Report No. 18041 World Bank Support for Small and Medium Industries in the Philippines An Impact Evaluation June 18, 1998 Operations Evaliation Department Document of the World Bank Thi; document has a restricted distribution and may be used by recipients onl% in the performance of their official duties. Its contents may not other\%ise be disclosed i th'ut World Bank authorization. Currency Equivalent Currency Unit: Philippine Peso 1976-79 P 7.4 1980-81 P 7.7 1982-89 P 17.3 1990-92 P 25.8 1993-96 P 26.4 1997 P 29.5 Fiscal Year January 1 - December 31 Abbreviations ADB Asian Development Bank APEC Asian Pacific Economic Co-operation BSP Bangko Sentral ng Pilipinas DBP Development Bank of the Philippines DTI Department of Trade and Industry GFSME Guarantee Fund for Small and Medium Enterprises IGLF Industrial Guarantee Loan Fund IMF International Monetary Fund MASICAP Medium and Small Industries Coordinated Action Program NCR National Capital Region NSO National Statistics Office OECF Overseas Economic Cooperation Fund OED Operations Evaluation Department PFI Participating Financial Institutions SBAC Small Business Advisory Center SBGFC Small Business Guarantee and Finance Corporation SME Small and Medium Enterprise SMI Small and Medium Industry SMIIES Small and Medium Industry Impact Evaluation Survey USAID United States Agency for International Development Acting Director-General, Operations Evaluation : Mr. Roger Slade Acting Director, Operations Evaluation Department : Mr. Roger Salde Manager, OEDCR : Mr. Ruben Lamdany Task Manager : Mr. Robert Buckley The World Bank Washington, D.C. 20433 U.S.A. Office of the Director-General Operations Evaluation MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: World Bank Support for Small and Medium Industries in the Philippines: An Impact Evaluation Attached is the Impact Evaluation Report on the World Bank Support for Small and Medium Industries (SMIs) in the Philippines, prepared by the Operations Evaluation Department (OED). Between 1976 and 1992 the World Bank provided the Philippines with US$180 million to finance SMIs through four lines of credits. The Asian Development Bank contributed US$100 million to the last loan. All four projects were retailed as term sub-loans to 2,600 SMIs. Besides lending, the Bank also supported technical assistance for SMIs, and, through sector work, the Bank supported broader economic reforms that improved the economic environment for SMIs. The last project was fully disbursed in FY93. This evaluation provides an empirically based perspective on the impact of the Bank's strategy on SMIs. It makes use of a survey of more than 300 firms, half'of whom were beneficiaries during the 1988-92 period, while the other half served as a control group. A survey of 17 participating financial institutions (PFIs) was 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 results indicate first that support to SMIs in the Philippines resulted in more middle-income jobs. However, these jobs were not as productive as those provided by larger firms. Second, although the rationale for SMI-targeted lines of credit may no longer be justified, it is clear that Bank support for SMIs in the Philippines was part of a broader financial sector dialogue that was quite fruitful. The financial system became stronger and deeper and almost certainly these improvements helped the economy sustain the shocks which its neighbors experienced. More directly, an unexpected and unusual by-product of the support for SMIs was that the credit risk of the commercial banking sector was reduced: the repayment record of assisted SMIs was considerably better than that of larger firms. Hence, SMI lending was less risky, and helped improve the soundness of the banking sector. A final finding is that SMI lending was an inexpensive way to support the creation of new jobs. The study of these operations offers four lessons. First, SMIs can be effective job creation programs. However, they are not likely to have much, if any, effects on directly reducing poverty or increasing exports. Nor should they be thought of as programs that can shift the location of industry. Second, SMI lending through financial intermediaries can support banking sector soundness. Directed credit programs that can improve banking sector soundness are possible. Third, SMIs could use more flexible financial products. SMIs already confront higher real borrowing costs because they have higher transaction costs. One way to reduce some of these costs would be first to establish basic banking links, such as bank accounts or short-term credit. These links can be seen as building blocks towards a more complex banking relationship such as that implied by long-term loans. Finally, SMIs could participate in more market-driven technical assistance. SMIs already invest in acquiring technical assistance so that there is room to take a more demand-driven and more private sector-oriented approach to technical assistance. Attachment Robert Picciotto by Roger Slade Contents E xecutive Sum m ary ....................................................................................................................... 3 H ighlights from Borrow er's Response............................................................................................ 6 1. In tro d u ction ............................................................................................................................... 7 2. SM Is and the L abor M arket.................................................................................................... 8 3. The Financial Sector and SMI Lending................................................................................ 17 4. T he C osts of SM I Support...................................................................................................... 24 5. L essons L earned ..................................................................................................................... 27 Annexes A . C om m ents from B orrow ers .................................................................................................... . 29 1. National Economic and Development Authority ............................................................... 29 2. Wholesale Banking Sector- PMD II (DBP)........................................................................ 33 B . D efining SM Is in the Philippines............................................................................................ 36 C . E fficiency of Sm all Firm s ....................................................................................................... 37 B ib liograp hy ................. ............................................................................................................... 38 Figures 2.1 Labor Absorption by Major Economic Sector, The Philippines, 1960-96...............................9 2.2 Manufacturing Employment by Firm Size, Korea, Japan and The Philippines ............ ..... .....9 2.3 Capital-Labor Ratios by Firm Size, The Philippines, 1994.................................................... 12 2.4 Protection and SM Is in A sia, Early 1990s.............................................................................. 13 2.5 Manufacturing Value Added and IGLF Loans, Early 1990s.................................................. 14 This report was prepared by Messrs. Robert Buckley (Principal Evaluation Officer) and Edgard Rodriguez (Economist). Messrs. Dipak Mazumdar and Albert Berry (consultants, University of Toronto) helped to design early versions of the firm survey, applied in similar studies in Ecuador and Sri Lanka. Prof. Gwendolyn Tecson (consultant, University of the Philippines) was a major contributor to the earlier drafts of the report (macro and financial sectors), and Mr. Federico Mini (consultant, Georgetown University) provided statistical work. The report benefited from comments by the Region (Messrs. Vinay Bhargava, Zafar Khan, and Aloysius Ordu), by the Networks (Ms. Kristin Hallberg, Private Sector Development), and by the Borrower (Ms. Margarita Songco, Director, Trade, Industry and Utilities, NEDA; and Mr. Orlando Pineda, Vice-President, Wholesale Banking, DBP). The latter are included in their entirety under "Borrower's Response" in this report. Ms. Norma Namisato and Ms. Brigitte Wittel provided administrative assistance. Contents (continued) 2.6 SMI Workers in Income Distribution, 1994-5........................................................................ 15 3.1 Long-Term Loans in the Banking System, June 1993 .......................................................... 17 3.2 Real Interest Rates and Spreads, the Philippines, 1976-96 .................................................... 18 3.3 IGLF Interest Rates and Market Lending Rates, 1977-96...................................................... 19 3.4 Credit to SM Es, The Philippines 1991-6................................................................................ 20 3.5 Long-Term Debt: IGLF Firms within an International Perspective.......................................21 3.6 IGLF and Total Arrears Rates, Commercial Banks, 1989-95 ................................................ 22 4.1 Government Consumption, Exports and Financial Deepening, 1975-95...............................24 Tables 2.1 Graduation of IG LF Firm s, 1985-95 ...................................................................................... 10 2.2 G raduation of N ew Firm s, 1990-95........................................................................................ 11 4.1 Effectiveness of Directed Credit Programs: Selected Characteristics.................................... 25 3 Executive Summary 1. The Bank has supported small and medium industries (SMIs) based on the assumptions that SMIs are more labor intensive than larger firms, that they are often more efficient, and the banking sector, particularly in heavily controlled financial systems, provides insufficient funding for the high return investments that SMIs undertake. In addition, SMIs have often been supported by the Bank to encourage employment of lower paid wage earners in a way that helps diversify and broaden the base of an economy. Finally, a recent OED study (1 997a) argues that SMIs can be an effective vehicle to support economies undergoing fundamental economic restructuring. The study also suggests that, in some cases, it appears that selective SMI lending can improve the prudential soundness of the financial system. 2. On the other hand, in many places SMIs have been shown to be less labor intensive and efficient than larger firms. In addition, reduced access to finance by SMIs has in some places been shown to be problematic. Finally, Bank financial sector policy, particularly since Operational Directive 8.30 was implemented in 1992, has made SMI lending a virtually extinct line of business for the Bank. The presumption underlying the directive is that in most cases it is better to let commercial banks decide who their clients should be rather than restricting the size of firms to whom Bank funds should be onlent. 3. Thus, the question of whether Bank support for SMIs can create new jobs and contribute to financial sector strengthening is ultimately an empirical one. Do the circumstances of a particular economy make it desirable to channel funds to smaller enterprises? Or should Bank support rely on the banking sector to identify and assume the risks of its lending decisions? The answers to these questions require evidence on whether smaller firms are likely to be either more labor-intensive or efficient than larger firms. If such evidence is adduced, then the necessary conditions for supporting SMIs are in place. If, in addition, SMIs are found to have limited or mispriced access to credit, then there is a sufficient condition to support SMIs. 4. This study examines the effects that Bank support for SMIs had in the Philippines. This topic is of concern for two reasons. First, and most importantly, unemployment in Metro Manila is now over 14 percent. With the prospects that the East Asian financial crisis could lead to the return of many of the Filipinos working in nearby countries, it is easy to envision the capital city's unemployment rising sharply even if the economy performs well. Hence, the efficacy of job creation programs, such as SMIs, is of some immediate interest. Second, an empirical analysis such as this one allows us to trace through the results on the ground as to how many jobs were created for a specific form of lending. Analysis of the behavior of firms provides significantly more robust results than the simplified accounting estimates that characterize most Bank reviews and appraisal reports of SMI lending. 5. Because the Philippines has just been the focus of a broader Country Assistance Review by OED (World Bank, 1998b), this note focuses mainly on empirical aspects of SMI lending. The approach is to examine the evidence on the performance of Bank-supported SMIs in the Philippines and a control group of SMIs which did not receive support during the 1988-92 period. Recent manufacturing census and survey data are also reviewed. 4 Findings SMIs and Jobs 6. Bank support for SMIs in the Philippines resulted in more jobs. They were jobs whose productivity was indistinguishable from those generated by SMI firms which did not receive credit. However, the productivity of firms increases slightly by firm size (Mini and Rodriguez, 1998), as it does in Indonesia, Malaysia and Taiwan (Tan and Batra, 1995). Consequently, there appears to be a price to be paid in lower productivity from supporting SMIs. In addition, our results indicate that SMIs did not provide a great many jobs for the very poor, or for exporting. Assisted SMIs did increase their exports, but not by a great deal. There are far more effective ways to support export growth than providing credit to SMIs. Cumulatively, then, our results indicate that support to SMIs in the Philippines resulted in more "middle class" jobs. However, these jobs were not quite as productive as those provided by larger firms. SMIs and the Financial Sector 7. One of the World Bank's central concerns with lines of credit, such as those that support SMIs, is the issue of why prudent lenders, motivated by profits and constrained by their capital, should not be the best judge of which firms should receive credit. Bank Operational Directive 8.30 requires that any line of credit supported by Bank lending should be accompanied by policy discussions to address the impediment which created the limited access to credit by high return investors. Ideally, according to the directive, a directed credit line should be undertaken only as long as progress is being made on addressing the imperfection that justifies the line of credit in the first place. That is, the rationale for such lines of credit should ultimately be eliminated. 