Report No. 33030 OED Review of Bank Assistance For Financial Sector Reform July 22, 2005 Country Evaluation and Regional Relations Operations Evaluation Department Document of the World Bank Contents Foreword ............................................................................................................................. i Preface................................................................................................................................. v EXECUTIVE SUMMARY ............................................................................................ vi1 .. 1. Objectives. Inputs. and Organization o f the Review ................................................ 1 2. What constitutes good practice? ................................................................................. 3 . . perspective ............................................................................................... Histoncal 3 Review o f the literature............................................................................................ 3 Financial market structure ............................................................................ 4 Ownership o f banks ..................................................................................... 5 Incentive framework for banking................................................................. 6 Causes o f Crises ........................................................................................... 7 Bank guidelines and strategies ................................................................................. 7 Past OED recommendations and management response ......................................... 9 A framework for evaluation..................................................................................... 9 PART IASSESSING INPUTS: BANK ASSISTANCE .............................................. 11 3. Trends in lending and non-lending.......................................................................... 11 Overview ................................................................................................................ 11 Bank lending for financial sector reforms: trends ................................................. 12 Focus o f financial sector reforms ........................................................................... 14 Bank non-lending assistance for financial sector reforms: trends ........................ 16 4 . Regional patterns o f Bank assistance....................................................................... 17 E C A dominates ..................................................................................................... -17 Afiica and LCR: early reformers .......................................................................... 18 E M : mostly crisis-driven ..................................................................................... 20 MNA and S A R : conservative approach to reforms .............................................. 20 5 . Quality.atentry of Bank assistance.......................................................................... 22 Overview ................................................................................................................ 22 Quality o f non-lending services ............................................................................. 25 Consistency o f approach within countries ............................................................. 26 Coherence o f Bank approach to financial sector reforms across countries ........... 27 . 6 Bank assistance: outcomes o f loans and credits ..................................................... 30 Overview ................................................................................................................ 30 Financial sector loans versus components o f multi-sector loans ........................... 31 . . Country charactenstics ............................................................................... 32 Trends and sequence o f adjustment lending .......................................................... 33 Does the provision o f technical assistance help outcomes? ....................... 33 . 7 Bank support to countries experiencing a crisis ..................................................... 35 Overview.. .............................................................................................................. 35 Did the Bank anticipate the crisis? ......................................................................... 37 .. Bank response to cnsis ........................................................................................... 39 Objectives and design o f the loans......................................................................... 40 Relevance o f objectives......................................................................................... 42 Achievement o f objectives: below average .......................................................... 43 Collaboration with the IMF ................................................................................... 44 I s a centralized unit in the Bank specializing on crises effective and sustainable?46 Bank management, dealing with crisis .................................................................. 47 Recommendations ................................................................................................. -47 PART I1 ANALYZING OUTPUTS, OUTCOMES, AND IMPACT AT A COUNTRY LEVEL ................................................................................................... 48 8. Outputs at a country level: ownership and incentives........................................... 48 Overview ............................................................................................................... -48 The shift to private ownership ............................................................................... 48 Considerable progress ........................................................................................... 49 ...But privatization i s far from complete in many countries ................................. 50 Quality matters ....................................................................................................... 51 Financial restructuring prior to privatization: better outcomes., ........................... 52 Other forms o f bank restructuring.......................................................................... 54 Avoiding buildup o f N p L s ..................................................................................... 54 Altemative to bank privatization: closure ............................................................. 55 Privatization took longer than expected .............................................................. -55 ....and sometimes led to unanticipated problems .................................................. 55 Restructuring banks without privatization: seldom successfbl ............................. 56 Recommendations on restructuring and privatizing banks .................................... 56 Getting incentives in place: legal and regulatory reforms. strengthening .. supervision ....................................................................................................... 57 Overview ................................................................................................................ 57 Improvements in the regulatory regime: mixed picture on the details .................57 Regulatory framework for banking and capital markets in ECA transition countries ........................................................................................................... 58 Implementingthe laws and regulations and banking supervision: little information., ..................................................................................................... 59 Special topic: legal immunity for supervisors ..................................................... -60 Special topic: deposit insurance.. .......................................................................... 60 Recommendations on improving the incentive framework.. ................................. 61 9 . Outcomes at a country level: market structure, contestability, efficiency, and health........................................................................................................................... 62 Overview ................................................................................................................ 62 Changes in market structure: bank concentration................................................ -62 Changes in Contestability ..................................................................................... -63 Interest Rate Spread .............................................................................................. -66 Health o f the financial system................................................................................ 66 . 10 Impact at a Country Level: Financial sector depth and stability ............. 68 Overview ................................................................................................................ 68 Financial sector depth: positive findings ............................................................. 68 .....but financial systems are s t i l l very shallow in many Bank clients ..................70 Credit to the private sector ..................................................................................... 70 Financial sector depth: capital markets................................................................. 72 Stability: did Bank borrowing improve the stability o f banks? ............................ 73 . 11 Findings and Recommendations ............................................................................... 74 . . Findings.................................................................................................................. 74 Recommendations., ................................................................................................ 78 References...................................................................................................................................... 8 1 Boxes B o x 3.1 Identifying Bank assistance o f financial sector ............................................................... 13 B o x 3.2 The Financial Sector Assessment Program (FSAP) ........................................................ 16 B o x 4.1 Financial sector reform in the E C A Region: Bank Strategy, analysis, and lending ............................................................................................................................. 19 B o x 4.2 Bank assistance to China ................................................................................................. 20 B o x 4.3 Pakistan and Bangladesh government’s commitment explains in patterns o f Bank lending.............................................................................................. 21 B o x 5.1 Examples o f high relevance of objectives o f financial sector reforms ............................ 24 Box 5.2 Examples o f strong consistency between Bank products within countries ..................... 26 B o x 6.1 What a difference a (near) crisis (sometimes) makes ...................................................... 34 B o x 7.1 Objectives o f crisis lending: mostly ambitious reforms .................................................. 42 B o x 7.2 M i x e d outcomes .............................................................................................................. 44 B o x 7.3 Coordination between Bank and IMF in crisis: needs improvement ............................... 45 B o x 8.1 Problems comparing results in countries with and without Bank borrowing.................. 49 B o x 8.2 Data on bank ownership can be misleading .................................................................... 5 1 B o x 8.3 Privatization i s no guarantee o f good banks - the quality o f the buyer matters ..............52 B o x 8.4 N o financial restructuring prior to privatization: didn’t work well ................................. 52 B o x 8.5 Asset Management Companies (AMC) - empowerment ................................................ 53 B o x 8.6 Liquidation has been supported by the Bank in most Regions, but has proved difficult 55 B o x 8.7 Restructuringbanks with no commitment to change ownership: examples of Bank support .................................................................................................................... 56 B o x 8.8 Lack o f political support: Algeria .................................................................................... 60 B o x 9.1 O E D D E C on constructing a “counterfactual” ................................................................ 65 B o x 9.2 Financial reforms can affect measures o f banking health in both directions ................... 67 Figures Figure 2.1 Evaluation framework ................................................................................................... 10 Figure 3.1 Bank loans classified as finance ................................................................................... 12 Figure 3.2 Number o f Bank loans with financial sector reforms ................................................... 14 Figure 3.3 Focus o f reforms as percent o f all Bank projects with financial sector components ................................................................................................................... 14 Figure 3.4 Investment loans supporting financial sector reforms, by year ..................................... 15 Figure 3.5 Number o f ESW reports containing financial sector analysis. FY93-03 ...................... 15 Figure 4.1 Bank support for bank restructuring and privatizations. by country ............................. 22 Figure 6.1 Outcomes o f adjustment and TA loans. financial sector and other. FY93-03. by number and net commitments ....................................................................................... 30 Figure 6.2 Outcome ratings and sector classification, by number, FY93-03 ................................. 3 1 Figure 6.3 Outcome ratings by sector and by CPIA rating, by number, FY93-03 ......................... 31 Figure 6.4 Outcome ratings and country characteristics, FY93-03 ................................................ 32 Figure 6.5 Outcomes o f adjustment loans with and without technical assistance.......................... 35 Figure 7.1 Outcome o f adjustment lending. crisis lending with financial sector components versus non-crisis ........................................................................................................... 43 Figure 8.1 Changes in government ownership o f banks. with and without Bank lending for privatization .................................................................................................................. 49 Figure 9.1 Bank concentration in countries that borrowed from the Bank for financial reforms, 1993-2001 ....................................................................................................... 62 Figure 9.2 Changes in foreign ownership, with and without borrowing from the Bank ................64 Figure 9.3 Median interest rate spreads: in countries that borrowed from the Bank. 1992-2002 ..................................................................................................................... 66 Figure 10.1 Financial sector depth (M2/GDP) and liquidity preference (cash as a percent o f money supply), in countries that borrowed from the Bank for financial reforms. 1992-2002....................................................................................................... 69 Figure 0.2 Credit to the private sector as a percent o f GDP in countries that borrowed from the Bank for financial reforms. 1992-2002 .......................................................... 70 Figure 0.3 Market capitalization and value o f stock traded in countries that borrowed for capital market reforms, 1992-2002 ............................................................................... 73 Tables Table 3.1 Lending for financial sector reforms. FY93-03 ............................................................ 13 Table 3.2 Investment lendingwith financial sector components, by category, FY93-03 .............14 Table 4.1 Lendingcategorized as finance, percent o f total, FY93-03 .......................................... 17 Table 4.2 Lendingwith financial sector components, FY93-03 ................................................... 17 Table 4.3 Regional concentration o f reforms................................................................................ 19 Table 5.1 Quality o f ESW, Q A G assessment, FY97-03 ............................................................... 25 Table 5.2 Bank loans supporting deposit insurance, by FY.......................................................... 28 Table 6.1 Outcome ratings o f financial sector lending and components, FY93-03 ...................... 31 Table 6.2 Outcomes ratings and timing, sequence o f adjustment loans ....................................... 33 Table 7.1 Response to crisis: international rescue o f efforts and Bank response ......................... 39 Table 7.2 L i s t o f crisis Country Loans with Financial sector components ................................... 41 Table 7.3 Post-crisis adjustment operations with financial sector components ............................ 43 Table 8.1 Change in government ownership, by number o f adjustment loans and with and without TA ............................................................................................................. 50 Table 8.2 Change in bank ownership ............................................................................................ 50 Table 8.3 Capital adequacy and loan classification, changes between 1998 and 2003, with and without Bank lending..................................................................................... 58 Table 8.4 Indicators on strength o f financial regulations, transition countries ............................. 59 Table 9.1 Annual growth rate o f changes in bank concentration: with and without Bank lending for financial sector reforms .............................................................................. 63 Table 9.2 Changes in contestability .............................................................................................. 63 Table 9.3 Annual growth rate o f changes in interest rate spread: with and without Bank lending for financial sector reforms .............................................................................. 66 Table 9.4 Measures o f banking health in borrowing versus non-borrowing countries .................68 Table 10.1 Annual growth rates in financial sector depth and confidence in the banking system: with and without Bank lending for financial sector reforms ........................... 69 Table 10.2 Distribution in changes in measures o f financial sector depth, borrowing countries ........................................................................................................................ 70 Table 10.3 Annual growth rates for credit to the private sector: with and without Bank lending for financial sector reforms .............................................................................. 71 Table 10.4 Distribution in changes in access to credit, borrowing countries.................................. 71 Table 10.5 Distribution in changes in capital market measures...................................................... 73 Table 10.6 Number o f countries with and without systemic insolvency ........................................ 74 Annexes Annex 1 Data on trends and patterns in lending and non-lending ............................................... 88 Annex 2 Outcomes o f Bank loans ............................................................................................... 94 Annex 3 Outputs at a country level. with and without Bank lending .......................................... 97 Annex 4 L i s t o f counties for desk studies, value o f operations, by Region. and as percent o f lending ..................................................................................................... 100 Annex 5 Outcomes at a country level.......................................................................................... 101 Annex 6 Definitions and sources o f information, OEDiDEC model .......................................... 104 Attachments Attachment 1 Management Response ............................................................................................... 107 Attachment 2 CODE Chairman Summary ....................................................................................... 116 i Foreword After more than a decade o f borrowing from the Bank for financial sector reforms, most o f the 96 borrowing countries have witnessed improvements in their financial sectors, in terms o f ownership o f banks, efficiency measures, financial sector depth, and access to credit. These improvements can be associated with Bank borrowing: financial sector outcomes in countries that borrowed from the Bank are generally significantly better than in countries that did not borrow from the Bank for financial sector reforms. Nevertheless, in most o f the countries, although the trend has been in the right direction, the financial sectors remain relatively shallow, and private sector access to credit remains low. These findings suggest that although reforms supported by Bank lending over the past decade can be associated with improvements, they have not been sufficient to bring about the ultimate objective o f well-developed financial systems. Between FY93 and FY03, Bank assistance for financial sector reforms was supported by some US$56 billion dollars in lending, or 24 percent o f the Bank’s total commitments; most o f this lending was embedded in multi-sector loans. Over this period, lending for financial sector reforms declined, due mainly to the sharp drop in lines o f credit (LOC); apart from LOC, support for financial sector reforms has declined only slightly. This Operation Evaluation Department (OED) review o f World Bank assistance for financial sector reforms finds that the objectives o f Bank assistance generally followed good practice in terms o f reducing government ownership o f financial intermediaries, improving prudential regulation to be consistent with intemational norms, and strengthening banking supervision to adhere more closely to intemational principles. This review also finds, however, that consistency within a country - for example, advocating the privatization o f banks while simultaneously supporting the expansion o f government ownership o f banks - should be improved, as could the coherence o f the Bank’s approach to financial sector reforms across countries - for example, advocating rapid privatization in one transition country while recommending a slow, gradual approach to privatization in another transition country. Other areas where there has been wide variation in Bank support and what seems to be an ad hoc approach to the priority for Bank support include payments systems, deposit insurance schemes, and capital market development. The combination o f on-going debates within the Bank (e.g., whether and how to support deposit insurance schemes), absence o f “good policy” notes, and the decentralized nature o f Bank operations have all contributed to a situation in which the Bank speaks with many voices on important matters o f financial sector policy, a difference which cannot be fully explained by differences in country circumstances or willingness to reform. Outcomes o f loans under the financial sector board were significantly better than outcomes o f financial sector components o f multi-sector loans, which points to the need 11 for a stronger role in quality assurance by the sector board as well as the need to ensure strong support from financial sector officials in the client country. Bank assistance for financial sector reforms to countries experiencing crisis constitute some 50 percent o f the lending reviewed here. Crisis lending differs from non- crisis lending in several important respects: the former i s prepared under stressful conditions; speed i s important; sometimes without prior analysis or dialogue with the government about issues; and i s typically part o f large, publicly announced international rescue package. Because o f these exceptional factors, OED examined crisis lending separately, in 14 countries. OED found that the Bank was ill-prepared to respond quickly in the earlier crises in Mexico (1994), and Thailand, Korea, and Indonesia (1997); and better prepared in Argentina, Russia, and Turkey. Even in countries where it recognized signs o f vulnerability (Indonesia, Turkey), official Bank documents gave sanguine assessments o f risks. Although the stated objectives o f the loans were similar in scope and nature to financial sector reforms pursued in non-crisis situations, outcome ratings o f these closed operations (US$18 billion) are lower by some 15 percentage points than outcomes o f non-crisis lending. This i s a somewhat surprising finding given the high relevance o f the objectives and the fact that crises often induce or strengthen commitment o f governments to addressing the problems. I t is likely the result o f the need to state overly-ambitious objectives to justify the large loans that are necessary to fulfill the pre-announced assistance package. Collaboration with the IMF in countries that experienced a crisis was not always smooth, particularly in Indonesia, Mexico, Russia, and Thailand. Following the Asian experience, the Bank and the IMF reached agreements in principle to improve collaboration, although the boundary between the two institutions i s not always clear. I n addition, regional development banks often play a role in the rescue, which needs to be coordinated as well. Collaboration among the I F I s in countries experiencing a crisis remains a challenge. Finally, recommendations from an earlier, high-level internal Bank review suggested that the Bank prepare guidelines for crisis situations on triggers for actions and clear lines o f responsibility; these recommendations have not been implemented and remain valid today. Recommendations 0 The Bank’s financial sector anchor should provide more guidance for Bank staff and client countries, in areas such as restructuring o f banks (if, when, and how); asset management companies (if, when, how); privatization o f banks; promotion o f capital markets (if, when, and how, in conjunction with IFC on this); and for topics related to the strengthening the legal, regulatory, and supervisory environment, a particular focus o n implementation. I naddition, the financial sector network should become more pro-active in quality control o f financial sector components in multi-sector loans. ... 111 0 The Bank should develop monitorable indicators to assess progress o n objectives in the area o f prudential regulations and supervision for financial intermediaries. 0 On support for countries prior to and following crisis, the Bank should develop a rating system, in partnership with other relevant institutions, for vulnerability to crisis, making use o f readily available information that can be used to engage countries in crisis prevention measures and issues in crisis response. The Bank should also do a better j o b than in the past o f presenting assessments more candidly in documents. Finally, the Bank should make internal arrangements to respond better to crisis by developing guidelines for dealing with crisis, which should include the possibility, if circumstances warrant, o f lending liquidity support to countries experiencing a crisis without stipulating ambitious reforms (that may not be realized) as justification for the loan. Aj ay Chhibber Acting Director-General Operations Evaluation iv V Preface This evaluation presents an independent assessment o f the Bank’s support for financial sector reforms over the period FY93-03. I t is the second part o f a two-part evaluation; the first part o f the assessment covered lines o f credit. This volume focuses on Bank lending for financial sector reforms, including both lending categorized under the financial sector and financial sector components o f multi- sector loans. The assessment examines trends in lending, the quality at entry o f Bank assistance; and the outcomes o f individual loans and components addressing financial sector reforms. I t also assesses the extent to which the objectives o f Bank assistance were achieved, including reducing government ownership o f financial intermediaries, decreased market concentration, increased competition and efficiency, healthier and more stable financial intermediaries, and deeper, more developed financial systems. Finally, the assessment examines Bank support for financial sector reforms in countries under crisis. The basis for the evaluation consists o f a data base developed by OED on all Bank’s lending for financial reforms, background papers on selected topics, and case country studies based on desk reviews. The report has been circulated to Bank management involved in financial sector support, the Financial Sector Board o f the Bank, and the Financial Sector Operations and Policy Department (OPD). This evaluation was discussed at the Committee o f Development Effectiveness (CODE) meeting held on March 30,2005, and the Chairman’s Summary i s attached as Attachment 2. The Management Response i s attached as Attachment 1 to this report. This evaluation was written by Laurie Effron, with the assistance o f Robert Cull (DEC), Nicolas Dujovne (consultant), Ilka Funke (OEDCR), Jeremy Ghez, Manuel Hinds (consultants), Gulmira Karaguisheva (OEDCR), Miguel Kiguel, Fred Levy, Millard Long, Dan Mozes (consultants), Brandie Sasser, and Barbara Yale (OEDCR). Roziah Baba (OEDCR) provided administrative assistance. The evaluation has also benefited from the comments of three peer reviewers: Patrick Honohan, Roberto Rocha, and Stijn Claessens (OPD), and an External Advisory Panel o f outside reviewers: Andrew Sheng, Charles Goodhart, and Narayan Vaghul. vi vii EXECUTIVE SUMMARY 1. This Operations Evaluation Department (OED) review o f World Bank assistance for financial sector reforms finds that the objectives o f Bank assistance generally followed good practice in the areas o f reducing government ownership o f financial intermediaries, improving prudential regulations to be consistent with international norms, and strengthening banking supervision to adhere more closely to international principles. This review also finds, however, that consistency within a country and most especially coherence o f the Bank’s approach to financial sector reforms across countries should be improved, particularly with respect to the priority for Bank support for payments systems, deposit insurance schemes, and capital market development. The combination o f on-going debates within the Bank (e.g., whether and how to support deposit insurance schemes), absence o f “good policy” notes, and the decentralized nature o f Bank operations have all contributed to a situation in which the Bank speaks with many voices on important matters o f financial sector policy, a difference which cannot be fully explained by differences in country circumstances or willingness to reform. 2. After well over a decade o f borrowing from the Bank for financial sector reforms, most o f the 96 borrowing countries have witnessed improvements in their financial sectors. These improvements can be associated with Bank borrowing (see paragraph 7 below). Nevertheless, in most o f the countries, the financial sectors deepened only modestly and remain relatively shallow, and private sector access to credit remains low. 3. Between FY93 and FY03, Bank assistance for financial sector reforms was supported by some US$56 billion dollars in lending, or 24 percent o f the Bank’s total commitments; these figures take into account lending that i s categorized by the Bank under the financial sector board as well as components o f multi-sector lending categorized under other boards (mostly Economic Policy). The support aimed at bank restructuring and privatization, strengthening prudential regulations and banking supervision, improving the regulatory and institutional framework for capital markets and insurance, and capacity building in specific financial intermediaries. 4. Most o f the lending for financial sector reforms was embedded in components o f multi-sector loans; out o f 385 loans containing support for these reforms, only 36 percent (137 loans) were in the financial sector, and the remainder, were components o f adjustment and technical assistance loans and lines o f credit in other sectors. Over the period FY93-03, lending for financial sector reforms has declined, due mainly to the sharp drop in lines o f credit (LOC). Apart from LOC, support for financial sector reforms through adjustment and technical assistance lending has declined only slightly, with a more noticeable drop in (formal) non-lending assistance. 5. Excluding LOC, which are analyzed in a separate OED review, outcomes o f all lending for financial sector reforms (adjustment plus technical assistance (TA) loans) averages 75 percent satisfactory, slightly below the 79 percent average for all (adjustment and TA) lending excluding financial sector. However, the outcomes o f loans under the financial sector board were significantly better than outcomes o f financial sector ... Vlll components o f multi-sector loans, which points to the need for a stronger role in quality assurance o f financial sector components by the sector board as well as the need to ensure that the financial sector reforms embedded in multi-sector loans have strong support from financial sector officials in the client country. 6. In addition, adjustment loans and components o f adjustment loans have better outcomes in countries with modest institutional capacity when they are accompanied by TA loans than when TA loans are absent. I nhigher capacity countries, however, adjustment loans have worse outcomes when TA loans accompany them than when they don’t. One explanation for this is that a TA loan in a higher capacity country may be a signal that the government i s not fully committed to carrying out the reforms. 7. At a country-level, OED examined whether Bank borrowing could be associated with changes in outputs, outcomes, and impact. Output was defined as a decrease in government ownership of banks and stronger regulatory and supervisory frameworks for banking. Outcomes were defined as: (i) market structure measured by concentration rates; (ii) contestability measured by ease o f entry and absence o f restrictions on activities - freedom to compete - in banking; ( iii ) efficiency measured by interest rate spreads; and (iv) health o f the banking system measured by capital adequacy and non-performing loans. Finally, impacts were defined as: (i) financial sector depth in banking, measured by the money supply as a proportion o f GDP and preference for cash as an indicator o f the lack o f confidence in the banking system; (ii) size o f the capital markets, measured by capitalization and turnover as a proportion o f GDP; ( iii) credit to the private sector, and (iv) financial sector stability (absence o f systemic banking insolvency). Because financial sector developments are so closely linked to other country characteristics, for much o f this analysis, an econometric model was used to control for country conditions, including growth rates, inflation rates, fiscal deficit, and institutional capacity. OED also tested whether the results were different for countries that borrowed from those that did not borrow for financial sector reforms over the period under review. Because countries that borrow from the Bank may be self-selecting, and more likely to b e reform-oriented than those that don’t borrow, the results o f the econometric analysis show association o f Bank borrowing with outcomes, rather than causality, although further econometric tests (including treatment effects regressions that explicitly account for self-selection, and propensity score matching techniques) provided evidence that reinforce the main findings. 8. Output at the country-level. Between the early 1990s and 2003, Government ownership decreased dramatically in countries that borrowed for bank privatization, and by more than in Bank client countries that were also privatizing their banking system without borrowing from the Bank. Official data mask the full picture o f government control o f financial intermediaries, however, because governments often retain significant minority ownership in banks that are considered private and many countries have state owned non-bank financial intermediaries that do substantial lending. Thus, reducing governments’ role in financial intermediation remains a challenge. Although the Bank often and appropriately supported financial restructuring prior to privatization o f banks, Bank support has not consistently focused on the quality o f the new owners, and this has contributed to poor results. In addition, the Bank has supported financial restructuring o f ix banks in the absence o f government commitment to change their ownership, and this has led to poor results (re-appearance o f poor loan portfolios and insolvency). 9. Improvement in laws and regulations governing the financial sector was uneven in borrowing countries. Between 1998 (the earliest year for which systematic information was available) and 2003, capital requirements remained about the same, while rules o n loan classification were stricter; the opposite was true for non-borrowing countries (stricter capital requirements, less stringent loan classification). Among transition countries, the regulatory frameworks for banks and capital markets show more improvement since 1998 in borrowing than in non-borrowing countries. O n the critical aspect o f implementation o f the laws and regulations, there was little information, and thus it was not possible to assess the extent to which laws and regulations were in fact observed. Strengthening banking supervision remains a priority. A number o f countries that borrowed from the Bank to strengthenbanking supervision are s t i l l far from complying with Base1 core principles. 10. Outcome at the country-level. Concentration levels decreased significantly since the early 1990s for all countries, although more so in non-borrowers, while contestability since 1998 (earliest year for which data are available) increased in borrowing countries as measured by lower restrictions on banking activities and decreased in non-borrowing countries. Interest rate margins (since the early 1990s) narrowed significantly in borrowing countries and did not change in non-borrowing countries. Finally, data on health are not sufficient for a comparative analysis (of “with” and “without” borrowing), but they do point to an improvement (non-performing loans decreased; capital adequacy increased) in the borrowing countries. Thus, overall, Bank borrowing i s associated with good outcomes and, where information permits comparisons, to mostly better outcomes than in non-borrowing countries. 