analysis of India and Other Economies. A World Bank Research Publication. New York, NY: Oxford UP. Little, I.M.D., and others. (1993). Boom, Crisis, andAdjustment. The Macroeconomic Experience of Developing Countries. New York, NY: Oxford UP. Murphy, Kevin M., Andrei Shleifer, and Robert W. Vishny. (1989). "Industrialization and the Big Push." Journals ofPolitical Economy, vol. 97, no. 5. Chicago, IL: University of Chicago. National Development Bank. Report & Accounts 1995. Colombo, Sri Lanka. OECD. (1985). "Employment in Small and Large Firms: Where Have the Jobs Come From." Employment Outlook. Paris. September. Patel, Urjit R. and Pradeep Srivastava. (1996). "Macroeconomic Policy and Output Comovement: the Formal and Informal Sectors in India." World Development, vol. 24, no. 12, pp. 1915-1923. People's Bank. Annual Report 1991-1992. Colombo, Sri Lanka. Prywes, Menahem. (1995). "Unemployment in Sri Lanka: Sources and Solutions." South Asia Region Internal Discussion Paper. Report no. IDP- 154. Ramey, Garey and Valerie A. Ramey. (1995). "Cross-Country Evidence on the Link Between Volatility and Growth." The American Economic Review. December, pp. 1138-1151. Roberts, Mark J. and James R. Tybout (eds.). (1996). Industrial Evolution in Developing Countries. Micro Patterns of Turnover Productivity, and Market Structure. New York, NY: Oxford UP. Schiantarelli, Fabio and Alessandro Sembenelli. (1996). "Form of Ownership and Financial Constraints." Policy Research Working Paper no. 1629. Washington, D.C.: World Bank. Schneider, R. (1996). "An Interim Review of the Quality of Economic Analysis in Staff Appraisal Reports for Projects Approved in Fiscal Year 1996." 79 Bibliography (cont.) Stiglitz, Joseph E. and Marilou Uy. (1996). "Financial Markets, Public Policy, and the East Asian Miracle." Research Observer, vol. 11, no. 2. Washington, D.C.: World Bank. Webster, Leila M., Randall Riopelle and Anfie-Marie Chidzero. (1995). "World Bank Lending for Small Enterprises 1989-1993." Private Sector Development Department. Weston, J. Fred and Eugene F. Brigham. (1978). Managerial Finance. Hinsdale, Illinois: Dryden Press. Wignaraja, Ganeshan. (1997). Trade Policy, Technology and Manufactured Exports: Sri Lanka's Liberalisation Experience. Basingstoke, UK: MacMillan Press. Forthcoming. Winter, Sidney G. (1995). "Small and Medium-Size Enterprises in Economic Development. Possibilities for Research and Policy." Policy Research Working Paper no. 1508. Washington, D.C.: World Bank. World Bank. (1953).,Industrial Sector Study. . (1957). Some Problems Encountered in the Establishment ofDevelopment Banks. Washington, D.C.: World Bank. . (1979). "Sri Lanka. Issues & Prospects for Industrial Development." South Asia Region, Report no. 2261-CE. . (1981). Staff Appraisal Report for Sri Lanka Second Small and Medium Industries Project. Report no. 3480-CE. . (1984). "Selected Issues of Industrial and Trade Policies in Sri Lanka." South Asia Region Office. Report no. 4795. . (1987). "Sri Lanka and the World Bank. A Review of a Relationship." A World Bank Operations Evaluation Study. _ . (1989). "Sri Lanka. Recent Macro-Economic Developments and Adjustment Policies." Asia Country Department 1, Report no. 8193-CE. Washington, D.C.: World Bank. . (1991 a). "World Bank Support for Small and Medium Industry in Selected Countries." Operations Evaluation Department, Report no. 9530. . (1991b). "Sri Lanka. Financial Institutions Study." Asia Region, Report no. 9339-CE. . (1991c). Project Performance Audit Report on Sri Lanka First Small and Medium Industries Project, Second Small and Medium Industries Project and Industrial Development Project. Operations Evaluation Department, Report no. 9724. 80 Bibliography (cont.) ___. (1991d). "OED Study of Bank Support of Industrialization in Newly Industrializing countries: Case Studies of Korea, India, and Indonesia." Report no. 9899. __. (1992). "India. Small Scale Industry. Performance, Evaluation and Agenda for Reform." Report no. 10390-IN, vol. I. __. (1993a). "Sri Lanka: Country Economic Update FY93. Public Sector Rationalization for Private Sector Development and Poverty Alleviation." South Asia Region, Report no. 11862-CE. __. (1993b). Staff Appraisal Report for Sri Lanka Private Finance Development Project (PFDP), Report no. 11572-CE. __. (1995a). "Sri Lanka. Private Sector Assessment." South Asia Region, Report no. 12514- CE. . (1995b). "Sri Lanka. Poverty Assessment." South Asia Region. Report no. 13431-CE. . (1996a). "Sri Lanka In the Year 2000 - An Agenda for Action." Report no. 15455-CE. . (1996b). Performance Audit Report on Sri Lanka Economic Restructuring Credit and Public Manufacturing Enterprises Adjustment Credit. Operations Evaluation Department, Report no. 15820. . World Development Report 1996. 