8. Recent Bank research, while in agreement with this perspective, suggests that the time periods over which financial market imperfections can be eliminated are likely to be lengthy. This work, by Caprio and Klingebiel (1996), indicates that the historical record is that financial reforms are rarely if ever implemented quickly. The implication of this perspective has been developed in the World Bank's new Financial Policy Framework (1997) which makes it clear that the Bank now views financial reform and development to be a process rather than an event such as a project with policy covenants. From this longer-term perspective, we find that Bank support for SMIs in the Philippines was part of a broader financial sector dialogue that was quite fruitful. The system became stronger and deeper and almost certainly these improvements helped the economy weather better the shocks which its neighbors experienced. 9. More directly, an uncommon by-product of the support for SMIs was that the credit risk of the commercial banking sector was reduced: the repayment record of SMIs was considerably better than that of larger firms. Today, the observed interest rate differentials between small and large firms may reflect more the high costs of borrowing among small firms than discrimination against SMIs. The latter was more likely to have prevailed in the pre regulation days when Bank support to Philippine SMIs first appeared. The effects of the loans on the corporate financing patterns of SMI firms is much clearer. Today, Philippine beneficiaries have greater indebtedness than firms in other parts of the world. Certainly, by the time of our survey, 1997, it does not appear that these SMIs had much difficulty in accessing credit. 5 The Costs of SMI Support 10. The direct effect of SMI programs on government costs was minor. Direct budget costs were low and did not expose the government to risks. Indeed, the projects helped reduce government risk exposure. On the other hand, the technical assistance provided directly to SMIs had no significant impact, but this weakness is more the result of program scale and structure than content. Nevertheless, the government's apparent enthusiasm for the development of a technical training center for SMIs, as indicated by the Philippines'1996 APEC initiative, must be considered an as yet untested idea. 11. The allocational effects of the directed credit channels used to provide the credits do not appear to be large. Technical efficiency is slightly lower at SMIs than at larger firms, so the restrictions on credit use do result in a misallocation of resources. However, when the better portfolio performance and diversification of the banking sector is considered, the foregone technical efficiency is a small price to pay (see Stiglitz, 1998). Lessons learned 12. The study of these operations offers a number of lessons: (i) Create jobs. Under the right circumstances, SMIs can be effective jobs programs. They were in the Philippines, and, with policy reforms and careful design, they can be more effective. However, they are not likely to have much, if any, effects on directly reducing poverty or increasing exports. Nor should they be thought of as programs that can shift the location of industry. (ii) Improve banks 'performance. SMI lending through financial intermediaries (FILs) can support banking sector soundness. Good directed credit program can exist, particularly in countries suffering from an inability or recalcitrance to foreclose on large borrowers. (iii) Use moreflexiblefinancial products. SMIs confront higher real borrowing costs because they are seen as bigger risks --or simply have higher transaction costs- and are priced accordingly. One way to reduce some of these costs would be to establish basic banking links, such as bank accounts or short-term credit. These links can be seen as building blocks towards a more complex banking relationship such as that implied by long-term loans. (iv) Design more market-driven technical assistance. SMIs already invest in acquiring technical assistance so that there is room to take a more demand- driven and more private sector-oriented approach to technical assistance than previously taken. 6 HIGHLIGHTS FROM THE BORROWER'S RESPONSE The SMI lending from the World Bank to the Philippines extended from 1976 to 1992 in four consecutive operations (Loans 1120, 1727, 2169, and 3038). The Borrower was the National Economic and Development Authority (NEDA), and the Development Bank of the Philippines (DBP). OED is grateful for their contributions to our final draft which underscore OED's efforts to bring our Borrowers to the forefront of our evaluation work. Their views are, in general, similar to OED's. However, their specific comments helped clarify the report. Areas of Agreement On basic findings of the report. NEDA finds "the study useful and the findings as expected. There were no surprises. As 'owner' of the funds, it is heartening to note that the facility has to some extent served some of its purposes and at no great cost when, perhaps, compared to other similar facilities." Similarly, DBP "generally agree[s] with your observation and evaluation of SMIfinancing in the Philippines. " On "Lessons Learned". While OED's report calls for more flexible financial instruments for SMIs, DBP notes that the IGLF Review Committee has recently approved "... a short-term rediscounting program precisely to provide facilities for micro and cottage-sized enterprises. Only countryside banks such as thrift and rural banks are eligible participating financial institutions ... under this program. ... Our feedback on this facility is very encouraging provided we relax some terms. This is the kind of program fund needed in the countryside where small banks ... do not have the capability to evaluate long term projects." Areas of Disagreement or Clarification On lending outside Manila after 1992. The Borrower (NEDA and DBP) notes that there has been substantial progress in SMI lending outside Manila since 1992 with second generation funds. For example, DBP points out the existence of a "Special ... Facility for Countryside Projects sourced from earnings of IGLF to encourage [banks] to look for projects located in the countryside... As ofDecember 31, 1997 the ratio of loans granted in the regions was 60% in term of amounts ... granted. " The report finds that, while SMI credit is a good instrument to assist firms to grow and prosper, it is an inefficient instrument to attempt to affect firm location. On clarifying firm size. Comments from NEDA -and the Region- asked to clarify the definitions of firm size used in this report which, in principle, does not include the World Bank microenterprise lending (covered in the recent OED's Country Assistance Review, World Bank, 1998b: 21). This was done in the present draft. Moreover, wherever possible the report has explicitly indicated the employment-based definition used, and it has added an annex to explain the correlation of employment-based and assets-based definitions. Other Borrower comments that lead to clarifications include comments on the role of credit versus other non-lending services, a comment also made by the Network, credit access versus credit availability, low use of guarantees, and reasons for higher interest rates for SMIs. NEDA and DBP's complete responses are attached to this report. 7 1. Introduction 1.1 In the 1980s the Bank actively supported SMI projects around the world based on the assumptions that SMIs were more labor intensive than larger firms, that they were often more efficient, and the banking sector, particularly in heavily controlled financial systems, provides insufficient funding for the high return investments SMIs wanted to undertake. In addition, SMIs have often been supported by the Bank to encourage employment of lower paid wage earners in a way that helps diversify and broaden the base of an economy. Finally, one recent OED study (World Bank 1997a) argues that SMIs can be an effective vehicle to support in economies undergoing very fundamental economic restructuring. The study also suggests that in some cases SMI lending can improve the prudential soundness of the financial system. 1.2 On the other hand, in many places SMIs have in fact been shown to be less labor intensive and efficient than larger firms (see Little, Page and Mazumdar 1987; and World Bank 1998a). In addition, the evidence that SMIs suffer reduced access to finance has in some places been shown to be problematic (see Jaramillo, Schiantarelli and Weiss 1993). Finally, Bank financial sector policy, particularly since Operational Directive 8.30 was implemented in 1992, has made SMI lending a virtually extinct line of business for the Bank. With 8.30's implementation, Bank lending for SMIs tailed off sharply, falling to almost zero. The presumption underlying the directive is that in most cases it is better to let commercial banks decide who their clients should be rather than restricting the size of firms to whom Bank funds could be onlent. 1.3 Thus, in many respects, the question of whether Bank support for SMIs can be expected to be an effective instrument is ultimately an empirical one. Do the circumstances of a particular economy make it desirable to channel funds to smaller enterprises? Are there gains to be realized from placing size restrictions on the firms Bank credit supports? Or, alternatively, should Bank support rely on the banking sector to prudently identify and assume risks? The answers to these questions require evidence on whether smaller firms are likely to be more labor-intensive and efficient than larger firms. If such evidence can be adduced that they are, then the necessary conditions for supporting SMIs are in place. If, in addition, SMIs are found to have reduced access to credit, then there is a sufficient condition to support SMIs 1.4 In this study we examine the effects that Bank support for SMIs had in the Philippines. The impact of the Bank support for Philippine SMIs is of some concern for two reasons. First, and most importantly, unemployment in Metro Manila had already reached 14.5 percent in October 1997 (NSO 1998). With the prospects that the East Asian crisis could lead to the return of many of the more 180,000 Filipinos working in near-by countries, it is easy to envision the capital city's unemployment rate exceeding 18 percent even if the economy performs well. Second, an empirical analysis such as this one allows us to trace through the results on the ground as to how many jobs were created for a specific form of lending. Analysis of firm behavior provides significantly more robust results than the simplified accounting estimates that characterize most Bank reviews and Appraisal Reports. See OED (World Bank, 1991) for a criticism of the costs ofjobs created estimates in Bank reports on SMIs. 1.5 Given the fact that the Philippines has just been subjected to a broader Country Assistance Review, OED (World Bank, 1998b), this note focuses mainly on the empirical 8 evidence on the performance of Bank-supported SMIs in the Philippines and a control group of SMIs which did not receive support during the 1988 to 1992 period. We also examine the recent manufacturing census and survey data on the performance of larger firms. This rich data set enables us to address the empirical rationales for supporting SMIs in the Philippines. To provide some comparative perspective, we also rely on evidence and literature of how SMIs have performed in a number of other countries. 1.6 The plan of the paper is as follows. In the next section we briefly describe the dimensions of Bank support for SMIs in the Philippines. Then, we consider how the credit affected aspects of labor market performance--employment, productivity, and poverty. In the third section we review how the Bank's support for SMIs affected financial markets and the corporate financial situation of firms. In a fourth section, we focus on the important interstices of labor market and financial market policies, policies that Krugman (1998) suggests, have resulted in enormous contingent liabilities being realized by a number of East Asian countries. In a final section, we spell out a number of lessons from the Bank's experience in the Philippines. 2. SMIs and the Labor Market Bank's Support for SMIs1 in the Philippines 2.1 Financing of urban-based small industry in the Philippines started in mid- 1970s and continued until early 1990s; a total of US$180 million was disbursed as part of four SMI loans (for an analysis of the total Bank's assistance to the Philippines, see World Bank, 1998b). Other World Bank loans have onlending operations to rural firms, but only some of these rural beneficiaries are SMIs.2 Two other important contributors to SMI lending in the Philippines have been the ADB and the Government of Japan.3 By 1996, however, SMI and micro lending disappeared completely from the World Bank's Philippine portfolio. 2.2 The Bank's Philippine SMI lending followed the different modalities used by the Bank worldwide. The first SMI loan used a direct credit allocation by a government agency, simultaneously with an apex credit allocation by commercial banks. The former was administered by the Development Bank of the Philippines (DBP) and the latter-known as Industrial Guarantee Loan Fund or IGLF4-by the Central Bank. An apex system was used during the following three SMI operations. In late 1980s, the World Bank recommended and The report focuses on SMEs in manufacturing, or SMIs, defined as having less than 20 million pesos in assets (an assets-based definition). Most SMI literature defines SMIs as those having fewer than 99 workers. Also, see annex I for a correlation between the assets-based and the employment-based definition. 2 For example, the Second Rural Finance Loan has funded only a fifth of rural firms with less than 20 million pesos in assets (equivalent to three quarters of a million dollars). The latter corresponds to a definition of "medium firm", see annex 1 for a discussion of assets-based and employment-based definitions in the Philippines. First, by 1996, ADB had disbursed US$100 million for SMIs, and US$38 million for microenterprises. Second, the Government of Japan-through its Overseas Economic Cooperation Fund-contributed 22.5 billion Yen (US$200 million) for credit to SMEs through the DBP. Also, USAID has participated and it has now focused on nonlending services to SMEs in less developed regions of the country, such as Mindanao. 4 The IGLF facility supported by the World Bank has been in existence since 1952. 9 finally transferred its apex facility from the Central Bank to DBP. In contrast to many SMI operations supported by the Bank (see Webster, 1996, and Dessing, 1990), the Philippine operations were all successful. All were rated as having satisfactory outcomes, with sustainability likely. Figure 2.1. Labor Absorption by Major EconmicSector, The Philippines, 1960-95 2.3 The SMI projects 100% - were implemented in a labor 90% i market in which 0% - seru.ces - manufacturing employment 70% grew relatively slow. Figure 60% 2.1 shows that the Manufacturing Philippines has undergone a 4t i limited transformation away 40% from agriculture into 3%riculiure -- industry and services. 