11. Impact at the country-level. The positive results on outcomes discussed in the previous paragraph do not translate into equally positive findings on impact over the last decade, although developments have been in the right direction. Financial sectors became deeper in countries that borrowed for financial sector reforms over the period, although not significantly more than in non-borrowing countries. In any case, they remain, on average, relatively shallow - M2/GDP, for example, was below 40 percent in the Bank borrowers in 2002 (it i s about 80 percent in the Organization for Economic Co- operation and Development (OECD) countries). Liquidity preference (cash as a proportion o f the money supply - considered the inverse o f public confidence in the banking system) decreased significantly (at roughly the same rate as in non-borrowing countries), which could be the result o f the reforms aimed at downsizing, restructuring, and privatizing banks and pro-active efforts by governments to regulate and supervise them. 12. Credit to the private sector (as a percent o f GDP) grew at an annual rate o f 0.4 percent per year in the countries that borrowed from the Bank for financial sector reforms, less than it did in countries that did not borrow from the Bank (where i t grew by about 1.7 percent per year). One explanation o f the modest growth in credit i s that the process o f strengthening both governance and prudential regulations could lead to greater X prudence in lending; thus, although the growth i s slower than in non-borrowing countries, i t may be more prudent lending. But o n average, credit to the private sector remains very low, below 30 percent o f GDP in the 62 borrowing countries for which information was available (and in 17 countries, i t was below 10 percent; in OECD countries, as a point o f comparison, i t was over 110 percent). Finally, OED found no pattern in terms o f improved stability o f the financial system in countries that borrowed from the Bank relative to those that didn’t. 13. The findings on financial sector depth and credit to the private sector suggest that the reforms supported by Bank lending over the past decade are closely associated with improvements in the financial systems, but they have not been sufficient to bring about well-developed financial systems. 14. Bank assistance for financial sector reforms to countries in crisis constitute some 50 percent o f the lending reviewed here. The circumstances surrounding crisis lending are different from non-crisis lending: the former i s prepared under stressful conditions; speed i s important; sometimes without prior analysis o f or dialogue with the government about issues; as part o f large, publicly announced international rescue packages. Because o f these exceptional factors, OED examined crisis lending separately, in 14 countries. 15. OED found that the Bank was ill-prepared in Mexico in 1994, and in Thailand, Korea, and Indonesia in 1997 to respond quickly; and better prepared in Argentina, Russia, and Turkey. Even in countries where it recognized signs o f vulnerability (Indonesia, Turkey), official Bank documents gave sanguine assessments o f risks. Although the stated objectives o f the loans were similar in scope and nature to financial sector reforms pursued in non-crisis countries, outcome ratings o f the 3 1 closed operations (USS18 billion) are lower by some 15 percentage points than outcomes o f non-crisis lending. This i s a somewhat surprising finding given the high relevance o f the objectives and the fact that crises often induce or strengthen commitment o f governments to addressing the problems. I t is likely the result o f the need to state overly-ambitious objectives to justify the large loans that are necessary to fulfill the pre-announced assistance package (Chapter 9). 16. Collaboration with the International Monetary Fund (IMF) in countries that experienced a crisis was not always smooth, particularly in Indonesia, Mexico, Russia, and Thailand. Following the Asian experience, the Bank and the IMF reached agreements in principle to improve collaboration, although the boundary between the two institutions i s not always clear. In addition, regional development banks often play a role in the rescue, which needs to be coordinated as well. Collaboration among the International Financial Institutions (IFIs) in countries experiencing a crisis remains a challenge. Finally, OED found that prior recommendations for the Bank to prepare guidelines for crisis situations on triggers for actions and clear lines o f responsibility have not been implemented and remain valid today. xi Recommendations 0 The Bank’s financial sector anchor should provide much clearer guidance for Bank staff and client countries and the financial sector network should become more pro-active in quality control o f financial sector components in multi-sector loans. This involves producing good practice notes on a range of topics, in areas where there i s a cohesive internal Bank view on reforms. In areas where debate continues, i t needs to provide a review o f issues and options for Bank support. Subjects where guidance i s needed include restructuring o f banks (if, when, and how); asset management companies (if, when, how); privatization o f banks; promotion o f capital markets (if, when, and how, in conjunction with IFC on this); deposit insurance (what to do if government seeks support; issues to consider) and for topics related to the strengthening the legal, regulatory, and supervisory environment, a particular focus on implementation. 0 The Bank needs to focus assistance on: (i) the process o f preparing banks for privatization (financial restructuring) and ensuring that banks are sold to fit and proper owners; (ii) implementation o f laws and regulations governing the financial sector; (iii) strengthening supervision o f financial intermediaries; and (iv) increasing access to credit by improving collateral laws, creditor rights, providing technical assistance and training. 0 The Bank should develop monitorable indicators to assess progress on objectives in the area o f prudential regulations and supervision for financial intermediaries. 0 O n support for countries prior to and following crisis: 9 The Bank should develop a rating system, in partnership with other relevant institutions, for vulnerability to crisis, making use o f readily available information that can be used to engage countries in crisis prevention measures and issues in crisis response, The Bank should also do a better j o b than in the past o f presenting assessments more candidly in documents. 9 The Bank should make internal arrangements to respond better to crisis by developing guidelines for dealing with crisis, which should include the possibility, if circumstances warrant, o f lending liquidity support to countries experiencing a crisis without stipulating ambitious reforms (that may not be realized) as justification for the loan. 9 Coordination with the IMF and other I F I s in crisis assistance needs to be improved, and at the outset o f a crisis, the I F I s should reach quick agreement on division o f responsibilities. 1 1. Objectives, Inputs, and Organization o f the Review 1.1 Background. The importance o f the financial sector is widely recognized for the role it can play in the development o f a country. Although i t s impact o n poverty alleviation is not as obvious as investments in, say, rural infrastructure, financial sector development i s essential for mobilizing resources, channeling them to productive investments, managing risks, and thereby contributing to economic growth. Other key services provided by a well-hnctioning financial sector include efficient payment and settlement systems, which lower transaction costs, and effective monetary policy. In addition, the forces o f globalization and changes in technology have affected the roles as well as the vulnerabilities o f financial sectors as never before; the last decade provides many examples o f the devastating impact that financial crises can have on countries in terms o f lower growth and increased poverty. A well diversified, robust, and stable financial sector can better withstand the forces that induce crises-although it may not be able to prevent them entirely-which negatively affect economies for years afterward. 1.2 For more than fifty years, the Bank has supported financial sectors in client countries, initially through helping to set up and strengthen development finance companies and then, starting in the late 1980s, through supporting sector-wide reforms, particularly in banking, but also in capital market, pension,' and insurance reforms. 1.3 The present review examines Bank assistance to financial sector reform over the past decade. Between FY93 and FY03, the Bank made financial sector loans (excluding pension reforms) totaling some US$24.8 billion, representing 11 percent o f total Bank commitments. If all loans and credits with financial sector reform components are included, a total o f US$56 billion involved some financial sector reforms, representing 24 percent o f Bank lending over this period. 1.4 Objectives of the review. This review answers a series o f questions: 0 Are Bank policies o n financial sector reforms well-defined and do they follow good practice as defined by the literature? 0 What has Bank lending been in the aggregate over time and by Region in Bank assistance for financial sector reforms, including such support in multi-sector loans? 0 Has the Bank followed good practice, as defined by the literature and Bank policies? 0 Do outcomes o f Bank loans show any patterns over time, by type o f instrument, by sector classification, by country characteristics? 0 Were the objectives o f Bank lending met at a country-level, using both quantitative and qualitative indicators o f financial sector performance; and i s ' Pension reform was, until 2002, under the Financial Sector Board. 2 there any difference between countries that borrowed from the Bank for financial sector reforms and those that did not in terms o f these performance indicators? What i s the assessment o f Bank assistance to countries that experienced crises? The period covered by this review, FY93-03, i s characterized by severe financial sector crises-in Asia, Latin America, and Europe-that prompted the international community, including the Bank, to mobilize large amounts o f assistance. What was the Bank’s role in these countries before and after the crises, what have the outcomes been, and what lessons can be drawn for the future from this experience? 1.5 Caveats on scope ofreview. This review i s one o f a series o f recently completed and on-going OED reviews that cover financial sector issues. OED has just completed a review o f lines o f credit (LOC),’ which frequently had financial sector objectives. Most o f the analysis in this current report therefore does not include analysis o f LOC. In addition, OED i s currently reviewing the Financial Sector Assessment Program (FSAP), a major joint initiative o f the Bank and the International Monetary Fund (IMF), which is the most significant form o f Bank non-lending assistance since 1999 (in terms o f resources and use o f Bank staff) in the financial sector. This present review therefore does not cover the FSAP. Finally, Bank support for pension reform i s the subject o f a separate on-going OED review and is not discussed here. 1.6 Although the importance o f legal and judicial institutions to financial sector development has been recognized in the literature and in the Bank, this review touches on these issues only tangentially in order to limit the assessment to a manageable scope. In addition, Bank support to the financial sector has included corporate restructurings and out-of-court arrangements, which are also not covered here in any detail. 1.7 A final caveat i s that the Bank is only one source o f support for financial sector reforms and not always the most important one; thus, distinguishing the Bank’s contribution in the context o f j o i n t efforts by other donors is a challenge. Given OED’s mandate to evaluate Bank activities, i t was beyond the scope o f this review to examine the extent o f cooperation within the Bank Group or with other donors, although cooperation or lack o f it can be a critical factor in the success o f the Bank’s efforts. Nevertheless, given the scope o f the Bank’s lending over the past decade in support o f financial sector reforms, i t i s important to examine results o f these efforts. 1.8 Inputs. Nine background papers were commissioned for this review; they are listed in the References and will be available o n OED’s website. These papers, combined with desk reviews o f Bank assistance (lending and non-lending) to the financial sectors in 37 countries, form the major inputs for examining patterns o f Bank assistance. Data on outcomes at a sector-level come from standard sources such as International Financial Statistics, central banks, and Fund and Bank sector reports. ’OED (2005). 3 1.9 Organization o f the review. Chapter 2 summarizes the literature on factors associated with financial sector development, reviews Bank guidelines and strategy for assistance to the financial sector, draws conclusions on benchmarks for assessing the quality o f the Bank’s interventions, and sets out a framework for the evaluation. The remainder o f the review i s divided into two parts: Part I analyzes Bank assistance as an input to financial sector reforms: Chapters 3 and 4 review trends and Regional experience, respectively, o f financial sector assistance. Quality at entry o f Bank assistance is the subject o f Chapter 5, and Chapter 6 analyzes Bank assistance in terms o f outcome ratings. Part I concludes with Chapter 7, which examines Bank assistance for financial sector reforms in countries experiencing crisis. Part I1focuses on results at a country-level: Chapters 8 examines changes in bank ownership and the prudential and regulatory regime o f financial sectors; Chapter 9 looks at outcomes o f Bank assistance for financial reforms in terms o f market structure, measures o f contestability and interest rates as indicators o f competition, and health o f the banking system; and Chapter 10 examines the impact at a country-level in terms o f financial sector depth, liquidity preference, access to credit, and stability. Chapter 11 draws conclusions and presents recommendations for the future. 2. What constitutes good practice? Historical perspective 2.1 Ideas about the basic ingredients o f a sound financial system have evolved over time. In the early 1900s free banking was popular, in which banks could be set up and operate without government oversights3Until the late 1980s, most OECD countries had substantial government ownership o f banks; more than a few still do todaym4 2.2 Capital requirements related to risk assets were introduced on an international scale only in 1988 and have recently been modified. Deposit insurance i s a relatively new instrument (in the United States, introduced in 1934 following widespread bank failure), and is the subject o f debate and research onits impact o n financial sector stability (paragraph 2.1 1); and the emphasis o n regulatory requirements and supervision o f financial institutions may be shifting toward a greater reliance on the role o f “market forces” (paragraph 2.12). Review o f the literature’ 2.3 There are, however, certain tenets on which theoretical and empirical literature agree. One i s that macro-stability is important forjkancial sector development. Both In the United States, for example, privately owned central banks were established twice in the late 18” and early 19” century, but their existence was controversial and challenged at the Supreme Court, and their charters allowed t o lapse; the country had n o central bank for most o f the 19” century. In 2003 in Germany, for example, 42 percent o f banking sector assets were i n state controlled banks; in Greece and Portugal, it was 23 percent; i n Switzerland, 14 percent (Clarke, Cull, and Shirley, 2004). Unless otherwise indicated, the discussion in t h i s section i s based o n a background paper for t h i s review, Cull (2004), which will b e available o n OED’s website. 4 theory and empirical evidence support the view that financial depth tends to increase with stability. 2.4 A second tenet for which there i s empirical support i s that government- administeredfinancial systems involving fixed interest rates and directed credit lead to financial repression and ineflcient allocation o f credit, and that less direct government control over the financial system will over time result in deeper, more stable, and more efficient systems (Caprio, Honohan, Stiglitz, 2001; World Bank, 1989). 2.5 A third generally accepted view is on the importance o f a well-fbnctioning and properly supervisedpayments system that can effect efficient, fair, and safe payments in domestic and cross-border markets (Bossone and Cirasino, 2001). Financial market structure 2.6 Research on the best mix offinancial institutions, in terms of bank-based systems versus market-based (capital markets) has a striking lack o f results. The debate on the issue started in the early 1960s (Gerschenkron, 1962) and continues to this day.6 Although theoretical arguments have been advanced for one type over the other, recent empirical research suggests that neither bank-based nor market-based financial systems are associated more with higher growth rates for firms, industries, or the economy over the other (Levine (2002) examines GDP growth rates; Beck and Levine (2002) look at industry growth rates; and Demirguc-Kunt and Maksimovic (2002) focus on firms’ sales growth). Rather, it is the overall level o f financial sector development, regardless o f which structure dominates, that matters for growth. Thus, whether to promote the establishment or expansion o f capital markets in a country will depend on the circumstances, including the ability o f the country to reduce informational asymmetries. 2.7 For banking systems, the findings from research are ambiguous o n whether market concentration or more competition leads to more efficiency and/or more stability. Theory suggests that more concentrated market share could lead to greater economies o f scale, efficiency, and access to credit (Demsetz, 1973; Peltzman, 1977); or that i t could lead to market power and greater inefficiency. Empirical research on cross-country data in developing countries, which i s not very extensive, suggests that concentration has a negative effect on access to finance, although the results don’t hold for countries with well-developed institutions (Demirguc-Kunt, Laeven, and Levine, 2003 , and Beck, Demirguc-Kunt, and Maksimovic, 2002). In addition, research indicates that more concentrated banking systems are less prone to banking crisis, which i s likely due to diversification in banks’ lending and products rather than reduced competition. Finally, recent empirical research suggests that measures o f concentration are less relevant to assessing competitive forces in the banking industry than measures o f contestability, including restrictions on banking activity, ease o f entry, and foreign bank ownership (Claessens and Laeven, 2004). Thus there is no compelling argument for reducing banking concentration; the appropriate degree of concentration depends on institutional 6 A discussion of that literature can be found in Allen and Gale (2000), Boot and Thakor (1997), Goldsmith (1969), Levine (2002), Rajan and Zingales (2001), and Stultz (2001). 5 capacity and other objectives. By contrast, there is some evidence that reducing entry requirements and restrictions on activity, and allowing foreign ownership is positively associated with competition. Ownership of banks 2.8 Although theoretical arguments exist for state control o f banks (see, for example, Calomiris and Himmelberg, 1994, Greenwald and Stiglitz, 1986; Stiglitz, 1994, and World Bank, 2001), empirical researchJinds that state ownership is associated with poorerflnancial sector performance than privately dominated systems: less financial sector development; slower growth, lower productivity, and greater tendency to banking crisis (Barth, Caprio, and Levine, 2001a, b, and LaPorta, Lopez-de-Silanes, and Shliefer, 2002, and Beck, Demirguc-Kunt, and Levine, 2003). At the same time, however, privatization o f state banks has not always been successful. Studies show post- privatization efficiency gains in Argentina’s provincial banks (Berger et al, 2003); Nigeria’s banks (Beck, Cull, and Jerome, 2003), and in a sample o f banks from eleven transition countries in Central and Eastern Europe (Bonin et al, 2003) but also include cautionary instances o f unsuccessful privatization, most notably in Chile in the 1970s (Brock, 2000) and Mexico in the 1980s (Haber and Kantor, 2003). Chile privatized banks without first cleaning their balance sheets and sold them to their previous owners, while in Mexico, the government sold banks only to domestic buyers and prohibited n foreign ownership or any large foreign banks from ~ o m p e t i n g .I~ both countries, the banking system experienced subsequent crisis: in both countries, the banks were either closed or re-nationalized and privatized a second time with more success. 2.9 The research on foreign banks in developing countries shows mainly positive impacts: greater efficiency and better quality portfolios, and lower probability o f systemic banking crisis (see Cull, 2004, for a fuller discussion o f the literature covering the impact o f foreign banks on these aspects). 2.10 On the question o f access to credit, the limited empirical studies suggest that access is no better in banking systems that are predominantly state-owned than in privately-dominated banking systems. In Argentina and Chile, for example, public banks lend less to small businesses than other banks (Clarke, Crivelli, and Cull, 2003, Clarke, Cull, and Peria, 2001, and Clarke et al, forthcoming). The literature on foreign banks in developing countries suggests a complicated relationship between the foreign banks and access to credit. Work by Clarke et a1 (forthcoming) finds that large foreign banks lend more to small f i r m s than large domestic banks, although on average foreign banks lend less than domestic banks to small firms. In addition, studies have found that foreign banks may concentrate o n certain market segments, so that increasing foreign ownership might result in less access in certain sectors (Barajas, Steiner, and Salazar, 2000, and Cull, 2004). Overall, the research argues for private ownership o f banks, and allowing entry o f foreign banks; but also that the quality o f the purchaser matters for outcomes. I t has been argued that the Mexico experience was not a failed privatization, but a privatization that did not go far enough. The point i s the same: that privatization per se does not lead to good outcomes. 6 Incentive framework for banking 2.1 1 This rubric includes regulation and supervision, safety nets such as deposit insurance, legal framework for creditors’ rights, and market forces for monitoring. There are theoretical arguments why both regulation and supervision o f banking are important. However, the empirical research for developing countries on the effectiveness o f regulatory requirements such as minimal capital, loan classification, and liquidity ratios found no association between such requirements and better banking sector performance. The same conclusion applies to banking supervision: neither supervisory powers nor independence is statistically associated with banking development, while private monitoring i s strongly associated with more banking development and healthier banks (Barth, Caprio, and Levine, 2001b). Private monitoring includes the requirement to be externally audited; rating o f banks by international rating agencies; and the quality o f disclosed accounts and other disclosure requirements. I naddition, restrictions on bank activities and entry restrictions for domestic and foreign banks are associated with worse performance o f banking systems. Furthermore, theory and empirical research concur that private monitoring - incentives such as no explicit deposit insurance and requirements for accounting and auditing, rating o f financial institutions by private rating agencies-is strongly linked to banking sector development. Thesefindings suggest that banking regulations and banking supervision need to be carefully tailored to the conditions o f the countries and that, reforms should focus as a priority on creating the incentives and tools (accounting, auditing, disclosure requirements, rating agencies) for market participants to monitor financial institutions. 2.12 Theory argues both for and against deposit insurance. I t can make depositor runs less likely and therefore serve as a stabilizing influence on banking systems. In addition, if governments are already providing an implicit guarantee o n all deposits, establishing an explicit system can both protect some depositors while limiting the cost for government by setting caps on the insurance. But i t can also introduce moral hazard: depositors have less incentive to monitor banks because they know they are covered in the event o f a crisis; and banks have an incentive to take higher risks with depositors’ money (Diamond and Dybvig, 1983). Recent evidence suggests that in weak institutional environments, explicit deposit insurance i s associated with a higher incidence o f banking crisis, higher fiscal cost o f resolving a crisis, and slower recovery. There is, however, also evidence that uninsured depositors do monitor the riskiness o f banks (Demirguc- Kunt and Detragiache, 2002, Honohan and Klingebiel, 2003, Martinez, Soledad, and Schmukler, 2001), which would argue for a system where at least some depositors are not insured. The policy implications are that deposit insurance should be designed to exclude coverage o f some deposits, and f i equity considerations matter, these should be the larger deposits that belong presumably to wealthier clients, and inter-bank deposits; in this way, it w i l l be the generally wealthier clients who are uncovered and who have an incentive to monitor the banks. 2.13 Although there i s some debate over which types o f legal systems are more conducive to financial sector development, most empirical work points to the importance o f creditor rights and, more broadly, property rights o f financiers external to the enterprises, as well as enforcement o f contracts, for financial sector development. Legal 7 systems o f different origins tend to offer varying levels o f protection to the different categories o f stakeholders and as a result may influence the sort o f financial development that occurs (debt versus equity markets; external financing versus self-financing), but the literature is unambiguous in finding that protection o f the rights o f debt and equity holders is associated with more developed financial systems (Levine, 1998, and Levine, 1999). Causes of Crises 2.14 Finally, this review examines the Bank’s role in assisting countries experiencing crisis. Although financial and currency crises have existed for decades, the crises experienced by developing countries in the 1990s were arguably transmitted more widely and rapidly across countries and proved more costly, both economically and politically, than in the past. These crises in turn followed a wave o f liberalization in the 1980s and early 199Os, supported by the international financial institutions (IFI), that included, to varying degrees, opening current accounts and capital accounts, freeing exchange rates, freeing interest rates, and lifting restrictions on entry into the financial sector and restrictions o n lending by domestic banks, thereby creating the conditions for rapid credit growth, and larger and more volatile global capital flows. 2.15 The impact o f liberalization and the ensuing financial integration on growth and volatility in many developing countries i s the subject o f considerable controversy in the literature (see Claessens, 2005, for discussion o f literature on this subject, forthcoming). Many authors examining the causes o f crisis agree, however, that liberalization per se has not been the underlying cause o f the crises.* Mishkin (1999), Feldstein (2002), for example, point to a lending boom characterized by excessive risk taking and poor banking regulation and supervision as the root causes for the crisis o f the 1990s. Demirguc-Kunt and Detragiache (2002) found that the certain features o f deposit insurance relate to the incidence o f crises. Although there i s general agreement that the quality o f institutions matter for the success o f reforms, the speed and scope o f domestic deregulation - interest rates, entry, restrictions o n activities - as well as the appropriate sequence o f reforms aimed at moving toward a more open economy are s t i l l the subject o f some debate (Claessens, 2005, forthcoming). The question o f whether the IF1 encouraged liberalization prematurely, that is, in the absence o f adequate safeguards and strong institutions is an interesting one, but would involve examining reforms outside o f the financial sector, for example, in exchange rate policies and current and capital account policies. Such an assessment is beyond the scope o f this review. Bank guidelines and strategies 2.16 The Bank’s 1989 World Development Report (WDR) on Financial Systems and Development was the first public document setting out the Bank’s views on the financial ~ In fact, Hinds (2003) argues, in a background paper for this review that countries had little choice in the late-1980s but to liberalize: retaining fixed exchange rates and closed capital accounts was no longer a policy option for dealing with the transformations occurring in the world economy. 8 sector. At the time, the importance o f the financial sector for developing economies was not widely understood or appreciated, and the first part o f the report examined the ways in which the financial sector could contribute to economic growth; i t also gave a brief overview o f how financial systems had evolved in the Bank’s client countries. The last h a l f o f the WDR was devoted to outlining the essential ingredients o f a healthy financial system; its underlying theme was the importance o f an enabling environment with well- governed institutions, where market participants would perform their functions o f mobilizing resources, allocating credit, and managing risks in an efficient manner. The ingredients included: a legal framework which ensures creditor rights and a functional court system to enforce them; information flow based on sound accounting and auditing; and strong and independent regulation and supervision o f financial institutions. Many o f these ingredients have been shown by subsequent research to be associated with more developed financial systems (see paragraphs 2.11-2.13). 2.17 These views were codified in the Bank’s 1992 Operational Directive (OD) 8.30 o n Financial Sector Operations, although far more attention was given to the macroeconomic environment and financial sector policies on interest rates, directed credit, and credit subsidies than on other aspects. Given the environment in most client countries at the time, this emphasis was mostly appropriate, although the focus on interest rates was arguably premature in systems largely dominated by state-owned banks lending to many state-owned enterprises, whose behavior (both banks and enterprises) was not much influenced by interest rates. OD 8.30 also contained guidance on bank restructuring and resolution o f bad debts, but very little on privatization o f banks.g In addition, Development Economics (DEC) issued three Notes in 1995, on directed credit, lending rates, and restructuring banks, and although the Notes never had the formal standing o f directives, they were intended to provide guidance to staff on these issues. Given the surge in Bank support for privatization in the 1990s (see Chapter 3 and Annex l), the Bank should have provided more guidance on this important reform. 2.18 In 1998, O D 8.30 was replaced by Operational Policy (OP) 8.30, which dealt primarily with lines o f credit, leaving a vacuum in Bank policy on financial sector reforms.’O In 2001, the Bank issued a financial sector strategy, containing the pre- requisites for a well-developed financial system. The emphasis o f the strategy i s interesting for its contrast to the 1992 OD 8.30, reflecting the shift in the political environment that had occurred in the intervening decade. Where the OD had focused on the macro environment and policies on interest rates and subsidized credit, the strategy focused on the importance of: (i) a reliable legal and judicial environment; ( ii) strong banking systems, including a good incentive, regulatory, and supervisory environment; adequate governance o f banks; and a well-functioning payment systems; ( iii) promotion o f capital markets and other non-bank financial intermediaries: and (iv) finding market- based solutions to expanding access to credit. The strategy does not constitute Bank policy, but it i s the closest thing to a statement o f priorities and guidance to Bank staff on OD 8.30 had only this to say about bank privatization, “Opportunities should be explored for attracting new equity investments, including from foreign banks, and for selling government-owned shares to private investors” (paragraph 45). lo The 1997 Strategic Compact declared the Bank’s intention to work with the IMF o n financial sector issues (paragraph 3.3), but this did not constitute an internal Bank guideline. 9 financial sector reforms that exists at present. The strategy indicated that sound practice notes would be prepared on key topics, but as yet, none has appeared, although the Bank's research department and financial sector network have an active research, policy, and dissemination program which have provided intellectual guidance on a range o f topics." Nevertheless, there i s currently no written guidance for Bank staff on good practice for support for bank privatization, for example, for resolution o f non-performing loans, or under what circumstances prudential regulations should be aligned with those contained in the 1988 Basle Accord or the more recent Basle 1 1, or for the entire gamut o f reforms that address constraints to financial sector development. 2.19 The financial sector strategy draws on the literature in arguing for strong banking systems based on good governance o f banking institutions and a reliable legal and judicial environment. I t i s also consistent with research findings o n competition, when the strategy points out that increasing competition in the financial sector may be inappropriate for the small financial systems that characterize many o f the Bank borrowers. I t i s arguably less consistent with the literature in promoting capital market development, to the extent that the literature i s ambiguous on this point. Past OED recommendations and management response 2.20 A previous OED review o f Bank support for financial sector reforms (OED, 1998) recommended that: (i) the Bank follow OD 8.30 (OP 8.30 had not yet replaced the OD); ( iieconomic and sector work (ESW) precede lending, as outcomes at a country ) level were better when this was the case; ( iii) financial sector staff have a greater role in quality control; (iv) more resources be used for systematic monitoring o f financial sector outcomes; (v) technical assistance loans be used more judiciously than in the past (only where there i s clear government commitment to reform and the Bank puts in the necessary resources for designing and supervising the operations); and (vi) the Bank collaborate more actively with both the IMF and the IFC. Management agreed to these recommendations, and gave them prominence in the financial sector strategy that followed in 2001. A framework for evaluation 2.21 This evaluation o f Bank assistance to the financial sector follows the fiamework set out in Figure 2.1 below. The inputs examined in this review are Bank lending and non-lending assistance although it i s recognized, as noted in Chapter 1 , that inputs such as other donor assistance and, in particular, the government's o w n reform programs, are also highly relevant to the picture. These include seminars and conferences o n topical issues, as w e l l as papers and web-notes o n selected issues, including interest rate deregulation, asset management companies, and deposit insurance. The Bank's pro-active dissemination o f i t s research o n financial sector issues t o both B a n k staff and clients m a y w e l l have affected the thinking, diagnostic approaches, loan designs, policies, and reforms for b o t h Bank staff and clients, but OED did n o t carry out specific tracer studies t o assess this. I t could be an interesting area for self evaluation for DEC. 10 Figure 2.1: Evaluation framework Bank’s Lending and Non-Lending assistance 1 Ownership Incentives: ,* Laws and Regs * Prudential Norms * Supervision I Market Structure Contestability Interest Rate Spreads Health - Financial depth Liquidity Preference Access to credit Stability Exogenous factors Exogenous factors I I L I 2.22 Part I o f this report that begins in the next chapter, analyzes inputs. Apart from trends and Regional pattems o f lending, inputs are assessed in two ways: quality at entry o f Bank assistance (Chapter 5) and the outcomes o f Bank loans (Chapter 6). For quality at entry, OED presents findings o f the background papers for this review and desk reviews o f 37 case study countries carried out by OED, supplemented by findings from the Quality Assurance Group (QAG). OED examined the relevance o f the objectives and designs o f the loans in light o f good practice, derived from the literature and the Bank’s intemal guidelines, to the extent that they existed during the period under review. The main elements o f good practice consist o f promoting incentives to sound risk management in financial intermediation through a combination o f strong prudential environment, consistent with international norms; supervision consistent with international principles; decreased government control o f banks and non-bank financial institutions; and putting in place the tools and the incentives for monitoring financial intermediaries by market participants. For lending outcomes (Chapter 6), OED analyzed financial sector loans as well as outcomes o f components supporting financial sector reforms in multi-sector loans. 2.23 Part I1(Chapters 8-1 0) analyzes the rest o f the results chain in Figure 2.1 ,at a country-level. Outputs are intermediate achievements that may be necessary, although by themselves are not sufficient, for realizing the ultimate objectives for the financial sector. Based on the discussion above on good practice, Chapter 8 examines changes in private ownership o f banks and changes in the legal and regulatory environment o f countries that borrowed from the Bank for financial sector reforms and, with appropriate caveats, compares these outputs to those in countries that did not borrow from the Bank for financial sector reforms during the period under review. 2.24 Chapter 9 reviews outcomes at a country level: market structure, contestability (competition), efficiency, and health o f the financial system, particularly the banking system where most o f the reforms were aimed.” Although market structure i s an ” I t could also be argued that one o f the objectives o f a financial system i s to intermediate eficiently, and that therefore financial sector efficiency and profitability should be considered impacts. T h i s review follows recent literature, however, that uses financial sector depth and stability as measures o f financial sector performance (see, for example, Barth, Caprio, and Levine, 2001a and 2001b, and Cull, Senbet, and Sorge, forthcoming), and thus, as the definition o f “impact”. 