81 Annexes 83 Annex I Methodology and Data Tabulations on the Survey of the Small- and Medium-scale Industry Lending Scheme The World Bank/Asian Development Bank-sponsored Small- and Medium-scale Industry (SMI) lending scheme became operational in Sri Lanka in the year 1979. To date, Sri Lanka has obtained credit under four successive lines of credit. The last line of credit known as SMI IV commenced in 1992 and was fully committed in 1996. The number of loans under each scheme and the value of each line of credit is given in Table 1. Table 1. Loan Approvals Under SMI Schemes SMI I SMI II SMI III SMI IV Total Period 1979-81 1982-87 1988-91 1992-95 1979-95 No. of loans 1,741 2,491 2,439 9,065 15,736 Facility (Rs.mn) 229 1,059 1,114 5,323 7,725 Facility (US$'000) 12,729 36,467 28,593 106,695 184,484 Source: National Development Bank. Eligibility criteria Activities. Projects eligible for credit under the scheme should fall into the following sectors: Manufacturing Mining Construction contracting Agricultural Industries Fish Processing Industrial Services Horticulture Commercial Transport of goods Animal Husbandry Although originally this list was much shorter, it has been subject to regular review and, over time, more sectors have qualified for SMI lending. In addition to the above, the National Development Bank has the discretion to approve projects under any other sector/activity. Project size. While there is no minimum level of investment, projects qualify for lending only if their fixed assets are below the maximum allowed. At the commencement of each line of credit this limit was reviewed and revised where necessary. 84 Annex I Table 2. Size of Enterprise to Qualify for SMI Funding Total investment in fixed assets at original book value on completion ofproject SMI I Rs. 2 mn excluding land & buildings (leased or owned) SMI II Rs. 4 mn excluding land & buildings (leased or owned) SMI III Rs. 8 mn including land & buildings (leased or owned) SMI IV Rs. 16 mn excluding land & buildings (leased or owned) Source: National Development Bank Loan size. At the commencement of SMI I the maximum loan amount was Rs. I million. With each line of credit this limit was increased : SMI II Rs.2 million SMI III Rs.4 million SMI IV Rs.8 million The debt equity at the completion of a project should be, at most, 3 (i.e., 75:25) with the project proposers making a minimum contribution of 25 percent of the project cost. Test survey. Originally it was intended to collect financial information annually for the past 10 years. However, a test run on 15 companies with the original questionnaire revealed that collection of financial information for 10 years was not practical. This questionnaire was then revised to obtain financial information for 3 years, 1995, 1992 and 1985 over the 10-year period. The revised questionnaire was also tested on 3-4 companies prior to the commencement of the survey proper and was found to be more acceptable. Enumerators. Undergraduates and recent graduates from the disciplines of business and finance and final students of accountancy carried out the survey. From over 100 applicants, about 35 were selected after review of curriculum vitae and interviewing. In selecting enumerators, in addition to checking on their knowledge and communication skills, efforts were made to select enumerators from the districts which were selected for the survey. Enumerators residing in the same district found it easier to locate the industries in that area and were more familiar with the customs, practices and people of that area. It also helped the enumerator to travel from his/her own residence. The enumerators selected for the survey were provided training on carrying out the survey. This included familiarizing them with the questionnaire