20% Indeed, the manufacturing 10% wid!sector's share of 0% employment in 1995 was % of labor force 1960 1970 1980 1995 lower than it was in 1970. Source: NSO (1995, 1992) Figure 2.2. Manufacturing Employment by Firm Size, Korea, Japan and the Philippines in the early 1990s 2.4 As early as the 1970s, the Bank's economic and sector work had already noted % os Manufacturig LA that the Philippines had a 70 lower-than-expected share of - SMIs in employment and 40 value added compared to so Malaysia or Korea (World 20 Korea (1993) Bank, 1975: vol. II). On the , Japan (1903) other hand, the Philippines ,a- -Philppines (1994) was a country that gave-and .iLargeippineedium still gives-considerable (200.) 100. (50-99) sall attention to SMIs, even if its Firm Size by Number of Workers manufacturing sector tends to mu Sources: NSO-Philippines (unpublished); NSO-Korea (1995); Statistics Bureau-Japan have a greater domination by larger firms (Berry and Mazumdar, 1991: 47). Figure 2.2 presents data on the industrial structure of the Philippines, 5 Korea and Japan . Despite the relatively slow growth of manufacturing and SMI employment, the Government of the Philippines has continued to be interested in SMIs. For instance, during the 1996 APEC Leaders Summit in the Philippines, the government recommended the creation of a technology transfer and training center for SMIs. 5 Most literature on SMEs uses an employment-based working definition to facilitate cross-country and over-time comparisons (see Chee, 1992, for definitions in Asian countries). "Small" firms tend to be defined as those with 10 to 49 workers, and "medium" firms, as those with 50 to 99 workers. This report follows this definition. See annex 1 for a discussion of assets-based and employment-based definitions of firm size in the Philippines. 10 Question 1: Did Bank Support for SMIs Help Create Jobs? 2.5 Yes. Based on a firm-level survey, firms receiving credit from the Bank, what we term IGLF firms, generated more jobs than did similar firms without such credit. Between 1985 and 1995, IGLF firms generated more jobs than did nonbeneficiaries, and they did it faster. IGLF firms in the sample created 4,821 new jobs over a decade, representing 27 new jobs per IGLF firm while nonbeneficiaries lost 1,400 jobs, representing, on average, 10 fewer jobs per firm, during the same period. Moreover, employment generated by IGLF firms grew more than twice as fast as the Philippine labor force over this period: 5 percent per annum, versus 2.4 percent 6 throughout the economy. 2.6 In Table 2.1. Graduation of IGLF Firms, 1985-95 addition, not (as percentage of 1985 cohort) only were jobs created, but the Size in 1995 Micro Small Medium Large firms that (1-9) (10-49) (50-99) (100+) created the jobs Size in 1985 Micro (1-9) 35 61 4 0 were more Small (10-49) 1 69 24 6 Medium (50-99) 0 12 42 46 and expand. For Large (100+) 0 0 0 100 the most part, Note: Based on 164 IGLF firms with information on employment for both years. Shaded graduation of cells indicate firms that remained the same. beneficiary firms into the Source: SMIIES-NSO (1997). next size category occurred at a faster pace (Table 2.1) than it did for other firms in the Philippines (Table 2.2): 61 percent of all IGLF microenterprises grew into small firms between 1985 and 1995, and one- quarter of all IGLF small firms became medium-sized firms over the same period while 6 percent became large firms. In contrast, other firms in the Philippines displayed a much less pronounced dynamism: only one percent of microenterprises founded in 1990 became small firms after 5 years, and 4 percent of small firms became larger. Hence, these figures suggest that IGLF firms represent a very promising group of firms.7 6 High growth rates are usually expected from participating firms in any particular program: entrepreneurs tend to come from certain areas, sectors or firm sizes which makes them more likely to succeed, and so that they outperform the general population. A fairer comparison was provided by two comparable groups of entrepreneurs (from the same geographical area, sector, and firm size) to control for some observed firms' characteristics. However, other sources of self-selection are harder to determine (such as selection because of unobserved characteristics like entrepreneurial motivation) and some others are inherent to retrospective data collections (such as selectivity due to natural attrition). The high rate of graduation among IGLF firms does not necessarily suggest that these firms performed better because of IGLF support. The good performance of these firms may simply reflect the pick-the-winner strategy of commercial banks, the fact that only surviving firms were interviewed in the 1997 survey, or both. 11 Table 2.2. Graduation of New Firms, 1990-95 (as percent of 1990 cohort) Size in 1995 Micro Small Medium Large (1-9) (10-49) (50-99) (100+) Size in 1990 Micro (1-9) 99 1 0 0 Sall (10-49) 45 51 2 2 Mdium (50-99) 14 0 14 71 Large (100+) 0 0 0 11)( Note: Based on 2,447 firms founded in 1990 and still operating in 1995. Shaded cells indicate firms that remained the same. Source: NSO (unpublished listings of establishments between 1990 and 1995). 2.7 Much weaker graduation rates come from recent firm-level data for Africa. Biggs and Srivastava (1997) report that one third to 82 percent of African small firms (those with 10 to 49 workers) in their survey remained the same size or became smaller between their startup date and 1992-3. Graduation of microenterprises was even less likely: 72 to 95 percent of all microenterprises remain microenterprises between their startup date and 1992-93. 2.8 Furthermore, more than 80 percent of Philippine beneficiary firms remained in operation five years after they received credit. This is a very high survival rate given the international experience on the rate at which firms fail, especially small firms. For instance, smaller firms in Ecuador and Sri Lanka experienced failure rates of 40 and 47 percent, respectively (World Bank, 1997a, 1998a), and the five-year failure rates for IGLF firms were similar to those of larger firms after the first year of operation in Colombian and Chilean firms (21 and 27 percent, respectively, see Roberts and Tybout, 1996), and among new German firms (23 percent after two years, see Audretsch, 1991).How did SMIs expand employment? The experience of African microenterprises matches the weak role of microenterprises in general. 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. 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 (1997) report higher graduation rates for microfirms in Ghana (41 percent) and Kenya (50 percent). 12 Question 2: Are SMIs in the Philippines More Labor Intensive? 2.9 Yes. In Figure 2.3. Capital-Labor Ratios by Firm Size, the Philippines, 1994 general, the lower r600 capital intensity of the ifirm, the larger the possibilities for 400 expanding total ('000 pesos) employment. We -200 found that size is a good predictor of Capital Goods 0 labor intensity. We Intermediate Goods also found greater Consumer Goods M L V Total Manufacturing M e a ea variation in labor m a y a intensity across firm e e sizes than across 0 m Some: NSO (special tabulations based on the 1994 census of establislmts). different industries so that small firms in the Philippines can in general be said to have greater labor intensity (i.e., lower capital intensity). Figure 2.3 illustrates this point. It shows that microenterprises (defined as those with fewer than 10 workers) producing consumer goods are twice as labor-intensive as microenterprises producing intermediate goods, but they are ten times more labor intensive than very large firms (defined as those with more than 200 workers) in the same industry. Question 3: Are SMIs More Efficient? 2.10 No. Although central to the argument for special support to SMIs, Bank beneficiaries have not been more efficient than firms which did without the credit. Moreover, as is the case in Malaysia, Indonesia and Taiwan, SMIs are slightly less efficient than larger firms (Tan and Batra 1995). A rigorous analysis of efficiency-measured by total factor productivities of the two groups, IGLF and non-IGLF firms, as well as across sizes-shows no statistically significant difference between beneficiaries and the control group: The method of TFP takes into account the level of inputs used by firms (i.e., capital and labor) to generate total sales. At the same time, across firm size categories, small firms -the most numerous group of firms in our sample- have lower TFPs than larger firms. Relying on census data also confirms that SMIs are significantly, albeit slightly, less efficient than larger firms.'0 2.11 These findings indicate that SMIs have generated more jobs, but these jobs were slightly less productive than the jobs generated by larger firms. Thus, one of the two necessary Little (1987) finds that size could be "a poor indicator of any attribute of social importance" when there is more variation in labor intensity among industries than across size categories. 10 See background paper by Rodriguez and Mini (1998). The paper finds a monotonically increasing relationship between size and efficiency, which weakens the evidence in support of SMI-targeted policies. The increase in technical efficiency from small to large firms was only 3.5 percent (from 0.46 to 0.495 percent), very similar to Tan and Batra's findings for Indonesia, Malaysia, Taiwan, and other countries (see Tan and Batra, 1995). 13 conditions for assisting SMIs is fulfilled-SMIs were more labor intensive. They were not more productive, however, and, so there is a price to be paid in supporting such firms in terms of lower productivity. Question 4: Are There Any Non-Lending Policy Instruments to Improve the Efficiency of SMIs? 2.12 Yes. When supporting SMIs, the size of firm matters, but so do economy-wide factors. Under inward-looking policies that attempt to isolate an economy from external factors, SMIs tend to perform worse than in more open economies. Of course, trade liberalization can damage SMIs because of increased competition (as Parker et al., 1995, show for Africa), but liberalization also brings undeniable long-term efficiency gains by reducing the cost of inputs and through the technology advances embodied in the capital that can be imported in freer trade regimes. 2.13 Figure 2.4 provides a simple perspective on trade policy and the share of SMI labor force in a number of Asian economies. It shows that the success stories of Hong Kong and Japan are associated with lower tariffs and a larger presence of SMIs, while South Asian manufacturing employment was highly protected as well as dominated by large firms. It also shows that in countries where Figure 2.4. Protection and SMIs in Asia, Early 1990s manufacturin is % of manufacturing labor force mauatuigi 50 on Kong highly protected, "Japan the share of SMIs 45 in total manufacturing 40 employment is much lower, such 35alaysia as the Philippines, Korea, Rep. of India, and Sri Lanka. Of course, Indial trade policy is by Php ns no means the only 25 or even the most Sri nka important factor 20~ 1 -__ _ i__ _affecting SMIs. 0 10 20 30 40 50 60 Note: The share of SMis is calculated by using emp ffis trswLal employment ofall firmswith 10 Regulatory and tax workers or more (microenterprises are excluded). Sources: World Bank (tariffs, 1997b); Statistical Offices (Philippines, Korea, Sri Lanka. India, Japan, and Hong Kong; various years). arrangem ents m ay also hinder SMI development. For instance, in the Philippines, an institutional disincentive for more flexible labor markets arises from the outright prohibition of "labor-only contracting" which influences the existence of SMIs and their potential to create jobs. The Labor Code expressly prohibits the contracting out of labor to protect the rights of workers. It also prevents employment creation through subcontracting. Both restrictions reduce opportunities for SMI employment. Also, fiscal incentives are biased against SMIs. For instance, many large firms receive VAT exemptions to fertilizers, pesticides, herbicides including chemicals for formulation of pesticides (industrial chemicals). Similarly, large firms in the transport equipment industry enjoy indirect tax breaks under the car 14 manufacturing program. Because of these and other weaknesses of the tax system, there has been a comprehensive reform of the tax system recently approved in Congress. 2.14 Ultimately, leveling the tariff, regulatory, and taxplaying fields will improve productivity in general, and this general improvement in the economic environment will result in more efficient SMIs. Implementation of such a policy would also reduce the concentrations of people and employment in large cities. It would also obviate the attempts of policymakers to relocate investment away from the overcrowded cities and particularly the capital city. This kind of strategy was one of the rationales given for a number of the early Bank-supported SMI projects in the Philippines. It raises the question: Question 5: Did SMI Support Contribute to Regional Dispersion of Industry? 2.15 No. The assumption that support for SMIs could help decentralize economic activity does not take into account the benefits from economies of agglomeration that SMIs can derive (from infrastructure, closeness to markets, to suppliers, etc.). Indeed, it is the economies of such locational advantages that are the central gain to firms that cannot gain from economies of scale.11 Programs to relocate small firms which have little or no internal scale economies-as in the Philippines-are bound to be expensive and ineffectual. It should not be surprising therefore that SMI credits are even more concentrated than is the actual concentration of total manufacturing value added in the capital region. 2.16 This pattern of locational concentration is exactly what occurred. The first appraisal report of SMI credit in the Philippines expected the trend of future industrial growth to be more decentralized because of the SMI support. However, in the second half of the 1970s, IGLF firms followed Philippine patterns of manufacturing geographical distribution: almost 48 percent of manufacturing value added was concentrated in the National Capital Region (NCR) and so was 45 percent of the sub-loans Figure 2.5. Manufacturing Value Added and IGLF Loans, Early 1990s from the first SMI loan. By early 1990s, Figure 2.5 Mindanao shows that the geographical pattern had UVA(1992) OIGLF(1990-2)wosndthNC worsened: the NCR Visayas absorbed even more loans than its share of manufacturing value Luzon added, while other regions received less than their share in value added. NCR NCR had 52 percent of value I Iadded and 73 percent of the o uas 20 30% 40% 50% 60% 70% sub-loans disbursed Note: 238 sub-loan from SMI IV. Sourme: NSO O992); DBP (special tabulaions) between 1990 and 1992. Finally, an interest rate In Sri Lanka, Ecuador and the Philippines, econometric results consistently reject the existence of increasing returns to scale among SMIs. 15 incentive under SMI IV offered to IGLF loans outside NCR failed to create any significant impact on the geographic dispersion of IGLF loans. 