11 imperfect (and, some economists argue, outdated) measure o f competition, i t is included here because more than a dozen borrowing countries had, at the beginning o f the period, concentration rates (percent o f banking assets held by the top three banks) o f 100 percent, and one o f the objectives o f Bank assistance, sometimes implicit, sometimes explicit, was to reduce the concentration o f market power these situations reflected. Contestability is also examined as a measure o f banking competition, that is, the extent to which entry restrictions and restrictions on banks’ activities were changed over the period under review. Finally, interest rate spreads are examined, although they too are imperfect measures o f either competition or efficiency, as they can be heavily influenced by other factors, including the inflation rate, fiscal deficits, reserve requirements, and tax rates on financial institutions. Some o f these factors are taken into account in the analysis. Measures o f health are also examined, to the extent the data permit, in terms o f capital adequacy, non-performing loans, and profitability. 2.25 Finally, Chapter 10 examines the extent to which the ultimate objectives for financial sector development have been achieved at a country level: how well i t serves as an intermediary between savers, mobilizing relatively large amounts o f resources, and efficient investors, lending the resources to the private sector, and the extent to which it has remained stable and avoided costly crises. The measures used are: (i) progress toward greater banking depth (M2, which consists o f cash, demand deposits, and time deposits, as a proportion o f GDP); for capital market reforms, size and turnover o f the market; (ii) increasing confidence in the banking system, measured by the inverse o f the preference for liquidity (cash as a proportion o f M2); ( iii) credit to the private sector, as a percent o f GDP; and (iv) stability o f the financial system in terms o f the absence o f a major systemic banking crisis. PART IASSESSING INPUTS: BANK ASSISTANCE 3. Trends in lending and non-lending 0verview 3.1 Beginning in the late 1980s, in recognition o f the important role the financial sector could play in growth,I3 the Bank shifted i t s focus from support o f individual financial institutions, which in any case had had disappointing results, to supporting sector wide improvements in the financial sectors o f client countries. In the first half o f the 199Os, the Bank dramatically increased analysis o f financial sectors as i t sought to understand the constraints to better financial sector performance and to provide an underpinning for its adjustment lending for financial sector reforms. 3.2 By the time o f the Strategic Compact in early 1997 (prior to the Asian crisis) the Bank expressed its intention to work with the IMF to build capacity in client countries to regulate and supervise their financial systems, with particular focus on banking, and to l3See the 1989 World Development Report, devoted to the financial sector. 12 develop a set o f core monitoring indicators to identify vulnerability to crisis. The Asian crisis in the second half of 1997 gave urgency to supporting financial sector reforms as well as for obtaining more timely information on financial stability. The Bank responded by providing exceptionally large amounts o f lending for financial sector reforms to the Asian countries in crisis; it also began a joint diagnostic process with the IMF, the Financial Sector Assessment Program (FSAP), to be examined separately by OED (forthcoming). Thus, although driven mainly by the Asian crisis and subsequent macroeconomic and financial sector crises in other countries, the increased focus on the financial sector expressed in the 1997 Strategic Compact was realized in both lending and diagnostic work in the second half o f the period under review. 3.3 Whether in response to the Strategic Compact or to the Asian crisis and i t s aftermath, financial sector issues also received greater focus in country assistance strategies (CASs). According to the CAS retrospective o f 2003, more than 80 percent o f the FYOO-01 CAS had some discussion o f recent progress in the financial sectors, and there had been a significant improvement in both quantity and quality o f coverage o f financial sector issues over the earlier periods. Figure 3.1: Bank loans classified as finance, by amount, as Bank lending for financial sector reforms: percent of total Bank commitments,Fy93-03 trends loans classified as finance, about half, or 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 US$12.1 billion, was for countries Adjusbmnt 0 hvestmnt I#Qsis 3.5 Aside fkom crisis lending, discussed in Chapter 7, there has been a slight upward trend over the period FY93-03 in the proportion o f annual Bank commitments classified as finance, driven by an increase in adjustment lending, which began in the late 1980s, and which has offset the drop in lending for lines o f credit (LOC).14 The non-crisis adjustment lending, which total some US$9.2 billion in commitments over the entire period, has roughly doubled between the first and second h a l f o f the period. ~ ~ For a l4 detailed discussion o f lines of credit, their trends, designs, outcomes, and issues, see OED (2005). 13 3.6 Lending withfinancial sector components. The picture changes, however, when loans in other sectors are included that contain financial sector components (see Box 3.1 on how these loans were identified): US$56.1 billion, or 24 percent o f total Bank lending, included some support for the financial sector over the period. O f this total, some US$43 billion was in adjustment (169 operations), and US$13 billion was in investment lending (2 16 operations), including L O C with financial sector objectives. Excluding LOC, investment lending was only US$3 billion. A total o f 96 countries borrowed from the Bank for financial sector reforms, if L O C are included (87 countries, if L O C are excluded). Box 3.1: Identifying Bank assistance for financial sector Most Bank lending for financial reforms has been through multi-sector adjustment or technical assistance loans and credits. OED read through over 2,000 Bank loan documents to identify support for reforms or investments in the financial sector, finding some 385 operations (excluding pensions, which account for an additional 130 or so), that contained conditionality or f h d i n g related to the financial sector. O f the total number o f Bank operations containing financial sector components, only 36 percent o f them were classified as finance; another 23 percent were under economic policy; private sector development accounted for about 14 percent; and the remainder were other sectors (see Annex 1, Figure 6 for breakdown). Because it i s not possible to allocate lending amounts in multi-sector adjustment loans to specific sectors and it was difficult to distinguish multi-sector adjustment loans that focused primarily o n the financial sector from those where it was a minor aspect, the data give a general, rather than precise, picture o f how lending assistance for the financial sector has evolved in the past decade. Table 3.1: Lending for financial sector reforms, FY93-03 Sectoral Loan Classification Adjustment Investment Totals As percent o f Bank No Amount No Amount No Amount No Amount US$m US$ m US$ m Financial Sector 54 19,683 83 5,122 137 24,805 5 11 Other sectors 115 23,356* 133 7,912* 248 31,268* 9 13* Total 169 43,039* 216 13,034* 385 56,073* 14 24* * These figures show the total amount o f lending that that include some focus on financial reforms; they overstate the amount deiicated only t o financial reforms. Annex 1 has more details on Bank-wide and Regional trends. 3.7 As shown in Figure 3.2, financial reforms in multi-sector operations classified under other sectors outnumbered loans classified as finance in most years (for breakdown by sector, see Annex 1,Figure 3). I naddition, there has been a notable downward trend in the last decade in the proportion o f loans containing financial sector reforms, although this is due mostly to a dramatic drop in the number o f L O C approved. Without LOC, the downward trend, although still evident, is less strong. Because L O C are discussed in a separate OED report (OED, 2005), they are not included in the remainder o f this review unless otherwise indicated. 14 Focus of financial sector reforms 3.8 Banking reform^'^ dominate the agenda Figure 3.2: Number of Bank loans with financial sector in most Bank loans, compared to non-banking reforms, as percent of total loans, FY93-03 reforms such as those covering capital 20% markets, insurance, and pensions. This reflects 18% 16% the fact that for most Bank clients, banks and 14% bank-like institutions are far more important 12% 10% than other forms o f intermediation. 8% 8% 4% 3.9 Within banking, restructuring and 2% privatization dominated the agenda. Whether 0% I993 1994 1995 I996 1997 1998 I999 2000 2001 2002 2003 borrowing countries were centrally planned, [ classified as finance classified under other sectors 1 socialist states or more market-driven, the vast majority of Bank clients had, at the beginning of Figure 3.3: Focus of reforms as percent of all Bank the 1990s, banking systems heavily dominated projects with financial sector components by government-owned banks, many o f which E 100% 1 were characterized by an accumulation o f non- 7 1 6 90% (Y performing loans (NPL), inadequate capital, and 80% : ;; 70% l o w profitability. As a result, most countries undertook, at a minimum, to restructure their - .S 40% banks, and many also moved toward more f 30% fundamental solutions, including consolidation, = c, 20% ‘i 10% liquidation, and privatization. Bank lending reflects these trends. Figure 3.3 shows that out f O% Regulation Supervision Supervlsion Restructuring Capital hsurance F - and and Markets o f 280 Bank operations (excluding LOC), almost ’s Leclisiation Lecllsiation Rivatzation 70 percent contained reforms aimed at bank I 2 banklng and bank-llke financial Instltutlons non-banking financlal lnstltutions restructuring and/or privatization (and often both Table 3.2: Investment lending with financial 3.10 By contrast, lending for capital market sector components, by category,* FY93-03 was some 22 percent o f loans and insurance Amount reforms comprised only 10 percent o f loans over Category Number $’OOO the period,” and tended to be concentrated in TA 69 1,112 Guarantees 7 631* middle income countries, mostly in Europe and Specific Investment 35 1,359 Central Asia (ECA), Latin America and Total 111 3,102 Caribbean (LCR), and Middle East and North * Excludes LOC; categories are somewhat arbitrary; most Africa ( m A ) regions (see Chapter 4). Support specific investment loans were TA. See para. 3.12. l5The definition o f “banks” varies by country. Development finance companies; savings and loan associations; and even specialized banks are often considered non-bank financial institutions. For the analysis in this report, reforms aimed at bank-like institutions are categorized under banking reforms. l6M a n y loans are under more than one category o f reforms. l7In addition, pension reform would account for about 38 percent o f total loans if they were included. 15 markets may have been in part the result o f IFC's greater role in this area, but it may also have been in recognition o f the small part played by the capital markets relative to banking, and the large unfinished agenda in banking. diagnosis or audits o f banks or other financial 16 institutions, prepare them for privatization, or, in 14 general, to provide technical support to the reforms y) 12 supported through adjustment lending. Slightly f less than half o f these operations (50 o f the 111) were in lower income countries, and most o f those g= : approved in middle income countries were either 0 in transition economies or in countries 1993 1994 1995 I 9 8 6 1997 1998 I 9 9 9 2000 2001 2002 2003 experiencing crisis. classnled as finance classified as other sectors 3.12 Among the specific investments, three large projects accounting for over half o f the total commitment are really adjustment loans disbursed as time slices connected with privatization. In Pakistan, the Banking Sector Restructuring and Privatization Project for us$300 financed severance payments n i Figure 3.5: Number of ESW reportscontaining bank downsizing prior to privatization; and in two financial sector analysis, FY93-03* Brazil loans. Rio de Janeiro and Minas Gerais I State Privatization projects for US$250 and 35 30 US$170 million, respectively, funds were { 25 disbursed as time slices, but not for severance payments related directly to banking reforms. The E 20 15 ' 10 third category o f investment lending i s guarantees: 5 six were approved over the period, for US$630 0 1883 1894 1885 1898 1887 I 9 9 8 1889 2000 2001 2002 2003 million, o f which one was a loan to Argentina for 1 lESW lFSAP1 - US$500 million. ~~ ~~ *Includes only formal sector reports '*There are very few TA loans that were approved i nthe absence o f an adjustment loan and it was usually because the planned adjustment loan did n o t materialize (Togo, Uzbekistan). China i s the exception t o this (Box 4.2). 16 Bank non-lending assistance for financial sector reforms: trends 3.13 As noted in paragraph 3.1, the shifl in focus in the late 1980s to sector-wide reforms was accompanied by the need for a better understanding o f the constraints and issues in the sector. This i s reflected in the surge in formal economic and sector (ESW) reports containing financial sector analysis from single digits per year in the 1980s to 25 - 30 reports per year in the first half o f the 1990s. In the first five years o f the period under review, FY93-97, some 126 formal ESW reports were produced. In the last half o f the period, by contrast, some 74 ESW reports were produced, or 48 apart from the highly specialized FSAP reports (Figure 3.5 for number o f reports and B o x 3.2 for description o f FSAP). 3.14 The drop in formal reports i s likely due to several factors. In the late 199Os, there was a shift to informal sector work, such as policy notes (Box 4.2 o n China as an example), which do not show up in these numbers. I n addition, the resources for the FSAP and for the more recent anti-money laundering and combating terrorist financing (AMLKFT) a c t i ~ i t i e smay ' ~ have replaced (or displaced, depending on one's point o f view) other financial sector work in a given country or within a Bank unit covering several countries. Box 3.2: The Financial Sector Assessment Program (F'SAP) The FSAP i s a major initiative undertaken jointly by the W o r l d Bank and the IMF in response to the financial crises o f the late 1990s. It was initiated in M a y 1999, initially as a 12-country pilot exercise, to be expanded to other volunteering countries, to facilitate early detection o f financial sector vulnerabilities and identification o f financial sector development needs, as well as to support the dialogue among the national authorities, the Bank, and the IMF. As o f July 2004, assessments have been completed or initiated in over 80 countries and o f those reassessments have been completed or initiated in eight countries; the program has involved a significant use o f Bank resources. Reassessments were initially planned to take place every four to five years, but later the frequency was set at seven to ten years because o f resource constraints and their implications for the pace o f the program. The FSAP i s being assessed by OED and the IMF's Independent Evaluation Office, scheduled for FY06. l9 Since FY02, AMWCFT activities have taken o n increasing prominence in the Bank's non-lending n several Regions. financial assistance and have been incorporated as components in lending assistance i 17 4. Regional patterns o f Bank assistance 4*1 came to the Table 4.1: Lending categorized as finance, percent of total, FY93-03 Drocess o f financial sector reforms Lending in the last decade at different Lending Amount Percent o f times, with different initial Amount Percent o f Excluding RegionBank conditions, capacity, and most Region US$m RegionBank Crisis Lending, Lending Lending US$m particularly, degrees o f AFR 630.6 2 630.6 2 commitment, and there are EAP 6,357.0 11 524.0 1 Regional patterns to these ECA 5,257.2 11 2,029.4 4 differences; Bank lending reflects LCR 7,499.8 13 4,49 1.8 8 these patterns, which this Chapter mA 845.5 7 845.5 7 briefly describes. SAR 659.8 2 659.8 2 Total 21,249.9 9 9,181.1 4 4.2 For lending categorized as finance (excluding LOC), three Regions, Latin America and Caribbean (LCR), East Asia and Pacific (EM), and Europe and Central Asia (ECA), account for 90 percent o f commitments (Table 4.1 and Annex 1 , Figures 1 and 2), although the large sums are due mainly to the crisis lending (as defined here) which i s concentrated in these three regions. Excluding the crisis lending, EAP has had almost no lending under the financial sector board. Africa (AFR) and South Asia ( S A R ) regions also had little, in absolute terms or as a percentage o f the Region’s own lending. In S A R , Pakistan is the only country that had financial sector adjustment loans, although financial sector reforms have been introduced recently in TA projects in Bangladesh and Nepal. By contrast, even aside from the crisis lending, E C A and L C R have had financial sector adjustment loans in most years and in a fairly large number o f countries, reflecting the generally strong trends toward reform in those Regions, while in Middle East and North Africa (MNA) region, financial sector adjustment lending has been concentrated in a few years to a few countries (Jordan, Morocco, and Tunisia). ECA dominates Table 4.2: Lending with financial sector components, FY93-03 Number Percent o f Total amount Percent o f 4.3 When the lending is of RegionaVBank o f lending* Regionamank $m lending* expanded to include all lending with Region Projects Projects AFR 69 10 3,926 13 any financial sector components EAP 29 7 11,558 21 (again, excluding LOC), E C A far ECA 99 16 14,018 31 outnumbers the other Regions in LCR 59 10 12,845 22 terms o f both number (99 projects) MNA 14 7 2,22 1 18 and proportion o f i t s loans (16 SAR 10 4 1,575 5 percent) that dealt with financial Total 280 10 46,141 20 reforms over the FY93-03 period * See note on Table 3.1 (Table 4.2 and Annex 1 , Figure 4). At i t s peak in FY95-96, as many as 25 percent o f ECA’s loans contained financial sector components, and out o f 28 borrowers in the Region, only one (Estonia) had no loans with 18 financial sector reforms included” (see B o x 4.1 on lending in ECA). This clearly reflects the focus o f the countries’ commitment to transition from state-controlled mono-banking to an entirely different banking structure and governance; financial reforms were often accompanied by other reforms to establish private ownership and market mechanisms. The incentive o f accession to the European Union provided further impetus to reforms. Lending for financial reforms in E C A has decreased in recent years compared to the early part o f the period (Annex 1,Figure 4). Africa and LCR: early reformers 4.4 Some o f the earliest borrowers for bank privatization were in AFR and LCR. Ghana, for example, had an adjustment credit in FY88; Cameroon, FY89; Senegal, FY90. nLCR, Bolivia, Chile, Mexico, and Venezuela borrowed in the 1980s for financial I sector reforms. 4.5 nAfrica, about 10 percent o f lending operations have contained financial sector I components over the period FY93-03, with the majority o f projects in sectors other than finance, which may reflect the need in smaller countries to package reforms across sectors into one operation. Out o f about 40 active borrowers at any given time in the Africa Region, 24 have borrowed for financial reforms, with a heavy emphasis o n bank restructuring and privatization (Table 4.3 and Figure 4.1). Regulation and supervision were less o f a focus, and were included in fewer than h a l f o f the operations, possibly because the banks in West African countries are supervised by Regional central banks. Somewhat surprisingly, given the modest size o f the economies in Africa, about one fifth o f the operations in Africa that touched o n the financial sector included support for capital market reforms.” 4.6 In LCR, a significant portion o f lending for financial sector reforms i s connected to crisis support; out o f a total o f 59 loans with financial sector components, about one- third o f them are crisis-related. The non-crisis loans form a heterogeneous group, tailored to the conditions and commitment o f the borrowing country. In addition to the focus on banking - restructuring and privatization (Table 4.3 and Figure 4.1); regulatory and legislative changes, such as aligning prudential regulations with Basle standards; and introducing deposit insurance schemes or reforming existing schemes, there was a more intense focus on capital market reforms than in most other regions (see Annex 1, Table 2) - Argentina, for example, had a US$500 million adjustment loan devoted to capital market development. In Brazil, lending for financial sector reforms started relatively late, in FY97, and took an unusual form, with large TA loans for privatization o f state banks. Most o f the other non-crisis lending to L C R countries that supported financial sector reforms consisted o f only one adjustment loan and one TA loan per country over the period under review. 2o Estonia had a Rehabilitation Loan with financial sector coverage; and a line o f credit that played a catalytic role in commercial bank restructuring, although most o f the finds under the project were not used. 21 See background paper by Mozes (2003) on Bank lending for financial sector reforms in Africa. 19 Table 4.3: Regional concentration of reforms Total AFR EAP ECA LCR MNA SAR Number o f loans with financial sector components 280 69 29 99 59 14 10 percent o f loans that focused on reforms Regulation and Legislation 43 76 7 1 53 64 50 Supervision 33 62 54 49 50 3 0 Restructuring and Privatization 71 7 2 72 61 36 90 Capital Markets 14 28 18 29 50 0 Insurance 22 0 8 8 36 10 Payment system 14 14 2 1 5 21 20 Box 4.1: Financial sector reforms in the ECA Region: Bank strategy, analysis, and lending Strategies. In most countries, financial sector reform was a priority in the assistance strategies throughout the decade, reflecting not only i t s importance but also the gradual nature o f the progress being made. In some country programs, the priority on financial sector development was reduced in later CASs, either because the job was perceived to have been largely completed (e.g., Hungary, Poland, and Kyrgyz Republic - in this latter case, a judgment that tumed out to be wrong), or because progress was so slow (e.g., Romania, Russia, Uzbekistan). In other countries, after significant progress was achieved (Macedonia, Lithuania), the Bank shifted i t s focus to diagnosis (e.g., FSAP) and policy dialogue, with further reforms financed by other agencies, including the IFC. Analysis. A considerable body o f financial sector analytic work was embedded in economic reports or produced as informal pieces o f work, with formal financial sector reports emerging in the latter half of the decade. For some E C A countries, it i s surprisinghow late in the decade the first pieces o f formal financial sector work appeared - e.g., Armenia (ZOOO), Georgia (1999), Kyrgyz Republic (1999) where substantial Bank lending in the financial sector had already been undertaken. N o dedicated formal financial sector work at all was found for Albania despite 14 loans (four adjustment; four TA; six LOC) aimed at least partially at financial sector objectives or Bosnia Herzegovina, with 16 loans with financial sector components (three adjustment; one TA, 12 LOC). Lending. In many E C A countries financial sector components were included in a series o f structural adjustment operations that took a gradual, but steady, approach to reforms. In Armenia, for example, an Institution Building Project (FY93) and a Rehabilitation Credit (FY95) addressed banking supervision, and between FY96 and 03, five SACs and two accompanying TA credits supported restructuring and then privatizing banks, promoting a capital market, and introducing deposit insurance. Georgia’s Rehabilitation Credit (FY95) supported strengthening o f prudential regulations, a diagnostic review o f five state-owned banks, and development o f restructuring or privatization plans, followed by two SACs and two TA credits (in FY96, 98) with conditions on privatizing the majority o f shares o f former state banks, and after their sale, meeting agreed performance targets. SAC I1also had measures to support capital market infrastructure. Latvia, Lithuania, Kazakhstan, Moldova, Tajikistan, and Ukraine each had several multi-sector adjustment loans addressing banking reform, although in some the financial sector components were relatively minor. By contrast, Poland had one adjustment loan, the EFSAL (FY93, preceded by a FY91 loan focusing on financial institutional reforms) that included recapitalizing state owned banks and empowering them to reduce their non-performing loans by restructuring enterprises, and then privatizing the banks. Similarly, Hungary, Croatia, Slovenia, and Slovak Republic each had one adjustment operation addressing mainly banking reforms. In Hungary, the FY97 EFSAL took several years to prepare and negotiate, but was a wide-ranging operation that addressed most issues identified in prior sector work. Source: Fred Levy (2003), and O E D data base. 20 EAP: mostly crisis-driven 4.7 In EAP Region, with the exception o f the Philippines, the larger countries had few loans dealing with reforms o f the financial sector until after the Asian crisis. In the Philippines, a financial crisis at the Central Bank drew Bank support in FY93 to help i t restructure; this was followed by an adjustment loan to provide continued support for banking regulation and supervision and for privatization o f one large state bank. In Vietnam, an FY95 Structural Adjustment Credit (SAC) included a condition for auditing two state banks, and a tax reform on banks’ net income; and no further financial sector reforms until FYO1. In China, the Bank made only one loan for TA (Box 4.2). Mongolia and Lao PDR, by contrast, have each had two adjustment credits and accompanying TA operations (in Lao, it was an Institutional Development Fund grant) for banking reforms. Box 4.2: Bank assistance to China In the past ten years, although several loans were prepared, the only Bank loan approved and disbursedto China for financial reforms was an FY93 Financial Sector TA Project for US$60 million. I t aimed to make improvements in accounting and auditing o f banks, supervision by the central bank, and buildinga modem payments system, but i t s underlyingpurpose was to begin a substantive dialogue on reforming the banking system. The scope o f the task was huge, given that the central bank itself had over 2,400 branches and 180,000 employees, supervising a banking system with more than US$1trillion in assets. The preparation and supervision o f the project enabled the Bank to engage govemment officials in policy issues, leading to a reorganization o f the central bank and a diagnostic audit o f several branches o f a state bank, revealing worrisome operational procedures, but the larger purpose was not accomplished. The project’s outcome was considered satisfactory, but the Bank, for a variety o f reasons related both to reluctance on the side o f the Chinese authorities and disagreement within the Bank on the approach, has made n o other loans in the financial sector in decade since the TA loan was approved and for about five years, between 1995 and 2000, had no effective dialogue. Starting in 2000, the Bank ramped up i t s non-lending activities, producing four (informal) policy notes (on interest rate liberalization; deposit insurance; bank supervision; and reforms o f state banks). Source: S. Ramachandran (2003). MNA and S A R : conservative approach to reforms 4.8 MNA and S A R trail the other Regions in the proportion o f each Region’s loans that contain financial sector reforms (Table 4.2), and in the proportion o f countries that have borrowed from the Bank for banking privatization (Figure 4. l), a reflection o f the relatively conservative approach o f the countries in these two Regions to financial reforms. In MNA, several countries have pursued stronger prudential regulations and modest restructuring and privatization (Morocco and Tunisia), while other borrowers (Algeria, Egypt) have not borrowed from the Bank to pursue significant banking reforms. 4.9 In S A R , only Pakistan has borrowed frequently during the period for financial sector reforms (Box 4.3), although the Bank has recently resumed lending to address financial issues in both Nepal and Bangladesh. The Bank carried out analytic work during this period in both Bangladesh and Nepal, however, even in the absence o f lending. In Nepal, the FY03 TA operation was the first Bank credit approved in almost fifteen years (since FY89) to address financial sector reforms. 21 Box 4.3: Pakistan and Bangladesh: government commitment explains differences in patterns of Bank lending Pakistan began to reform its financial sector in the late 198Os, supported by a FY89 adjustment loan and a FY95 LOC with substantial policy content (which followed a series o f earlier LOC in the 1980s with mostly unsatisfactory outcomes). Although some measures were taken (partial privatization o f t w o state banks; liberalization o f interest rates, stronger prudential regulations), they failed t o make significant improvements and in 1996 Pakistan experienced a banking crisis. After this, the govemment began t o tackle the more serious issues facing the sector, including poor governance, rampant default by large, well-connected borrowers, over-staffing, and undue interference by labor unions in bank operations. The Bank supported the reforms with a series o f policy loans (three financial sectors for US$766 million, including funding o f severance payments, and three multi- sectors) and one TA loan. The pace o f reforms has been uneven, but significant progress has been made in downsizing and restructuring the large state banks; the asset share in govenunent owned banks has dropped from 92 percent in 1990 to 45 percent in 2002. Central bank supervision has improved, and is considered to meet 22 o f the 25 core principles o f good supervision; and prudential regulations have been strengthened. Weaknesses remain, particularly in state dominated non-bank financial intermediaries, and the legal and judicial process for enforcing legal contracts. Bangladesh also borrowed f r o m the Bank in the early 1990s for financial sector reforms (through both LOC and adjustment), but the poor results discouraged the Bank f r o m pursuing further reforms for about a decade. The Bank considered the government insufficiently committed t o addressing the corruption and governance plaguing the sector, which by any standards are quite serious. In the late 199Os, the Bank estimated that 50 percent o f loans were non-performing; there were several hundred thousand defaulters and a pervasive “culture o f default”; the large state-owned banks were essentially dysfunctional (insider lending, fraud, negligence) and enforcement o f prudential regulations by the central bank was lax. Bank lending to Bangladesh for finance between 1992 and 2002 concentrated o n supporting micro-finance, which was intermediated by specialized institutions outside o f the banking sector, and n o t plagued by the same ills. The Bank nevertheless carried out analytic work (with a 1996 report o n rural finance and a 1998 report o n the financial sector), and lending for financial reforms resumed in 2003, with a multi-sector credit addressing prudential regulations and bank restructuring with a v i e w to eventual privatization and a TA credit. Although stronger prudential regulations have been passed, political opposition to bank privatization has been stronger than expected and the process o f preparing banks slower than planned. Source: L o n g (2003b). 22 Figure 4.1: Bank support for bank restructuring and privatization, by country - 0 d i d n ’ t b o r r o w for b a n k r e s t r u c t u r i n g or p r i v a t i z a t i o n b o r r o w e d for r e s t r u c t u r i n g only-no p r i v a t i z a t i o n b o r r o w e d for r e s t r u c t u r i n g andlor p r i v a t i z a t i o n 0 n o t B ank clients (high i n c o m e ) 5. Quality at entry o f Bank assistance Overview 5.1 This chapter reviews quality at entry in lending and quality o f non-lending assistance. In addition to reviewing assessments o f individual products (loans and sector reports), OED relied on background papers and desk reviews o f 37 country case studies22 to address, first, the consistency o f Bank assistance within a country, between diagnosis and lending and across lending operations and second, the question o f whether Bank assistance across countries reflects a coherent strategy for the sector, after taking into 22 The country case studies were selected t o capture the bulk o f the (non-crisis) lending and t o represent a l l Regions (see Annex 4 for list o f countries and amount o f lending reviewed). 23 account specific conditions in borrowing countries. This chapter also reviews whether past OED recommendations for the financial sector are reflected in Bank assistance. Quality at entry in lending 5.2 Since 1998, QAG has carried out six quality at entry assessments (QEAs) o f loans and credits, using a random sample o f operations approved shortly before the assessment, and examining eight dimensions o f quality. Operations receive an overall score, from 1 to 4, corresponding to highly satisfactory, satisfactory, marginal, and unsatisfactory. Across all six QEAs, 32 financial sector operations were assessed, representing about 25 percent o f total financial sector lending covered by this OED review. The loans received an average overall score o f 2.0, corresponding to a satisfactory rating, which i s exactly the same as the average rating for loans from all other sectors over the six years (Table 5.1 on page 30). QAG’s assessment i s consistent with OED’s own review in background papers and case countries (paragraphs 5.6-5.9), with several important caveats. 5.3 The first caveat i s that the quality at entry o f LOC, only some o f which were in the financial sector but most o f which had financial sector objectives, were found in a separate OED review to be poor and to deviate frequently and in significant ways from the Bank’s guidelines o n L O C (see OED, 2005 for details). 5.4 The second caveat i s that most o f the support for financial sector reforms, both in numbers o f operations and in amounts lent, has occurred over the last decade in components o f multi-sector loans (see Chapter 3 o n trends in lending), so it is not possible to get the fillpicture o f the quality o f Bank support for financial reforms by reviewing only financial sector operations. The next chapter reviews outcomes o f both financial sector loans and components o f multi-sector loans. 5.5 The OED review o f country case studies found that the objectives o f reforms supported-bythe Bank have been consistent with the literature in areas where there i s widespread agreement in the literature and within the Bank: reducing government ownership o f banks and other financial intermediaries; improving prudential regulations consistent with international standards; and strengthening bank supervision, to be consistent with international principles.23 Examples o f good practice exist in every Region, even where outcomes were unsatisfactory (Box 5.1). 5.6 Even where the objective o f the reforms was consistent with good practice, however, the specific conditionality or design o f the loan was not always appropriate for achieving the objective. For example, the Bank sometimes aimed to strengthen the health o f the financial sector without addressing the underlying reasons for the poor situation o f the banks. Thus, the Bank supported recapitalization o f state banks in the absence o f any 23The 1998 OED review o f B a n k assistance had recommended that the B a n k pursue reforms as advised in OD 8.30, but for most o f the period under review, the OD was becoming outdated: it emphasized directed credit and administered interest rates, while Bank lending was become more focused o n reducing govemment’s direct role in controlling banks and other financial intermediaries and o n bringing the prudential and supervisory framework in line with international norms. The OD was relatively sketchy in these areas (see paragraph 2.16 for discussion o f OD). 24 government commitment to change their governance, particularly through privatization, in Algeria, Lao PDR and Vietnam. Although the Bank i s constrained by what the government i s willing to do, there i s ample evidence that new investment in banks which in practice have political mandates is not a sustainable solution to improving the health o f the banking system and generally results in a re-accumulation o f bad debts (this issue is discussed further in Chapter 9). Box 5.1: Examples of high relevance of objectives of financial sector reforms I In Burkina Faso, the Bank took a broad view o f the troubled banking system by focusing on consolidation, financial rehabilitation, privatization and, if necessary, liquidation. In Pakistan, the efforts included improving prudential regulations to align them with international norms, undertaking an ambitious program o f downsizing and restructuring public banks to prepare them for privatization, and improving the quality o f banking supervision. In Lithuania, the Bank addressed a wide-ranging reform agenda inthe financial sector, including collateral law, accounting standards, and concurrent enterprise privatization. 5.7 In addition, there are cases where Bank lending, in pursuit o f reducing the role o f government as owner o f banks, has been overly focused on privatization as an end in itself, and too little focused on the ultimate objective o f having well-managed banks whose owners have incentives to both manage risks and realize returns. Thus, in Mozambique and Georgia, for example, the Bank did not discourage privatization o f a bank or banks to inappropriate owners, which in Mozambique, led to considerable expense for the government and in Georgia, led to concern about the quality o f the banking assets, I nUganda, the Bank encouraged privatization o f banks to inappropriate owners, which led to a re-nationalization and re-privatization, also at considerable expense to the g~vernment.’