in detail, and role-playing. Each and every question was discussed with the enumerators, and the terms used in the questions were defined and explained. The pilot survey helped structure training so that the enumerators were able to overcome the practical problems in obtaining information and data. The enumerators were remunerated on the number of surveys completed by each person. In addition, an incentive scheme was developed based on the number of questionnaires completed, accuracy, and the quality of their surveys. The enumerators were required to complete a minimum of 10 surveys to qualify for the incentive, and the survey was targeted to be completed within three months of commencement. Survey. Prior to the commencement of the survey, letters were sent to the enterprises selected for the survey. These included a letter from the World Bank representative's office in Sri Lanka explaining the nature and purpose of the survey, and a covering letter from 85 Annex I ECONSULT informing the firm that a representative of ECONSULT would be calling over to carry out the survey. At times the World Bank letter did not suffice and had to be augmented by a letter from the National Development Bank. The sending of letters was found to be very useful as many recipients called back to inform us their willingness to participate in the survey, thereby easing the task of getting appointments. On the other hand, the letters sent to firms which were closed down or to firms which had changed their address, were returned to our firm. This saved the enumerators time and trouble to locate these industries. As the exit rate of the beneficiary group was very high, returned letters marked "cannot be found" were treated as closed. We commenced the surveys with the beneficiary group as the information on the nonbeneficiary group was not available at the time of commencement. The enumerators were allocated to the four districts and a supervisor was placed in charge of each district. The supervisors also participated in the two-day training program and were expected to monitor and coordinate the work of the enumerators. The allocation of firms, issue of letters and sorting out of problems for a particular district rested in the hands of the supervisor for that district. Once an enumerator completed a survey, the supervisor checked the questionnaire for accuracy and completion, prior to approval for payment. Where necessary, the supervisors rechecked the information obtained by enumerators by either revisiting the firm or more commonly, by contacting them over the telephone. Survey results On completion of the survey 304 enterprises had been surveyed, 149 beneficiaries and 155 nonbeneficiaries. A comparison of the proposed and actual surveys done for the two groups is given below. (a) By sector Table 3: Comparison of Proposed and Actual Number of Surveys by Sector Beneficiary Group SMIli SMI III Total Proposed Actual Proposed Actual Proposed Actual Sector No. % No. % No. % No. % No. % No. % Food & beverage 38 50 36 53 37 49 39 48 75 50 75 50 Textile 5 7 4 6 5 7 6 7 10 7 10 7 Garments 12 16 11 16 12 16 13 16 24 16 24 16 Construction material 9 12 7 10 10 13 11 14 19 12 18 12 Metal products 11 15 10 15 11 15 12 15 22 15 22 15 Total 75 100 68 100 75 100 81 100 150 100 149 100 86 Annex I Table 4: Comparison of Actual vs Proposed Nonbeneficiary Group Proposed Actual Sector No. % No. % Food & beverage 72 48 74 48 Textile 10 7 13 8 Garments 27 18 26 17 Construction material 19 12 18 12 Metal products 22 15 24 15 Total 150 100 155 100 In carrying out the survey priority was accorded to meeting the sector targets and then the district targets respectively. This sector classification is based on the original classification of firms by sector in the respective databases. When the survey was carried out it was found that some of them did not belong to the categorized sector. Nevertheless, these firms were surveyed as: * we commenced the survey with the beneficiary group and all enterprises in the NDB database had applied for loans under the SMI scheme, and * the exit rate was high and there was a shortfall of firms available for survey. The total number of firms surveyed which were outside the five sectors selected for the survey amounted to about 5 percent of the