2.17 Thus, SMI support in the Philippines did not help to deconcentrate employment away from the capital city. Nor should it have attempted to have done so. SMI support quite simply is not an effective spatial policy instrument. It can assist firms to grow and prosper, particularly if they are credit constrained, but it is an inefficient instrument to attempt to affect firm location. Question 6: Did SMIs Reach The Low Income Wage Earners? 2.18 No. As shown in Figure 2.6. SMI Workers in Income Distribution, 1994-5 Figure 2.6 workers in the 300000 - -_--- Philippines tended to be middle-income employees in 250000 --------- -- the sixth and seventh income 200000 deciles. Thus, Bank support 2 for SMIs did not reach the 150000S-- - -- -n poorest. Of course SMI SMI Workers' Annual Wages lending can, in principle, 100000 " support lower income wage 00oo --__ 1 -earners, as it did in Sri Lanka, see OED (World Bank, Average Annual Income (Pesos) 1997a), but in few cases is it Bottom 2nd 3rd 4th 5th 6th 7th 8th 9th Top likely to be an effective B: NSO om5.p.44bae o 4 FanyncedEpedr Su.ey): 1997 SMd4ES.N4O-fS. poverty reduction instrument. 2.19 In addition, attempts to make it a more effective poverty instrument are likely to make it a less effective jobs generating policy. The central effect of SMI lending on poverty stems from its potential to reach somewhat lower-paid urban workers, not poor people. Our survey reveals that workers are paid, on average, US$2,277 to US$2,851 a year, while the average laborer in the National Capital Region (NCR) earned an average of US$1,900, according to the annual survey of the Department of Labor and Employment (NSO 1995). Finally, SMI lending could not have a greater impact on poverty because of the concentration of the employment it generated was in the NCR where less than 8 percent of households were under the poverty line in 1994. Even if IGLF had employed the lowest paid workers in the NCR, poverty in the Philippines remained high in rural areas at 47 percent. (See Borrower's response for information on the new IGLF approach to rural areas). Question 7: Did Bank Support for SMIs promote exports? 2.20 Yes, but not by much. IGLF firms were almost twice as likely to be exporters -20 percent of the firms in the sample- as the control group. Among our sample of SMIs, as occurs in other countries such as Sri Lanka, it was the larger firms that were most likely to be exporters. Thus, while our evidence indicates that SMIs supported by the Bank did export and that the credits increased the likelihood of their doing so, only a relatively small share of them produced for external markets. Consider why this might be the case. 16 2.21 Trade in manufactures from the Philippines has indeed expanded, but not necessarily because of greater support to SMIs. Manufactured exports-as percentage of total exports- climbed steadily from 42 percent in 1970 to 87 percent in 1995, but they are concentrated around particular products: apparel and electronics dominate with 66 percent of all manufactured exports in 1995. Until 1980, apparel was the principal manufactured export.12 Since then, electronics13-included under capital goods-overtook exports of apparel in 1980 as the country's primary source of manufactured exports. This shift occurred because a number of large multinationals undertook semiconductor production exclusively for export markets. Therefore, most manufactured exports are exported by very large firms14. 2.22 However, there are some export sectors where SMIs can and do dominate, such as furniture and non-electrical machinery. Also, and more importantly, SMIs can take the indirect route to export markets via subcontracting with large firms. A large number of SMIs, especially in the apparel industry, started out as indirect exporters, by supplying parts and other intermediate inputs to large direct exporters (Tecson et al 1990). Similarly, in Japan, SMIs managed to produce for the export markets mainly through the intermediation of large trading houses (the so-called sogo-shoshas) which linked them to the world through their global information and marketing network. 2.23 As the most recent Country Economic Memorandum (World Bank, 1997c) says, subcontracting can become a promising area through which SMIs can more readily obtain access to export markets, provided a sector has high export orientation and that it can use components.15 The study links the need for greater technological support to SMIs with the need for improvements in SMI competitiveness. The potential for subcontracting appears large given the fact that available evidence indicates a quite limited level of subcontracting in the Philippine manufacturing: The ratio to total output of "industrial service providedfor others"-a gross indicator of subcontracting-shows that the average share is only slightly around 3 percent of output in 1993. Thus, there would appear to be considerable potential for expansion. 2.24 However, a World Bank study on technical support to SMIs by Levy et al. (1994) qualifies the speed at which SMIs could actually benefit from promoting SMI exports. The study argues that, in some circumstances, private mechanisms (i.e., personal connections, ethnic groups, etc.) are more important than collective mechanisms (i.e., formally organized groups) for helping firms improve their technological capability. As production becomes more complex, collective mechanisms are more important. Only, at this later stage, do they recommend a "high- intensity approach." Another study by Berry and Levy (1994) shows that, at the present stage of 12 Growth of the apparel export industry resulted partly from the a protective environment created by the Multi-Fiber Arrangement as it did in the case of Sri Lanka. 13 Electronics include four sub-industries: consumer electronics, telecommunications equipment, computers and peripherals, and components (semiconductor) manufacturing. The first sub-industry is basically domestic-market oriented, while the rest are export-oriented. 14 The 1993 survey of manufacturing gives a breakdown of manufactured exports by firm size: Very large firms (those with 200 workers or more) export 88 percent of all consumer exports, 79 percent of all intermediate-good exports, and 60 percent of all capital goods. 15 Industries with high or low export orientation can have low subcontracting activity such as textiles, non-ferrous metals, etc. if the possibilities for components and sub-assembly operations are not large. 17 Indonesia's development-and, probably, the Philippines-any direct support by the public bureaucracy to find foreign buyers-large or small- is bound to fail. They argue that Indonesian exporting SMIs (in garment and furniture industries) only needed to rely on private support mechanisms to sustain a substantial fraction of SMI exports. But these mechanisms are more readily accessible to larger firms, to educated entrepreneurs who can take advantage of a network of linkages among an extended Chinese community beyond Indonesia's borders (the Philippines has a similar situation with a thriving Chinese business community). Summary 2.25 We find that in the Philippines the Bank's support to SMIs resulted in more jobs. They were jobs whose productivity was indistinguishable from those generated by firms of similar size which did not receive the credit, but they were of slightly lower productivity than that of larger firms. In addition, the jobs were located largely in labor-intensive firms which operated in the same places as existing industry. Thus, the projects did not disperse employment away from the capital city. Nor were the jobs taken by the poor. SMI support is, in general, not an effective social safety net instrument. It is unlikely to provide a great many jobs for the very poor. Finally, assisted SMIs did increase their exporting, but not by much. There are more effective ways to support export growth than providing credit support to SMIs. 3. The Financial Sector and SMI Lending 3.1 As the recent Figure 3.1. Long-Term Loans in the Banking System, 3 Cn Assisance the Philippines, June 1993 OED Country Assistance Savings&Loans Review (World Bank, Associations 1998b) details, domestic Savings Banks 1% financial markets in the 24% -Philippines, have become considerably stronger, deeper, and more deregulated since the crisis in the mid-1980s. Bvate Dvopment Commercial Banks The ratio of broad money 1o 65% to GDP has more than Source: Bangko Sentral ng Pilipinas, Philippine Financial System Fact Book, 1995. doubled to 40 percent, and the Financial Sector Adjustment Loan supported a series of reforms that strengthened the Central Bank's autonomy and independence. Like many other developing countries, the largest bulk of long-term credit in the system remains in the commercial banking system (Figure 3.1). The commercial banks have also been the main conduit of the World Bank's SMI loans through an apex modality. Non-bank financial institutions have been restricted from supporting SMIs as a result of a strict accreditation program-introduced by the World Bank in the first SMI operation. Given the role played by undercapitalized financial 18 intermediaries in financial crises in a number of countries, such caution appears to have been quite wise. 3.2 One of the World Bank's fundamental concerns with lines of credit, such as those that support SMIs, is the issue of why prudent lenders, motivated by profits and constrained by their capital should not be the best judge of which firms should receive credit. Bank Operational Directive 8.30 requires that any lines of credit supported by Bank lending should be accompanied by policy discussions to address the impediment which created the limited access to credit in the first place. Ideally, according to the directive, a directed credit line should be undertaken only as long as progress is made on addressing the market or policy imperfection that creates the need for the line of credit. That is, in the limit, such lines of credit should ultimately eliminate their reason for existing. 3.3 Recent Bank research, by Caprio and Klingebiel (1996), indicates that the historical record for both developed and developing countries that financial reforms are rarely if ever implemented quickly. The implication of this perspective has been developed in the World Bank's new Financial Policy Framework (World Bank, 1997d) which makes it clear that the Bank now views financial reform and development to be a process rather than an event such as a project with policy covenants. From this longer term perspective, the following question arises: Question 1: Did Bank support for SMIs help to expand financial markets? 20% Figure 3.2. Real Interest Rates and Spreads, Philippines, 1976-96 3.4 Yes. But, N Real Lending Rate again, it has not 15% 1Spread been a rapid process. Over the A period of Bank support for SMIs, 0% A total credit (and 0% long term funding) _ _%__ for SMIs not only -5% s1I I Ski i SNII Hf iW IV increased, it -10% Eark 80s 1987-8. - increased in an Unsucce;s- Successful 9- environment which 15%- ful SALs i Economic Fmlancial nd 11 Reco% erN djustmen ultimately-after 5 20% Ln Lnyears-moved to 19S3-5 Balance ot positive interest -25% '976 1980 Payrment Crisis rtes, and then Source: IMF. IFS D.aabase. 1984 finally-after an additional 8 years- to lower interest rate spreads. While total Bank SMI lending was too small an amount to account for such fundamental changes in financial markets, as part of the ongoing the World Bank support to financial reform-a process the Country Asssistance Review (World Bank, 1998b) has described as being highly satisfactory-the Bank's SMI support has both helped broaden the financial sector and developed SMI access to credit. It has, in short, helped the Bank maintain a broad dialogue on the importance of a well-functioning financial sector for achieving and maintaining stability and growth. 19 3.5 However, again, this process has not been linear and straightforward. For example, for many years the government supported a range of lines of credit to support favored access to credit by specific actors in the economy, such as SMIs. As early as the 1950s, it tried various approaches to direct credit towards microenterprises and SMIs, and, by the mid 1980s, there were more than 100 different direct lending programs, mainly to SMIs or microenterprises (see OECF, 1995). Finally, in 1990 a "Magna Carta for SMIs" was announced by the government to rationalize government support for SMIs, including a mandatory minimum percentage of bank loans to SMIs (see the answer to question 3). Unfortunately, the government instead created another institution the Small Business Guarantee and Finance Corporation (SBGFC) that has added to the already large number of institutions specialized in credit to SMIs. Question 2: Did Support for SMIs Lower Interest Rate Spreads in the Financial Sector? 3.6 Ultimately, yes. Until the early 1980s, the Philippines had a Usury Law in effect which put a cap on all lending rates, on the simplistic assumption that lower interest rates would encourage more investments, especially among SMIs. In such a repressed financial environment, credit rationing was bound to take place, with SMIs crowded out by large firms. Then, the influential economic and sector work by the Bank (Anderson and Khambata, 1981) showed how the inability to price risks led to capital market failures which are particularly relevant for SMI credit markets. This work presented first hand evidence on the risks and administrative costs of SMI lending, making the case for a relaxation of administrative constraints on the structure of interest Figure 3.3. IGLF Interest Rates and Market Lending Rates, 1977-96 rates. 