~ 5.8 One type o f assistance that will never show up in QEA but which deserves positive recognition consists o f situations where the Bank reduced the amount o f a loan, or delayed lending, or did not lend at all, because the government was not committed to reforms. These include the preparation o f Economic Competitive Adjustment Loan (ECAL I )in Tunisia, where the financial sector component was removed from the loan and the amount cut in half during preparation because the government was not ready to make reforms sufficient to justify lending for them. The Bank returned two years later with E C A L I1focused only on financial sector reforms. In the Slovak Republic, the Bank postponed Enterprise and Financial Sector Adjustment Loan (EFSAL) for six years, from a planned operation in FY95 until FYO1, when the government was ready to reform. In Bangladesh and Nepal, the Bank had no adjustment operations for over ten years, yet the 24 Assessment o f designs o f Bank operations, particularly adjustment loans, i s difficult because o f large differences between what the program document stated was expected during implementation and legal conditions for disbursement. Thus it was difficult to know the specific reforms agreed i n the context o f the loan. Examples o f these differences were found in Algeria, Cameroon, Chad, Cote d’Ivoire, Kazakhstan, Madagascar, Pakistan, and Poland, to name a few. 25 dialogue continued in both countries until FY03, when a TA credit addressing financial sector reforms was approved in each country. Quality o f non-lending services 5.9 QAG has also carried out assessments o f E S W over five years (FY98-02), using a random sample of ESW completed prior to each assessment, and examining five dimensions o f quality. As with lending operations, ESW reports receive an overall score, from 1 to 4, corresponding to highly satisfactory, satisfactory, marginal, and unsatisfactory. Combining all five QEAs, twenty-two financial sector reports, including FSAP reports, received an average overall score o f 1.7, which is between a satisfactory and a highly satisfactory rating. By contrast, the average score for all other ESW reports i s 2.1 (Table 5.1), significantly lower than the financial sector work.25 Table 5.1: Quality of ESW, QAG assessment, N97-FY03 Financial sector network Other networks Number Average score* Number Average score* Difference Lending Operations 32 2.0 483 2.0 ESW, including FSAP 22 1.7 322 2.1 0.4** ESW, excluding FSAP 18 1.8 322 2.1 0.4 * Lower score i s higher quality **Statistically significant at 5 percent 5.10 Several background papers for this review also noted the strong quality o f financial analysis. In S A R , for example, an extensive ESW program supported lending in Bangladesh, including rural finance reviews that supported lending for micro-finance, as well as in India and Pakistan. The Country Assistance Evaluation for India (OED, 2001a) gave particularly high marks to the financial sector ESW. I nthese countries plus Nepal, the ESW provided the basis for continued policy dialogue, and helped to define nECA, “policy the issues and the policy alternatives, even in the absence o f lending. I papers and ESW reports., . were o f very high quality, and the issues and options involved in financial sector development were well understood and set out” and the priorities, coverage, and content o f the recommendations were consistent with good practice and international standards (Levy, 2003, page 42). Nevertheless, different views o n major issues sometimes emerged in ESW within a country, which sent mixed signals to the borrower (paragraph 5.14). 25 One could argue that F S A P reports should n o t be included, because the underlying analysis i s a j o i n t effort with the IMF, with a standardized approach and scope, so that their quality i s not attributable solely to Bank effort. On the other hand, the report produced by the Bank i s part o f the Bank’s diagnostic work. The results with and without F S A P are therefore presented. In addition, procurement and financial management assessments (CFAA and CPAR) are also somewhat standardized E S W products, so OED analyzed the results both with and without these products and found the same results in b o t h cases. 26 5.1 1 In addition, OED had recommended in i t s 1998 review that E S W precede lending and in most countries and for most loans this was the case. O f the 37 country case studies, recent ESW - defined as dated within four years preceding or one year after the year o f loan approval - was available in 3 1 o f them. Although the designs o f the loans were typically not able to take on board all the recommendations in the ESW, the reforms addressed in the loans had usually been identified as important in the diagnosis. 5.12 Exceptions to this pattem occur particularly in countries that experienced a crisis during the analyzed period (Colombia, Jamaica, Korea, Thailand, and Uruguay), where loans were put in place rapidly without benefit o f recent ESW. Similarly in post-conflict countries (Bosnia and Herzegovina, Democratic Republic o f Congo, and Sierra Leone), the Bank provided assistance relatively quickly without benefit o f prior diagnosis. Other situations included countries where the Bank had a number o f loans addressing financial sector reforms, with an on-going dialogue through implementation and supervision (Algeria and Tunisia). ,As noted in B o x 4.1, the Bank supported major financial sector reforms in E C A countries (Albania, h e n i a , Georgia, Kyrgyz Republic, and Poland) without the benefit o f formal financial sector reports; OED’s 2000 Country Assistance Evaluation for Albania (OED, 2000) called the absence o f a sector strategy early o n a mistake, and suggested that one o f the reasons for the failure o f the early attempts at sector reform was lack o f adequate diagnosis, focus, or prioritization. Consistency o f approach within countries 5.13 Synergies among ESW, adjustment lending, and TA loans (and, on occasion, LOC) in a given country have been good, with mutually reinforcing messages such as the importance o f well-governed financial institutions, stronger prudential norms, better legal framework, creditor rights, and extemal audits. Examples o f this are in B o x 5.2. Box 5.2: Examples of strong consistency between Bank products within countries In Yemen, the financial sector note was prepared specifically as a way o f identifying m a i n areas for financial sector reform and as a result, the design o f the Financial Sector Adjustment Credit (FSAC) followed closely from the recommendations o f the ESW. In Brazil, in addition t o identifying the large and problematic role o f state banks, several sector reviews in FYOO also identified the need t o improve collateral rights and sharing o f credit information and this analysis fed directly into the design o f the programmatic Financial Sector Adjustment Loans (FSALs) that followed. In Hungary, the F Y 9 7 E F S A L included virtually a l l the main issues that had been identified in the sector work that preceded it by two years. 5.14 But in some countries the Bank has sent mixed signals across different but closely timed strategy and diagnostic work, between ESW and lending, or within lending. I n Russia, for example, an early banking sector study focused on the need to restructure the large state banks, while the country assistance strategy that followed soon thereafter mentioned only that government should assign high priority to privatizing state banks and consolidating private ones, while focusing Bank lending on providing L O C to private 27 banks (and leaving the larger issues untouched). In a number o f countries, the Bank advocated closing or privatizing state banks while at the same time supporting expansion o f government ownership o f banks: in Albania, for example, the Bank supported within the same credit closure o f a state-owned rural bank and establishment o f a new one, which then closed down four years later after accumulating a poor portfolio o f loans. In Mongolia, the Bank supported liquidation and privatization o f public banks while concurrently helping the government to establish a new state owned commercial bank and a savings bank. In both Morocco and Cameroon, the Bank supported developing the post office as a lending agency at the same time it was encouraging privatization o f commercial banks. 5.15 On deposit insurance, the Bank has also sent mixed signals within a country: a sector report for Ukraine in FY95 recommended that creation o f a deposit insurance scheme should be an objective only for the long term, to be established only after other reforms were in place and the banks were strong enough to give such a scheme credibility, yet the introduction o f deposit insurance was a condition o f the FY99 FSAL. These inconsistencies may reflect disagreements within the Bank (which in turn reflect international disagreement) on good practice or on the appropriate approach in a given country, but they suggest the absence o f a coherent approach to financial sector development in a specific country. 5.16 In addition, the Bank supported the establishment o f stricter prudential regulations, which were followed by Bank funded LOC; although some o f the L O C involved non-bank financial intermediaries, there were no requirements for these intermediaries to meet any prudential regulations. In Kyrgyz Republic, for example, a special rural credit agency had no prudential requirements for participating in the Bank LOC, and in Russia, an enterprise restructuring project involved credit guarantees from commercial banks, with no eligibility requirements. The Bank could have used the L O C to reinforce the relevance and importance o f prudential norms, even if the intermediary was not formally considered a bank; by failing to make use o f them in its own lending, the Bank undermined i t s message that prudential regulations matter. Coherence of Bank approach to financial sector reforms across countries 5.17 Bank support has followed international norms and principles in support o f prudential regulations and banking supervision and, to a lesser extent, with respect to government control o f financial intermediaries (see Figure 3.3 for breakdown o f Bank lending by objective). These elements were central to most loans and other features such as improving the accounting and auditing frameworks, introducing or improving bankruptcy law, and ensuring the independence o f the supervisory authority were also frequently included in Bank loans addressing financial sector reforms. I n addition, financial sector ESW across countries i s characterized by a focus on similar issues. 5.18 There were, however, significant differences in the process o f reforms (how); sequencing (when), and the selection o f specific reforms, which cannot be explained by 28 initial conditions in the borrowing country, reform momentum, willingness and ability o f the government to address constraints, or the coverage by other donors.26 5.19 Bankprivatization. In ECA, although the Bank was consistent in recommending that if privatization was to be pursued, ownership should be concentrated in the hands o f strategic investors, and preferably reputable foreign banks, Bank lending in ECA, as elsewhere, did not always support this approach (for example, in Georgia, Uganda, Mozambique). Second, there were inconsistent approaches on whether to privatize or liquidate large state-owned banks, as well as on how quickly to proceed, even within ECA, where there was acknowledged urgency to reforming both the banking and enterprise sectors in the context o f transitioning to a market economy. In Azerbaijan, for example, the Bank recommended that any state bank not privatized within 18 months should be liquidated (except for the savings bank), while in Kazakhstan and Albania, the Bank called for a gradual approach to privatization, to be pursued only after sound regulations and strong banking supervision were in place.27 5.20 Payment systems. There i s wide agreement that an efficient, reliable payment systems, i s an important building block for financial sector development (paragraph 2.5). Over the ten year period under review, however, the Bank addressed issues o f payment systems in only 28 countries and 2 regional systems, with a total o f 43 lending operations (3 1 investment; 12 adjustment) and a relatively heavy concentration in E C A (14 countries, 2 1 operations). This limited involvement, particularly outside o f ECA, cannot be explained by the adequacy o f the systems in most o f these countries or by support from other donors, which would indicate little need for Bank assistance (see next paragraph). 5.21 Instead, support for improving payment systems came late in the cycle o f Bank assistance in a number o f countries. Improvements to Pakistan’s payment systems, for example, were addressed for the first time in FY03, although the Bank has been supporting financial sector reforms in the country since 1989. In Uganda, the Bank first addressed payments system upgrade in FY99, although it has been involved in financial sector reforms since the early part o f the decade; in Albania and Mongolia, the pattern i s similar, where the Bank supported reforms in FY93 (Albania) and FY97 (Mongolia) but did not finance investments in payment systems in either country until some six years later. In a number o f countries, payment systems improvement appears to be (appropriately) the focus o f reform efforts when there i s little or limited agreement on other, more politically charged reforms: the Bank made such loans in Algeria, Angola, China, S r i Lanka, Tajikistan, Ukraine, Uzbekistan, and Vietnam, where Bank support for 26Clearly a one-size f i t s all approach i s inappropriate; at the same time, however, the specific reforms supported should be the result o f a combination o f analysis o f country conditions, what other donors are doing, and a consistent view o f the critical elements needed for an efficient, effective financial system. 27The Bank’s FY95 economic report o n Albania stated, “ASthe banks become healthier and more experienced in commercial banking practices, plans for the restructuring and eventual privatization o f the state banks should be developed” (see page 59). T h i s recommendation may have been based on perceptions o f limited country commitment to privatization at the time, although i t i s unclear why gradual privatization was appropriate for Albania and Kazakhstan but not for Azerbaijan. 29 bank privatization (in systems dominated by state banks), for example, was not on the agenda. Table 5.2: Bank loans supporting deposit insurance, by FY 5.22 Deposit insurance schemes. By contrast with FY 93 94 95 96 97 98 99 00 01 02 03 payment system strengthening No.ofloans 1 1 2 5 3 10 12 7 7 8 4 where there is widespread agreement on its importance, deposit insurance is a more controversial area (paragraph 2.12). Yet the Bank has supported deposit insurance schemes in 35 countries and 60 operations (mostly adjustment), considerably more than for improving payment systems (Table 5.2). Most o f these loans aimed to improve other components o f a financial safety net: 28 support for the supervisory agency and the prudential framework, and restructuring and/or privatization o f banks. 5.23 Eighty percent o f the operations involving deposit insurance schemes were approved in FY98 or later. The timing o f the support coincided with either crisis (all o f the EAP countries, six in LCR, and four in ECA) or with h t u r e prospect o f accession to the European Union, where deposit insurance systems have been mandatory since 1994. Although the timing o f setting up deposit insurance has not been optimal,29governments have apparently been more interested in establishing them in times o f systemic banking crisis, with i t s attendant political and social costs. Nevertheless, given the on-going debate within the Bank on the impact on a financial system o f deposit insurance schemes, the extent o f Bank support for such schemes i s somewhat surprising. 5.24 Capital market development. Finally, the Bank has had an ad hoc approach to the priority that capital market development should be given in financial sector reforms, and under what country conditions it i s appropriate to support capital markets. This i s perhaps reflective o f the differing views within the Bank on this issue and the priority given to i t by governments. Some 48 Bank operations (21 adjustments; 27 investments) in 30 countries have supported capital market development, half o f which were approved over a four year period (FY95-99), and concentrated in E C A and LCR. O f the 30 countries where the Bank supported capital markets, most (19) are middle income countries, but a number o f the countries have very small economies and financial systems (e.g., Bolivia, Georgia, Guyana, Kyrgyz Republic, Lesotho, Mali, and Mongolia), with little clear potential even in the medium term for capital market development. I t i s in this area in particular that the absence o f guidelines or good practice on the relevance and priority o f capital market development, and under which country conditions, is most evident. 5.25 I nconclusion, the Bank has followed good practice where there is widespread agreement o n the importance and the nature o f reforms, with some exceptions. In addition, within many countries, support for specific reforms has been consistent, although there are exceptions to this as well. Across countries there i s a much wider ’ 28A country’s financial safety net consists o f a lender o f last resort, insolvency regulations, a framework o f rudential regulations and supervision, and a deposit insurance scheme. Garcia (2001) says that a deposit insurance scheme should be installed only in a country with a sound banking system and other components o f a safety net that are hnctioning well. 30 variation o f approach, particularly in support for payment systems, deposit insurance - schemes, and capital market reform. The combination o f on-going debates within the Bank (e.g., whether and how to support deposit insurance schemes), absence o f “good policy” notes (paragraph 2.19), and the decentralized nature o f Bank operations have all contributed to a situation in which the Bank speaks with many voices on important matters o f financial sector policy.3o 6. Bank assistance: outcomes o f loans and credits Overview closed and been rated by OED. By 100% value, US$35 billion out o f a total o f 00% f z 80% US$46 billion had been rated.31 4 50% 6.2 For financial sector adjustment 40% operations, outcome ratings are better f than overall adjustment ratings, both by 10% 0% number and by commitment amount. For Financial Sector TA and Investments Loans Financial Sector Adjustment Loans financial sector TA operations, outcomes All Adjustment Loans(exc1uding financial sector) All Investment Loans(exclud1ng TA) are similar to outcomes o f other TA 6.3 Because closed multi-sector loans addressing financial reforms outnumber those categorized as finance (130 closed and rated multi-sector versus 60 closed and rated financial sector adjustment loans), OED rated the financial sector components o f multi- sector 10an.s.~’ This provided a more complete database o f ratings o f financial sector ~ 30 The Bank’s research department has been active in exploring a number o f subjects and producing articles o n experience with different elements o f a financial system, o r o f reforms (for example, deposit insurance; asset management companies), but research findings do n o t emanate f r o m the same authorizing environment nor bear the same weight as would good practice notes f r o m the network anchor. 31 Thischapter excludes outcomes o f LOC, which are analyzed i n OED, 2005. I t also excludes loans focused solely o n pension reform, which are the subject o f an on-going OED review. 32 Of the 99 component ratings, 12 came f r o m implementation completion reports, validated by a n internal OED review; 27 came f r o m OED’s assessment reports. O f the remaining 60 rated by an OED desk review, outcomes o f 9 components were rated better than the overall project outcome rating and 11 were rated 31 components and allowed more robust testing o f trends over time and characteristics that might be associated with success, as discussed in the next section. Table 6.1: Outcome ratings of financial sector Financial sector loans versus components o f lending and components, FY93-03 multi-sector loans Percent satisfactory 6.4 W h e n financial sector loans and financial Financial sector + components 75 All Bank lending 79 sector components are combined, the rate o f satisfactory outcomes drops below outcomes o f all Adjustment: financial + components 75 other Bank lending (Table 6. l), driven mainly by all other adjustment 79 the poor outcomes o f the components. Outcomes o f TA: financial + components 78 the financial components in multi-sector adjustment all other TA 80 loans have only a 69 percent satisfactory rating (by For further details, see Annex 2, Table 2. number), which is some twenty percentage points lower, than outcomes o f adjustment loans under the Figure 6.2: Outcome ratings and sector financial sector board. Among TA loans, outcomes for lassification. bv number. ,FY93-03 components o f multi-sector loans are slightly lower than for I I financial sector loans (Figure 6.2 and &ex 2, Table 2). 90% 80% $ 70% 8 60% 6.5 These results cannot be explained by differences in 50% the reforms or conditionality, as they were similar in 3 40% 5 30% 1 financial sector and multi-sector loans; nor do the financial 20% 10% sector loans tend to be made in non-crisis situations, while 0% multi-sector loans are for the crisis situations - there i s a Adjustment mixture o f both types o f loans in crisis and non-crisis Financjal Sector Loans Financlal Components of Multisector Loans lending. But the poorer outcomes for multi-sector lending may be the result o f other country characteristics - if multi-sector loans are clustered in smaller countries with poorer institutional and policy . capacities - and lower incomes, this could explain the poorer results. To test this, OED Figure 6.3: Outcome ratings by sector and examined outcomes in countries with different ratings on by CPIA rating, by number, FY93-03 Country Policy and Institutional Assessment (CPIA) and I 100% , different per capita income levels.33 The results in Figure 1 1 6.3 show that even among countries with similar l o w * 80% E 70% 00% CPIA ratings, outcomes o f financial sector components in 2 60% 5 50% multi-sector loans are much lower (by about 20 40% 6 30% + percentage points) than outcomes o f financial sector loans 20% and among higher CPIA countries, the difference i s 13 10% 0% percentage points (for details, see Annex 2, Tables 2-4). LOW CPlA High CPL4 These differences persist between countries categorized w Financial Sector Loans Bl Financial Components of Muitsector Loans ’ worse. For all 99 component ratings, the net “downgrade” relative to overall project outcome ratings was 3 ercent. The component ratings by source are in Annex 2, Table 1. The CPIA (Country Policy and Institutional Assessment) i s a composite indicator that measures the capacity o f a country to manage i t s resources efficiently and carry out policy reforms, comprised o f an unweighted average o f 20 indicators, o f which only 2 are related to the financial sector. The degree o f circularity in this analysis i s therefore quite modest. 32 by income level as well, with the largest difference in outcome ratings among middle income countries, where component outcomes were 23 percentage points lower than those o f financial sector adjustment loans. Most o f these results are statistically significant. 6.6 These findings suggest that financial sector reforms under the control o f financial sector staff in the Financial Sector Network have better outcomes than such reforms under other Networks. This may be the result o f having specialized Bank staff prepare the loans; the review process within the Network prior to loan approval; or the quality o f the Bank's supervision, all o f which may focus more resources and more effort on pursuing reforms. Better outcomes may derive from factors on the Borrower's side, such as having specialist counterparts from Central Banks or Ministries o f Finance, who may also focus more intently on financial sector issues than in situations where reforms cover many sectors and ministries. These findings could be a proxy for stronger ownership: when reforms are concentrated in a sector, the extent o f government commitment to reforms in that sector may be more apparent than when reforms are dispersed across a number o f sectors and ministries. Whatever the reason behind the differences in outcomes, these findings suggest that if financial sector reforms are considered a priority by client country officials, and are to be supported by Bank lending, the financial sector board should be closely involved in quality control at the preparation stage; counterparts from finance in the client country (from the ministry or the central supervisory authority) should be closely involved; and financial sector specialists should be assigned to supervise the component. Figure 6.4: Outcome ratings and country Country characteristics characteristics. FY93-03 6.7 N o t surprisingly, country characteristics 100% 80% 1 1 1 802 mattered for outcomes, particularly when measured by 70% 2002 or 2003 characteristics (Figure 6.4 and Annex 2, 60% 5096 Table 3). In addition, outcomes o f Bank lending for 40% 30% financial reforms in transition countries were higher 20% 10% than in other countries; and when the transition I 0% Income. CPW' Transition countries are examined separately, the differences in 'Excluding Transtion outcomes between the remaining l o w and middle income countries are significantly larger. 6.8 For CPIA ratings (available for most countries in the sample only as far back as 1996), the pattem i s similar. The difference in outcome ratings between l o w and high CPIA countries is 16 percentage points (Figure 6.4). The relatively good outcomes in transition countries are probably due to the strong reform movements in many o f them. For almost h a l f o f the transition countries (the Baltic and Central European countries), the incentive o f accession or association to the European Union may have driven both the direction and speed o f reforms, and the financial sector reforms were part o f a larger program o f reforms aimed at enterprises as well, which may have contributed to better outcomes. 33 Trends and sequence o f adjustment lending 6.9 The Bank has been lending for policy reforms for almost 20 years and many lessons have emerged, e.g., on the importance o f government commitment; on keeping the design o f the adjustment loans relatively simple, on setting realistic timeframes for conditionality. In addition, over the period under review, many countries had more than one adjustment loan, so it could therefore be expected that outcomes o f adjustment lending for financial sector reforms would show improvement over time. 6.10 Outcomes o f loans approved in the second half o f the period, however, are not much higher than in the first half. By contrast, adjustment loans that built on a prior loan for financial reforms had better outcomes than the first loan (Table 6.2).34 This finding may be the result o f perseverance by the Bank, or as likely, a crisis or near-crisis in the banking sector. Governments that were initially reluctant reformers became more convinced o f the need (or were forced out o f office) once they faced either crisis: near- crisis, or widespread banking insolvencies, which happened at different times in different countries over the period under review. Following the (near) crises, the Bank was often able to engage in more active dialogue on the financial sector. There are countries, however, where crisis or near crisis had little impact on Government’s views toward govemance reforms (Box 6.1), and a third group o f countries, such as Latvia and Lithuania, where the government undertook reforms, particularly bank privatization, in the absence o f or prior to a (near) crisis. Table 6.2: Outcomes ratings and timing, sequence o f adjustment loans Timing, Sequence Number o f Number o f satisfactory Percent loans loans satisfactory Year of approval FY93-FY97 74 53 72 FY98-FY03 68 53 78 Loan sequence First loan addressing financial reforms 51 35 67* Not first loan addressing financial 91 71 78* * Significant at the 10 percent level Does the provision of technical assistance help outcomes? 6.1 1 Conditionality in adjustment loans aimed at the financial sector often involves highly technical issues, such as passage o f banking laws, stricter prudential regulations, and privatization o f banks. If a country doesn’t have the relevant in-house experience or expertise to carry out the reforms, it i s reasonable to expect that the provision o f technical assistance (TA) may be the difference between timely and successful implementation and failure. The analysis that follows compares outcomes o f Bank loans for financial sector reforms that were accompanied by Bank-financed TA loans with outcomes where no Bank funding for TA was provided. An important caveat o f this analysis i s that TA may have been provided by other donors, and thus the results here may obscure the importance o f timely assistance from other sources. 34OED examined the period prior to FY93 for adjustment loans addressing financial sector reforms; investment lending with reforms were n o t captured in this analysis. 34 Box 6.1: What a difference a (near) crisis (sometimes) makes Financial crisis, near crisis, or widespread insolvency was often followed by a change in government or, at least a change in government’s willingness to undertake reforms in i t s financial sector. In Albania, for example, the Bank h a d supported reforms through two adjustments and one TA credit which did not address underlying governance issues. I t was only after the widespread pyramid crisis in F Y 9 7 followed by c i v i l unrest that the new government was ready t o engage in real reforms. The Bank supported them with three adjustment and two credits which aimed t o resolve the pyramid scheme fallout; liquidate or privatize banks, and establish an asset management company t o handle b a d debts. By mid-2004, a l l banks had been privatized and the banking system was fairly healthy. In the Slovak Republic, two adjustment loans similarly made little progress; and an E F S A L planned for 1995 was postponed because o f lack o f Government interest. After a near financial crisis in 1999, the new government was ready to address fundamental problems in the sector, supported by an FYOl loan. Other examples o f this were found in Brazil, Burkina Faso, Cameroon, Croatia, and Romania. M a n y countries that had full-blown crises (Chapter 7) were also reluctant t o reform their financial sectors prior to crisis. Thailand, for example, had n o Bank lending and n o dialogue with the Bank o n financial sector issues prior t o its crisis, and in Korea and Indonesia, Bank lending was limited to lines o f credit. Argentina agreed t o privatize provincial banks with Bank support only after the banks became a serious drain o n the provincial governments’ budgets in the early 1990s; there has been notably less interest o n the part o f the authorities in Argentina in privatizing national banks. Only after i t s 1999 crisis did Colombia begin t o consolidate the weak cooperative system and to address privatizing its national banks. In Mexico, the Bank had supported early and only moderately successful banking reforms prior t o the Tequila crisis o f 1994, but thereafter M e x i c o re-nationalized (with Bank support) and re-privatized i t s banks, allowing foreign banks to participate. Similarly, in Turkey, outcomes o f Bank’s adjustment lending in the financial sector were unsatisfactory until the crisis in 2000. ...but not always Mongolia began i t s transition t o a market based economy i n 1991, and had banking crises in 1992, 1994, 1996, and 1998. The F Y 9 7 F S A C and TA credit supported liquidation o f two banks and establishment o f two n e w public banks, debt recovery mechanisms, and the establishment o f a credit information bureau, but n o change in governance. Only in the FYOO F S A C and TA credit did the government agree t o divest one state bank and to put nplace a clear exit policy for troubled banks. In L a o PDR, the F Y 9 6 S A C 1 i 11, aiming t o strengthen the prudential framework and accounting o f banks and carry out audits o f state owned banks, was considered unsatisfactory o n a l l components, and even after state banks reached total insolvency at the end o f the 199Os, the government was willing only t o restructure the banks, supported by an FY02 F S A C and TA credit, with n o change in governance. In Algeria, the state banks served for years as channels for Treasury support t o unprofitable state enterprises and, according to detailed diagnosis i n the early 199Os, had reached a level o f insolvency that implied negative capital. T w o adjustment loans were approved (FY95, FY96) that included restructuring state banks and introducing private capital. The reforms were attempted at a very difficult political juncture in Algeria and little progress was made. No further adjustment lending for financial sector reforms has been made to Algeria since FY96. 6.12 The results o f the analysis show no difference in outcomes overall between adjustment loans that had associated TA loans and those that did not. I t might be expected that TA would make more o f a difference for lower income countries than in middle income countries, but outcomes on adjustment lending are similar here as well, whether or not transition countries are examined separately (Figure 6.5). In l o w CPIA countries, however, outcomes are better in l o w CPIA countries when a TA loan accompanies the adjustment loan (details are in Annex 2, Table 4). To the extent that the CPIA rating i s a good proxy for institutional capacity, this finding makes sense: where capacity i s limited, the provision o f TA has a measurable value added for the outcomes o f adjustment lending. 35 6.13 By contrast, for the higher CPLA Figure 6.5: Outcomes of adjustment loans with countries, with better institutional capacity and and without technical assistance policies, the difference in outcomes is the opposite o f what would be expected, that is, 1 100% 80% , I outcomes are better when there is no TA loan in 80% 70% thepicture. This finding hold whether 1996 or 8 60% 2003 CPIA measures are used, whether or not 50% 5 40% transition countries are included, and whether “high” CPLA i s defined as over 3.0 or 3.5 10% 0% (although in none o f these i s the difference A II CPIA‘ LOW High CPIA‘ Transition statistically significant). This suggests that the *ExcIudmg TranSitlOn Iw lth TA E¶ w l t h o u t TA provision o f TA by the Bank has little positive impact on outcomes o f adjustment lending for financial reforms and may even be a signal that the adjustment loan is quite risky, although these findings may simply reflect the failure to measure the TA provided by other donors. Or it may be that in countries with better institutional capacity, the provision o f a TA loan by the Bank is an attempt to address some other, non-technical constraint, such as lack o f widespread ownership or presence o f political obstacles, in the hope that the presence o f outside technical specialists may be able to overcome these obstacles. Whatever the explanation, these results suggest that in high capacity countries, the provision o f technical assistance in conjunction with an adjustment loan does not appear to carry much value added for the achievement o f the objectives o f the adjustment loan, although it may add value for other reasons (such as establishing or improving a payment system). 7. Bank support to countries experiencing a crisis35 Overview 7.1 M u c h has been written o n the financial crises that have occurred in the developing world in the last decade - their causes, their costs, their consequences, and their aftermaths. The causes have been complex and varied across countries. The costs have been high, in terms o f both the increased fiscal burden (as high as 55 percent o f GDP in Indonesia to recapitalize the banks) and the drop in output, not only in the year o f the crisis but in subsequent years. The consequences in terms o f corporate bankruptcies, unemployment, increased poverty, access to international capital markets, and political and social upheaval have been serious; and recovery from the crisis has taken many years. The financial sectors in some o f these countries, such as Ecuador, Indonesia, Russia, and Thailand, have arguably not yet h l l y recovered. 7.2 There i s no agreed definition o f what constitutes a country in crisis. The one used here i s a country that experienced both a banking crisis and a macroeconomic crisis, either simultaneously or in quick ~ u c c e s s i o n The . ~ ~ run on banks resulted in illiquidity 35This chapter is based o n the background paper by M i l l a r d L o n g (2003b) and OED’s project assessments. 36M a n y countries h a d one sort o f crisis but n o t the other. Brazil, for example, had a macroeconomic crisis that did not result in a banking crisis. Caprio and Klingebiel(2003) list 83 countries that had technically 36 and required govemment action, and the macroeconomic crisis led to a large devaluation; the combination o f events created problems for the corporate sector, which could no longer service i t s loans, creating further pressure o n the banks and affecting outputs and investments. Growth dropped and poverty increased. 7.3 Using this definition, fifteen countries in three Regions experienced crises over the FY93-03 period. The 1994 Tequila crisis in Mexico spread to Argentina; and the 1997 crisis that started in Thailand quickly spread to Korea and Indonesia and then to Russia and Bulgaria; Bolivia and Ecuador had crises in 1998 and 1999, and in 2000-02, Argentina, Colombia, Guatemala, Jamaica, Turkey, and Uruguay had crises.37 7.4 The rationale for assessing Bank lending to these countries separately from other financial sector support is twofold: one i s the importance o f this lending. Financial sector loans to countries experiencing or following a crisis represents over 50 percent o f total financial sector lending over the period (US$12 billion out o f US$21 billion); all loans, including multi-sector, to these countries that include financial sector reforms also account for almost 50 percent o f total loan amounts approved by the Bank that had any financial sector components (US$21 billion out o f US$46 billi~n).~’ Thus crisis lending looms large in the Bank’s portfolio o f financial sector support.3g 7.5 The second reason for considering crisis lending separately i s that such lending i s usually prepared and approved quickly, under emergency situations, in the context o f large financial aid packages put together by intemational financial institutions (IFI). I t may not benefit from prior diagnostic work on the sector or from a close dialogue with government o n reforms. On the other hand, governments that were reluctant reformers prior to crisis may become more willing adherents. All o f these factors may affect, in different directions, the nature and quality o f the reforms undertaken, and the outcomes in ways that do not apply, or apply to a much lower degree, under less urgent conditions. 7.6 The next section reviews the Bank’s record on predicting crisis and assessing vulnerability; the following section reviews the Bank’s response to the crises, and how i t s assistance fit into the larger intemational rescue efforts. The chapter then discusses the objectives and outcomes o f loans that focused o n financial sector reforms. Finally, the chapter examines cooperation with the IMF during crisis, whether a centralized approach within the Bank worked well and i s sustainable for responding to crisis, and whether the Bank’s organization is adequately structured to handle crises. The chapter concludes with lessons drawn from the experience o f the past decade on dealing with crises. insolvent financial systems between 1990 and 2002, and thus were labeled as a systemic o r borderline banking crisis country. Unless these countries also experienced a macroeconomic crisis, they are not discussed in this chapter. 37 Venezuela h a d a crisis, but n o Bank lending, and i s n o t discussed here. 38 These figures exclude LOC. 39 Other adjustment loans made t o countries experiencing crisis that did n o t involve the financial sector are not discussed here. In addition, loans reviewed here were approved within two years o f the crisis. 37 Did the Bank anticipate the crisis? 