total enterprises surveyed, a detailed analysis of which is given in Table 5. Table 5: Firms Surveyed Beyond the Five Sectors Selected for the Survey Industry Beneficiary group Nonbeneficiary group Total Leather products - 1 I Wood products 5 - 5 Rubber products -1 1 Plastic products 1 1 2 Nonelectrical machinery I - I Other manufactures 3 3 6 Total 10 6 16 % of surveys completed 6.7 3.9 5.3 87 Annex I (b) By district Table 6: Comparison of Proposed and Actual Number of Surveys by District Beneficiary Group SMI II SMI III Total Proposed Actual Proposed Actual Proposed Actual District No. % No. % No. % No. % No. % No. % Colombo 42 56 28 41 42 56 41 51 84 56 69 46 Gampaha 17 23 20 29 17 23 18 23 34 23 38 26 Galle 10 13 12 17 10 13 12 15 20 13 24 16 Kurunegala 6 8 9 13 6 8 9 II 12 8 18 12 Total 75 100 69 100 75 100 81 100 150 100 149 100 Table 7: Comparison of Actual vs Proposed Nonbeneficiary Group Proposed Actual District No. % No. % Colombo 84 56 92 59 Gampaha 34 23 31 20 Galle 20 13 18 12 Kurunegala 12 8 14 9 Total 150 100 155 100 Of the 149 surveys in the beneficiary group, 9 enterprises had closed down after 1992, and the financial information for the year 1995 was not available for these enterprises. Despite the closures, these companies were surveyed to obtain a better understanding of the problems existing with SMIs. An additional question with reasons for the closure was also filled. The total number of closures was very much higher. As stated earlier, most of these could not be traced. Summary of Statistics: SMI Survey of Firms, Sri Lanka, 1996 Mean [Standarad Deviation] Firn Size Beneficiaries Nonbenefciaries SmallI-49) Medin 150-991 LargefL00+/ Total, Benefidar Smal 11-49/ Medium/50-991 Large [190+/ Tota NoabenefcadWa No. of Firms 95 22 23 140 76 30 49 155 No. ofFirms with Loans 95 22 23 139 50 25 35 110 %of Firms with Loans 98.95% 100.00% 100.00% 99.33% 65.79% 83.33% 71.43% 70.96% Mean No. of Workeis in 1995 18.86111.221 67.09 (11.81 197.87 (13.611 55.85 (80.261 25.37 112.22 71.67 116371 216.881 (189.471 95.32 1136.471 Years ofOperation 16.22 (10.621 19.18 [18.8] 20.0 [12.66] 17.31 [12.55] 21.4 [17.38] 22.73 [17.11] 26.37 [19.03] 23.24 [17.891 Average Capacity (%) 71.91 [29.88] 67.59 [31.32] 79.74 [22.9] 72.51 [29.111 73.23 [20.78] 69.27 [23.85] 77.67 [17.191 73.87 [20.45] Average Production Workeres Wage 2109.38 [16861 1855.91 [1474] 2231.17 [15341 2089.56 [1624] 2475.9611606] 3166.1 [15851 2639.37 [1563] 2662.4 [15991 (1995 rupees) Average Loan Interest (%) 18.08% [3.83] 17.03% [4.111 17.13% [3.46) 17.6% [3.881 19.34% [6.02] 19.84% [5.7] 18.58% [4.85] 19.2% [5.55) 00 % of Sales in Foreign Markets 7.25 [24.0] 44.32 [46.0] 37.96 [47.24] 18.12 [36.291 16.52 [34.43] 37.67 [45.291 44.16 [46.53) 29.44% [42.49) % of Firms with Exports 10.53% 54.55% 47.83% 22.15% 19.74% 46.67% 55.10% 36.13% Material Costs/Sales (1995) .53 [.221 .54 [.231 .55 [.2] 0.53 [0.221 .57 [.19] .50 [.251 .53 [.22] 0.54 [0.211 Long-Term Debt/ Total Debt (1995) .42 [.33] .481.32] .34 [.261 0.4210.32] .23.J.291 .31 [.35] .19 1.261 0.23 10.301 Total Debt fotal Assets (1995) .22 [.20] .34 [.24] .44 [.26] 0.28 [0.23] .27 [.23] .28 [.25] .38 [.30] 0.31 [0.26] Total Assets/Total Employment (1995) 524.17 [707.21 279.85 [211.3] 703.53 [1925.8] 515.24 [972.9] 424.06 [4191 469.02 [634] 571.48 [964] 479.73 [676] Value Added per Worker (1995 rupees) 225.41 [464.0] 246.15 [576.91] 152.15 [177.56] 216.63 [449.11 152.41 [200] 160.91 [188] 247.38 [415] 184.28 [286] Source: 1996 Econsult survey. Projects Related to the SMI Credits L/C Approval Closing Loan Percent . Instit. Bank Bank Bank Total SW Sustam. Performance Performance Performance spent on Number Project title date date ($m) Utilized Outcome Sustain. Devel.