30 C3F Sra IGLF to PFI Rate 3.7 Before 1987, 25 Market Lending Rate IGLF rates were mostly below market 20 rates. After 1987, IGLF followed market _s_- - - rates more closely. N Figure 3.3 illustrates 10 this evolution of IGLF nominal rates vis-A-vis market rates. In the early 1980s, average a -. -- -lending rates remained 1977 1981 1984 1987 1990 1992 1996 unchng at 12 Source: World Bank (various years). percent although effective lending rates were much higher. Before 1987, PFI borrowing rates from the IGLF facility were kept so low that even after a maximum gross spread of 5 to 7 percent, depending on firm size, PFIs could still lend to small borrowers at less than the prevailing market rate: The period 1984-5 illustrates such a case when lending rates soared to 28 percent. Starting in 1987 banks were finally allowed to charge their own rates for IGLF loans, subject to approval by the IGLF Review Committee. After 1989-90, under SMI IV-that is, after IGLF's transfer to DBP-IGLF rates finally reflected "market" rates (i.e., prevailing domestic interest rates). 20 3.8 Pricing long-term loans is of course difficult when there are no fully-developed capital markets in a country and when there are potential foreign exchange losses in economies with regulated interest rates. During the 1980s, when the peso was devalued, the government was left with a large foreign exchange loss towards which the accredited financial institutions and sub- borrowers made no contributions (World Bank, 1989). Once financial markets become fully deregulated, interest rates reflect inflationary expectations and foreign exchange adjustments (such as the latest 35-percent devaluation of the peso in late 1997) which explains why programs, such as those administered by IGLF, are expected to reflect these changes by using market-based lending rates. Ultimately, however, significant progress on interest rate spreads has occurred. They have declined to around 5 percent, while real interest stayed below 10 percent until the financial crisis in late 1997. Question 3: Did SMIs in the Philippine Have Less Access to Credit? 3.9 Almost certainly in the years before financial liberalization (1989-92) access to credit was limited, simply because the breadth and scope of the financial sector itself were deeply constrained. However, since financial liberalization, the answer to this question appears more and more to be that SMIs Figure 3.4. Credit to SMEs, The Philippines, 1991-6 % of total ortfolio have gained more access to 16.0% SME Eligible Loans credit, at least, short-term 14.0% Mandatory Rate (RA 6977) credit. . In the 1990s, credit IGLF Loans as % of all to SMIs increased even 12.0% SME Eligible Loans beyond the mandatory levels prescribed by law, as 1o.0% indicated in Figure 3.4, e.0% although term credit to SMIs-for the most part 6.0% from IGLF-explained a 4.0% small and shrinking portion of this credit. 0.0% 3.10 As the figure 1991 1992 1993 1994 1995 1996 shows, credit to SMIs Source: Bangko Sentral ng Pilipinas; Development Bank of the Philippines exceeded the Government's mandatory credit allocation to SMIs in 1990. These mandatory targets were spelled out in the previously mentioned "Magna Carta" for SMIs which required lending institutions to set aside a portion of 16 their total loan portfolio for SMIs, aiming to expose banks to the experience of lending to SMIs with a view of reducing, if not eliminating, their reluctance to lend to such borrowers. The law was expected to create awareness among lending institutions of the profitability of extending credit to SMIs, to improve their informational base on the creditworthiness of such borrowers, and to help develop structures within banks to handle SMI accounts. In the event, the mandatory allocation of credit was not at all restrictive because SMIs ended up receiving even more credit from the system than the percentage prescribed by the government. At the same time, IGLF 16 The portion mandated "at least be 5 percent by the end of the first year, 10 percent by the end of the second year through the end of the fifth year, and 5 percent by the end of the 6th year and may come down to zero by the end of the 7th year" (Magna Carta for Small Enterprises, R.A . 977, Chapter III, Sec. 13). 21 credit-the single largest source of long-term credit to SMIs-declined as a percentage of total credit to SMIs, as most of the increase of credit to SMIs was in the form of short-term loans. 3.11 Moreover, beneficiaries of World Bank term credit appear to have reached a debt mix similar to other countries. Demirguc and Maksimovic (1996) studied firm debt maturities in thirty developed and developing countries between 1980 and 1991. They found that long-term debt represents a higher portion of total firm debt in more developed economies and in countries with larger commercial banking systems. When IGLF beneficiary firms are compared to their sample of firms, we find that beneficiaries had a higher-than-expected debt maturity mix for the Philippines' level of financial development. For instance, Bank-supported firms in the Philippines have almost half of their debt as long term debt, and this figure is higher than what one would expect given the size of the Philippine commercial banking system and the income level of the country. Among firms with IGLF term credit, the debt maturity mix reached the levels of firms in countries with more developed banking systems. Similar results were found for Sri Lanka17 (World Bank, 1997a). Figure 3.5. Long-Term Debt: IGLF Firms within an International Perspective Long-term over Total Debt 0.7 Canada Norway 0.6 Philippines (IGLF) New U.S. Finland Gernlany A )NwZealand A 0.5-A Sweden Switzerland Sri Lanka (NDB) Australia A A Singlanore 0.4 South kdrica Sp nce Austria Malaysia Nethe5tlands 0.3 KoreaARep.of a Japan Ingla UA4 Hog Kong a Brazil B TArkey Thailand J6rdan A A A Jra 0.2 Zimbabwe Pakistan 0.1 0 0 0.5 1 1.5 2 2.5 3 Deposit Money Banks' Domestic Assets (as % of GDP, 1980-91 averages) Sources: Demirguc and Maksimovic (1996); World Bank (1997a); SMIIES-NSO (1997). Question 4: Did SMI Lending Affect Commercial Bank Risk Exposure? 3.12 Yes. It appears to have improved commercial bank risk exposure. Unlike the situation in developed economies where arrears rates for all small firms are multiplicatively higher than those of large and mid-size firms, arrears rates for Bank-supported SMIs were much lower than those on other bank lending. We do not have figures on loan losses by firm size which would be 17 Firms with term credit by the National Development Bank (NDB) had an above-average long-term debt ratio. 22 conclusive evidence of relative riskiness. As a result, we only have a proxy measure for credit risk. However, the performance of SMIs on this measure was more than eight times better than that of large firms. Thus, given the scale of this difference in performance, arrears rates are probably at least an accurate monotonic comparator of relative risk.18 3.13 Risk of default and arrears changed depending on the modality of the credit and the time of appraisal. The first line of credit to Philippine SMIs had two very different conduits. The Development Bank of the Philippines (DBP), the first implementer, operated as a direct lending to the SMIs without private financial intermediation. As Webster et al. (1995) show, this modality has been proven to have a higher incidence of arrears. Nevertheless, in many countries, where privately-owned banks did not exist, government banks became a common channel for the direct distribution of World Bank loans to SMIs. 3.14 Reliance Figure 3.6. IGLF and Total Arrears Rates, Commercial Banks, 1989-95 on an apex (as % of total outstanding loans) 9 ---- institution was the second approach applied to SMI lending in the Philippines. The Al Loans results of this shift in 4- intermediation 3 _ channels resulted in slightly lower IGLans arrears than under - - the direct lending ~ M program with 1989 1990 1991 1992 1993 1994 1995 DBP. The Bank's Sources: BSP (based on data from banks, unpublished); ADB (based on DBP, 1996: table 2, p. 22). conclusion was to switch, after 1989, SMI support exclusively to IGLF in the next SMI loan (SMI II). But, regardless of the modality of SMI credit, arrears of SMI loans have been lower for a long period of time compared to all other loans in the system, implying either a very cautious selection of SMI sub-borrowers by commercial banks or less underwriting seriousness on the part of the banks with regard to their larger loans. During the crisis years of 1984 and 1986, arrears ratios for SMIs were extremely low while economy-wide indicators showed that more than one fifth of all loans in the system were in arrears. After 1989, the difference-less pronounced than in the mid 1980s-still persisted, and arrears among SMIs remained much lower. Figure 3.6 shows arrears rates of IGLF 18 We do not have evidence on default rates or the costs of default. However, in the most recent estimates, the arrears rates for SMIs is one-eighth than that of all firms (large or small). Hence, unless the default rates of all firms are not at least eight times larger per dollar lent for a given arrear rate, then the default costs for SMIs must be lower. 19 SMI I concluded that both DBP and IGLF had reasonable overall levels of 5 to 6 percent arrears ratio to total loans, but one would have expected DBP to have an even lower arrears ratio since most of the DBP loans were to large clients. 23 and all loans in the banking system after 1989.20 Surveyed banks in 1997 also confirmed the findings shown in this graph. Our survey of PFIs shows that IGLF arrears rates were actually almost negligible while the arrears rates on their other non-IGLF loans reached an average of 3 percent to 8 percent (for commercial banks and non-commercial banks, respectively). 3.15 Thus, while we are not sure of the mechanism that caused SMIs to be such relatively good credit risks, the evidence is clear that in the Philippines-as well as in Sri Lanka and Ecuador-small firms receiving IGLF loans are much less likely than are larger firms to experience loan repayment problems. The explanations for this result lies beyond the scope of this report, but it may well be that loans to large firms are made on the basis of "crony capitalism" in which loans represent a form of insider trading. Alternatively, in small open economies, larger, and, hence, more export-oriented, firms may have differential sensitivity to shocks than smaller, more inward-oriented firms. Ultimately, the result would be considered unusual in well-developed financial markets where most SMIs are actually bound to fail. However, an immediate implication of this unusual result is that term lending to SMIs causes private lenders to incur less risk. In other words, it means that, unless the interest rates charged to SMIs (and adjusted for higher administrative costs) were lower than those charged to larger firms (also adjusted by an amount that offset the lower credit costs of their borrowing), then SMIs were discriminated against in the competition for credit. Unfortunately, our data do not allow us to price the credit risk or administrative cost premia across large and small firms, but we find that, within our sample of SMIs, larger SMIs paid an interest rate 15 percent lower than smaller SMIs. This interest rate differential may still reflect higher borrowing cost of small firms (also, see Borrower's response). Summary 3.16 From a longer-term financial sector perspective, we find that Bank support for SMIs in the Philippines was part of a broader dialogue that was quite fruitful. The system became stronger and deeper and almost certainly these improvements helped the economy weather better the shocks recently experienced by many of its neighbors. More directly, an unexpected and unusual by-product of the support was that the credit risks of the commercial banking sector was reduced: the repayment record of SMIs was considerably better than that of larger firms. Today, interest rates differentials between small and large firms may just suggest higher costs of borrowing among small firms. This is in contrast to discrimination against SMIs, which was more likely to have prevailed in the pre deregulation days when Bank support to Philippine SMIs first appeared. 3.17 The effects of the loans on the corporate financing patterns of SMI firms appear clearer. Philippine beneficiaries have greater indebtedness than do firms in other parts of the world. Certainly by the time of our survey, prior to the crisis in 1997, it does not appear that these SMIs had any difficulty in accessing credit. 20 Between 1982 and 1987, total arrears are calculated as a ratio of outstanding loans, not total initial loans. 24 4. The Costs of SMI Support 4.1 The costs of supporting SMI lending in the Philippines must take into account direct government expenditures, the possible resource costs for the misallocations encouraged by the program, and the guarantee costs that the government incurs-either directly or indirectly-as a result of government supported banks undertaking such lending. Question 1: Was the SMI Job Creation Expensive for the Government? 4.2 No. The direct costs were a small share Figure 4.1. Government Consumption, Exports and Financial of overall program Deepening, The Philippines, 1975-96 (as % of GDP) costs. In SMI IV, for 45%- example, government 40% - expenditures in support _ of assisted SMIs amounted to $47 30%. million dollars-out of 25%- a total credit cost of 20% .4Government $355 million- 15enmn representing less than 1I consu Wion 12 percent of total 10 project costs. However, 5 as discussed under 0% question 4 below, it does not appear that 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 Source: IMF, Infemalional Finanial Stalistics these expenditures were very productive. Figure 4.1 shows that the trend in government consumption over the past 20 odd years has been essentially flat. However, it also shows that financial deepening has followed a rapid growth path, increasing almost as rapidly as exports as a share of the economy. Thus, since direct costs were relatively low, it is with respect to the resource costs or costs of guarantees-either explicit or implicit-that the chief costs of SMI support is likely to occur. Question 2: Did the SMI Credit Line Have High Resource Costs? 4.3 No. The SMI support-particularly in the later loans-took the form of a directed credit program that was on-lent by private lenders at market determined interest rates. The general costs of this kind of policy was considered in a recent article by Stiglitz and Uy (1996). They identify some characteristics of directed credit programs that are likely to be effective in the sense that they fulfill program objectives while carrying low resource costs. 