7.7 All o f the countries that had financial crises over this period had systemically weak financial systems, but not all countries with weak financial systems have had crises. T w o other elements have been involved in most o f these countries examined in this review. One was an economic or political shock (deterioration in terms o f trade; contagion from other crises; assassination o f a presidential candidate) that led to an initial run on the banks. The second was a government response that the markets deemed inadequate, which in turn led to a larger run and crisis. W h i l e it was feasible for the Bank to analyze the weaknesses o f financial systems in most countries, it was and i s not possible to predict shocks, nor in most cases, government’s response or the reaction o f market participants. Thus, it would be unrealistic to expect the Bank, or any other institution, no matter how well-informed, to predict timing o f crises.4o It i s reasonable, however, to expect the Bank to assess the vulnerability o f its clients to crisis and therefore to be prepared to respond quickly once a crisis hits. 7.8 In a number o f the countries under review here, however, the Bank was not well- informed, in part because it had not been active in the financial sector in the years leading up the crisis. In Mexico, after supporting financial liberalization in 1989-90, the Bank considered the reforms successful, and the Bank’s dialogue lapsed. As a result, the Bank had little recent work to draw upon prior to the crisis. An internal high-level review o f the Bank’s handling o f i t s post-crisis assistance to Mexico concluded that given the warning signs o f potential trouble in the banking system - a lending boom, a rapid increase in non-performing loans (NPLs), a weak legal and regulatory framework for banks - the Bank should have been better prepared to respond to a crisis. The OED Country Assistance Evaluation on Mexico (OED, 2001b) also noted, “The inadequate high-level attention to the financial system during 1992-93 was by far the most serious omission in the Bank’s agenda in Mexico during the period under review.” 7.9 In Thailand, the Bank’s 1990 sector report o n the financial sector was the most recent analysis prior to the 1997 crisis, although there were several economic reports produced between 1994 and 1997 which did not mention the financial sector. In addition, the Bank had not made any financial sector loans in many years prior to the crisis. I n Korea, the Bank had produced a report on the financial sector in 1993, at the request o f the government, but had not had a dialogue since then, except for supervision o f an FY94 line o f credit. In spite o f warning signs o f increasing vulnerability in these two countries, the Bank had little current financial sector analysis relevant to the crises that hit both o f them. In Indonesia, the Bank had an active line o f credit and had produced a financial sector review in 1996 that identified weaknesses in the financial sector, but the government was not interested in adjustment lending to address them prior to the crisis. By contrast, the Bank had been heavily involved in adjustment and/or investment lending in Argentina, Russia, and Uruguay and was both aware o f and trying to address weaknesses in the financial systems. 40A 2003 analysis by the US. General Accounting Office concluded m u c h the same about I M F ’ s ability t o anticipate crises. 38 7.10 The degree to which the Bank’s assessments found their way into internal papers, formal sector work, and lending documents varied in candor, according to the primary audience for the analy~is.~’ In Indonesia, for example, a financial sector report that was discussed within the Bank but not formally with government raised concerns about the health and vulnerability o f the financial system and the need to introduce reforms (these issues were discussed, however, at meetings between the Bank and the Central Bank o f Indonesia). At the 1995, 1996, and 1997 meetings o f the Consultative Group for Indonesia prior to the crisis, the Bank pointed out the risks to the macro-economy o f the financial sector’s vulnerability to shocks. Yet the assistance strategy for Indonesia discussed at the Bank’s Board in the summer o f 1997 was sanguine about Indonesia’s risks. In Turkey, although the Bank was well aware o f the fragile situation o f the banks in Turkey and the pressures on them, the formal country economic report o f September 2000 and the country strategy presented an optimistic scenario for the reforms and likelihood o f success.42 7.1 1 Two reasons cited by proponents o f providing a sanguine treatment in public documents o f the vulnerability o f a country’s financial system to crisis are: (i)publicizing high vulnerability in the financial sector o f a client country could precipitate a crisis that might not occur otherwise; and ( ii ) if client countries know that the Bank will make i t s assessments public, it would be unwilling to provide the confidential information required to make the assessments. OED disagrees with both o f these arguments. 7.12 First, assessing vulnerability to crisis i s not the same as predicting a crisis. The Bank has identified high NPLs, weak supervision, poor governance, concentrated risks, rapid credit growth, poor accounting, and other factors associated with vulnerability in many countries that have not had crises. I t i s possible to use available information to assign risk categories to financial systems without precipitating a run o n the banks. Second, governments have allowed information on these factors to be available in Bank documents as well as to other market participants (like rating agencies) for years; pulling this information together into an assessment o f r i s k would be n o more revealing than what is currently available in the public domain. On the other hand, drawing conclusions from publicly available information on risks could help both the Bank and the client government focus on contingency planning. 7.13 Since the 1997 Asian crisis, the Bank and IMF have started ajoint program o f financial sector assessments (FSAP) that i s intended to identify more systematically the resilience o f the financial systems to r i s k and the adequacy o f the supervisory and prudential framework. A s o f July 2004 more than 80 assessments are completed or on- going. The details o f the assessment are confidential, but both institutions produce summary assessments to their Boards. On the basis o f these summary assessments, the Bank could develop risk categories for financial systems, which would signal to the Bank, other donors, and stakeholders as well as the government (if it hadn’t gotten the 41 Internal documents the most frank, documents t o the Board the least, and sector w o r k somewhere in between (the degree o f candor may depend o n whether the documents are disclosed t o the public). 42 The President’s report o f the FSAL h a d an underlined section that noted the risk o f a banking crisis if weaknesses were n o t addressed. 39 message from the FSAP itself) the priority that should be given to financial sector reforms and resources devoted to contingency planning (what the best course o f action would be if a crisis were to occur). I t would also provide a more candid basis for assessing whether proposed assistance programs are focusing o n the most relevant issues. Bank response to crisis 7.14 The Bank made post-crisis loans to all but two o f the fifteen crisis countries. In Russia, the Bank approved a large Structural Adjustment Loans (SALS) for US$1.5 billion as part of a US$23 billion rescue package in the month before the crisis, to try to avert one. The Bank did not lend to Venezuela. In the other countries, the Bank was part o f a larger rescue effort by the intemational financial institutions (IFI) and G-7 countries (Table 7. l), and the amounts pledged and lent by the Bank were relatively small compared to the IMF. In Mexico, for example, following the 1994 Tequila crisis, the Bank committed roughly 4 percent o f the US$49 billion pledged by the intemational community; the IMF committed 35 percent. I nThailand, the Bank lent a total o f US$2.1 billion out o f an IF1package o f US$17 billion; the IMF, US$4 billion. In Korea, although Bank lending reached a record high o f US$7 billion lent over six months to one country, i t was a modest portion o f the US$58 billion emergency package put together by the IF1(although the full amount never materialized - see note to Table 7.1); the IMF’s share was US$21 billion. I nArgentina, in the third round o f crisis support, the Bank’s lending was under 5 percent o f the total package, compared to the IMF’s share o f over 50 percent. Table 7.1: Response to crisis: international rescue efforts and Bank response IMF Bank Rescue package As percent of country’s Standby or EFF actual commitments US$ billion* GDP * * US$ billion US$ billion Argentina, 1995-96 3.7 1 1.9 1.66 Argentina, 1999 8.3 3 2.8 3.03 Argentina, 2001 40.0 15 22.7 1.85 Ecuador 1999-2000 2.0 12 0.3 0.43 Indonesia, 1997-99 38.0 18 10.0 2.45 Jamaica 1996-1997 2.0 33 0.0 0.23 Korea, 1997-98 58.0 12 21.0 7.05 Mexico, 1995 48.8 17 17.8 1.95 Russia, 1998 22.5 8 12.5 1.50 Thailand, 1997-99 17.2 11 4.0 2.08 Turkey, 2001-2003 22.2 15 19.0 3.23 Uruguay, 2002 3.3 27 2.2 0.40 * Announced; full amount includes bilateral pledges, which were not typically committed; for example, the U S 5 8 billion for Korea included U S 2 0 billion “second line of defense” from bilaterals that was never used. **GDP in first year o f crisis; a more appropriate measure might be rescue package as percent of capital outflow, but this information was not readily available for most countries. 7.15 The Bank often pledged lending amounts prior to any dialogue with the government concerned. Thus the Bank’s intentions on both timing and amount o f hnding were publicly announced, without benefit o f discussion o n the scope o f the reforms or negotiations with the governments. The first adjustment loan approved immediately after the crisis was often made under emergency and difficult conditions, where speed was essential and the need for comprehensive understanding o f the issues or 40 government’s capacity to address them, secondary. These factors provide perspective on the ensuing discussion o f the outcomes o f Bank lending in crisis. 7.16 On the other hand, many o f the governments in these 15 countries had been unwilling to undertake reforms o f their financial sectors. Nine o f the countries had had no Bank adjustment lending, or none within the previous 10 years prior to the crisis in support o f financial sector reforms. The crises changed either the governments themselves or their attitudes about reform, or both, thus underlining again the oft- repeated finding that government ownership i s critical to successful pursuit o f reforms. Thirteen o f the countries agreed to address financial sector problems following the crisis (and Russiajust before the crisis), and o f these, seven countries also accepted TA loans accompanying the adjustment loans to help implement the reforms. The countries and loans containing financial sector reforms are listed in Table 7.2. Objectives and design of the loans 7.17 Although all the loans were timed and sized to address liquidity problems, to try to contain the currency runs, and to restore market confidence, the loans also addressed underlying structural problems, particularly in the banking and corporate sectors. The loans included analysis o f banks’ financial condition, establishment o f asset management companies and/or deposit insurance institutions to take over troubled financial institutions, restructure them, and re-privatize them and dispose o f loans and other assets, and establishment or support to corporate bankruptcy and restructuring. Other reforms addressed fundamental legal and regulatory issues, banking supervision, and accounting (Box 7.1). I nother words, the reforms supported under these crisis adjustment loans were very similar in nature and scope to the financial sector reforms discussed previously in this review, but many o f them were prepared under emergency conditions, and some without benefit o f recent diagnostic work or extensive dialogue with the government. 7.18 The TA loans were often also prepared quickly; in Bolivia, Indonesia, and Thailand, they preceded the adjustment loans. I nthe absence o f detailed knowledge about priorities and local capacity to implement quickly neededreforms, these TA loans were appropriately flexibly designed, to adjust to the circumstances as they developed. At the same time, several o f them suffered during the early years o f implementation from inadequate attention to “mundane” issues such as Bank guidelines on procurement and on hiring consultants, and experienced delays which were all the more frustrating in a situation where speed was critical to stem the bankruptcies and further deterioration in the economy. 41 Table 7.2: List of Crisis Loans with Financial sector components Country Loan Commitment Approval Loan Name Type Amount (US% m) FY Outcome Provincial Bank Privatization Argentina SAL 500.0 1995 Satisfactory Argentina Bank Reform SAL 500.0 1996 Satisfactory Argentina Special Structural Adjustment Loan SSAL 2525.3 1999 Unsatisfactory Argentina Special Repurchase Support Facility SSAL 505.1 1999 Highly Unsatisfactory Bolivia Regulatory Reform Sector Adjustment Credit SAL 40.0 1999 Satisfactory Bolivia Regulatory Reform and Privatization TA 20.0 1998 Active Bulgaria Rehabilitation SAL 30.0 1997 Unsatisfactory* Bulgaria Financial and Enterprise Sector Adjustment Loan FESAL 100.0 1998 Satisfactory* . Bulgaria Critical Imports Rehabilitation SAL 40.0 1997 Satisfactory* Colombia Financial Sector Adjustment Loan FSAL 505.6 2000 Moderately Sat Colombia Programmatic Financial Sector Adjustment Loan PSAL 150.0 2003 Satisfactory Ecuador Financial Sector Technical Assistance TA 10.0 2000 Unsatisfactory Ecuador Structural Adjustment Loan SAL 151.5 2000 Unsatisfactory Guatemala Financial Sector Adjustment Loan SAL 150.0 2002 Active Guatemala G T Financial Sector T A Loan TA 5.0 2002 Active Indonesia Banking Reform Assistance TA 20.0 1998 Unsatisfactory* Indonesia Policy Reform Support (PRSL I) SAL 1000.0 1999 Moderately Unsat* Indonesia Second Policy Reform Support (PRSL 11) SAL 500.0 1999 Moderately Unsat* Jamaica Bank Restructuring & Debt Management PSAL 75.0 2001 Satisfactory Jamaica Bank Restructuring & Debt Management I1 PSAL 75.0 2003 Moderately Sat Jamaica Jamaica Emergency Recovery Loan SAL 75.0 2002 Moderately Sat Korea, Rep Structural Adjustment SAL 2000.0 1998 Satisfactory Korea, Rep Structural Adjustment I1 SAL 2000.0 1999 Satisfactory Korea, Rep Financial and Corporate Restructuring Assistance TA 48.0 1999 Satisfactory Korea, Rep Economic Reconstruction SAL 3000.0 1998 Satisfactory Mexico Financial Sector Restructuring FSAL 1000.0 1995 Unsatisfactory Mexico Financial Sector Technical Assistance TA 37.4 1995 Satisfactory Russian Fed. Structural Adjustment Loan 111 SAL 1500.0 1999 Unsatisfactory Thailand Finance Companies Restructuring SAL 350.0 1998 Satisfactory** Thailand Financial Sector Implementation Assistance TA 15.0 1998 Satisfactory Thailand Economic And Financial Adjustment Loan EFAL 400.0 1999 Satisfactory** Thailand Economic and Financial Adjustment Loan I1 EFAL 600.0 1999 Moderately Sat** Turkey Financial Sector Adjustment Loan FSAL 771.8 2001 Moderately Sat Programmatic Financial and Public Sector Adjustment (PFPSAL I ) PSAL I Turkey SSAL 1100.0 2002 Satisfactory Second Programmatic Financial and Public Sector PSAL / Turkey Adjustment (PFPSAL 11) SSAL 1350.0 2002 Moderately Satisfactory Uruguay Structural Adjustment Loan SSAL 151.5 2003 Active Uruguay Special Structural Adjustment Loan SAL 101.o 2003 Active Total 37 operations 21408.2 Ratings as o f July 16,2004. * based on an OED assessment review. ** PPAR pending, ratings are not final. 42 Box 7.1: Objectives o f crisis lending: mostly ambitious reforms In Colombia, prior t o the 1999 crisis, the only Bank lending over the period for financial sector reforms was a TA loan that was not proceeding well. After the crisis, two adjustment loans (FYOO and FY03) addressed a large program o f bank restructuring, downsizing, liquidation, a n d o r privatization o f state banks, and closing or restructuring financial cooperatives, as w e l l as strengthening banking regulation and supervision (including anti- money laundering), deposit insurance, housing finance, insurance regulation, regulation and supervision o f capital markets, and Government debt and money markets. In Korea, there had been n o adjustment lending for financial sector reforms prior t o the crisis; the f i r s t adjustment loan after the 1997 crisis explicitly stated that the primary objectives o f the US$3 b i l l i o n loan (the largest ever approved by the Bank) were the provision o f emergency liquidity t o restore confidence in the economy and the development o f a framework for medium-term structural reform, which was to be pursued under subsequent adjustment lending. The two subsequent adjustment loans, for US$2 b i l l i o n each, and the accompanying TA loan (US$48 m i l l i o n approved, US$26 m i l l i o n disbursed) had extensive and detailed objectives, focused o n the financial sector, the corporate sector, the labor market and the social safety net, including improved transparency of Government support t o a l l financial institutions and corporations; capital market reform covering government auctions o f debt instruments; and improved competition policies. In Turkey, early B a n k support in the 1980s for financial sector reforms were n o t successful; but by the late 199Os, the Bank and Government had agreed o n a four pillar strategy for reforming the sector: creation o f a strong regulatory and supervisory agency for banks; aligning prudential regulations with international norms; strengthening the bank failure resolution agency (deposit insurance entity); and restructuring and privatizing state owned banks. The FYOl F S A L was approved prior to the crisis incorporating these pillars, but once the crisis hit, the F S A L was restructured t o allow for a series o f programmatic loans addressing these objectives. T w o programmatic FSALs were approved in subsequent years (FY02 and FY03), embracing these four reform areas and adding public sector reforms as well. Relevance o f objectives 7.19 OED assessments o f these loans did not question their relevance or design. As they addressed fundamental problems in the banking and corporate sectors, as well as legal and regulatory issues that were at the core o f the crisis, they were considered by OED to be highly relevant for the return to economic growth and stability. Nevertheless, many critics have questioned whether the Bank and other IF1should be providing large rescue packages and liquidity during crisis,43thereby creating perverse incentives. For lenders and investors, particularly from the foreign private sector, the rescue packages have not required them to “take a haircut”, that is, to forgive debt or negotiate write- downs, and thus, they have not borne the costs o f the risks o f committing funds to developing countries. For wealthy and well-connected domestic investors, particularly in the case o f Indonesia and Russia, the liquidity provided to banks enabled them to get their money out o f the country. And finally, for governments, because rescue packages were announced based o n the promise o f reform rather than after reforms have been undertaken, the large financial flows have been no guarantee that the reforms would be undertaken and may in fact have served as a disincentive to undertake them.44 43See, for example, Kenen (2002). 44The IMF examined the possibility o f “bailing-in” the private sector, t o make investors share losses in the case o f crisis. A p i l o t case was used in Ecuador, with mixed results, and the IFIs have since moved away 43 Achievement of objectives: below average 7.20 Given the high relevance o f the objectives o f the crisis loans, their outcomes were mainly a h c t i o n o f whether those objectives were achieved. As shown in Table 7.3, o f the 30 adjustment operations included in this review, 27 have closed and been rated, for a volume o f US$21 billion in gross commitments. Sixty-seven percent by number and 68 percent by net commitment amount had satisfactory outcomes, averages that are below all other adjustment lending and below financial sector lending, by number and by commitment amounts (Figure 7.1). O f the seven TA loans that were put in place, four have closed and been rated; o f these, three were rated ~ a t i s f a c t o r y . ~ ~ Table 7.3: Post-crisis adjustment operations with financial sector components Adjustment Technical Assistance Total Number Net Number Net Number Net o f loans Commitment o f loans Commitment o f loans Commitment Total, o f which: 30 21,253 7 155 37 2 1,408 Closed and rated, o f which: 27 17,931 5 77 32 18,008 Satisfactory 18 12,109 3 64 21 12,173 Percent satisfactorv 67 68 60 83 66 68 7.21 These outcomes are somewhat surprising, given the later finding (Chapter 8) that banking distress O r near Crises often fOcused Figure 7.1: Outcome of adjustment lending, crisis lending attention on the need for reforms that with financial sector components versus non-crisis authorities had been unwilling to tackle prior to the banking crisis. In a substantial number o f case study countries, outcomes o f Bank loans that came after the onset o f systemic banking problems had better results than Bank loans that preceded them. But these two sets o f findings are not mutually exclusive: in an emergency situation, when both significant resources and speed are by Roject b. bym - essential to stem the crisis, the ambitious I finacid Sector Adjustrnnt Lending (excluding crisis) objectives set out in Bank documents can I All Adjustrnnt W i n g (excluding financial sector and Crisis) often not be realized in the short timeframe 0m s Lwdng i o f a single adjustment operation. Korea i s a good example o f a series o f adjustment loans under crisis conditions that started out with a first adjustment loan that sought only to supply liquidity and establish the framework for fbture reforms; subsequent operations then relied on that framework to specify the reforms. 7.22 Most o f the initial adjustment loans that had unsatisfactory outcomes had ambitious and, in the end, unrealistic objectives (Box 7.2). This may be due to two from further consideration o f t h i s approach. International pressures on the Bank w i l l be strong to continue to participate in emergency rescue operations, and i t i s highly likely the Bank w i l l continue to play a role. 45 Out o f the twenty-nine closed and rated operations (including TA loans), nine have had assessment reports; three in Thailand are on-going and the ratings for these operations are not yet final. 44 factors: (i) an over-estimation o f government’s commitment to reform; and (ii) a perceived need to assure the Bank’s Board that the measures being undertaken are sufficiently deep and broad to justify such a large loan. ’ Box 7.2: Mixed outcomes In Argentina, the first round o f financial reforms after the 1994 crisis focused o n privatization o f provincial government-owned bank, which were a considerable fiscal drain o n the provinces. The outcomes of the loans involved were considered satisfactory, and the process was used in the Bank as example o f good practice. These reforms strengthened the banking sector, which may have helped Argentina withstand the 1998 shocks and, along with the IF1 lending, avert a crisis at that time (see and K i g u e l and Dujovne, 2003). The 1999 crisis was followed by two adjustment loans; one aimed t o strengthen banking supervision, reduce public involvement in banks by privatizing the mortgage bank, and improve regulation o f the capital market; and the second loan was t o provide liquidity to stem a banking run. B o t h o f these had unsatisfactory outcomes, mostly because the reforms implemented were necessary but ultimately insufficient t o redress the cumulative impact o f the series o f shocks that confronted Argentina in 1999 and 2000. Once the 2001/2002 crisis ensued i t quickly undermined the improvements in banking supervision and other reforms carried out under the projects. The impact o f the crisis was magnified by the Government’s decision t o concentrate crisis- related losses in the bank’s balance sheets through asymmetric pesification. In Indonesia, the series o f loans following the 1997 crisis addressed resolution o f the banking crisis and corporate restructuring. One TA loan and two adjustment loans were approved in support o f these objectives. The outcomes o f a l l three loans were considered by a n OED assessment t o be unsatisfactory. In the years immediately following the crisis, the government was not fully committed t o resolving the problems in the banking and corporate sectors, and the agency established to deal with the resolution and re-privatization of the banks and disposal o f assets made little progress. By 2003, the pace o f reforms had improved, but government s t i l l controlled over 60 percent o f the banking system, disposal o f assets moved slowly, and the banking sector remained vulnerable t o m e r shocks. 7.23 Within the Bank, the case o f Bank support to post-crisis Thailand was one o f the most contentious: the three adjustment loans to Thailand that contained financial sector reforms had been rated satisfactory by both the Region’s self-evaluations and OED’s desk reviews carried out shortly after the loans closed. But in the course o f this current OED assessment, i t became clear that many knowledgeable staff in the Bank (and outside observers) thought that the Bank’s assistance had been misguided and unsatisfactory, particularly with respect to i t s role in closing virtually all o f Thailand’s finance companies. As a result, OED undertook an assessment on three adjustment loans, the Finance Companies Restructuring (FY98), and the two Economic and Financial Adjustment Loans that followed (both, FY99); their report i s f ~ r t h c o m i n g . ~~ The experience in Thailand raises difficult questions about coordination and cooperation with the IMF, the subject o f the next section. Collaboration with the IMF 7.24 The division o f responsibility between the Bank and IMF o n financial sector work i s not clear. Pre-crisis diagnostics, monitoring, post-crisis lending, and TA all lie within the mandates o f both organizations. On substance, macroeconomic policies, fiscal, and 46 F o r t h i s assessment, OED hired two finance professors who had n o t been involved in the Asia crisis bailout, but were familiar with the issues. 45 financial areas are all covered, albeit to different degrees, by both organizations and are also the areas that, if weaknesses exist, can lead to crisis. The absence o f a clear division o f responsibilities has in some cases led to duplication o f efforts, confusion, and disagreements between the Bank and the Fund in post-crisis assistance efforts, in some cases in a public forum (Box 7.3), which only added to the uncertainties o f the crisis. 7.25 Since 1999, the Bank and IMF have been collaborating on the Financial Sector Assessment Program, assessing the vulnerability o f financial systems (this is the subject o f a separate OED review). In addition, the IMF has primary responsibility for on-going surveillance and for containing a crisis when one occurs; the Bank’s lending in a crisis i s contingent on the IMF’s having a program in place. Following the experience in the Asian crisis, the Bank and the IMF reached agreements in principle to improve c~llaboration.~’ The latter i s to focus on the immediate aftermath o f a crisis, on shorter- term actions to stem the crisis, such as devaluation o f the currency, government guarantees o f financial liabilities, and government intervention in specific institutions. The Bank i s to tackle the longer t e r m reconstruction o f the financial system, including bank restructuring and re-privatization, disposal o f banking assets, corporate restructuring, and improving the legal, regulatory, and accounting structures for both banking and corporations. 7.26 In practice, however, the boundary between these roles i s s t i l l not clear. The way in which the IMF oversees government’s actions to guarantee financial liabilities and intervene troubled financial institutions will have repercussions on subsequent restructuring efforts supported by the Bank. In addition, the roles o f regional development banks need to be coordinated. The most practical way o f approaching these issues may well be on a case-by-case basis, but from the outset o f a crisis, there needs to be agreement on basic approaches and the respective roles o f each institution to avoid the sorts o f problems that have complicated crisis management in the past. 1 Box 7.3: Coordination between Bank and IMF in crisis: needs improvement I In Mexico and Russia, the Bank and IMF disagreed o n the extent to which the currencies were over-valued. As a result, the Bank carried out i t s own macro-economic analysis. In Thailand and Indonesia, there was confusion over the division o f responsibilities in the early stages o f the crises among the Asian Development Bank, the World Bank, and the IMF. In Thailand, even after an agreement was reached that the IMF would focus o n banks and the Bank on finance companies, the agreement was not kept. In Indonesia, Bank staff did not have access to data concerning the financial sector obtained by the IMF, because the IMF was concerned about maintaining the confidentiality o f the information. And the public criticism by the Bank’s Chief Economist o f IMF’s approach in Indonesia drew wide press coverage, adding to the confusion in the midst o f an already difficult situation. An IMF evaluation o f i t s role in crisis (IMF, 2003) noted that the degree o f cooperation depended mostly on the personalities o f the mission leaders. Source: Long (2003b). 47 IMF 2001, IMF 2002a, and IMF 2002b. 46 I s a centralizedunit in the Bank specializing on crises effective and sustainable? 7.27 “Crisis tests government officials as few other events in their career will.. .few will have the prior experience to be well preparedto face it. The role o f multilateral institutions such as the Bank and the IMF in helping the authorities overcome a crisis, bringing to bear their extensive experience in other countries.. .can be pivotal in influencing the outcome.’y48As the Asian crisis unfolded, the Bank created a specialized central unit in January 1998, Special Financial Operations (SFO), to oversee the Bank’s assistance to the Asian crisis countries. The SFQ was generously funded, from a special budget allocation from the Bank, from a trust fund from industrialized countries, and from regular Bank budget connected with the Region’s TA loans. 7.28 Because i t s budget was substantial, the SFO was able to provide services to Thailand, Korea, and Indonesia that the Regional units were not in a position to finance. The SFO had full time staff based in the field over several years focused o n a single country, assuring both close contact with developments and continuity o f staff. The SFO was also able to hire people with specialized skills to support the different tasks involved in resolving financial crises. In Korea, the SFO hired a senior former government official who had good access to political decision makers, which was considered a key factor in the Bank’s ability to work at the political as well as the technical level. 7.29 O n the negative side, the newly hired staff o f the SFO lacked experience with Bank procedures, which was a handicap for speedy implementation o f projects involving procurement and hiring consultants; this handicap was overcome in time. In addition, and more fundamentally, the centralized unit with responsibility for managing the Bank’s lending for the crisis was the source o f friction with the Regions, who had been handling all lending work since the late 1980s. The work o f the SFO was not w e l l integrated with the rest o f the Bank’s program in the country; and the existence o f the SFO was contentious. Its generous budget a source o f frustration for the Regions, which wanted to handle the assistance to their countries even in crisis; and there were disagreements between the SFO and other central Bank staff on substantive issues, such as procedures involving bad loans and emphasis on banking supervision. 7.30 No other Region agreed to use the SFO’s services for the subsequent crises in Russia, Argentina, or Uruguay and the SFO structure was not sustainable in the Bank’s organizational structure. The SFO was disbanded in 2001, i t s budget and staff allocated to the Regions, mostly to EAP. Although there is a small central unit responsible for Banking and Financial Restructuring whose mandate includes contributing to fiture crisis work, the Bank no longer has a team specialized in crisis response. 7.31 Deep crisis.of the sort discussed in this review is too rare to justify a dedicated group. It does make sense, however, to identify experienced staff within the Bank who could be mobilized o n short notice, as a sort o f “virtual” crisis response team; if that proved insufficient to deal with a multi-country crisis - as occurred in Asia - the Bank 48 Scott (1999). 47 could again put together the resources and external staff to work with the virtual, experienced Bank staff. Bank management, dealing with crisis 7.32 Because the Bank deals with almost all sectors and themes touching economic development, its top Regional managers are seldom specialized in financial sector issues, and normally lack the background to deal with financial crises. Dealing with top IMF, bilateral, or govemment officials over policy issues or agreeing on division o f responsibilities in crisis situations has proved problematic. Bank staff working on these countries reported that their positions on issues were not adequately represented or defended by management. The Bank needs to articulate a clear line o f responsibility for representing the Bank in the event o f crises, to work with Regional managers in dealing with governments, the IMF, other IFIs, and bilaterals, and in ensuring intemal Bank-wide coordination o f efforts. 7.33 The 1996 internal review o f the Bank’s response to the 1995 Mexico crisis concluded that the Bank was ill-prepared and i t s response ad hoc to the crisis. I t recommended preparing guidelines with triggers for action, clear lines o f responsibility, and procedures for concentrating resources, putting in place a core team, and providing a framework for debating and agreeing expeditiously o n recommended actions. These recommendations were not acted upon; as a result, the Bank remained unprepared for the next round o f crises. The recommendations are s t i l l valid today. Recommendations 7.34 Although the Bank cannot predict crisis, it can do a more systematic job o f assessing vulnerability to crisis, particularly now that the Financial Sector Assessment Program is on-going. In addition, the Bank should change i t s approach to presenting risks in i t s documents, to provide a more candid assessment o f low, medium, and high r i s k countries based, in part, on i t s assessment o f the financial sector vulnerability to crisis. OED does not think this will affect the Bank’s access to information in the client countries nor the behavior o f the markets. 7.35 I t i s likely that intemational pressure o n the Bank to lend in crisis situations will continue and that the Bank will be called on to play its role in any intemational rescue package. The Bank should be more candid in the objective o f its lending; it should be clear that in the first instance it i s primarily to provide liquidity and restore market confidence. Second, it should frame i t s objectives based on a realistic assessment o f what the government i s willing and capable o f doing in a short time period, regardless o f the size o f the loan. The timing and size o f subsequent adjustment loans, after the initial frenetic, “emergency” phase, should be based on progress to date on reforms and likelihood o f continued progress. If TA loans are part o f the package, special arrangements should be made at the time o f approval to expedite procurement and selection o f consultants. 48 7.36 Coordination with the IMF and other I F I s needs improvement. At the outset o f a crisis, the Bank, the IMF, and any other IF1involved should reach an agreement on the basic approach and respective role o f each institution. The Bank should also better prepare itself to handle crises, appointing a top manager to be responsible for coordinating the Bank’s response and dealing with governments and external agencies. Just as the Bank now has guidelines for post-conflict assistance, the Bank should develop similar guidelines for dealing with crises. PART I1 ANALYZING OUTPUTS, OUTCOMES, AND IMPACT AT A COUNTRY LEVEL 8. Outputs at a country level: ownership and incentives Overview 8.1 Three main pillars o f Bank lending for financial sector reforms were privatization o f banks, establishing or improving prudential regulation, and strengthening supervision o f banks (see Figure 3.3 and discussion in Chapter 3). This chapter reviews changes in measures at a country-level o f these reforms supported by the Bank, as well as lessons emerging from the quality o f the reform processes. The shift to private ownership 8.2 Although the empirical literature i s fairly unambiguous in i t s findings on the benefits o f private ownership compared to state ownership o f banks in the Bank’s client countries, the Bank sometimes focused o n privatization as an end in itself rather than on improving the governance o f banks that was the underlying objective o f the process (this issue i s discussed further below). Nevertheless, privatization was an objective or the means to achieving a deeper objective in Bank lending in some 40 countries. This chapter thus examines progress in privatization, measured by the change in assets in government-owned banks4’as a percent o f total banking assets (Annex 3, Table 5, has the list o f countries where information was available for this analysis). Although this definition has serious drawbacks as a measure o f government ownership (see B o x 8.2), it was the only one that provided a consistent data series across countries and over time. 8.3 The chapter also draws on background papers and case study countries to gain insights into data limitations, factors associated with success (or failure), and experience with different approaches to bank restructuring and privatization, including support for asset management companies. 49 Defined as banks in which the government owns at least 50 percent o f the capital. 49 Considerable progress. .. Figure 8.1: Changes in government ownership o f banks, with and without Bank lending for privatization 8.4 At the beginning o f the period under review, assets in government-owned banks comprised an average 79 percent o f total banking assets in the 40 countries that subsequently borrowed from the Bank for bank privatization (where information was available). By the end o f the period, assets in government owned banks had dropped to about 21 percent o f total banking 8 w Ith Bank lending w Lhout Bank OECD countries for privatnatsn lending for assets. By this measure, Bank support for m prN atnat K)n privatization can be considered, on the whole, *Latest year available successful (Figure 8.1). 