(Pprane Pervon) spervson I (Ident.) (Appraisal) (Supervision) supervision L0520 DFCC 1 11/16/67 6/30/72 4 45% Not Rated Not Rated Not Rated L0634 DFCC II 7/15/69 4/30/76 8 36% Satisfactory Not rated Not rated- C0512 Program credit project 9/17/74 12/30/75 15 100% Satisfactory Not rated Not rated C0566 DFCC III 6/26/75 9/30/79 5 92% Satisfactory Not rated Not rated C0742 DfCCIV /f13/f7 12/31/81 8 94%o Satisfactory Not rated Not rated CO92 S6/26/79 6/30/85 16 94% Unsatisfactory Uncertain Negligible Satisfactory Satisfactory Satisfactory SMIII 10/13/81 12/31/87 30 99% Unsatisfactory Uncertain Negligible Satisfactory Unsatisfactory Unsatisfactory 85.5 Cl401 IDP I71/12/83 9/30/88 25 100% Satisfactory Likely Substantial Satisfactory Satisfactory Satisfactory 79 C1692 IDP II- 5/8/86 6/30/94 20 98% Satisfactory Likely Substantial Satisfactory Satisfactory Satisfactory 82.6 C 1860 SMI III 12/15/87 6/30/93 20 96% Satisfactory Uncertain - Modest Satisfactory Satisfactory Satisfactory 53.8 C948IP 7/26/88 6/30/95 44, 100% Satisfactory Likely Modest Satisfactory Satisfactory Satisfactory 99.6 C2128 Economic Restructuring Credit 5/1/90 12/31/94 213 99% Satisfactory Uncertain Negligible Satisfactory Unsatisfactory Satisfactory 50.8 C2185 Public Mfg. Enter. Adj. Credit 11/27/90 5/30/96 252 100/; Satisfactory Likely Modest Satisfactory Satisfactory Satisfactory 100.9 C2250 SMI IV 5/28/91 6/30/97 45 92% 1 ONGOING 59.5 C2484 Private Finance development 4/20/93 6/30/99 60 77% ONGOING 50.4 TOTALS: Total committed amount ($m) 764 Percent utilized 96% Percent utilized satisfactory (Bank wide average = 74%) 93% Percent utilized sustainable (Bank-wide average = 55%) Percent utilized with substantial ID (Bank-wide 7% average = 35%) Average staff weeks on 74 supervision (last 9 projects) Chronology of Events Year GOSL Actions/Events Bank Interventions 1967 June - DFCC I approved (credit line) 1969 July - DFCC II approved (credit line) 1970 Left-wing coalition government under Sirimavo Bandaranaike; import substitution policies introduced. 1972 Land Reform Act; Limited Housing; Compulsory Savings Scheme; Business Acquisition Act 1975 June - DFCC III approved (credit line) 1977 UNP/Jayawardena rule begins. Liberalization measures implemented including September - DFCC IV approved (credit line) 1978 1979 Free Trade Zone established September - Bank report "Issues and Prospects for Industrial Development" (No. 2261); June - SMI I approved (credit line, TA, and policy studies) 1981 October - SMI II approved (credit line, TA, policy agreement) 1983 Military conflict with LTTE escalates July - IDP I approved (credit line, TA, policy studies) 1984 Tariff reform implemented January - Bank report "Selected Issues of Industrial and Trade Policy in Sri Lanka" (No. 4795) 1985 May - Bank report "Recent Economic Developments and Policies for Growth" (includes assessment of tariff regime) (No. 5628) 1986 May - IDP II approved (credit line, TA, policy studies); July - Joint IMF/IDA Financial Sector 1986 ~~~ ~~~~~~~~~~Report __________________________ March - Industrial Policy Statement provides first comprehensive strategy for industrial R eport 1987 development; July - Indo-Lanka Peace Accord; JVP activity begins; Treasury bills April - Bank report "Issues in Macroeconomic and Industrial Development Policy" (No.6701) assesses impact of 1984 tariff reform; December - SMI III approved __ introduced; JVP resurgence begins_________ ____ _______________________ May - Bank report "A Break with the Past: The 1987-90 Program of Economic Reform and Adjustment" (No. 7220) outlines measures for second wave of liberalization; July- IDP III 1988 December - Premadasa (UNP) government begins (with a focus on tariff reform; export incentive structure; debt recovery policies and procedures) 1989 December - GOSL's Industrial Strategy Statement issued; December - "Bank report Recent Macro-economic Developments and Adjustment Policies" (No. 8193) May - Economic Restructuring Credit approved; November, 1990 - Public Manufacturing 1990 JVP activity ends; debt recovery legislation passed Enterprise Adjustment Credit approved 1991 February - Bank report "Financial Institutions Study" reconfirmed the existence of constraints in financial sector found in 1986 study; May - SMI IV approved 1992 January, Bank report "Strengthened Adjustment for Growth and Poverty Reduction" (No. 10079) April - Private Financial Development Project approved; June - Bank report "Public Sector 1993 May.- Premadasa assassinated; D.B. Wijetunge president; NDB privatized Rationalization for Private Sector Development and Poverty Alleviation" (No. 11862) 1994 Chandrika Bandaranaike government (SLFP) elected; SCB's recapitalized 1995 March - Bank report "Private Sector Assessment" (no. 12514) 1996 , March- Bank report "Sri Lanka in the Year 2000: An Agenda for Action"