25 It is instructive to use these Table 4.1. Effectiveness of Directed Credit Programs: Selected characteristics to compare the Bank- Characteristics supported SMI credits in the Philippines. As shown in Table 4.1, Philippine Philippine since SMI III, the Philippine SMI SMI I to II SMI III to operations conform with all the IV characteristics of an effective directed 1. Moderate or no subsidy no yes credit program. At real interest rates of ltow 2. Institutions monitor performance yes yes less than 10 percent, and with loweffectively administrative and credit risk cost estimates, there is no subsidy to 3. Ability to adapt rapidly yes yes borrowers; recovery is at commercial 4. Directed towards private firms yes yes rates; interest rates and terms can be 5. Limited amounts yes 5. Lmite amuntsyesyes varied by the lender during the course of a loan; private firms use the credit 6. Credible Development Financial no yes and because they have to repay and the funds are not subsidized, they only 7. Targeting SMIs yes yes take the credit if it can be used productively. Thus, the credit is targeted on high return concerns. Further, if the targeting is wrong in the sense that it goes to lower-return borrowers, it is not the government, but the private owners of the firms or banks who bear the consequences of the targeting decisions. Finally, SMI lending has not been binding since commercial bank lending exceeded government credit restrictions. As a result, there are no domestic resource costs implied by the credit lines. Question 3: Was Government Risk Exposure Affected by Bank Support for SMIs? 4.4 Yes, it appears to have been improved. Krugman's (1998) analysis is helpful in considering this question. He argues that one of the central aspects of the recent financial crisis in East Asia was that financial institutions were seen to be insured by the government, even if they were not, until in the crisis, it became too expensive to honor the guarantees. In other words, banks, even if not explicitly insured or if insured only up to some limit, were, in fact, perceived as carrying a government guarantee. In this case, the operations of banks carry either a contingent liability for the government or a resource cost in terms of the investment incentives generated. In the former case, banks' credit risk is, as a result, shared with the government. Hence, lending decisions that lower commercial bank risks also lower government contingent liabilities. Accordingly, lower credit risk for SMIs vis-A-vis other lending often implies lower overall government risk exposure. Question 4: Did the Explicit SMI Credit Guarantees Result in a Better Risk Allocation? 4.5 No. The explicit guarantees for SMIs in the Philippines have been little used. As in a number of other countries, guarantee schemes were tried in the Philippines to attempt to protect banks from SMI default of loans.21 But these guarantees are not only not needed, they may be perverse if they encourage more risk-taking than is optimal. The results of SMI guarantee schemes in other countries have proven to be mixed (Levitsky and Prasad 1989; Levitsky 1997). 21 The requirement of collateral - usually in the form of real estate - compensates for the real and perceived risks of lending to this sector. 26 While in Japan and Korea guarantee schemes have had a good coverage (7.5 percent of SMIs in Japan and 13.1 percent in Korea), in countries such as Indonesia,22 these schemes have experienced a low coverage and high default rate. 4.6 In the Philippines, IGLF used two kinds of guarantee schemes, but their use remained disappointingly low: only 3.8 percent of the IGLF loan approvals were covered.24 The main deterrent to the use of guarantees by the banks lay in the claim procedure. In case of default, the claim procedure required PFIs to foreclose and liquidate the collateral before the IGLF would process the claim for payment. Actual payments would come only after about four to seven years, because legal foreclosure on real assets could take 2 to 4 years and their liquidation another 2 to 3 years. In the meantime, interest charges would continue until the foreclosure is completed, while determination of the actual amount of loss could be determined only after liquidation of the asset (World Bank, 1989: 17). Thus, four intermediaries interviewed claimed that they do not use the IGLF guarantee scheme, mainly because of its unattractive payout feature.25 See the Borrower's response for additional reasons on the tighter guarantee control by IGLF. Question 5: Did Technical Assistance Increase SMI Efficiency? 4.7 No. The effectiveness of technical assistance to SMIs has been disappointing in terms of impact. We were unable to discern any statistical effect of technical expenditures on firm performance. However, in many respects this result is not surprising as teasing out such empirical effects is difficult at best. Nevertheless, even by qualitative standards the effects seem weak. For example, by the last SMI operation, the emphasis of technical assistance had shifted away from blanket extension services to focus on exporting SMIs. It offered a number of services to exporters such as pre-shipment export guarantee. But, according to our firm-level survey, less than 15 percent of IGLF beneficiaries used any of this technical assistance. This is a very low overall coverage to expect any significant impact on efficiency stemming from these projects. The fact that almost four-fifths of these users found this assistance efficient and agreed 22 For instance, in Indonesia, the scheme covers less than 0.1 percent of all SMIs, but it paid 50.3 percent of guaranteed claims to these covered SMIs (Hatakeyama et al., 1997). 23 The first, known as the collateral-short guarantee, was made available for small loans to cover collateral deficiency up to a amount equivalent to 25 percent of the loan amount. It was a fixed absolute amount guarantee which covered first losses to the extent of the guarantee or the loan outstanding, whichever was lower. The second scheme, known as the credit-risk guarantee, covering up to 60 percent of small loans and 40 percent of medium loans. A straight guarantee up to 80 percent of the loan amount was also originally offered to PFIs that used their own funds. A fee of 2 percent of the guaranteed percentage of the loan outstanding was passed on to the borrower in the case of the collateral-short guarantee and to the PFI for the credit-risk guarantee. 24 Another guarantee plan remained widely underutilized (the Pre-shipment Export Finance Facility provided by Philguarantee). After enacting the Magna Carta of SMIs, it was expected that the newly created institution, SBGFC would provide large-scale guarantees for SMIs. But, since the beginning of its operations (1993-96), SBGFC only had an average volume of guaranteed loans of approximately 400 million pesos, equivalent to only 5 percent of IGLF loans during the same period. 25 Some admitted preference for and actual recourse to the guarantee scheme under Guarantee Fund for Small and Medium Enterprises (GFSME), which pays the amount guaranteed after the PFI files a claim even before foreclosure of assets of the defaulting sub-borrower. If it were not for the present IGLF's claim procedure, its guarantee scheme would have been preferred to the alternative guarantee schemes such as the GFSME facility which is quite limited in amounts and the Philguarantee scheme which is limited to exports. 27 that the assistance had helped them to increase their sales is at least suggestive that the problem was more one of outreach than content of the assistance. 4.8 The performance of technical assistance in the Philippines is consistent with international experience. The general experience is that technical assistance provided by central governments is not impressive. Drawing from the experience of marketing and technical support to SMIs in Colombia, Indonesia, Japan, and Korea, a World Bank study by Levy et al. (1994) finds that the most effective marketing support for SMIs came from decentralized organizations such as industry associations or local governments. Also, SMIs themselves found technological support overseas. A number of case studies -including India, Japan, Korea and Singapore- illustrate how SMIs seek foreign sources of technology to adopt them to their national conditions (Buckley et al., 1997). Evidence from Taiwanese firms reinforces the fact that technical efficiency rises, not only, with investment in training, R&D, but also with informal contacts with foreign purchasers through export sales (Aw and Batra, 1998). But the interest of governments to give technical support to SMIs has not waned. In 1996, during the APEC Leader Summit, hosted by the Philippines, the government presented its official initiative for a technological transfer and training center for SMIs. The rationale of greater investment in SMIs lies in the fact that greater technology and SMIs may generate faster growth fueled by exports. Summary 4.9 To sum up the effect of SMI programs on government costs and risks, we find that they were not an expensive way to support the creation of new jobs. Nor did this support expose the government to risks. Indeed, it appears that the projects helped reduce government risk exposure. While the technical assistance provided had no significant effects it seems that this weakness appears to be more the result of program scale and structure than, content. Nevertheless, the government's apparent enthusiasm for the development of a technical training center for SMIs must be considered an as yet untested idea. Finally, the allocational effects of the directed credit channels used to provide the credits do not appear to be costly. Indeed, an argument can be made that their use improved the distribution of risks. 5. Lessons learned 5.1 The study of these operations offers a number of lessons: (i) Create jobs. Under the right circumstances, SMIs can be effective jobs programs. They were in the Philippines, and, with policy reforms and careful design, they can be more effective. However, they are not likely to have much, if any, effects on directly reducing poverty or increasing exports. Thus, they should be considered as part of a social safety net in only the broadest possible terms, for example, as instruments that can help address high urban unemployment rates. Nor should they be used as programs to support locational policy. (ii) Improve Banks 'performance. SMI lending through financial intermediaries (FILs) can support banking sector soundness. In a word, there is such a thing as a good directed credit program particularly in places suffering from the inability or recalcitrance to foreclose on large borrowers. In environments with fragile 28 trust in the financial system and large contingent liabilities realized by the government on behalf of financial institutions, this can be a significant gain. (iii) Use moreflexiblefinancial products. SMIs confront a fundamental problem of higher real borrowing costs for credit quite simply because they tend to be or are seen as bigger business risks, and are priced accordingly. These costs, and often the associated spreads, tend to rise under macroeconomic imbalances. In such environments, establishing basic banking links, such as bank accounts or short- term credit, is one way to reduce some of these high costs faced by SMIs with little or no credit history. These links can be seen as building blocks towards a more complex banking relationship such as that implied by long-term loans. While much of Bank supported SMI lending has emphasized access to long-term maturities, microenterprise finance lending has shown the importance of short term maturities to small firms. (iv) Design more market-driven technical assistance. SMIs already invest in acquiring technical assistance so that there is room to take a more demand- driven and more private sector-oriented approach to technical assistance, such as emphasis of technology upgrading and market linkages with large private firms. 29 Annex A Comments from Borrowers REPUBLIC OF THE PHILIPPINES NATIONAL ECONOMIC AND DEVELOPMENT AUTHORITY NEDA sa Pasig. Amber Avenue, Pasig City Cable Address: NEDAPHIL P.O. Box 419. Greenhills Tels. 631-09-45 to 64 1 June 1998 Mr. Ruben Lamdany Manager Country Evaluations and Regional Relations Operations Evaluation Department The World Bank 1818 H Street N.W. Washington, D.C. 20433 U.S.A. Attention: Mr. Edgard Rodriguez Dear Mr. Lamdany: As requested, transmitted herewith are our comments on the OED report entitled "World Bank Support for Small and Medium Industries in the Philippines: An Impact Evaluation." These reflect only the views of this Staff and have not undergone inter- agency discussions in the IGLF Technical Advisory Group (IGLF-TAG). Mr. Edgard Rodriguez in a later communication to us has suggested that we integrate our comments with those of the Development Bank of the Philippines. However, we find this time consuming. Nevertheless, perhaps we can include a discussion of this report in one of the forthcoming meetings of the IGLF-TAG and subsequently the IGLF-Review Committee. We hope that you will find the attached in order. Very truly yours, MARGXRITA R. SONG Director 30 NEDA SECRETARIAT COMMENTS ON WORLD BANK SUPPORT FOR SMALL AND MEDIUM INDUSTRIES IN THE PHILIPPINES: AN IMPACT EVALUATION General Comments 1. In general, we find the study useful and the findings as expected. There were no surprises. As "owner" of the funds, it is heartening to note that the facility has to some extent served some of its purposes and at no great cost when, perhaps, compared to other similar facilities. 2. There is, however, need to put the study in its proper perspective for the guidance of all those who get to read the study. It is our understanding that the study covers only the IGLF program in the Philippines, i.e., the firms who were included in the survey have availed themselves of the IGLF. It this is true, then this has to be explicitly stated in the paper whether at the Introduction section or somewhere else. We know of another World Bank supported facility (i.e., Cottage Enterprise Finance Project) which was at the outset pushed by the Bank and implemented in September 1991 but had to be pre-terminated in April 1993 because of implementation difficulties. It is recognized that this letter facility was directed at cottage enterprises,. However, one who is not familiar with the definitions of cottage, small and medium enterprises can say that this falls under the category of small enterprises. 3. If our interpretation as mentioned in item 2 above is correct, it should also be clarified that the IGLF facility has been in existence for more than four (4) decades now, i.e., starting in 1952. The period covered by the survey is already the second half of the life of IGLF. Given the more then twenty (20) years experience prior to the period covered by the assessment (i.e., 1976-1992), it is possible that the lessons learned could have contributed to the favorable performance of the facility. Moreover, as gleaned from the comments of the Development Bank of the Philippines, remedial measures are being undertaken along the way to make the facility more effective. This, of course, does not imply that the facility was less effective in the previous years. 