8.5 In addition, the average change in government ownership is higher in countries that borrowed from the Bank in support o f bank privatization, than in countries that did not borrow for this purpose (Figure 8.1 - the average for OECD countries is also shown for information and not as a the standard against which Bank client countries should be assessed; the l i s t o f OECD countries i s in Annex 3, Table 6).50 There may clearly be some bias in the sample o f countries that did borrow from the Bank, as they may have been more willing to privatize than countries that did not borrow. Although OED made an effort to avoid this (Box 8. l), it i s likely that some bias still exists which explains part o f the difference. Another explanation for the difference, however, could be that the process o f negotiating loans with the Bank and the subsequent requirement to adhere to loan conditionality within a certain time frame may exert pressure to show results that is missing in countries with no Bank lending. Box 8.1 :Problems comparing results in countries with and without Bank borrowing In an effort to avoid obvious problems in comparing the two groups o f countries, those that borrowed for bank privatization with those that didn’t, only countries were included that had an active bank privatization program. Thus, countries were excluded if they had banking sectors already substantially privatized, such as Botswana, Lebanon, Senegal, and Swaziland, or if they had no active privatization program, such as Algeria, China, Iran, Syria, and Vietnam. This, o f course, begs the question o f why countries with active programs would not want to borrow from the Bank in support o f privatization. The reasons likely include no need for balance o f payments support, an unwillingness to negotiate conditionality, general avoidance o f adjustment lending, or absence o f policy dialogue on financial sector issues. Any o f these reasons could introduce a bias n the “with” and “without” groups for this analysis i s in Annex 3, Table 5. in the results. A l i s t o f countries i 8.6 Neither the number o f loans nor the inclusion o f TA loans affected the results, as shown in Table 8.1.51 By contrast, country characteristics mattered: Table 8.2 shows that progress in transition countries stands out as particularly successful, where the banks 50 Because o f the possibility o f a skewed distribution, median values were also examined; they have a smaller difference, but the same pattern: countries that had Bank support for bank privatization showed a larger drop in government ownership than did countries that had no such support. 51 Because o f the limited number o f observations, it was not possible to test whether outcomes for bank privatization in l o w CPIA countries might have been better with TA than without, as was the case for outcomes o f individual loans found in Chapter 6. 50 changed from virtually totally government-owned (except for Hungary, Poland, and the Slovak Republic) at the beginning o f the period to almost totally privately-owned by 2002. For non-transition countries, differences between groups are not as great: low income countries did (somewhat ~~ ~ Table 8.1: Change in government ownership, by number o f surprisingly) better than middle income adjustment loans and with and without T A countries, excluding transition (45 percentage point change versus 36, Privatization: government ownership respectively), while l o w CPIA borrowing No of Change in countries did exactly the same as the higher countries percent One Bank loan 12 -59 CPIA borrowers. Countries with larger financial systems compared to smaller More than one Bank loan Significantly different? 27 __-58 no systems were also somewhat behind in Countries w Bank funded TA 23 -59 terms o f reducing government's role. This Countries wno Bank funded TA 16 -58 latter result m a y reflect the greater Significantly different? -_ no difficulty in selling very large public banks, which are sometimes preceded by the social and political hurdles involved in downsizing and laying-off large numbers o f people. ...But privatization i s far from complete in many countries 8.7 Data o n commercial Table 8.2: Change in bank ownership bank ownership presented Assets of government Change in above are only part o f the owned banks as percent percent Number of total banking assets ownership story, because they may 1991-93* 1999-2002* understate the extent to which Countries with no Bank lending 23 64 29 35 the govemment has reduced Countries w Bank lending, of which: 40 79 21 58 i t s role as intermediary. First, some governments retain a Transition countries 17 94 1s 79 large minority ownership in Low income countries, w/out transition 16 70 25 45 banks that are considered Middle income countries, " " 7 62 26 36 legally private, and thus retain Low 2003 CPIA countries, " " 9 71 28 42 effective control. Second, High 2003 CPIA countries, " " 14 66 23 43 some banks are owned by Larger financial systems, " ** " 6 64 31 33 state-owned enterprises or Smaller financial systems, " " 17 69 23 46 public utilities and are *latest year for which data available **Argentina, Brazil, Colombia, Morocco, Pakistan, and Philippines. therefore de facto controlled by government. Third, near-banks, using deposits or other sources o f funding to make loans, are not counted as part o f the commercial banking system, and are therefore excluded from the statistics on government ownership. These can include specialized banks, like housing or agricultural banks, and development banks, that may account for a substantial portion o f total banking assets more broadly defined, and that can introduce distortions by non-market based lending and represent considerable contingent liabilities for the government (Box 8.2). 51 Box 8.2: Data on bank ownership can be misleading Restructuring and privatization o f commercial banks were supported in Cameroon by three adjustment and t w o TA operations. At the beginning the 199Os, government ownership accounted for 37 percent o f the shares o f the top banks; by 2002, a l l commercial banks were considered private. Government, however, has retained ownership o f between 25 to 45 percent o f the top three banks that account for over two-thirds o f the assets o f the banking sector and a m u c h higher percentage o f retail banking in the country. Government does not appear t o be actively involved in the daily management or policies o f these banks; nevertheless, in one o f them, government agreed t o sell one-third o f i t s shares t o local businessmen, but has been arguing with the bank’s management o n an acceptable list o f buyers for over two years. In Cape Verde, after privatization o f the largest bank BCA, government retained 20 percent equity stake and “Golden” share rights, i.e. privileged voting rights. “Golden Shares” were created in order for the government t o maintain control over strategic industries. In Cote d’Ivoire the government retains some 15-25 percent o f the capital in the privatized banks. In Tunisia, the government can retain up to 49 percent o f shares in banks that are considered private, and privatization has been mostly through selling equity shares in the market, with the government retaining effective control. In addition, there are a number o f public development banks (at least seven as o f end-2003) that account for a significant share o f term lending and that are not part o f the statistics o n commercial banks. 8.8 In addition, the averages mask wide variations among countries. In most o f the transition countries, state ownership has shrunk to close to zero, starting from 100 percent ownership. In Pakistan, by contrast, it was still over 50 percent in 2002. In Tunisia, Morocco, and Yemen, state ownership has shrunk by an average o f (only) 23 percentage points and remained (in 2003) over 30 percent o f banking assets, as i t did in Argentina and Brazil, due to a combination o f ambivalence by government and the difficulty in selling the banks. In the latter two countries, there was a clear pattern o f success at the sub-national level, but an inability to privatize the large federal banks. This does not diminish the relevance or the achievement o f the privatization objectives in those countries; it does, however, underline that satisfactory outcomes do not imply that the agenda o n privatizing banks is finished. I naddition, this discussion does not cover countries that did not borrow from the Bank and/or did not have programs to privatize, including Algeria, Belarus, China, Costa Rica, Iran, Nepal, Syria, and Vietnam, where the banking sector i s dominated by state owned banks. 8.9 I t is unrealistic to expect governments to have no involvement in financial intermediaries (see footnote 2 and Figure 8.1 for OECD average). Bank staff have reported that governments in most Regions express interest in continued Bank support for public banks, so it i s clear that much work remains to be done to engage governments in developing internally consistent policies on the role o f the public sector in banking sector intermediation. Quality matters 8.10 Although research shows that private banks and foreign banks often have a positive impact on banking performance in client countries (paragraphs 2.7-2.9), the Bank’s experience demonstrates that neither privatization nor foreign entry has been a guarantee o f better performance. Even apart from other factors that can affect the subsequent performance o f the banks (macroeconomic factors, market structure, investment climate), the quality o f the process mattered for the outcome in terms o f how 52 well the banks performed after privatization. The process can include financial restructuring prior to privatization, measures to prevent a re-accumulation o f N P L s before the sale o f the bank, speed o f privatization after restructuring, privatization to a strategic owner versus sale o f shares to the public, and whether government retained significant minority shares. The following paragraphs discuss the Bank experience with different types o f restructuring prior to privatization. The quality o f the investor(s) who bought the banks also made a difference to the performance o f the privatized bank (Box 8.3). Box 8.3: Privatization i s no guarantee of good banks - the quality o f the buyer matters In Mozambique, the Bank was closely involved in the mid-l990s, through an adjustment and a TA operation, in helping to privatize two large commercial banks ( B C M and BPD). B C M was sold to a foreign businessman with no banking experience and BPD to a s m a l l foreign banking group, with government retaining significant ownership i nboth. B C M continued to accumulate NF’Ls after privatization and went through several rounds o f recapitalization by the government before it was merged with another Mozambique bank (see background paper by Mozes, 2003, for details). Inthe case o f BPD, the combination o f government interference and adverse economic conditions in the bankers’ home country caused the foreign investors to stop making capital investments, and BPD was taken over by the central bank and re-privatized a second time, having been recapitalized four times by the government. In Macedonia, through a misunderstanding between the Bank and the government, the f i r s t “privatization” o f Stopanska Bank, supported by an FSAC in FY95, resulted in i t s sale to a former state owned enterprise; after at least four years o f further portfolio clean up, supported by a second FSAC (FYOl), it was sold to a foreign commercial bank, which then recapitalized it. Financial restructuringprior to privatization: better outcomes 8.1 1 In the majority o f case study countries, the Bank supported financial restructuring prior to privatization. These cases have better outcomes than in the few countries where financial restructuring was not undertaken and where the privatization did not go well: either the banks could not be sold, or at least not at a price acceptable to government, or they were sold to investors who were inappropriate, or at least who did not manage the bank well after taking control (Box 8.4). Box 8.4: N o financial restructuring prior to privatization: didn’t work well In Georgia, under the Bank’s FY97 SAC Ibanks were operationally restructured (branches closed), but not , recapitalized; NpLs were not dealt with: the banks were bought by employees, and remained unsound. In Morocco, neither o f the banks targeted for sale under the Bank’s FY96 F S A L were sold - one (CIH) because i t s portfolio had deteriorated so much it needed financial restructuring and the other (BNDE) because the two attempts to s e l l brought unacceptably l o w bids (the reasons for the l o w bid were not clear - they may have been unrelated to the quality o f i t s assets). A waiver was required on these banks’ sales prior to tranche release. In Togo, the FY98 TA credit financed consultants to prepare restructuring plans, including dealing with high NF’Ls; this effort was to be followed by an FSAC which never materialized. Government did not follow up on the bank restructuring and only one out o f seven o f the banks were sold, although this may have been due to lack o f government commitment as well as the poor financial situation o f the banks. 8.12 The scope and type o f Bank support depended o n h o w advanced the process was at the time o f the loan. In a number o f countries, the Bank provided TA for carrying out 53 audits or other diagnoses to identify the NPLs (which i s not a trivial task if either prudential norms or accounting practices are weak) and for developing a plan to deal with them. The Bank has also supported several methods for removing the banks’ N P L s (both in the context o f privatization and for restructuring alone; the latter i s discussed in paragraph 8.20 and B o x 8.8): taking them entirely o f f the books o f the banks and putting them into an asset recovery unit, which essentially shrank the bank, or replacing them with government bonds, which could provide a theoretically r i s k free asset at government expense.52 Other solutions to N P L s were pursued in Poland (EFSAL, FY93), where banks were given special legal powers to recover their loans; this met with some success, although a similar effort to create a special workout unit in a large bank in Mongolia, but with no special legal powers, did not lead to results in terms of recoveries. From the limited information available in Bank documents, it appears that special legal powers are key, whether for workout units in banks or independent asset management companies (AMC), discussed in the next paragraph. rates of loan recoveries Box 8.5: Asset Management Companies ( A M C ) -- empowerment when the NPLs were In Cameroon, the Bank supported the establishment o f an A M C in 1989, which owed by loss-making managed to recover only 3 percent o f the assets transferred to it; under SAC I1 state owned enterprises (FY96), the A M C was restructured and given more legal powers, and its or even defunct performance improved slightly, although it still has institutional weaknesses: the enterprises that couldn’t Bank i s currently providing TA to transform it into a for-profit debt collection agency. In Burkina Faso, the Bank-supported A M C achieved i t s loan recovery pay or politically well targets because it was exempt from going through the judiciary and because i t was connected borrowers able to publish a l i s t o f defaulters. that wouldn’t pay, and, in particular, when the In Albania, the Bank supported the establishment o f an A M C in 1997, but i n its AMC had no special f i r s t three years, i t recovered only 3 percent o f assets. Renewed World Bank support under a subsequent loan combined with new management and legal powers legal powers to collect have improved the AMC’s performance somewhat, so that by end 2003, about 7 on loan payments. By percent o f the initial stock o f assets had been recovered, 30 percent had been contrast, if the AMC was submitted to the courts for resolution, and another 30 percent sent to the Bailiffs’ given special judicial Office for execution. In the Slovak Republic, a work out scheme was established powers to recover the under an AMC, but was unsuccessful because o f attempts by the A M C to use the assets to become a real bank; the Bank intervention was successful in stopping this, loans (meaning it could bypass the normal court system), even in otherwise poor legal and judicial environments, they could meet targets for recovery o f N P L s (Box 8.5). Using AMC with special powers to pursue debtors had two other advantages: when AMCs were government-owned, and they usually were, the amounts collected could be used to defray part o f the costs o f bank restructuring. ” In Cameroon, the “risk” free asset proved to be high risk: the government was unable to service its bonds that had replaced N P L s in restructuring in the late 1980s; under SAC I1(FY96) government arrears were guaranteed by the Regional central bank and a second round o f restructuring was required. ’3 Countries are Albania, Cameroon, Romania, and Slovak Republic, as well as crisis countries (Indonesia, Korea, and Thailand). The support from adjustment lending was mostly through conditionality that specified transfer o f N P L s to the AMC; or recovery targets. TA loans provided more specific assistance. 54 Second, the process o f pursuing defaulters could serve as a signal that the default culture was no longer tolerated. Although some empirical research exists on the experience with A M C performance (Klingebiel, 2000) as well as with decentralized approaches to NPL recovery in banking crises (Dado, Klingebiel, 2000), this i s an area where the Bank could do more to provide guidance to staff on trade-offs in approaches (centralized versus decentralized) and on factors associated with effective loan recoveries. Other forms of bank restructuring 8.14 Downsizing. Pakistan i s the only case where Bank funds (US$300 million, Bank Sector Restructuring and Privatization Project, FY02) were used explicitly for severance payments in the context o f an ambitious program o f downsizing large state-owned banks. In Brazil, there were two sub-national investment loans supporting employee retrenchment in the context o f bank privatization, although the loans weren’t directly tied to this cost. Many other Bank loans supported downsizing prior to privatization, although it remains an open question whether the costs are an efficient use o f funds: new owners could have other ideas about the best size and structure o f their bank; o n other hand, new owners may not want to deal with political problems involved in laying o f f workers. In any case, there is little systematic evidence on whether downsizing is important prior to privatization; the Bank supported privatization o f banks in the absence of downsizing, apparently successfully, although the ability to do this may depend on the scale o f overstaffing and the ability o f an employer to fire workers and the political sensitivity o f doing so. 8.15 Twinning. Bank support o f twinning, matching foreign banks with weak domestic ones, has had mixed experience. I nPoland and Mongolia, twinning helped banks to restructure and reorganize prior to privatization, although in Poland, it was generally successful mainly for banks whose management was committed to the idea. The experience in Kazakhstan, where the Bank supported twinning for a large number o f banks, was less than satisfactory, as some o f the banks were uninterested while others were not sufficiently well organized to make the necessary arrangements. Avoiding buildup of NPLs 8.16 Credit ceilings don ’t work, or at least not for long. Bank loans sometimes included conditionality o n credit ceilings or suspension o f lending in the context o f bank restructuring as a precursor to privatization (and in other cases, as a way o f limiting the accumulation o f N PL in banks the government was determined to retain). Inboth Albania under FY94 Enterprise and Financial Sector Adjustment Credit (EFSAC), and Romania, under FY95 Financial and Enterprise Adjustment Loan (FESAL), the government imposed credit ceilings o n the state banks. I nneither case did the ceiling work for long: in both countries, the government undercut the agreement by allowing the state banks to exceed the ceilings. InYemen, under the FY98 Financial Sector Adjustment Credit (FSAC), the government suspended the state banks’ lending to public enterprises as agreed, but the Central Bank took over direct lending to the public enterprises instead. These few cases where information is available suggest that 55 governments or the banks themselves may not be able to resist the pressures from well- connected enterprises. Alternative to bank privatization: closure 8.17 The Bank has supported alternatives to privatization in all Regions except S A R , including increasing minimum capital requirements for banks and liquidation. In Armenia, for example, the Bank supported the closure o f private banks through a succession o f adjustment operations which introduced a gradual increase in minimum licensing requirements, thereby forcing the exit o f banks unable to meet them, and substantially reducing the number o f banks from 72 in 1991 to 30 in 2001. Under a series o f adjustment loans and TA operations, Kazakhstan also closed many banks, reducing the number from 184 in 1991 to only 22 in 2001. In other countries, the process o f bank liquidation has proved time-consuming and politically difficult (Box 8.6), but probably preferable to trying to privatize non-viable banks. Box 8.6: Liquidation has been supported by the Bank in most Regions, but has proved difficult Albania liquidated an agricultural bank twice; after closing it the f r s t time (under FY93 Agricultural Sector Adjustment Credit) the Bank helped the country to set-up a second rural bank, which was then closed in 1997 with Bank support when it too proved unviable. Under an adjustment operation (FY99 Enterprise and Bank Privatization Credit) in Bosnia, the Federation agreed in principle to liquidate all insolvent banks, which the Bank had identified through diagnostic work, but Bosnia’s own diagnosis found all banks to be solvent. InUkraine, two Bank adjustment loans have addressed the closure o f Bank Ukraina, which i s taking some time. In Cote d’Ivoire, the Bank supported the liquidation o f five development banks, and the transfer o f assets to an AMC; Guinea liquidated one public bank under FY95 FSAC, while the liquidation o f a bank in Burkina Faso (FY91 SAC) took over five years to accomplish. I ... I Privatizationtook longer than expected 8.18 The process o f bank privatization often took much longer than the two or three years envisaged at the outset, and in some countries, i t remains incomplete after more than a decade. Partly this was due to ambivalence o n the part o f governments in the early years, but in other cases, such as Burkina Faso, it was difficult to find buyers initially. Slow privatization, for whatever reasons, increased the costs, because o f the problem o f a re-accumulation o f NPLs. I nTanzania, for example, differences between the Bank and the IMF on how to split up the largest state owned bank took several years to sort out, and in the meantime, N P L s continued to accumulate. The delay due to this debate arguably ended up costing the government considerably more to resolve the N P L s than if the differences had been resolved expeditiously. ....and sometimes led to unanticipated problems 8.19 In Mozambique, the privatization o f the banks led to an unexpected concentration o f market shares. One o f the partners in a large Mozambique bank was a small foreign bank, which merged with a larger bank in the same country, which was also the partner o f a second large bank in Mozambique. After much discussion, the two large banks in 56 Mozambique, which now had the same foreign owner, were allowed to merge in Mozambique, creating one bank that held over two-thirds o f the assets o f the banking system. In retrospect, bank privatization should have been accompanied by safeguards against high level of concentration. Restructuring banks without privatization: seldom successful 8.20 Contrary to Bank guidance on restructuring banks in the absence o f a plan to privatize (DEC note 1995), the Bank has explicitly supported government recapitalization o f state-owned banks, with no government plan or commitment to privatize them. The most common outcome of these efforts has been deterioration in the financial situation o f the recapitalized bank and the need to repeat the exercise some years later, sometimes again with Bank support. In other cases, the government planned to privatize, but either the process was too slow or the attempt failed (no acceptable bidders) and new N P L s accumulated (Box 8.7). Box 8.7: Restructuring banks with no commitment to change ownership: examples o f Bank support Albania: The first round o f restructuring, supported by an F Y 9 5 Bank credit, involved credit ceilings; clearance o f inter-bank loans; action plans to strengthen the banks. T w o years later, a second round o f restructuring was necessary, supported by the Bank, involving the transfer o f NpLs, technical assistance, a change in management, and the re-imposition o f credit ceilings. The banks have since been privatized. Lao PDR restructured its state banks in the mid-1990s with ADB support and indirect Bank support through a parallel S A C 111; a second round i s again being supported through a Bank credit (FY02), but not in the context o f privatization Guinea: Under an F S A C in FY94, four private banks were recapitalized without changing their ownership or governance; four years later one bank was liquidated at considerable cost t o government and three banks were recapitalized again with interest free loans f r o m the government; information i s n o t available o n the current health o f these banks. Ghana: restructured in the early 1990s with Bank support and intended t o privatize, but privatization didn’t happen fast enough; needed t o restructure again under F Y 9 9 Economic Recovery Support Operation 1 1. Vietnam: F Y 0 3 Poverty Reduction Support Credit I1continues to support restructuring o f four biggest state banks even though the government has n o intention to privatize banks in the near future. Recommendationson restructuring and privatizingbanks 8.21 The recommendations that emerge from this review i s that the Bank needs to be involved in countries where capacity i s limited in helping in the process o f privatization to ensure that: (i)financial restructuring precede or accompany the privatization; in the absence o f financial clean up, the privatization process i s unlikely to attract good investors; (ii)recapitalization o f banks is in the context o f a government plan to privatize; ( iii) for debt recovery mechanisms, AMCs, if they are created, b e given special judicial powers; (iv) government sells all o f its shares in the banks to be privatized; continued ownership by the government may both discourage good investors as well as create 57 problems post-privatization; (v) any strategic investor involved be “fit and proper”;54 the Bank may need to provide support for due diligence o n potential owners; and (vi) appropriate competition policies are in place to avoid unanticipated mergers and creation o f exceptional market concentration. Getting incentives in place: legal and regulatory reforms, strengthening supervision Overview 8.22 The Bank has supported a wide range o f establishment o f and changes in laws and regulations affecting banks and bank-like institutions as well as capital markets. In banking, the basic thrust o f reforms supported in over 160 operations (representing some 60 percent o f all loans with financial sector reforms) in 74 countries has been to allow market forces to determine deposit and lending interest rates and allocation o f credit, and to bring client countries closer to Basle (international) norms for prudential regulations I and principles for bank s u p e r ~ i s i o n .n capital market reforms, the majority (about 80 ~~ percent) o f the 48 operations in 30 countries supported the passage o f laws, establishment o f a regulatory framework, and standards for securities markets, although the Bank was also active in providing assistance to strengthen the institutional capacity o f regulatory agencies and stock exchanges. Improvements in the regulatory regime: mixed picture on the details 8.23 To measure improvements in the regulatory regime for banks, OED compared data on changes in prudential requirements for banks between 1998 and 2003 for countries that borrowed from the Bank between FY98 and FY02 with changes in countries that did not borrow from the Bank for regulatory changes during this period (see paragraph 8.5 and B o x 9.1 for caveats to this Four variables were examined: capital adequacy, quality o f capital (requirements for items to be deducted from the definition o f capital), loan classification, and provisioning requirements for doubtful loans. 8.24 There are only eleven countries in the sample where the loans during this time period had specific conditionality for upgrading prudential regulations and where there were data points for 1998 and 2003. Overall, the average required capital ratio did not increase by much among the borrowing countries (Thailand increased it and Argentina, 54 The expression “fit and proper” means owners who have relevant banking experience, a good reputation, and no conflict o f interest through connections to companies that could benefit as bank clients. 55 The Core Principles for Effective Banking Supervision were issued i n September 1997, which i s about half way through the period under review; even prior to this, however, Bank loans supported many o f these principles. 56 The source o f the data was the Bank’s Database o n Prudential Regulation and Supervision, at: http://www. worldbank.org/research/interest/Drr stuff7bank regulation database.htm. OED included only those countries where the timing o f the loan was such that adoption o f new regulations should have shown up as differences in the data between 1998 and 2003. 58 after lowering i t temporarily after the crisis, i s now gradually increasing it again). However, all eleven countries were already requiring banks to be above the internationally recommended ratio o f 8 percent prior to 1998. By contrast, among the nineteen countries that did not borrow at all from the Bank during this period (and for which information i s available), the average capital requirement increased from 8.4 percent to 10.2 percent. Thus, by this measure, countries that borrowed from the Bank for prudential strengthening did not strengthen the capital adequacy requirement by as much as non-borrowers (Table 8.3). Furthermore, in terms o f the quality o f the definition o f capital, among the borrowing countries, two (Brazil, Tajikistan) upgraded their definitions, while three countries (Argentina, Bolivia, and Korea) lowered their standards. By contrast, among non-borrowers, the standards for measuring capital increased overall, and by a wider margin. Thus, in terms o f improving capital requirements, the borrowing countries did not do as well, overall, as the non-borrowing countries. Table 8.3: Capital and loan classification, 1998 and 2003, with and without Bank lending Countries that borrowed Countries that did not from the Bank borrow from the Bank -________--_ 2003 compared to 1998----------- Minimum Capital-Risk weighted assets ratio N o change Higher Quality o f capital (definition) Weaker Stronger Loan classification Stricter Less strict Note: For details, see Annex 3, Tables 1 and 2. 8.25 On loan classification, the picture i s different: four o f the eleven countries that borrowed during the period strengthened the classification o f loans by lowering the number o f days required before loans were downgraded and, on average, the requirements were stricter than for non-borrowing countries; among non-borrowing countries, two strengthened and two weakened the standards. Thus, countries that borrowed during this period have made better progress and now have stricter requirements for loan classijkation than countries that did not borrow (Table 8.3). On loan loss provisioning there i s no major difference between the two groups o f countries. 8.26 Overall, the data present a mixed picture, and one that is confirmed by the analysis o f the quality o f prudential regulations in the FSAPs. While the FSAPs found that almost half o f the twenty-four countries that had borrowed for Bank support for strengthening prudential regulations had strong systems, a little over half s t i l l had significant shortcomings, particularly with respect to exposure limits, insider lending, or ownership structures. Most countries had weaknesses in the measure o f capital adequacy. Regulatory framework for banking and capital markets in ECA transition countries 8.27 EBRD indicators exist (only) for ECA transition countries, for the regulatory framework for banks over the period 1998-2002, and for the regulatory framework for the securities market and non-bank financial institutions, over the period 1997-2002. 59 Assuming that reforms supported by Bank lending in years prior to FY97 would already be reflected in “baseline” indicators, OED compared average progress for countries that borrowed from the Bank for legal and regulatory reforms over the period FY97-01 and for capital market reform over the period FY96-02. Changes in indicators were compared with those for the transition countries that did not borrow from the Bank over the relevant period (see important caveat on selection bias in B o x 8.1). The results on banking indicators are in Table 8.4 and show that for the countries that borrowed over the period, there was an overall improvement in banking regulations averaging a little over one grade (0.36), whereas for the countries that didn’t borrow during this period, the improvement was more modest, at an average o f 0.2 (see Annex 3 for details by country). These results, for a relatively small sample o f countries, are consistent with the findings o n privatization: borrowing countries have done better than countries that didn’t borrow over the relevant period, although a closer look at the details (as in the previous paragraphs) reveals a more nuanced picture. Table 8.4: Indicators on strength o f financial regulations, transition countries Securities markets and non-bank financial Banking institutions Increase in quality o f financial regulations: 1998 - 2003 W i t h Bank lending for regulatory changes 0.36 0.28 Without Bank lending - - for regulatory changes 0.20 0.22 Note: The EBRD indicator for each country i s a composite measure, scaled from 1 to 4, with pluses and minuses; an increase from 2 minus to 2 was counted as an increase o f 0.33; from a 2 to a 2 plus was 0.33, etc. See Annex 3, Tables 3 and 4 for detailed indicators by country. Source: EBRD Transition Report, various years. 8.28 A similar analysis was carried out for the reforms in capital markets; the results in Table 8.4 show that by contrast with the banking sector, the improvements in both groups o f countries (with and without borrowing from the Bank for capital market reforms) were similar (at 0.28 in borrowing countries versus 0.22 in non-borrowing). In four o f the seven countries that borrowed from the Bank, there was no change in the regulatory framework for securities markets and non-bank financial institutions, while in over half o f the 18 countries that did not borrow there was improvement in the framework. Implementing the laws and regulations and banking supervision: little information 8.29 An equally important issue on prudential regulations and legislative reforms is . ~ ~ the thirty-seven case country studies, although virtually all their i m p l e m e n t a t i ~ n From o f them contained Bank support for strengthening the legal andor regulatory regimes, there was only sporadic information available on the extent to which the changes were being implemented. The story i s similar for strengthening supervisory capacity, which i s an integral part o f implementing prudential regulations: there was little evidence to 57 Theoretically, one set o f measures would be the changes in capital adequacy ratios and non-performing loans, which could be expected to improve (capital adequacy up, N P L s down) over time as a result o f stronger prudential regulations if everything else were constant. The obvious problem i s that there are far stronger economic as well as political influences at work that affect these ratios. 60 support whether supervisory agencies had been strengthened. The FSAPs found that out o f the twenty-four countries that borrowed for either legalhegulatory reforms or strengthening banking supervision, about half had improved in the quality o f the supervisory agency and in on-site and off-site supervision o f banks, but in 15 o f the 24 countries the FSAPs noted shortcomings in adherence to prudential regulations and lack o f enforcement o f the prudential framework by the supervisory authority. Thus, again, a mixed picture. 8.30 Inthe case studies and the FSAPs three constraints in particular are cited as hindering stronger implementation o f prudential regulations and better functioning o f banking supervision: (i) lack o f institutional capacity o f the supervisory agency, including an absence o f special enforcement power and legal immunity for the supervisors; (ii)a solid legal framework for bankruptcy framework; and ( iii) lack o f political support for the supervisory agency (see B o x 8.8). Box 8.8: Lack of political support: Algeria At the end o f 2002, public banks accounted for over 90 percent o f loans and 84 percent o f deposits. The banks still carry a significant volume o f nonperforming and poorly provisioned loans to the public sector. Although on-site supervision has been strengthened and off-site supervision i s being expanded, both human and financial resource constraints, “as well as the sometimes unresponsive reaction o f the authorities to instances of failure to observe the regulations, undermine the effectiveness o f the prudential system.” 1 Source: W o r l d Bank and IMF (2004). Special topic: legal immunity for supervisors 8.31 One issue that has been pursued in a number o f countries by the Bank i s establishing legal immunity for banking supervisors, which serves to insulate them from fear o f being sued by banks that didn’t like their findings. The Bank’s attempts to address this have met with mixed results. I nPeru, the Bank proposed including it in the FY92 FSAL, but Government didn’t agree. In both the Philippines and Brazil, introducing legal immunity for banking supervisors was a condition o f a recent adjustment loan, but in neither country was it met. O f the thirty-seven case countries examined, eleven countries have banking supervisors that were not immune from legal prosecution as o f 2002 (and information was not available for nine o f the countries). The FSAPs also cite this as an unresolved issue in many o f the borrowing countries. Special topic: deposit insurance58 8.32 Out o f the total o f thirty-five countries where the Bank lent for deposit insurance schemes (paragraph 5.21 and Figure 5.6), most (20 countries) involved creation o f a scheme, while the rest addressed reforms o f existing schemes (12 countries) or quite marginal changes (3 countries). Most o f the reforms creating deposit insurance schemes 58 This section i s taken f r o m a background paper by I l k a Funke (2004b). 61 (involving studies, legal reforms, the establishment o f the scheme, and establishment o f an agency to handle it) had satisfactory outcomes. L i t t l e information is available, however, on the quality, functioning, or impact o f the schemes. In three countries (Bosnia and Herzegovina, Bulgaria, and Poland), the Bank’s completion reports indicated that the deposit insurance schemes increased public confidence in the banking system, and in Argentina, the Bank reported that there was no evidence that trust had increased. 