4. From the paper, it seems to give the message that availability of credit to SME's with some technical assistance is the solution to the problems of SMEs and that of the country as a whole (at least from our reading) e.g., employment generation, technology upgrading, market expansion to rest of the world, industrial dispersion, poverty alleviation, among others- We recognize that availability of credit is just one measure or instrument to support/assist SMES. A package of measures is necessary from sourcing of raw materials to marketing of finished products. There is the issue of attribution or cause and effect. 5. "Access to credit "has been mentioned in the study a number of times. The way the term waS used (based on our perception) means making available to SMEs financing. On the other hand, we use this phrase in a dfferent manner, i,e., beyond availability of financing. While the funds are available to be on-lent to SMEs, these (SMEs) do not take advantage of the ftdlity because the documentary requirements and procedures maybe too cumbersome or complicated for the SMEs to comply with. Thus, their access to credit is hampered. In fact, lack of information about the existence of financing facilities is already a bottleneck. 6. Since 1992, the IGLF program has been implemented using second generation funds. This information, together with the initiatives undertaken by the IGLF Review Committee since 31 then to the present to address the perceived shortcomings of the program (some of which have been among the study's findings) may also be incorporated in the latter part of the paper, Note that mention was made of the ACTETSME, which was established in 1996 as part of our initiatives in APEC not only for the benefit of Philippine SMEs. 7. There are on-going studies on other financing facilities being administered by other entities such as non-government organizations (NGOs) with funding sources other than the World Bank. It will be interesting to compare the results thereof with subject study at the appropriate time. Specific Comments 1 . Page 9, para. 2.3. This is in terms of employment. However, in terms of share to gross domestic product, the shares of services and industry are larger than those from agriculture, fishery and forestry". 2. Page 11, para. 2.5. While the statement "....while nonbeneficiaries lost 1,400 jobs, representing, an average, 10 fewer jobs per firm ...." may be factual of the responses to the survey, there is danger in some of the interprotabons that can be derived from this. To put it simply, one can say, a firm lost jobs because it did not avail of an IGLF loan. Thus., to keep on generating employment, a firm has to avail of an IGLF credit. On the other hand, the IGLF beneficiary firms may have instituted measures to improve their viability other than borrowing from IGLF. Furthermore, the period covered by the study, included years when there was economic slowdown. This could have contributed to the performance of those firms that laid off workers. 3. Page 14, para 2.13. It has to be clarified what is meant by "larger presence of SMES." It is possible that there are more SMEs in terms of number but their share to total industry output is small or in terms of contribution to export value and/or volume. 4. Page 15, para 2.15 and 2.17. This is true not only for SMEs but even for large enterprises. Availability of infrastructure facilities is a major consideration in the location of business activities. Even without the study, the IGLF-TAG and IGLFRC already recognized the fact that the national capital region, Regions IV and III almost always accounted for the greater bulk of IGLF availments. In response to this observation, lower interest rates were allowed for the lagging regions. This move notwithstanding, these regions (NCR, Region IV and III) continued to be among the regions that maintained high level of availment of the IGLF program. Note that even as early as 1973, there was issued a Memorandum Circular which provides for, among others, that "There shall be no more factories, plants, industries and the like to be established within a 50-hectares radius of Manila." This goes on to show that there are other instruments employed to encourage location of industries outside of the capital city. 5. Page 16, para. 2.20. Does this refer to direct exports by SMEs? In exporting, there is need for volume and logistical support which many SMEs may not haveWhat has been done is to go into consolidation of exports. This is prevalent in the case of processed food. For example, a trader can put together sauces preserved fruits, ethnic food, etc. which are all products of SMEs and undertake the exportation. 6. Page 18, para. 2.25. We cannot agree more with the last sentence. Exporting is a complex process which involves a range of activities from product development to shipping, etc. 32 7. Page 19, para 3.1. The second to the last sentence also arises from a policy pronouncement to use financial institutions as conducts for delivery of creditHowever, there are existing financing schemes for SMEs funded by sources other than the World Bank which make use of government and nongovernmental organizations but are not financial institutions. 8. Page 28, para. 4.5. This may be examined from two (2) levels - the risk of the administering institution for IGLF which is the Development Bank of the Philippines or the national government and the risk of the participating financial institutions. At present, the credit risk is with the participating financial institutions and the foreign exchange risk is with the national government through the Department of Finance. 9. Page 30, para 4.8. The technical assistance component as a percentage of the total IGLF facility may not be substantial. Further, it may not have been the appropriate type of technical assistance required by the SMEs at that time These are worth looking into. At present, various types af technical assistance is provided to SMES: business consultancy, market encounter program, matching/linking program, networking, value orientation, productivity improvement, information dissemination, technology transfer, trade fairs. 33 S3cIDevelopment Bank of the Philippines Annex A Wholesale Banking Sector - PMD II May 28, 1998 Mr. Ruben Lamdany Manager Countryside Evaluation and Regional Relations Operations Evaluation Department World Bank 1818 H. Street N. W. Washington, D. C. 20433, USA Dear Mr. Lamdany: Re: World Bank Support for Small and Medium Industries in the Philippines: An Impact Evaluation Thank you for soliciting our comments on the aforecited report. We generally agree with your observation and evaluation of SMI financing in the Philippines. We would like, however, to comment on some of your observations insofar as the IGLF facility is concerned: 1. SMI did not provide a great many jobs for the very poor. The DBP through IGLF has been addressing this concern through a series of proposal to the IGLF Review Committee to perk up countryside lending as we feel that the location of projects dictates job opportunities. The IGLF Review Committee, upon the recommendation of DBP approved the Special IGLF Facility for Countryside Projects sourced from earnings of IGLF to encourage PFIs to look for projects located in the countryside. Also, only projects located outside of the National Capital Region, CALABAR and Bulacan province are eligible under this program and interest rates are generally 2% lower than regular IGLF loans. In order also to encourage new projects and therefore additional job opportunities under this program, the facility requires that at least 70% of the loan proceeds should be utilized for acquisition of fixed assets. The result is encouraging as over PI billion so far has already been released to various projects under this program. The end result of the aforecited project was the great improvement in countryside availments of IGLF compared to the previous years. As of December 31, 1997 the ratio of loans granted in the regions was 60% in term of amounts and 67% in terms of the number of loans granted. HEAD OFFICE MAKATI CITY, METRO MANILA. PHILIPPINES , P 0 BOX 1996, MAKATI CENTRAL POST OFFICE 1200 cABLE PHILOEBANK MANILA 'iON TELEPHONE (6321818-9511 In addition the DBP caused the approval by the IGLF Review Committee a short-term rediscounting program precisely to provide facilities for micro and cottage-sized enterprises. Only countryside banks such as thrift and rural banks are eligible participating financial institutions (PFIs) under this program. To date, DBP has already released a total of P80.00MM to rural banks all located in the south. While the amount is modest, we consider the program fairly successful since we did not aggressively market the fund. This is attributed to our policy of giving priority in loan releasing to the regular IGLF credit program. Funding for the rediscounting should only come from excess funds temporarily invested in treasury bills. Our feedback on this facility is very encouraging provided we relax some terms. This is the kind of program fund needed in the countryside where small banks (PFIs) do not have the capability to evaluate long term projects. 2. Smaller Firms were Charged Significantly Higher Interest Rates. The credit markets did discriminate against SMIs. The higher interest rates charged to smaller firms by banks is a normal practice and not discrimination against them. Smaller firms normally have higher credit risks compared to the big ones, hence, they are assigned higher rates in price tagging. Also, most smaller firms are inclined toward smaller banks such as thrift and rural banks since the latter generally do not ask for many credit requirements. These banks charge higher interest rates than commercial banks since their main source of income are from lending unlike in commercial banks where income-generating activities are much wider. Moreover, bigger firms could offer some compensating business (such as current account deposits, letter of credits, insurance coverage, etc.) with their banks and therefore can get concession in terms of pricing of loans. 3. In the Philippines, IGLF used two kinds of Guarantee scheme but their use remained disappointingly low. a) With the establishment by Government of the Small Business Guarantee Fund Corp. (SBGFC) and Guarantee Fund for Small and Medium Enterprises (GFSME) mandated to handle the guarantee of smaller firms, IGLF concentrated on lending activities although we do not refuse requests by PFIs to guarantee IGLF loans of cottage, small and medium scale industries. We also embarked on a complementation agreement with GFSME relative to the guaiantee of SME loans. b) IGLF has to adopt stricter claim procedures for the following reasons: * It would become necessary for us to thoroughly evaluate the viability of a project before a guarantee can be issued. However, IGLF agreed not to duplicate the evaluation made by PFIs in accordance with our Subsidiary Loan Agreement with them. Hence, we encourage our PFIs to use the guarantee facilities of GFSME for their high risk loans for control purposes, otherwise, the guarantee mechanism may be abused by some PFIs. * IGLF has limited resources to pay guarantee claims should procedure is relaxed. The fund allocated to pay guarantee claims comes from the accumulated collection of guarantee fees which as of December 31, 1997 totaled only P55MM. Very truly yours, ORLANDO B. PINEDA Vice President 36 Annex B Defining SMIs in the Philippines Economic literature defines SMIs based on the number of workers because it allows cross- country and over-time comparisons. Definitions based on assets are far more common among financial institutions. Both definitions have advantages and disadvantages. If fixed assets and number of workers are used simultaneously to define a small firm, then a third definition of "smallness" appears based on the capital-labor ratio (value of assets per worker). This definition could be more meaningful to the extent that SMIs are assumed to have a higher labor intensity than other firms. Previous OED reviews on SMI lending have supported this type of defining small firms for targeting (World Bank, 1991). Asset-based versus Employment-based Definitions, Manufacturing, The Philippines, 1988 Employment-based definition Micro Small Medium Large Very (average number of workers) (1-9) (10-49) (50-99) (100- Large 199) (200+) Asset-based definition (assets in 1988 US dollars) Micro ($250,000) 61,596 7,549 744 346 164 Small ($250,000-500,000) 5,539 1,349 176 184 210 Medium ($500,000-2million) 12 55 67 136 232 Large ($2-3million) 0 4 13 17 55 Very Large (>3 million) 0 3 5 15 164 TOTAL 67,147 8,960 1,005 698 825 Note: US$1=21 pesos in 1988. Source: NSO (special tabulation based on the 1988 census of establishments). Definitions are important because they determine who the beneficiary of a particular program is. In the Philippines, an assets-based definition gives a very wide range of firm sizes. According to the Magna Cart for Small Enterprises, approved in 1990, SMEs must have "total assets, inclusive of those arising from loans but exclusive of the land on which the particular business entity's office, plant and equipment are situated" under 20 million pesos (approximately, US$1 million in 1990). How workable is this definition? The table below shows that almost 90 percent of all the manufacturing establishments had less than $250,000 of total assets including land (the exclusion of land-according to the law-would give a much higher number of "eligible firms" for whatever special programs). Had SMIs been defined as those firms with less than half a million dollar, a simple exercise shows that almost all manufacturing firms in the country would be classified as SMIs (preliminary data on the 1994 Manufacturing census gives similar results). Recently, the limit of 20 million pesos has been increased even further: firms with assets below 60 million pesos (excluding land) -roughly corresponds to US$2 million by early 1998- are legally considered SMEs. 37 Annex C Efficiency of Small Firms A more rigorous analysis of efficiency involves measuring total factor productivities of the two groups (IGLF and non-IGLF firms) as well as across sizes. This method takes into account the level of inputs used by firms (i.e., capital and labor). The results show estimates of efficiency based on total sales. Measured by total factor productivity, technical efficiency between IGLF and non-IGLF is not different among small firms, and non-beneficiaries perform better than IGLF medium and large firms. Regression analysis also confirms that the policy variable "IGLF" is statistically insignificant. That is, IGLF firms are not statistically different from non-IGLF firms in terms of efficiency. This is an important finding since it weakens the rationale for targeted support in the first place. Total Factor Productivity by IGLF and non-IGLF, by size, 1995 No.of Firms IGLF Non-IGLF Small (1-49 workers) 228 100.4 99.7 Medium (50-99 workers) 79 - 102.8 125.8 Large (100+ workers) 67 91.8 128.1 TOTAL 374 97.1 109.5 Note: Based on 1997 SMIIES, NSO-ITS (1997). 