8.33 By contrast, the efforts to reform existing deposit insurance schemes did not achieve their objectives. Reforms included phasing out unlimited coverage that had been put in initially during a banking crisis, improving the schemes finances through raising premium levels or other means, or improving the functioning o f the deposit insurance institution. Out o f seven countries that tried to limit the insurance coverage, only Ecuador, Korea, and Mexico succeeded; o f those that were unable to limit coverage, the governments claimed that these reforms could not be implemented because o f the s t i l l l o w level o f confidence in the banking sector or the weak financial health o f the banks. Better progress was made in improving the financial health and operational efficiency o f the deposit insurance agencies, although even here, implementation has been uneven; eliminating automatic government guarantees for state banks was achieved only in Lithuania, but not in Bulgaria or Romania, where it was also supported; and legal immunity for deposit insurance agency staff was achieved only in Uzbekistan, not in Philippines (no information on Argentina). Recommendations on improving the incentive framework 8.34 I t i s important for the Bank to develop indicators to measure progress in the reforms i t supports, so that it has a means o f monitoring whether the reforms on paper are being implemented in pra~tice.~’ Indicators are necessary to measure progress toward objectives, for example, on the degree to which banking supervision adheres to Basle principles for good supervision, whether prudential regulations are consistent with international principles (Basle I), and, most important, the extent to which banks and other financial institutions are in compliance or moving toward compliance with the regulations. Especially in the context o f programmatic lending, which consists currently mostly o f support for actions rather than requiring progress o n outcomes, i t i s important to establish measurable, realistic, medium-termindicators which will enable all stakeholders to monitor whether targets are being met. 59 At present, the analysis o f the financial sectors carried out through FSAP provides some information, but n the sense o f providing the program does not cover all Bank borrowers; and i t i s not a monitoring tool, i information o n a regular basis. 62 9. Outcomes at a country level: m a r k e t structure, contestability, efficiency, and health Overview 9.1 This chapter examines whether outcomes at a country-level have been achieved in terms o f changing market structure,6ocompetition levels through greater contestability, efficiency, and health o f the banking system in countries that borrowed from the Bank for these purposes over the period under review. This chapter draws on both quantitative indicators and case studies for insights into the reforms and qualitative results. Changes in market structure: bank concentration 9.2 The change in market structure i s measured by the concentration ratio, which is the share o f total banking assets held by the three largest banks. Although the use o f this measure as an indicator o f competition has been contested in the literature (paragraph 2.7), the Bank has nevertheless sought to decrease concentration in many (particularly smaller) financial systems as a way o f decreasing market power and encouraging competition. I nmost o f the fifty-four countries that borrowed from the Bank for financial reforms and where information i s available on this measure, the data show a steady decrease in the share o f the top three banks over the period under review. The reforms pursued included deliberate downsizing, liquidation, and/or allowing entry o f new banks. Because larger systems61 might be significantly less concentrated than smaller ones, Figure 9.1 shows results separately for each group. By 2001 (latest year available), only Algeria had a concentration ratio over 60 percent among the larger systems, although among the smaller systems, twelve countries still had concentration ratios above 70 percent. By contrast, the larger financial systems had on average lower concentration ratios than OECD countries both at the beginning and at the end o f the period.62 9.3 In order to examine whether the yearly changes in banking concentration could be associated with Bank lending over the period under review, OED and D E C developed a model to compare annual changes in these indicators in countries that borrowed from the 60 A s noted in Chapter 2, the literature does n o t provide a consensus view o n a n efficient market structure, but Bank assistance t o concentrated financial sectors has nevertheless tried to increase competition; it was a n explicit objective in 23 out o f 37 case country studies (list is inAnnex 4). 61 Larger systems were identified as the 25 countries that accounted for 84 percent o f a l l banking system deposits in developing countries i n 2000; for the list o f these countries, see Hanson (2003). 62 OECD countries are shown for comparison only and n o t as a target or benchmark. 63 Bank for financial sector reforms with changes in indicators in non-borrowing countries (see B o x 9.1 for a discussion o f the challenges o f this analysis and how they were addressed). The results presented in Tables 9.1 and 9.3 and in Chapter 10 are for the model that includes macro-economic and institutional controls. Variations on this model include one with policy controls (specifying which policies were covered by the Bank loans) and a model with no controls. Results are qualitatively similar across the different variants o f the model presented here. 9.4 Table 9.1 shows that banking concentration decreased at an average rate o f 1.1 percent per year in countries that borrowed from the Bank for financial reforms, and by 2.2 percent per year in countries that didn’t borrow. Thus, banking sectors in developing countries have tended to become less Table 9.1: Annual growth rate o f changes in bank concentrated over the last decade. The concentration: with and without Bank lending for sector reforms decrease in banking concentration in countries without Bank lending, however, was Banking sector concentration With Bank lending -1.05* significantly larger than the decrease in the Without Bank ,ending -2.16* countries with Bank lending. The models also significantly different? Yes tested whether the number o f adjustment loans Number of countries 59 R2 0.33 or the presence o f TA lending had any * significantly different from zero at 1 percent level explanatory power for the results among Bank borrowers; they did not. Changes in Contestability 9.5 Recent literature has argued that contestability i s more important for competition in a banking system than concentration ratios (paragraph 2.7). Contestability can be measured by the ease o f entry and restrictions on banking activities, which measure the potential for competition. Using the Bank’s database on prudential regulation and supervision (reference, footnote 53), OED compared data on changes in entry requirements and restrictions o n activities for banks between 1998 and 2003 for 24 countries that borrowed from the Bank between FY98 and 02 with changes in 29 countries that did not borrow from the Bank during this period. Table 9.2: Changes in contestability Countries that Countries that 9.6 Entry requirements. T w o forms o f borrowed did not borrow entry requirements were examined: the 2003 compared to 1998 number of pieces o f information required for a bank to establish itself in a country, and the Entry: number same same o f licenses minimum capital requirement. A decrease in the average amount o f information required Minimum same same capital would indicate an increase in contestability. requirement Most o f the borrowing countries and the non- Restrictions on Less More borrowing countries had almost an identical banking restrictive restrictive number o f requirements at the beginning and activities end o f the period (eight items were required in Note: For details, see Annex 5, Tables 1-4. most countries); there was thus little change within groups or between groups (Table 9.2). For the minimum capital at entry 64 requirement, Table 9.2 shows again both sets o f countries changed very little in terms o f minimum capital required for entry into banking, although among countries that borrowed from the Bank for financial sector reforms, a slightly higher proportion increased the capital requirement than among non-borrowers. 9.8 Change in foreign ownership. Rather than examine the data on prudential changes in the de jure la0 5 9 3 ability o f foreign banks to establish partnerships or ownership, OED examined the de facto change in $ 8 0 foreign ownership o f banks, defined as share o f assets fp 70 held in banks that are 50 percent or more foreign- 2 5o owned, in the borrowing and non-borrowing countries, i.i 40 $ 3o .- because it i s the actual changes in ownership that Y i Q 10 indicate greater contestability rather than merely legal I B O changes which could be undermined by other s WithsankkndbJ WithoutE&&lendng administrative barriers. For twenty-six borrowing m t year avaibble 9.9 In sum, the picture i s mixed on the indicators o f contestability, but combining no change in some indicators with a change toward greater contestability in others, borrowing countries seem to have slightly increased competition levels in banking compared to non-borrowers. 63Data are available only up to 2000; the situation has evolved further in the direction o f more foreign ownership among a number o f these countries (with and without Bank borrowing) since then. 65 Box 9.1: OED/DEC on constructing a “counterfactual” In economic analysis, it i s very difficult to construct theoretically and statistically robust counterfactuals. T h i s evaluation i s no exception. As noted in paragraph 8.5, comparisons o f Bank borrowers with non-borrowers face the problem that countries that borrowed from the Bank may have had factors influencing the reforms that are not captured by the borrowhot borrow dichotomy. To address this, the OED/DEC models used a country-level fixed effect. The results o f the models should therefore be interpreted as departuresfrom a country’s typical valuefor the variable tested. Definitions and sources o f information for variables used and the models tested in this review are in Annex 6. Other factors could also drive financial indicators away from a country’s typical value. T h e regressions thus include variables measuring the quality o f the macroeconomic and institutional environment: growth rate, inflation rate, fiscal deficit (relative to GDP), and, as a measure o f institutional capacity, the CPIA. Variations o f the basic model include controls for the country’s financial sector reform program, a recognition that some types o f reform are more likely to spur short-term improvement on financial indicators than others. All controls are lagged one year relative to the financial indicators to help mitigate problems arising from the dependent and independent variables being simultaneously determined. Still, it i s possible that the borrowing countries were poised to make the most progress in reforms, in particular the transition countries, compared to countries that were not on the same reform path. Other countries might choose not to borrow because they had already reformed. Thus a bias (in terms o f observed changes) would be in favor o f the borrowers. On the other hand, countries that had been performing poorly and had more deeply entrenched banking weaknesses may have felt the most need to borrow in the hope that Bank assistance would bring about changes, and thus the bias would work against the borrowers. To address this issue, OED/DEC also used treatment effect regressions that explicitly account for self -selection and propensity score matching techniques; the results o f using both o f these techniques reinforced the main findings. In terms o f initial conditions (in the early 1990s) in borrowing and non-borrowing countries, they are presented in the Table below; on a number o f variables, they were not very different in the two groups, although the non-borrowers have somewhat better indicators. Table: Initial conditions in variables in borrowing and non-borrowing countries, early 1990s Indicator Borrowers Non- borrowers Government ownershit, o f - Assets in govemment owned banks as share o f total banking assets 79.0 64.5 banks Concentration ratio Share o f assets held by three largest banks as percent 74.8 76.8 oftotal assets Foreign ownership Share o f assets in foreign owned banks as percent o f 17.4 29.6 total assets Interest rate spread Difference between lending and borrowing interest rate 16.0 7.7 Financial depth M2IGDP 29.4 36.9 Liquidity preference Cash&l 2 24.5 18.3 Credit to private sector Banking credit to private sector/GDP 25.2 29.1 Additional variants o f the model reveal no strong statistical links between the timing o f loans and the outcomes. That is, post-loan growth rates for these indicators were not, for the most part, significantly larger than pre-loan rates, although in several altemative models, post-loan improvements in variables were larger than pre-loan growth rates. Some indicators, however, declined as the number o f adjustment loans increased, an indication that countries that received multiple loans tended to perform worse than others, or more probably that they needed additional loans because they were having difficulties. Taken together, these results suggest that Bank involvement in the financial sector i s a component o f successful reform programs, but not necessarily the driving force behind them. As a final caveat, the definitions o f “with” and “without” borrowing are not “pure”: in some countries, such as Nepal and Bangladesh, the Bank maintained an active dialogue, but made no loans addressing financial sector reforms (until FY03, so would not be included in the “with” for this analysis). Thus, although these countries are included in “without borrowing”, the dialogue may have nevertheless had an impact on the financial sector. In other countries, including Chile and Kenya, the Bank made adjustment loans addressing financial reforms prior to the period under review, whose impact may have emerged only in later years. What this discussion points to i s the difficulty o f constructing a counterfactual. 66 Interest Rate Spread Figure 9.3: Median interest rate spreads: in 9.10 Although the spread between interest rates on countries that borrowed from the Bank, 1992-2002 deposits and loans i s far from an ideal measure o f efficiency for a number o f reasons, i t is used here as an 25 I imperfect proxy to capture changes in efficiency and to serve as one more indicator o f the evolution o f the banking system in Bank client countries.64 Median interest rate spreads are in Figure 9.3, with OECD countries shown as a point of comparison. Consistent with the picture o f concentration ratios, interest rate I ' 1992 1993' 1994 1995' 1996' IS97 1998' 1999' 2000' 2001 200; ' ' ' - -- -Lar~e SystemBorrow ing Ccuntnes Smll System Borrow ing Countries spreads in the larger systems are about the same as those 'OaCountires in the OECD countries. The medians are used because o f the wide differences among countries, particularly at the beginning o f the period. Uganda, for example, had large negative spreads in 1992-94, while Peru and Zambia had spreads in triple digits in some years. By the end o f the period, spreads had converged, although Brazil s t i l l had spreads in excess o f 40 percent by 2002 and Georgia, Lao PDR, and Malawi in excess o f 20 percent; interest rate spreads in most other countries were in single digits. 9.1 1 The results o f the DEC/OED model on changes in interest rate spreads are shown in Table 9.3. There was a significant decrease in Table 9.3: Annual growth rate of changes in spreads, of 1.7 percent per year in borrowing countries, interest rate spread: with and without Bank versus no significant decrease in the countries that did lending for financial sector r e f ~ m s not borrow from the Bank, suggesting that Bank Interest Rate Spread borrowing can be positively associated with the Annual growth rate efficiency o f banking systems. As in the model on With Bank lending -1.14* changes in concentration ratios, the models on changes Without Bank lending -0.18 Yes in interest spreads showed no difference in results for Number of countries 41 the number o f adjustment loans or the presence o f TA ,Z . 1 0.21 ~~ operations. * significantly different from zero at 1 percent. Health o f the financial system 9.12 The trend in health indicators o f financial systems among borrowing countries, particularly for the last five years, i s generally upward. However, the measures o f health - NPLs, capital adequacy, and profitability - all proved difficult to measure over the full period under review, for a number o f reasons. First, data were hard to find in the early part o f the period (1992-93): only ten borrowing countries had information on NPLs, for example, at the beginning o f the period, and fewer on capital adequacy. Second, banking reforms can significantly affect the measures o f health without necessarily changing the underlying dynamics o f banking operations that led to poor health to start with. For ~ ~ 64Interest rate spreads are affected by inflation rates, tax rates, reserve requirements, unequal subsidies available to some banks, and the extent o f NPLs in the system. In addition, very l o w spreads may drive banks to insolvency and are thus not necessarily associated with long term efficiency. Finally, the reliability o f the information o n interest rates in a given country for a given year may not be great. 67 example, the introduction and implementation o f stricter prudential regulations can lead to an increase in the measure o f NPLs, provisioning requirements, and shortfalls in provisioning, and to a drop in the measure o f capital adequacy, even if nothing in the lending operations o f the banks change (see B o x 9.2 for an example). By contrast, restructuring banks by taking NPLs o f f their books and recapitalizing them obviously results in an immediate drop in the measure o f N P L s and an increase o f capital adequacy o f the banking system. The real test o f banks’ health is what happens to these ratios over time, after these reforms. Thus, the interpretation o f changes in NPLs, profitability, and capital adequacy depends on the nature and timing o f the reforms rather than on the inherent health o f banking system. Box 9.2: Financial reforms can affect measures of banking health in both directions In Tunisia, reforms in the early 1990s supported by Bank lending caused most measures o f health to worsen and then in the late 1990s, further reforms supported by the Bank caused most measures t o precipitously improve. At the beginning o f the 199Os, Tunisia introduced stringent prudential regulations, whereby banks had t o adopt loan classification, loan loss provisioning, and minimum capital ratios consistent with international good practice (Basle guidelines). For the first time, virtually a l l the commercial banks in the country, including subsidiaries o f foreign banks, showed large NPLs (31 percent o f assets) and shortfalls in provisions, and failed t o meet the minimum capital requirement. Banks drew up action plans (a condition o f a B a n k loan) t o meet the requirements within three years; most banks made progress, but not enough. In FY99, under ECAL 1 1, the Government agreed to restructure banks by replacing N P L s with zero interest bonds: by this action, the N P L s immediately f e l l f r o m 23 percent in 1997 to 13 percent in 2000; capital adequacy more than doubled, from 6 t o 13 percent o f risk assets, and profitability increased modestly. Over t h i s period there has been little change in governance o f commercial banks, and more recent data show that NPLs have increased again, t o pre-ECAL I1levels. 9.13 From the twenty-one case study countries for which some information was available, and based on both qualitative and quantitative assessments o f progress in the health o f the banking system, at least fourteen o f the countries moved in the right direction in terms o f decreasing N P L s as a percent o f loans, particularly in the last h a l f o f the decade. Most o f these countries reduced N P L s from well over double digits to well under, although in 2000 (last year available) Yemen s t i l l had N P L s o f 34 percent o f assets (down from 40 percent), and by 2001, Brazil had decreased NPLs from 23 percent (1995) to 11 percent. In most o f these countries, the reduction in N P L s came from bank restructuring - Albania being the most dramatic example o f reducing bank NPLs that had reached 91 percent following the pyramid scheme collapse in 1997 to 35 percent the following year and, afler another round o f restructuring, to 7 percent before the banks were sold. Looking only at the last five years, when more information i s available for 41 countries that borrowed from the Bank for reforms, slightly more countries improved than deteriorated (17 versus 15), and the improvement in NPLs was greater on average than for countries that did not borrow from the Bank during this period (Table 9.4). 68 Table 9.4: Measures of banking health in borrowing versus non-borrowing countries Number o f 1998 2003 Change Number o f countries that changed countries Ratio of NPLs to assets Countries that borrowed from 41 14.4 7.9 - 6.5 17 improved the Bank for financial sector 15 deteriorated reforms 9 no change Countries that didn’t borrow 19 7.1 6.3 - 0.8 10 improved from the Bank 5 deteriorated 4 no change Capital to risk adjusted asset ratio Countries that borrowed from 41 17.3 19.8 2.5 21 improved the Bank for financial sector 12 deteriorated reforms 8 no change Countries that didn’t borrow 19 14.2 14.9 0.7 10 improved from the Bank 7 deteriorated 2 no change 9.14 In terms o f capital adequacy, there was very little information for the whole period; for the few countries that had information, most increased to double digits. Between 1998 and 2003, capital adequacy in 41 borrowing countries increased in more o f them and by larger amounts than among countries that did not borrow from the Bank (Table 9.4). There has been no significant trend in profitability o f banks in borrowing countries (or in non-borrowing countries) over the period. 10. Impact at a Country Level: Financial sector depth and stability Overview 10.1 Development o f the financial sector is often measured by a set o f “bottom line” indicators, which include: (i) depth, the extent to which the financial sector mobilizes resources; (ii) credit to the private sector, the extent to which the financial sector uses i t s resources to finance productive investments; and ( i ii) stability, the extent to which financial sectors can resist systemic insolvencies. This chapter examines trends in these indicators and presents the findings o f the O E D D E C model, which takes country factors into account and compares results in countries that borrowed from the Bank for financial sector reforms with those that didn’t borrow over the period under review. The caveats in the previous chapter related to these comparisons apply to the results in this chapter as well (see B o x 9.1). Financial sector depth: positive findings.. .. 10.2 Two indicators are used to measure progress in financial sector depth: (i) which i s a measure o f the money supply relative to the size o f the economy; M2/GDP,65 65 M2 i s the combination o f cash, demand deposits, and time deposits (International Financial Statistics, lines 34 and 35). 69 the higher the ratio, the deeper the financial sector depth; and ( ii) cash as a percent o f M2, which is a measure o f liquiditypreference that declines when the public i s willing to put i t s fhds into the banking system, and thus i s inversely related to public confidence in the system. The lower the ratio o f cash to M2, the higher the level o f confidence. 10.3 Over the period o f 1992-2002, financial sector depth in countries that borrowed from the Bank for financial sector reforms grew from an average o f 29 percent o f GDP to 36 percent, as shown in Figure 10.1. Given the significant financial turmoil and subsequent restructuring that occurred over this period in many borrowing countries (Box 6. l), this increase in average financial sector depth can be viewed as reasonably good progress. 10.4 On the measure o f liquidity preference, which should decline as confidence in the banking system increases, Figure 10.1 shows that this measure also moved in the right direction: cash as a proportion o f the money supply declined from 25 percent to 22 percent, indicating an increase in confidence. Figure 10.1: Financial sector depth (M2GDP) and liquidity preference (cash as a percent of money supply), in countries that borrowed from the Bank for financial reforms, 1992-2002 I 3 1 I I 1 -Bank I Borrowers -- OECD Countries*l I -Bank Borrow ers -- oH3D Countries’ * Excludes countries in the Euro zone; Source: WDI, 2004. 10.5 The results o f the OEDiDEC model are in Table 10.1. They show that in both the “with Bank lending” and “without Bank lending” group o f countries, financial sector depth as measured by M2/GDP grew by about 1.7 percent per year. There was no significant difference between the two groups. 10.6 Preference for liquidity in the form o f Table 10.1: Annual growth rates in financial sector depth and cash dropped by about one-half percent per confidence in the banking system: with and without Bank lending for financial sector reforms year in both groups, thus indicating a greater willingness to put resources into the banking M2IGDP Cash/M2 system and an overall increase in confidence. Annual growth rates In this model, there i s no significant difference With Bank lending 1.73* -0.48* between the two groups o f countries, although Without Bank lending 1.65* -0.37’ Significantly different? No No in several variations o f the model (the simple Number of countries 69 77 model and the one including policy controls), R2 0.38 0.17 the non-borrowing group showed a * significantly different from zero at 1 percent level significantly lower increase in confidence. 70 This could indicate that reforms undertakenwith Bank support aiming at reducing govemment role and increasing competition may have also increased public confidence in the banks. One interesting finding on changes in public confidence i s that it was significantly and inversely related to the number o f adjustment loans. That is, the higher the number o f adjustment loans the lower public confidence was among borrowing countries. A plausible explanation o f this finding i s that countries in deep financial trouble have more loans from the Bank for financial reforms (see B o x 6.1 o n this point) than countries that are not experiencing banking problems, and that the public i s responding to the banking problems by keeping their money in cash. .....but financial systems are still very shallow in many Bank clients 10.7 The results presented above do not show the wide variations among clients and the very shallow systems that still characterize many clients. Figure 10.1 shows that M2/GDP is still on average only about one-half o f the level o f OECD countries. Table 10.2 shows the distribution o f changes in the two indicators for the borrowing countries where information was available. Although M2/GDP increased in the majority o f countries in the sample, i t f e l l for almost one-third o f them (18 out o f 62 countries), and remained below 20 percent in 18 out o f 62 countries. Public confidence actually f e l l in over a third o f the countries (in 19 out o f 57 countries, the ratio increased, indicating a fall in term deposits). Table 10.2: Distribution in changes in measures o f financial sector depth, borrowing countries Change in indicator between 1992-94 and 2001-02 Percentage point change in depth >20 10-19.99 5-9.99 0-4.99 O 0 - (-4.99) - (-5) (-9.99) (-10) - (-19.99) < (-20) Total Number of countries Cash/M2 19 17 13 6 2 57 Figure 10.2: Credit to the private sector as a percent o f GDP in countries Credit to the private sector that borrowed from the Bank for financial reforms, 1992-2002 10.8 Credit to the private sector, measured by claims on the private sector by the banking system as a percent o f GDP, i s considered one o f the keys for economic growth. I t i s the main objective o f a banking system’s mobilization o f resources. Credit to the private sector over the period 1992-2002 in countries that borrowed --. from the Bank for financial sector reforms i s shown in Figure 10.2, and shows a small -Bank Borrowers OECD Countries increase over the period (25.2 to 27.5 percent Source: IFC and WDI, 2004. o f GDP). 71 lo*’ The Of the show that Table 10.3: Annual growth rates for credit to the credit to the private sector as a share o f GDP increased private sector: with and without Bank lending for sector reforms in both borrowing and non-borrowing countries (Table 10.3), but here the growth rate i s larger in non- Private Credit‘GDP borrowing countries: 1.7 percent per year in non- Annual growth rate With Bank lending 0.37* borrowing countries versus 0.4 per cent per year in Without Bank lending 1.65** borrowing countries. One explanation may be that Significantly different? Yes Number o f countries 71 within the group o f Bank borrowers, the more rapid R2 0.19 bank privatization and establishment o f higher *significantly different from zero at 10 percent standards o f prudential norms - requiring stricter loan ** significantly different from zero at 1 percent classification and provisioning, higher capital ratios, and stricter rules on interest rate accrual - may have combined to foster more prudent lending. Thus the slower growth o f lending may not be, in the first instance, a bad thing. 10.10 Nevertheless, credit to the private sector remains at a l o w level in most Bank borrowing countries; i t i s still only about one-fourth the level in OECD countries; it fell in about 40 percent o f the countries that borrowed from the Bank (24 out o f 60 for which information is available) and increased by less than 10 percentage points for another 40 percent (Table 10.4). At end 2002, credit to the private sector remained at a very l o w 10 percent o f GDP in 16 out o f 60 countries.66 10.1 1 And again, as in the case o f public confidence, private credit as a percent o f GDP was inversely related to the number o f adjustment loans: the higher the number o f adjustment loans the lower the access to credit, which may reflect the degree o f distress in the banking systems that called for repeated Bank lending. Table 10.4: Distribution in changes in access to credit, borrowing countries Change in indicator between 1992-94 and 2001-02 Percentage point change in access to credit >20 10-19.99 5-9.99 0-4.99 10 0-10 The Bank should develop a rating system, in partnership with other relevant institutions, for vulnerability to crisis, making use o f readily available information that can be used to engage countries in crisis prevention measures 79 and issues in crisis response. The Bank should also do a better j o b than in the past o f presenting assessments more candidly in documents. 9 The Bank should make internal arrangements to respond better to crisis by developing guidelines for dealing with crisis, which should include the possibility, i f circumstances warrant, o f lending liquidity support to countries experiencing a crisis without stipulating ambitious reforms (that may not be realized) as justification for the loan. > Coordination with the IMF and other I F I s in crisis assistance needs to be improved, and at the outset o f a crisis, the IFIs should reach quick agreement o n division o f responsibilities. 80 81 References Allen, Franklin and Douglas Gale. 2000. Comparing Financial Systems, MIT Press: Cambridge and London. Barajas, Adolfo, Roberto Steiner, and Natalia Salazar. 2000. “Foreign Investment in Colombia’s Financial Sector,” in Stijn Claessens and Marion Jansen, Editors, The Internationalization of Financial Services, Kluwer Law International: The Hague, Netherlands. Barth, James, Gerard Caprio Jr., and Ross Levine. 2001a. “Banking Systems Around the Globe: D o Regulation and Ownership Affect Performance and Stability?” in Mishkin, Frederic (ed.) Prudential Supervision: What Works and What Doesn ’t, National Bureau o f Economic Research Conference Report, Chicago: Univ. o f Chicago Press. Barth, James, Gerard Caprio Jr., and Ross Levine. 2001b. “Bank Regulation and Supervision. What Works Best?” World Bank, Policy Research Working Paper 2725. Barth, James, Gerard Caprio, Jr., and Ross Levine. 2001c. “The Regulation and Supervision o f Bank Around the World - A New Database,” World Bank, Policy Research Working Paper 2588. Beck, Thorsten, Robert Cull, and Afekheina Jerome. 2003. “The Effects o f State Ownership and Privatization o f Nigerian Banks,” Paper presented at World Bank Conference on Bank Privatization, Nov. 20-21, 2003. Beck, Thorsten and Ross Levine, 2002. “Industry Growth and Capital Allocation: Does Having and Market- or Bank-Based System Matter?” Journal o f Financial Economics, 64(2): 147-180. Beck, Thorsten, Asli Demirguc-Kunt, and Ross Levine. 2003. “Bank Concentration and Crises,” Paper presented at World Bank Conference on Bank Concentration and Competition, April 3-4,2003. Berger, Allen, N., George R. G. Clarke, Robert Cull. Leora Klapper, and Gregory F. Udell. 2003. “Govemance and the Efficiency o f Commercial Banks: Evidence from the Argentinian Banking System.” Paper presented at World Bank Conference on Bank Privatization, Nov. 20-2 1, 2003. Bonin, John P., Ifiekhar Hasan, and Paul Wachtel. 2003. “Bank Performance, Efficiency and Ownership in Transition Countries,” Paper presented at the Ninth Dubrovnik Economic Conference, sponsored by the Bank o f Croatia, June 26-28. 82 Boot, Arnoud W. A. and Anjan V. Thakor. 1997. “Financial System Architecture,” Review of Financial Studies, 10 (3): 693-733. Bossone, Biagio and Massimo Cirasino. 2001. “The Oversight o f the Payments Systems: A Framework for the Development and Governance o f Payment Systems in Emerging Economies,” Payments and Securities Clearance and Settlement Systems Research Series No. 1. Brock, Philip L. 2000. “Emerging from Crisis: Bank Privatization and Recapitalization in Chile.” In Bank Privatization: Conference Proceedings of a Policy Research Workshop. Edited by Harvey Rosenblum, pp. 97-1 18. Dallas, TX: Federal Reserve Bank o f Dallas. Calomiris, Charles W. and Charles P. Himmelberg. 1994. “Directed Credit Programs for Agriculture and Industry: Arguments from Theory and Fact,” World Bank Economic Review, Proceedings o f the 1993 World Bank Annual Conference on Development Economics, pp. 113-137. Caprio, Gerard, Patrick Honohan, and Joseph Stiglitz (Eds). 2001, “Financial Liberalization, How Far, How Fast?” Cambridge University Press. Caprio, Gerard and Daniela Klingebiel. 2003. “Episodes o f Systemic and Borderline Financial Crises,” Financial Sector Network. April 15,2005: http://wbln00 18.worldbank.or~tml/FinancialSectorWeb.nsf/~attachmentweb~/Ta bleonsystemicandnon- systemicbankingcrises Januar~212003/$FILE/Table+on+s~stemic+and+non- svstemic+banking+crises Januarv+2 1+2003 .pdf. Claessens, Stijn. 2005. “Finance and Volatility,” Chapter in Joshua Aizenman and Brian Pinto (Eds), Managing Volatility and Crises: A Practitioner’s Guide, (forthcoming). Cambridge University Press. Claessens, Stijn, Asli DemirgQ-Kunt, and Harry Huizinga. 2000. “The Role o f Foreign Banks in Domestic Banking Systems.” In Stijn Claessens and Marion Jansen, eds., The Internationalization o f Financial Services: Issues and Lessons for Developing Countries. Boston: Kluwer Academic. Claessens, Stijn and Luc Laeven. 2004. “What Drives Bank Competition? Some International Evidence.” Journal of Money, Credit, and Banking, Vol. 36, No. 3. Clarke, George, Robert Cull, Laura D’Amato, and Andrea Molinari. 2000. “On the Kindness o f Strangers? The Impact o f Foreign Entry on Domestic Banks in Argentina,” in Stijn Claessens and Marion Jansen, Editors, The Internationalization of Financial Services, Kluwer Law International: The Hague, Netherlands. Clarke, George R.G., Juan Miguel Crivelli, and Robert Cull. 2003. “The Impact o f Bank Privatization and Foreign Entry on Access to Credit in Argentina’s Provinces,” 83 Paper presented at the NinthDubrovnik Economic Conference, sponsored by the Bank o f Croatia, June 26-28. Clarke, George R. G., Robert Cull, and Maria Soledad Martinez Peria. 2001. “Does Foreign Bank Penetration Reduce Access to Credit in Developing Countries? Evidence from Asking Borrowers.” World Bank policy research working paper 2716. Clarke, George R. G., Robert Cull, and Maria Soledad Martinez Peria. 2004. “Structural Reform in Latin American Banking in the 1990s: H o w Deep? H o w Effective?” presented at Stanford Center for Intemational Development Conference o n Latin American Financial Systems and the Challenge o f Economic Growth, November 2004. Clarke, George R. G., Robert Cull, Maria Soledad Martinez Peria, and Susana Shnchez. forthcoming. “Bank Lending to Small Businesses in Latin America: Does Bank Origin Matter?” Journal of Money, Credit and Banking. Clarke, George R.G., Robert Cull, and Mary Shirley. 2004. “Empirical Studies o f Bank Privatization: Some Lessons” paper presented at World Bank Sixth Annual Financial Markets Development Conference, April 2004. Cull, Robert. 2004. “The Effects o f Ownership, Incentives, and Institutions on Financial Development: A Survey o f Recent Literature,” background paper for OED review. Cull, Robert, Lemma W. Senbet, and Marco Sorge, forthcoming. “Deposit Insurance and Financial Development,” Journal o f Money, Credit, and Banking. Dado, Marinela and Daniela Klingebiel. 2000. “Corporate Restructuring. Cross Country Experience with Decentralized Solutions”. World Bank, mimeo. Demirguq-Kunt, Asli and Enrica Detragiache. 2002. “Does Deposit Insurance Increase Banking Sector Stability? An Empirical Investigation,” Journal of Monetary Economics, 49(7): 1373-1406. Demirgiiq-Kunt, Asli, and Harry Huizinga. 1999. “Determinants o f Commercial Bank Interest Margins and Profitability: Some Intemational Evidence.” World Bank Economic Review 13 (2): 379-408. Demirguc-Kunt, Asli, L u c Laeven, and Ross Levine. 2003. “The Impact o f Bank Regulations, Concentration, and Institutions on Bank Margins,” Paper presented at World Bank Conference on Bank Concentration and Competition, April 3-4, 2003. Demirguc-Kunt, Asli and Vojislav Maksimovic. 2002. “Funding Growth in Bank-Based and Market-Based Financial Systems: Evidence from Firm-Level Data,” Journal of Financial Economics, 65(3): 337-363. 84 Demsetz, Harold. 1973. “Industry Structure, Market Rivalry, and Public Policy,” Journal ofLaw and Economics, 16(1): 1-9. Denizer, Cevdet. 2000. “Foreign Entry in Turkey’s Banking Sector, 1980-1997.’’ In Stijn Claessens and Marion Jansen, eds., The Internationalization of Financial Services: Issues and Lessonsfor Developing Countries. Boston: Kluwer Academic. Diamond, Douglas and Philip Dybvig. 1983. “Banks Runs, Deposit Insurance, and Liquidity”. Journal of Political Economy, 9 l(3): 40 1-419. European Bank For Reconstruction and Development (EBRD). “Transition Report”, various years. Feldstein, Martin. 