38 Bibliography Asian Development Bank (1984). Project Performance Audit Report - NGO Microcredit Project in the Philippines. Report No. PPA:PHI-22186. Manila: ADB Post Evaluation Office. - (1990). The Role ofSmall and Medium-Scale Manufacturing Industries: the Experience ofSelected Asian Countries. Manila: ADB. p.423. _ (1996). Project Performance Audit Report - Small and Medium Industry Project. Report No. PPA:PHI-22188. Manila: ADB Post Evaluation Office. (1997). Microenterprise Development: Not by Credit Alone. Manila: ADB. p.157. Anderson, Dennis and Farida Khambata. (198 1). Small Enterprises and Development Policy in the Philippines: A Case Study. Staff Working Papers No. 468. Washington D.C.: World Bank. Audretsch, David. (1991). "New-Firm Survival and the Technological Regime." The Review ofEconomics and Statistics, vol. 73, no. 3, pp. 441-450. Austria, Myrna S. and Medalla, Erlinda M. (1996). A Study on the Trade and Investment Policies of Developing Countries. The Case of the Philippines. Discussion Paper Series no. 96-03. Manila: Philippine Institute for Development Studies, March. Aw, Bee Yan; Geeta Batra. (1998). "Technological Capability and Firm Efficiency in Taiwan (China". The World Bank Economic Review. Vol. 12, No. 1, pp.59-70. Washington, D.C.: The World Bank. Bangko Sentral ng Pilipinas (1995). Philippine Financial System Fact Book. Manila: BSP. Berry, Albert .(1997). SMI Competitiveness: The Power ofNetworking and Subcontracting. Washington, D.C.: Inter-American Development Bank, Social Programs and Sustainable Development Department, Infrastructure and Financial Markets Division. Berry, A. and D. Mazumdar (1991) "Small-scale Industry in the Asian-Pacific Region", Asian-Pacific Economic Literature, Vol. 5, No. 2 (September 1991), pp. 35-67. Berry, A.; B. Levy. (1994). Indonesia's Small and Medium-Size Exporters and Their Support Systems. Policy Research Department, Finance and Private Sector Development, Working Paper no. 1402. Biggs, T.; Oppenheim, J, (1986). What drives the size distribution offirms in developing countries? Employment and Small Enterprise Division, Discussion Paper No.6, Employment and Enterprise Policy Analysis Project, Washington, D.C.: USAID. Biggs, T. ; Srivastava, P (1997). Structural Aspects of Manufacturing in Sub-Saharan Africa: Findings from a Seven Country Enterprise Survey. World Bank Discussion paper No 346 , Washington, D.C. Bruch, M. and U. Hiemenz (1983). Small and Medium-Scale Industries in the ASEAN Countries: Agents or Victims ofEconomic Development? Boulder and London: Westview Press. 39 Bibliography (cont.) Brugger, E. and S. Rajapatirana (1995). New Perspectives on Financing Small Business in Developing Countries. San Francisco: International Center for Economic Growth and FUNDES. Buckley, P.; J. Campos; H. Mirza; E. White (1997). International Technology Transfer by Small and Medium-Sized Enterprises: Country Studies. London and New York: MacMillan Press and St. Martin's Press Inc. Caprio, G., Jr.; D. Klingebiel (1996). Bank Insolvencies: Cross-country Experience, World Bank Policy Research Working Paper No. 1620, Washington, D.C.: World Bank. Chee, Peng Lim (1992). "Statistical Profile of SMI". in Kim S.J. and Suh J-W. (eds.), Cooperation in Small and Medium-Scale Industries in ASEAN, Kuala Lumpur: Asian and Pacific Development Centre. De Leon, Hector S. (1988). The Law on Transfer and Business Taxation, Manila: Rex Book Store. Demirgtiq-Kunt, A. and V. Maksimovic (1996). Institutions, Financial Markets and Firms' Choice ofDebt Maturity. Policy Research Paper Paper No. 1686. Washington, D.C.: The World Bank. Dessing, Monyle (1990): Support for Microenterprises: Lessons for sub Saharan Africa. World Bank Technical Paper No. 122. Washington, D.C.: The World Bank. Esguerra, Emmanuel F. (1995). Financial Assistance for Small and Medium Enterprises in the Philippines: An Overview, mimeo. Quezon City: University of the Philippines School of Economics. Ghate, Prabhu et al. (1992). Informal Finance: Some Findings from Asia.. Hong Kong: Oxford University Press for the Asian Development Bank. Guevarra, Milwida. (1991). "Fiscal Policy for the Development of Small Enterprises - An Assesment and Proposals for Policy Directions." In G. Llanto, J. Viray, E. Hyman (eds.) Macro Policies Affecting Small Enterprise Development in the Philippines. Manila: International Development research Center (Ottawa) and Appropriate Technology International (Washington, D.C.). Grosh, B. and G. Somolekae (1996). "Mighty Oaks from Little Acorns: Can Microenterprise Serve as the Seedbed of Industrialization?" World Development, Vol. 24, No. 2, pp. 1879-90. Hatakeyama, M.; M. Yamamori; and H. Nakamura. (1997). "Credit Guarantee Systems for Small and Medium Enterprises in Some Asian Countries." The Financier, February-May, pp. 62-73. Hill, Hal. (1981). Subcontracting and Technological Difusion in Philippine Manufacturing. Discussion Paper B 112. Quezon City: School of Economics, University of the Philippines. Krugman, Paul (1998). What Happened to Asia?. MIT: mimeo. Lamberte, Mario B. (1990). Impact of Special Credit and Guarantee Programmes for SMIs on Employment and Productivity. ILO-JAPAN Dole Multibilateral Project on Strategic Approaches Toward Employment Promotion. Manila, August. Lamberte, Mario. (1991). "Small Enterprise Promotion Policy and Legislative Agenda: Focus on Monetary Policy, Finance and Credit Programs." In G. Llanto, J. Viray, E. Hyman (eds.) Macro Policies Affecting Small Enterprise Development in the Philippines. Manila: International Development Research Centre (Ottawa) and Appropriate Technology International (Washington, D.C.). 40 Bibliography (cont.) Lamberte, Mario. (1995). "Small Enterprises' Access to Formal Financial Services: A Review and Assessment." Discussion Paper Series no. 95-23. Manila: Philippine Institute for Development Studies. Lamberte, M. and G. Llanto. (1995). "A Study of Financial Sector Policies: the Philippine Case." In S. Zahid (ed.), Financial Sector Development in Asia: Country Studies, pp. 235-301. Manila: Asian Development Bank. Lapar, Maria L. (1994). Credit Constraints among Rural Non-farm Enterprises: Evidence from the Philippines. Ph.D. Dissertation. Ohio State University. Mimeo. Levitsky, Jacob. (1997). "SMI Guarantee Schemes: A Summary." The Financier, February-May, pp. 5-11. Levitsky, J. and R. Prasad. (1989). Credit Guarantee Schemes for Small and Medium Enterprises in Some Asian Countries. World Bank Technical Paper no. 58, Industry and Finance Series, Washington D.C. Levy, B. et al. (1994). Technical and Marketing Support Systems for Successful Small and Medium-Size Enterprises in Four Countries. Policy Research Department, Finance and Private Sector Development, Working Paper no. 1400. Washington, D.C.: The World Bank. Liedholm, C. and D. Mead (1992). The Structure and Growth ofMicroenterprises in Southern and Eastern Africa: Evidence from Recent Surveys. Department of Economics, Michigan State University, East Lansing, MI: mimeo. Little, I.M.D. (1987). "Small Manufacturing Enterprises in Developing Countries." The World Bank Economic Review, 1: 2. Little, I; D. Mazumdar; J. Page, Jr. (1987) Small Manufacturing Enterprises: A Comparative Analysis of India and other Economies, Oxford University Press: a World Bank Research Publication. Jaramillo, F.; F. Schiantarelli; A. Weiss (1993). The Effect of Financial Liberalization on the Allocation of Credit: Panel Data Evidence from Ecuador ". World Bank Policy Research Working Papers No. 1092. Washington, D.C.: The World Bank. McKean, C. and J.Fox (1994). Export and Investment Promotion Services: Do they work? USAID Program and Operations Assessment Report No. 6, Washington, D.C.: Agency for International Development. Medalla, E. et al. (1996). Catching Up with Asia's Tigers (in two volumes). Makati: Philippine Institute of Development Studies. Meyanathan, S.D. (1994). Industrial Structures and the Development ofSmall and Medium Enterprise Linkages. Economic Development Institute, The World Bank. Mini, F. and E. Rodriguez (1998). Are SMIs more efficient? Revisiting Efficiency Indicators in a Philippine Manufacturing Sector. OED background paper, Washington, D.C.: The World Bank. National Economic and Development Authority. (1984). Economic Impact study of IGLF Assisted Projects - Phase I. Industry and Utilities Staff, NEDA, Pasig City, December 1984. 41 Bibliography (cont.) National Statistics Office (unpublished). 1993 Annual Survey ofManufacturing, Quezon City: Industrial and Trade Statistics-NSO. _ (unpublished). 1994 Census ofManufacturing, Quezon City: Industrial and Trade Statistics-NSO. - (1998). Special Release: October 1997 Labor Force Survey (NSO website: www.nso.gov.ph), Manila: NSO. (1995). 1995 Philippine Yearbook, Manila: NSO. (1992). 1992 Philippine Yearbook, Manila: NSO. (1995). 1992 Annual Survey ofManufacturing, Quezon City: ITS-NSO. National Statistics Office (1995). 1995 Korea Statistical Yearbook, Seoul: NSO. Parker, R. et al. (1997). Small Enterprises Adjusting To Liberalization In Five African Countries. World Bank Discussion Paper (Africa Technical Department Series) No 271. Washington, D.C.: The World Bank. Pineda, V.S. (1997). Effects of the Unform Five Percent Tariff on Manufacturing, PIDS. Draft Report (January). Makafi: PIDS. Prystay, Cris. (1996). "Small Loans and Hard Work Spark Entrepreneurial Spirit." Asian Business (December 1996), pp. 62-69. OECF (1995). Policy-Based Directed Credit Programs in the Philippines. OECF Discussion Papers. Research Institute of Development Assistance. Tokyo, Japan (January). Roberts, M. and J. Tybout (eds.). (1996). Industrial Evolution in Developing Countries: Micro Patterns of Turnover, Productivity, and Market Structure. Washington, D.C.: Oxford University Press. Sachs, J.; A. Wagner (1997). "Fundamental sources of long-run growth", American Economic Review, Papers and Proceedings (U.S.); Vol. 87, No. 2:184-88, May. Saldafia, Cesar et al. (1993). "Liberalization in Directed Credit Programs for SMIs", Journal ofPhilippine Development, Vol. 20, No. 36, First Semester. Makati: PIDS. Sengenberger, Werner; Gary Loveman, Michael J. Piore (eds.). (1990). The Re-emergence ofsmall enterprises. Geneva, Switzerland : International Institute for Labour Studies. Statistics Bureau (1996). 1996 Statistical Yearbook, Tokyo: Government of Japan. Stiglitz, Joseph (1998). The Role ofInternational Financial Institutions in the Current Global Economy Address to the Council on Foreign Relations. Chicago, IL: February 27, 1998 Stiglitz, J. and M. Uy. (1996). "Financial Markets, Public Policy, and the East Asian Miracle." The World Bank Research Observer, Vol. 11, No. 2, Washington, D.C.: World Bank. Tan, H. and G. Batra. (1995). Technical Efficiency of SMIs: Comparative Evidence from Developing Economies. Private Sector Development Occasional Paper no. 19. Washington, D.C.: World Bank. 42 Bibliography (cont.) Tecson, Gwendolyn et al. (1990). "The Role of Small and Medium-Scale Industries in the Industrial Development of the Philippines." In The Role of Small and Medium-Scale Manufacturing Industries in Industrial Development: The Experience ofSelected Asian Countries, Manila: Asian Development Bank. Webster, L. and Charap, J., (1993). The Emergence ofPrivate Sector Manufacturing in St. Petersburg: A Survey ofFirms, Washington, D.C. World Bank Technical Paper No. 228. Washington, D.C.: The World Bank. Webster, L., R. Riopelle, and A. Chidzero (1996). World Bank Lending for Small Enterprises, 1989-1993. World Bank Technical Paper No. 311. Washington, D.C.: The World Bank. Webster, L. (1997). World Bank Support for Micro, Small and Medium Business. Private Sector Development. June 18, 1997. Draft. Washington, D.C.: The World Bank. Wignaraja, G. (1997). Trade Liberalization in Sri Lanka: Exports, Technology, and Industrial Policy. New York: St. Martin's Press. World Bank (1975). Philippines: Appraisal of a Small and Medium Industries Development Project. (in two volumens). Report no. 6672. Washington, D.C.: The World Bank. (1977). Employment Creation and Small Scale Enterprise Development. Industrial Development and Finance Department, Report no. 1543. Washington DC: The World Bank. (1979). Staff Appraisal Report for the Philippines Second Small and Medium Industries Development Project, Report no. 2417. Washington DC: The World Bank. _ . (1982a). Staff Appraisal Report for the Philippines Third Small and Medium Industries Development Project, Report no. 3838. Washington DC: The World Bank. (1982b). Project Performance Audit Report on the Philippines - Small and Medium Industries Development Project. Operations Evaluation Department, Report no. 3969. Washington DC: The World Bank. (1989). StaffAppraisal Report for the Philippines Fourth Small and Medium Industries Development Project, Report no.: 7531. Washington DC: The World Bank. (1990). Project Performance Audit Report on the Philippines Six DFC and SMI Loans. Operations Evaluation Department, Report no. 8781. Washington DC: The World Bank. (1991). World Bank Support for Small and Medium Industry in Selected Countries. Operations Evaluation Department, Report no. 9530. Washington D.C.: The World Bank. (1993). The Philippines: An Opening for Sustained Growth (in three volumes). Report no. 11061, Washington D.C.: The World Bank. (1995). Project Completion Report on the Philippines Fourth Small and Medium Industries Development Project, Report no. 14055. Washington D.C.: The World Bank. (1996a). Performance Audit Report on the Philippines Cottage Enterprise Finance. Operations Evaluation Department, Report no. 15834. Washington D.C.: The World Bank. 43 Bibliography (cont.) __.(1996b). A Strategy to Fight Poverty: Philippines. Country Operations Division, East Asia and Pacific Region, Report no. 14933. Washington, D.C.: The World Bank. (1997a). World Bank Support for Small and Medium Industries in Sri Lanka: An Impact Evaluation. Operations Evaluation Department, Report no. 16790, Washington D.C.: The World Bank. (1997b). World Development Indicators, Washington, D.C.: The World Bank. (1997c). Philippines - Managing Global Integration, Vol. I and II, Report No. 17024-PH, Washington, D.C.: The World Bank. (1997d). Financial Policy Framework, Washington, D.C.: The World Bank. (1 998a). World Bank Support for Small Scale Enterprises in Ecuador: An Impact Evaluation. Operations Evaluation Department, Report no. 17953, Washington, D.C.: The World Bank. (1998b). Philippines - CAR: A Review of the World Bank Assistance. Operations Evaluation Department, Report no. 17417, Washington, D.C.: The World Bank. (1998c). Financial Sector Review, Operations Evaluation Department, Report no. 17454, Washington, D.C.: The World Bank. Yoingco, A. and M. Guevarra. (1990). The Incidence of Taxes in the Philippines. Manila: National Research Council of the Philippines. Mimeo.