2002. “Economic and Financial Crises in Emerging Market Economies: Overview of Prevention and Management,” National Bureau o f Economic Research, Working Paper 8837. Funke, Ilka. 2004a. “Assessment o f information on bank privatizations in legal and appraisal documents o f the 37 case study countries” background paper for OED. Funke, Ilka. 2004b. “Deposit Insurance - Reforms supported by the World Bank during the period 1993-2004,yy background paper for OED. Gershenkron, Alexander. 1962. Economic Backwardness in Historical Perspective, A Book o f Essays. Cambridge, MA: Harvard Univ. Press. Garcia, Gillian G.H.. 2001. “Deposit Insurance: Actual and Good Practices”. International Monetary Fund. Goldberg, Lawrence G., and Denise Johnson. 1990. “The Determinants o f US. Banking Activity Abroad,” Journal of International Money and Finance, 9(2): 123-37. Goldsmith, Raymond W. 1969. Financial Structure and Development. New Haven, CT: Yale University Press. Gorton, Gary and Andrew Winton. 1999. “Liquidity Provision, the Cost o f Bank Capital, and the Macroeconomy.” Mimeo, Institute for Financial Studies, Carlson School o f Management, University o f Minnesota. Greenwald, Bruce C., and Joseph E. Stiglitz. 1986. “Externalities in Economies with Imperfect Information and Incomplete Markets,” Quarterly Journal o f Economics, lOl(2): 229-264. Haber, Stephen, and Shawn Kantor. 2003. “Getting Privatization Wrong: The Mexican Banking System, 1991-2002,” Paper to be presented at World Bank Conference on Bank Privatization, Nov. 20-2 1,2003. 85 Honohan, Patrick and Daniela Klingebiel. 2003. “The Fiscal Cost Implications o f an Accommodating Approach to Banking Crises,” Journal ofBanking and Finance, 27(8): 1539-1560. Hanson, James. 2003. “Banking in Developing Countries in the 1 9 9 0 ~ Policy ~” Research Working Paper 3 168, World Bank. Hinds, Manuel. 2003. “Bank Assistance to Financial Sector Development: A Glimpse into the Future,” background paper for OED. International Monetary Fund (IMF). 2000. “International Capital Markets: Developments, Prospects, and Key Policy Issues. International Monetary Fund (IMF). 2001. “Strengthening IMF-World Bank Collaboration in Country Programs and Conditionality”, SM/01/219. International Monetary Fund (IMF). 2002a. “Staff Report on Operationalizing Fund- Bank Collaboration in Country Programs and Conditionality”. International Monetary Fund (IMF). 2002b. “Strengthening IMF-World Bank Collaboration in Country Programs and Conditionality - Progress Report” SM/02/27 1. International Monetary Fund (IMF). 2003. “Evaluation o f the Role o f the IMF in Recent Capital Account Crises: Main Report”. Kenen, Peter B.. 2002. “The International Financial Architecture: Old Issues and New Initiatives,” International Finance 5: 1, 2002, pp. 23-45. Kiguel, Miguel, and Nicolas Dujovne. 2003. “An Evaluation o f the World Bank’s assistance to the financial sector in Latin American countries: the cases o f Argentina, Brazil, Colombia and Mexico during the nineties,” background paper for OED Review. Klingebiel, Daniela. 2000. “The U s e o f Asset Management Companies in the Resolution o f Banking Crises: Cross-Country Experience,” Policy Research Working Paper 2284, World Bank. L a Porta, Rafael, Florencio Lopez-de-Silanes, and Andrei Shliefer. 2002. “Government Ownership o f Banks,” Journal ofFinance, 57(1): 265-301. Levine, Ross, 1998. “The legal environment, banks, and long-run economic growth” Journal o f Money, Credit, and Banking 30,688-726. Levine, Ross, 1999. “Law, finance, and economic growth.” Journal o f Financial Intermediation, 8,36-67. 86 Levine, Ross, N. Loayze, and Thorsten Beck. 2000. “Financial intermediation and growth: causality and causes.” Joumal o f Monetary Economics, 46,31-77. Levine, Ross. 2002. “Bank-Based or Market-Based Financial Systems: Which i s Better?” Journal of Financial Intermediation, 1l(4): 398-428. Levy, Fred. 2003 “World Bank Assistance for Financial Sector Development in the ECA Transition Economies,” background paper for OED Review. Long, Millard. 2003a. “Bank Response to Crisis in the Financial Sector,” background paper for OED review. Long, Millard. 2003b. “Review o f Bank Assistance to the Financial Sector, 1993-2002, South Asia Region,” background paper for OED Review. Martinez Peria, Maria Soledad and Sergio L. Schmukler. 2001. “Do Depositors Punish Banks for Bad Behavior? Market Discipline, Deposit Insurance, and Banking Crises,” Journal ofFinance, 56(3): 1029-105 1. Miskhin, Frederic, S. 1999. “Lessons from the Asian crisis” Journal of International Money and Finance, 18 (1999) 709-723. Mozes, Dan. 2003. “OED Review o f Bank Assistance to the Financial Sector, 1992-2002, Sub Saharan Africa”, background paper for OED Review. Operations Evaluation Department. 1998. Financial Sector Reform: A Review of World Bank Assistance. Report no. 17454. Washington, D.C.: World Bank. Operations Evaluation Department. 2000. Albania Country Assistance Evaluation. Report no. 21390. Washington, D.C.: World Bank. Operations Evaluation Department. 2001a. India Country Assistance Evaluation. Report no. 22134. Washington, D.C.: World Bank. Operations Evaluation Department. 2001b. Mexico Country Assistance Evaluation. Report no. 22498. Washington, D.C.: World Bank. Operations Evaluation Department. 2005. OED Review o f Bank Lending for Lines of Credit. Report no. 3 113 1. Washington, D.C.: World Bank. Peltzman, Sam. 1977. “The Gains and Losses from Industrial Concentration”. Journal of Law and Economics, 20(2): 229-263. Rajan, Raghuram G. and Luigi Zingales. 2001. “Financial Systems, Industrial Structure, and Growth”. Oxford Review o f Economic Policy, 17(4): 467-482. Ramachandran, S.m. 2003. “China’s Financial System and the World Bank’s Role during the 1990s,” background paper to OED’s China Country Assistance Evaluation. 87 Scott, David. 1999. “The Korean Financial Sector Crisis: World Bank Policy and Operational Support”. Sherif, Khaled, Michael Borish, Alexandra Gross. 2003. “State-owned banks in the transition: origins, evolution, and policy responses”, World Bank report no. 25832. Stiglitz, Joseph. 1994. “The Role o f State in Financial Markets”. World Bank Economic Review, Proceedings o f the 1993 World Bank Annual Conference on Development Economics, pp. 19-52. Stulz, Rene. 2001. “Does Financial Structure Matter for Economic Growth? A Corporate Finance Perspective” in Asli Demirguc-Kunt and Ross Levine, eds., Financial Structure and Economic Growth: A Cross-Country Comparison o f Banks, Markets, and Development. Cambridge and London: M I T Press. pp. 143-88. U. S. General Accounting Office. 2003. “International Financial Crises - Challenges Remain in IMF’s Abiltiy to Anticipate, Prevent, and Resolve Financial Crises,” GAO-03-734. World Bank. 1989. World Development Report. Financial Systems and Development, New York, Oxford University Press. World Bank. 2001. Finance for Growth: Policy Choices in a Volatile World. Policy Research Report. New York: Oxford University Press. World Bank and IMF. 2004. Financial Sector Assessment Program. April 22,2005. http:llu7b11100 18.worldbank.orrz/FPS/fsapcountrycFb.nsf/FS~exte~~lcount~epo~ ts?OpenPaae&Start=l &Count=lOOO&ExpandView#A World Bank. 2004. “Practitioners o f Development Seminar Series: The Growth Experience: Lessons from the 1990s.’’ July 21,2005. http://info.worldbank.org/etools/vod/PresentationView.asp?PID=l3 15 &EID=328 88 Annex 1 Data on trends and patterns in lending and non-lending Lending classified as finance includingLOC Table 1: Lending by Region, including and excluding LOC, FY93-03 Lending Lending Amount Percent o f Amount including Region excluding Percent Of Region LOC, $M Lending LOC, $M AFR 806.6 3 630.6 2 EAP 6,764.0 12 6,357.0 11 ECA 6,386.4 14 5,257.2 11 LCR 8,145.8 14 7,499.8 13 MNA 990.5 8 845.5 7 SAR 1,711.7 5 659.8 2 Total 24,805.1 11 21,249.9 9 Figure 1: Lending classified as finance, including LOC, as percent of lending, by Region Africa East Asia and Paciflc 60% , I & 50% 50% 20% 10% 0% . . . . / I 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 I1 Adjustmnt 0 investment Adjustment 0 lnvestrrent L19 Crisis Ig Europe and Central Asia Latin America and Carribean 60% 7 I 60% 7 I 50% 50% 40% 40% 30% 20% 10% 10% 0% 0% I Adjustmnt 0 Investment 0 Crsis 1 11 I ~- Middle East and North Africa South Ada 60% 7 I 60% X E 50% 4 50% : E 20% 10% 0% , & \& & , $ ,+ & , $ , ,+ & & % e ," $8"& , =Adjustment 0 Investment I 89 Annex 1 Data on trends and patterns in lending and non-lending Figure 2: Lending classified as finance excluding L O C as percent of total lending, by Region Africa East Asia and Pacific 60% - 60% Y 5 50%- 50% n u iij f 40% - 40% 30% 20% - - 10% 0% Adjustmnt 0 hvestmnt 1 m Adjustmnt 0 hvestmnt Q ! Wiis 1 Europe and Central Asia Latin America and Carribean 60% , I Adiustmnt 0 hvestmnt Crisis I Middle East and North Africa South Asia 60% I 1 I 60% 7 1 E 20% P 0% EAdjustrrent o Investrent I I DAdjustrrent o hvestmnt I 90 Annex 1 Data on trends and patterns in lending and non-lending Figure 3: Breakdown of operations with financial sector components by sector board (including lines o f credit) Urban Other Rural Sector 8.6% Notes: Total number of Public Sector 9.6% Governance Social Protection-I .8% Energy and mining-1.3% Education-0.5% Private Sector Environment-0.3% Development HNP- 0.3% 22.6% Poverty Reduction-0.3% Water Supply and Economic Policy Sanitation-0.3% 91 Annex 1 Data on trends and patterns in lending and non-lending Figure 4: Number of loans, including multi-sector loans and lines of credit, by Region, as percent o f Region's loans Africa East Asia and Pacific 40% 1 1 u) 40% 5 35% .- g 30% 8 35% 'K al 30% 5 25% - ' m U 25% 0 20% c 5 20% U o 15% CI % 15% 5 : 10% 5 U 10% 5% er f 5% 0% P 0% classified as finance I i classified under other sectors ! /m classified as finance classified under other sectors] I 1 I I I Europe and Central Asia Latin America and Carribean 40% 7 1 40% I 1 u) ti 35% 'p30% 5 m 25% 5 * 20% 6 15% z 10% 5% 0% Im classified as finance B classified under other sector4 classified as finance Bl classified under other sector South Asia Middle East and North Africa 40% 40% 8 35% 3 % 35% 30% 30% e - *- 9, s 25% 25% 20% 2 20% r 0 15% 0 15% 1 CI 10% u 10% E 5% !! 5% g 0% g 0% classified as finance I !!!classified under other sectors 1 classified as finance E! classified under other sectors L 92 Annex 1 Data on trends and patterns in lending and non-lending Figure 5: Trends in lending in support o f specific financial sector reforms as percent o f all projects with financial sector reforms in that year Banking and non-banking financial institutions 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total I Regulation and Legislation 0 Supervision Restructuring and Privatization I 100% U 90% 80% 5 s 70% E $ E 1 8 60% 50% e 8 40% n u # 30% c 3 20% U 0% is 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total Q Capital Markets IInsurance 93 Annex 1 Data on trends and patterns in lending and non-lending Table 2: Bank lending for capital market reform, number of projects, by Region and total Region 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total AFR 2 1 2 1 2 8 EAP 1 1 2 1 5 ECA 4 3 3 1 2 13 LCR 1 3 2 2 2 2 1 1 1 1 16 MNA 2 1 1 2 6 SAR 0 Total 3 4 1 10 3 6 10 4 2 1 3 48 Figure 6: Number of ESW by region, FY93-03 70 _I 1 60 5 u) 50 40 c 2 P, 30 20 5 c 10 0 AFR E4P EC4 LAC MNA SAR . [ ESWB FSP AJ 94 Annex 2 Annex on outcomes o f Bank loans Annex on Chapter 6: Outcome ratings o f Bank loans Table 1 : Upgrades and downgrades* by source o f rating o f financial sector components of multi-sector loans ICR validated Grand Total Source of rating -------------------OED-------------------- by OED Desk Review PAR Total OED 27 60 87 12 99 upgraded 1 8 9 1 10 downgraded 4 7 11 2 13 disconnect 3 1 2 1 3 *Upgrade and downgrade refer to a change in outcome rating between the overall project rating and the component rating Table 2: Outcome ratings and sector classification o f loans Number o f Sector classification of loans Number of satisfactory Percent loans loans satisfactory Financial sector adjustment loans plus financial sector 142 106 75 components in multi-sector loans All adjustment loans (excl. financial sector) 243 192 79 Financial sector TA loans plus financial sector components 49 38 78 o f multi-sector TA loans All investment loans (excluding financial sector) 839 642 77 Sector classification Financial Sector adjustment loans 43 38 88** Financial Components o f multi-sector adjustment loans 99 68 69** Financial sector TA loans 17 14 82 Financial components o f multi-sector TA loans 32 24 75 Sector classification and 2003 CPIA rating Financial sector loans, l o w CPIA rating (CPIA between 1 17 15 88** and 3.5) Components o f multi-sector loans, l o w CPIA rating (" ") 51 34 67** Financial sector loans, high CPIA rating (above 3.5) 43 37 86** Components o f multi-sector loans, high CPIA rating (above 80 58 73** 3.5) Sector classification and 2002 p e r capita income Financial sector loans, l o w income 18 13 72 Components o f multi-sector loans, l o w income 54 37 69 Financial sector loans, middle income 42 39 93** Components o f multi-sector loans, middle income 77 55 71** * significantly different at the 10 percent level o f confidence ** significantly different at the 5 percent level o f confidence 95 Annex 2 Annex on outcomes o f Bank loans Table 3: Outcome ratings and country characteristics Country characteristics Number Number o f Percent satisfactory of loans satisfactory loans Income Levels 1993 per capita income Loans in l o w income countries 80 63 79 Loans in middle income countries 109 79 72 2002 per capita income Loans in l o w income countries 72 50 69* Loans in middle income countries 119 94 79* 2002 per capita income, transition countries separate Loans in l o w income countries, excl. Transition 56 34 61** Loans in middle income countries, excl. Transition 62 49 79** Loans in transition countries 73 61 84 CPIA Ratings*** 1996 ratings Loans in l o w CPIA (1-3.5) countries 127 94 74 Loans in high CPIA (3.6-5.0) countries 53 41 77 2003 ratings Loans in l o w CPIA (1-3.5) countries 68 49 72 Loans in high CPIA (3.6-6.0) countries 123 95 77 2003 ratings, transition countries separate Loans in l o w CPIA (1-3.5) countries, excl. transition 42 25 60** Loans in high CPIA (3.6-6.0) countries, excl. transition 76 58 76** Loans in transition countries 73 61 84 * simificant at the 10 uercent level o f confidence ** Zgnificant at the 5 percent level o f confidence * ** The index was changed from a five point scale to a six point scale, and this is reflected in the categories for 1996 and 2003, which show different ranges for CPIA values. The analysis was also carried out with different cuts for 1996 CPIA ratings (1-3.0; 3.1- 6.0) with similar results; for 2003, there were too few observations in the CPIA rating 1-3.0 to be meaningful. 96 Annex 2 Annex on outcomes of Bank loans Table 4: Outcomes of adjustment loans with and without technical assistance With and without technical assistance loans Number of Number of satisfactory Percent loans loans satisfactory All financial sector adjustment loans and Loans with associated TA loans 48 36 75 Loans without associated TA loans 94 70 74 Income level and TA L o w income countries: loans with TA (excl. 17 11 65 L o w income countries: loans without TA ( " 20 13 65 Middle income countries: loans with TA 12 9 75 Middle income countries: loans without TA ( 39 30 77 1996 CPIA and TA* L o w CPIA countries, loans with TA (excl. 11 10 91** L o w CPIA countries, loans without TA ( " ) 16 9 56** H i g h CPIA countries, loans with TA (excl. 17 9 53 H i g h CPIA countries, loans without TA ( " ) 40 31 78 Note: transition countries Transition countries: loans with TA 19 16 84 Transition countries: loans without TA 1 4 JJ 77 L I .. 77 97 Annex 3 Outputs at a country level, with and without Bank lending Table 1: Capital adequacy, 1998 and 2003, with and without Bank lending With Bank lending for legal and Without Bank lending for legal and regulatory reforms regulatory reforms Number o f countries 11 19 1998 2003 No. o f countries 1998 2003 No. o f countries with changes with changes 1. Average minimum capital-asset 9.50 9.60 2 up, 0 down 8.40 10.20 7 up, 0 down ratio requirement Number of countries requiring deduction Number o f countries requiring deduction 2. Items deducted from capital: Market value o f loan losses 5 5 2 up, 3 down 10 11 5 up, 4 down Unrealized securities losses 6 4 9 14 Unrealized for. exchange losses 6 6 13 12 Table 2: Comparison of classification of loans, by average number of days 1998_______________ ---------------2003 ________________ --_---________ Number o f countries Substandard Doubtful Loss Substandard Doubtful Loss with changes Countries with Bank lending 87 173 272 64 129 256 4up,2down Countries without Bank lending 74 153 318 79 164 332 2up,2down Table 3: Indicators on strength of financial regulations for banking, transition countries 1998 2002 Change 1998 2002 Change With Bank lending Without Bank lending Armenia 2 3- 0.66 Albania 2- 1+ -0.33 Azerbaijan 2- 1 -0.66 Belarus 1 2 1 Bosnia-Herzegovina 1 1 0 Croatia 3 2 -1 Bulgaria 3 3 0 Czech Republic 3 3 0 Hungary 4 3+ -0.67 Estonia 3 4- 0.66 Latvia 3 4- 0.66 Georgia 1 2+ 1.33 Lithuania 3- 3+ 0.67 Kazakhstan 2 3- 0.66 FYR Macedonia 2 3- 0.66 Kyrgyz Republic 2 2- -0.34 Romania 3- 3+ 0.67 Moldova 2 3 1 Russian Federation 3- 3- 0 Poland 4- 3+ -0.33 Tajikistan 1 3 2 Slovak Republic 3- 3- 0 Ukraine 2 2+ 0.33 Slovenia 3 3 0 Uzbekistan 2- 2- 0 Average 0.36 Average 0.20 Note: T h e EBRD indicator for each country i s a composite measure, scaled from 1 to 4, with pluses and minuses; an increase from 2 minus to 2 was counted as an increase of 0.33; from a 2 to a 2 plus was 0.33, etc. Source: EBRD Transition Report, various years 98 Annex 3 Outputs at a country level, with and without Bank lending Table 4: Indicators on strength of regulations for securities markets and non-bank financial institutions, transition countries W i t h Bank lending Without Bank Lending 1997 2003 Change 1997 2003 change Armenia 1 2 1 Albania 2- 2- 0 Croatia 2+ 3- 0.33 Azerbaijan 1 2- 0.66 Georgia 1 2- 0.66 Belarus 2 2 0 Kyrgyz Republic 2 2 0 Bulgaria 2 2+ 0.33 Romania 2 2 0 Czech Republic 3 3 0 Ukraine 2 2 0 Estonia 3 3+ 0.33 Uzbekistan 2 2 0 FYRMacedonia 1 2- 0.66 Hungary 3+ 4- 0.33 Kazakhstan 2 2+ 0.33 Latvia 2+ 3 0.67 Lithuania 2+ 3 0.67 Moldova 2 2 0 Poland 3+ 4- 0.33 Russian Federation 3 3- -0.34 Slovak Republic 2+ 3- 0.33 Slovenia 3 3- -0.34 Tajikistan 1 1 0 Turkmenistan 1 1 0 Average 0.28 Average 0.22 99 Annex 3 Outputs at a country level, with and without Bank lending Table 5: Change in government ownership of bai CS with Bank lending for privatization without Bank lendingfor privatization Albania Macedonia Bangladesh Argentina Madagascar Benin Armenia Malawi Chile Azerbaijan Mali Congo, Rep. Brazil Mongolia Costa Rica Bulgaria Morocco Czech Republic Burkina Faso Mozambique Dominican Republic Cameroon Nicaragua Egypt, Arab Rep. Chad Pakistan Ethiopia Colombia Philippines India Cote dIvoire Poland Kenya Croatia Romania Lao PDR El Salvador Slovak Republic Lesotho Georgia Slovenia Mauritania Ghana Tanzania Mauritius Hungary Togo Moldova Kazakhstan Tunisia Nigeria Kyrgyz Republic Uganda Oman Latvia Ukraine Panama Lithuania Yemen, Rep. Peru Sri Lanka Venezuela, RB Zambia Table 6: List of OECD countries Australia Austria Belgium Canada Denmark Finland France Germany Greece Iceland Ireland Italy Japan Luxembourg Netherlands New Zealand Norway Portugal Spain Sweden Switzerland United Kingdom United States 100 Annex 4 List of countries for desk studies, value o f operations, by Region, and as percent o f lending AFR Value of Loans, $M LCR Value of Loans, $M Burkina Faso 25 Brazil 808 Cameroon 545 El Salvador 50 Chad 85 Nicaragua 150 Cote d'lvoire 200 Peru 400 Ghana 340 Total 1,408 Guinea 23 As percent of regional lending 28% Madagascar 221 MNA Mozambique 420 Algeria 750 Tanzania 134 Morocco 600 Togo' __ Tunisia 412 Uganda 226 Yemen, Rep 82 Zambia 262 Total 1,844 Total 2,482 As percent of regional lending 85% As percent of regional lending 74% SAR EAP Pakistan 850 Lao PDR 57 Total 850 Mongolia 42 As percent of regional lending 74% Philippines 500 Vietnam 500 Total 1,099 As percent of regional lending 78% ECA Albania 100 Armenia 285 Georgia 195 Hungary 225 Kazakhstan 540 Lithuania 179 Macedonia 215 Moldova 120 Poland 450 Romania 1,230 Slovak Republic 257 Ukraine 1,410 Total 5,206 As percent of regional lending 61% Total number of case countries 37 Total value of loans in case countries 12,888 As percent of all countries borrowing for financial reforms** 54% As percent of total Bank lending for finance ** 59% * Togo had a TA operation that was to be followed by an adjustment credit that did not materialize. Included here because the TA operation was a substantial part o f Bank program in sector. ** Excluding crisis countries 101 Annex 5 Outcomes at a country level Tables 1-4: Measures o f contestability, 1998 2003 - Table 1: Entry requirements: average number of licenses 1998 2003 With Bank borrowing 7.7 7.6 Without Bank borrowing 7.4 7.3 Scale: No = 0; Yes = 1 Table 2: Minimum capital requirements: average change Change between 98-03 With Bank borrowing 1.04 Without Bank borrowing 1.oo Scale: decrease = 0; no change = 1; increase = 2 Table 3: Minimum capital requirements: number of countries that changed, 1998 and 2003 With Bank lending Without Bank lending Decrease 5 5 N o change 13 19 Increase 6 5 Total 24 29 Table 4: Restrictionson activities 1998 2003 With Bank borrowing 2.55 2.45 Without Bank borrowing 2.53 2.62 Scale: unrestricted=l; permitted = 2; restricted = 3; prohibited = 4; higher average indicates more restrictive 102 Annex 5 Outcomes at a country level Table 5: Foreign Ownership with Bank lending without Bank lending Argentina Benin Armenia Botswana Azerbaijan Chile Brazil Czech Republic Burkina Faso Estonia Cameroon Gabon Chad Guinea Colombia Kenya Cote d'lvoire Korea, Rep. Georgia Lesotho Ghana Mauritania Hungary Mauritius Kazakhstan Moldova Kyrgyz Republic Niger Lithuania Nigeria Macedonia, FYR Peru Madagascar Russian Federation Malawi Senega I Mali South Africa MBxico Swaziland Mozambique Tajikistan Poland Thailand Tanzania Venezuela Togo Zambia Turkey Zimbabwe Uganda ~~ Table 6: Concentration, interest rate spread, financial sector depth, liquidity preference, and credit I the private sect .* with Bank riding____________ -_-----_-- Bank lending---------- ______without Albania Lithuania Bangladesh Algeria Macedonia, FYR Benin Argentina Madagascar Botswana Armenia Malawi Cambodia Azerbaijan Malaysia Chile Bolivia Mali China Bosnia and Mauritania Costa Rica Herzegovina MBxico Czech Republic Brazil Moldova Dominican Republic Bulgaria Mongolia Egypt, Arab Rep. Burkina Faso Morocco Estonia Cameroon Mozambique Ethiopia Central African Nicaragua Gabon Republic Niger Gambia, The Chad Pakistan India Colombia Peru Iran, Islamic Rep. Congo, Democratic Philippines Kenya Republic of Poland Lebanon Cote d'lvoire Romania Lesotho Croatia Russian Federation Mauritius Ecuador Slovak Republic Nepal El Salvador Slovenia Nigeria Georgia Tanzania Oman Ghana Thailand Panama Guatemala Tunisia Papua New Guinea Guinea Turkey Paraguay Honduras Uganda Senegal Hungary Ukraine South Africa Indonesia Uruguay Sri Lanka Jamaica Vietnam Swaziland Jordan Yemen, Rep. Syrian Arab Republic Kazakhstan Zambia Togo Korea, Rep. Trinidad and Tobago Kyrgyz Republic Venezuela Lao PDR Zimbabwe Latvia *c initions and sources o f information for these variables are in Annex 6 103 Annex 5 Outcomes at a country level Table 7: CaDital Markets With Bank lending with Bank lending but not for capital markets without Bank lending at all Argentina Cote d'lvoire Bangladesh Bolivia Ecuador Botswana Brazil Hungary Chile Colombia Malaysia China Croatia MBxico Czech Republic Ghana Pakistan Egypt, Arab Rep. Indonesia Philippines India Jamaica Poland Iran, Islamic Rep. Jordan Russian Federation Kenya Korea, Rep. Slovak Republic Mauritius Morocco Slovenia Nigeria Peru Thailand Oman Romania Turkey Panama Tunisia South Africa Uruguay Sri Lanka Swaziland Trinidad and Tobago Venezuela Zimbabwe 104 Annex 6 Definitions and sources of information Table 1: Financial in Lability* and Bank borrowing for f iancial sector reforms, 1995-2002 Countries without systemic instability Countries with systemic instability Countries that didn't Count: 9 Count: 13 borrow from Bank Botswana Gabon Burundi Nigeria Costa Rica Gambia Clzech Rep. Parguay EaPt Senegal Djibouti Sao Tome & Principe Ethiopia Trinadad & Tobago Estonia Swaziland India Eritrea Venezuela Kenya Zimbabwe Liberia Borrowed from Bank Borrowed during instability and improved Count: 18 Count: 15 Angola T A only Lesotho T A only Armenia Mexico Belarus T A only Mauritania* Brazil Mozambique Central African Rep.* Mauritius T A only Bulgaria Nicaragua Chad Rwanda Burkina Faso Poland Ghana Tajikistan T A only Cameroon Russia Guatemala Togo* T A o n l y Croatia Slovenia Guinea* Tunisia Kyrgyz Rep. Tanzania Hungary* Ukraine* Macedonia Jamaica* Zambia* Borrowed during instability but didn 't improve Count: 23 Albania Latvia Argentina Malaysia Azerbaijan Niger Bolivia Romania Bosnia-Herzegovina Sierra Leone Cape Verde Slovakia Congo, Dem. Rep. Thailand Congo, Rep. T A only Turkey Ecuador Uganda Georgia Uruguay Indonesia Yemen Korea, Rep. o f Borrowed and instability followed Count: 2 Jamaica Ukraine * Financial instability as defined in Capri0 and Klingebiel(2003): banking systems in which much or a l l of the capital i s exhausted, based on official statistics or the estimationof experts familiar with the banking system in that country. 105 Annex 6 Definitions and sources of information OED/DEC model Indicators used to measure outputs, outcomes, and impact on the financial sector Variable D e j hition Sources Reference or comment Government Share of banking assets L a Porta et a1 (2002); Sherif et Definitions vary ownership o f banks held by government a1 (2001): EBRD Transition slightly by source: in Report, various years; Barth et Laporta et al, it was a1 (2001); Mozes (2003); share o f assets o f the Dujovne and Kiguel(2003) top 10 banks Foreign ownership Fraction o f the banking Barth et a1 (2001); IMF, 2000 o f banks system’s assets in banks that are 50 percent or more foreign owned Concentration Ratio Share o f assets held by Various Bank documents three largest banks as percent o f total assets Interest Rate spread Difference between SIMA lending and borrowing interest rate Financial sector M2IGDP SIMA depth Liquidity preference Cash/M2 I IFS I Lines (34+35) - Lines (24+25) Credit to private Banking claims on private SIMA sector sector/GDP 106 Annex 6 Definitions and sources o f information OEDDEC model OED/DEC models The basic model tested was: Where: Yit - M2/GDP, Private Credit/GDP, Cash/M2, Interest Rate Spreads, or Concentration Country i, Year t (I.. .I2) a - county-speciJicfixed effect wb - world bank lending for financial sector reforms between FY93 and 01 no-wb - no world bank lending for financial sector reforms between FY93 and 01 aa'j - number o f adjustment loans X - vector o f macro/institutional controls (inflation, CPIA, deficits, etc.) Variations on this model were as follows: Where the macro/institutional controls were excluded; Where the macro/institutional controls were included, along with: ref - vector of dummiesfor reform areas covered (privatization, regulation/supervision, microfinance, etc.) 107 Attachment 1 Management Response OED Review o f Bank Assistance for Financial Sector Reform April 28,2005 I.Introduction 1. Management welcomes OED’s comprehensive review o f Bank assistance for financial sector reform during the decade 1993 to 2003. The review provides a rigorous discussion o f the Bank’s financial sector program. W e are happy to note the review’s key finding that the objectives o f Bank assistance in the financial sector generally followed good practice. W e also appreciate the recommendations o f the Review. This response summarizes the main findings and conclusions o f the OED Review. I t then sets forth Management’s comments o n the analysis, conclusions, and recommendations. The Management Action Record i s attached. 11. Summary o f OED’s Findings and Recommendations 2. The main findings o f the review are: 0 Bank assistance in the financial sector to most borrowing countries in the past decade i s associated with improvements in bank governance, efficiency measures, financial sector depth and access to credit. Challenges remain in improving the impact o f reform programs o n financial depth and private sector access to credit. 0 The objectives o f Bank assistance generally followed good practice. Consistency in the approach to reforms should be improved, especially in the areas o f payments systems, deposit insurance schemes, and capital market development. 0 Outcomes o f operations under the responsibility o f Regional units that were members o f the Financial Sector Board were significantly better than Bank-wide averages for other sectors and also than outcomes o f financial sector components o f multi-sector loans. The latter points to a strong quality assurance role for the sector board as well as the need for strong support from financial sector officials in borrowing countries. 0 Interms o f Bank support for financial sector reforms in crisis countries, which account for 50 percent o f the lending reviewed, the review found that the Bank was ill-prepared to respond quickly in the earlier crises (Mexico, Thailand, Korea, and Indonesia) and better prepared in Argentina, Russia, and Turkey. OED outcome ratings o f closed operations in crisis countries were significantly lower than for non-crisis lending. Collaboration with the IMF in these crisis countries has not always been smooth. 3. Recommendations. The following are the recommendations for Management: 0 The Bank’s financial sector anchor should provide more guidance for Bank staff and client countries in areas such as restructuring o f banks, asset management companies, privatization o f banks, promotion o f capital markets, and strengthening o f legal, regulatory, and supervisory environment, with a particular focus o n implementation. The financial sector network should also be more pro-active in quality control o f financial sector components in multi-sector loans. 108 Attachment 1 0 The Bank should develop monitorable indicators to assess progress on objectives in the area o f prudential regulations and supervision for financial intermediaries. On support for countries prior to and following crisis, the Bank should: (i) develop a rating system, in partnership with other relevant institutions, for vulnerability to crisis, and present i t s risk assessments more candidly in its documents; (ii) make internal arrangements to respond better t o crisis by developing guidelines for dealing with crisis, including the possibility o f liquidity support to countries experiencing a crisis without stipulating ambitious reforms; and ( iii) coordinate better with the IMF and other IFIs, and at the outset o f the crisis, I F I s should reach quick agreement on division o f responsibilities. 111. Management Comments Impact 4. The OED Report documents that after a decade o f borrowing from the Bank for financial sector reforms, most o f the 96 borrowing countries have witnessed improvements in their financial sectors, in terms o f ownership and governance o f banks, efficiency measures, financial sector depth, and access to credit. These improvements can be associated with Bank borrowing, in that financial sector outcomes in countries that borrowed from the Bank for financial sector reforms are generally better than in countries that did not. Nevertheless, in most o f the countries, the financial sectors deepened only modestly and remain relatively shallow, and private sector access to credit remains low. 5. The Lagged Impact o f Financial Sector Reforms. The Bank’s recent work “Economic Growth in the 1990s: Learning from a Decade o f Reform,”69 shows the importance o f financial infrastructure and institutions for finance, especially in ensuring efficient credit allocation and better access to financial services. The same work showed that the greatest impact o f financial reforms o n the institutional changes in the sector occurred in the latter h a l f o f the 1990s with the growing movement away from state-owned banks. These reforms, along with improvements in market discipline and supervision and regulatory capacity, proved to take longer to carry through, and may have limited the gains from policy liberalization over the decade under review by OED. Moreover, reforms have often resulted in more conservative loan provisioning and write-off policies, which have caused the capital base o f banks to shrink, at least initially, thus reducing their capacity to lend. This, combined with institutional changes, has encouraged more prudent lending that has led to short-run reductions in private sector credit over the period o f transition following reforms. It i s likely, therefore, that expecting well-developed financial systems that provide increased outreach within a decade or less o f policy and institutional reforms i s unrealistic as a yardstick for assessing the impact o f financial sector reforms and associated Bank assistance.’’ 6. Macroeconomic Considerations. The OED review also shows that the ratio o f private sector credit to GDP in countries that borrowed from the Bank for financial sector reforms increased only slightly over the period. I t has also remained at a l o w level for most Bank borrowing countries. While this may indicate that the financial sector reforms take time t o achieve their full impact, as noted above, ~~ 69PREM Network, draf? for comment, http://www- wbweb.worldbank.org/prem/premcompass/know_learn/econo~cgro~h.h~ ’O F o r example, the many years o f financial sector reforms in M e x i c o appeared to have shown little improvement in the sector’s support for private sector development. However, recently, credit t o the private sector rose by 25 percent, albeit f r o m a l o w base. T h i s may be an indication that k e y institutional reforms, including with regard t o the judicial process, are finally taking hold. Another example i s Sub-Sahara Africa, which has undergone major financial sector reforms within the decade, and where the aggregate private sector credit to GDP ratio f e l l initially because o f greater prudence in lending but began to p i c k up (now o n a more sustainable basis) in 2002. 109 Attachment 1 Management would also like to note the importance o f macroeconomic influences. The Bank’s report on the lessons o f the 1990s cited earlier and other work shows that much o f the increase in bank deposits over the 1990s tended to be absorbed by government and central bank debt.” A major reason for the rise in government debt was post-crisis bank restructurings, involving replacement o f weak private sector credits, growing government deficits, and the Banks’ increased net holdings o f central bank debt and increased net holdings o f foreign assets for hedging purposes. It i s also conceivable that introduction o f the Base1 Capital Accord in 1988, with i t s favorable treatment (a zero risk weight) o f government securities for capital allocation markets, may have encouraged banks to hold larger quantities o f the latter. Scope 7. Knowledge and Learning Activities. The Review recognizes (footnote 11) the variety o f instruments from DEC that disseminate research findings for operational use. Management would like to note that the financial sector anchor unit, FSE, has produced a wide variety o f knowledge products, including global and regional learning events for client countries and staff, policy papers and numerous conferences on all manner o f issues relevant to financial sector reform and development, and a help desk for the financial sector. All o f these activities-which have put the Bank at the forefront o f policy analysis in the Financial Sector-are also geared toward raising policy maker and staff awareness o f lessons and good practices in financial sector reforms. 8. Role of IFC. The Review acknowledges the importance o f close coordination within the Bank Group, although it notes that a review o f IFC and MIGA activities i s beyond OED’s mandate. Nevertheless, Management would like t o note areas where W o r l d Bank Group support was instrumental in achieving good outcomes, notably bank privatization and restructuring. For example, in many cases IFC participated in the equity o f banks being privatized as part o f Bank supported programs, almost always helping to bring along a suitable strategic partner. A few examples include Tanzania, Madagascar, Cameroon, Burkina Faso, Zimbabwe, Macedonia (Stopanska), Bosnia, and Romania. The review cites some o f these privatizations (Macedonia and Tanzania), but does not mention IFC’s involvement and contribution to a successful outcome. 9. Analysis o f Bank Supportfor Crisis Countries. OED regards the Bank support for crisis countries highly relevant for their return to economic growth and stability. The review notes, however, that the proportion o f satisfactory ratings received by crisis operations i s lower than all other adjustment operations (the review period predates development policy loans) and below financial sector operations. I t attributes this performance in part to project objectives that were too ambitious to be realized in the short time frame o f single adjustment operations. The review also discusses the difficulties posed by Bank-Fund collaboration and the perception o f some staff that Management could have given greater weight to staff views o n some issues. Against this background, the review suggests ex-ante agreements between the Bank and IMF on approaches and respective roles and, within the Bank, clear lines o f responsibility for coordinating the Bank’s support in times o f crisis. Management appreciates these lessons o f experience in crisis country support, although it also notes that the stated objectives o f lending at the time o f crisis may not accurately capture the underlying motivation and may not h l l y reflect the realities o n the ground at the time o f crisis when quick decisions by multiple donors and policy makers may have to be made without full information. 71 For example, in the 25 developing and transition countries with the largest banking systems, the average ratio o f net government debt to bank deposits rose by more than 60 percent, from about 13 percent in 1993 to about 21 percent in 2000. See Hanson, James (2003) Banking in Developing Countries in the 1990s. World Bank Policy Research Paper, No. 3168. 110 Attachment 1 10. Readiness for Crisis Response. The Review notes that a 1996 internal review o f the Bank’s response to the 1995 Mexico crisis led to recommendations for the Bank t o prepare guidelines with triggers for action, clear lines o f responsibility, and procedures for concentrating resources, putting in place a core team, and providing a framework for debating and agreeing expeditiously o n recommended actions. The Review further states that these recommendations have yet to be acted upon by management. While factually correct, two initiatives undertaken by Management responded t o much o f the essence o f those recommendations. In 1996, the Bank created the Short Term Risk Monitoring Group (STRMG) as the forum to monitor regularly the short terms vulnerabilities o f the Bank’s client countries. The S T R M G pulls together the various sources o f systemic risks, including those f r o m the financial sector, ranks countries by risk categories, and reports i t s findings regularly to Senior Management. Regional management puts in place monitoring systems and contingency plans for countries in the highest risk categories. I n 1997, the Bank’s Strategic Compact enhanced the Bank’s resources for financial sector work, especially on i t s ability to respond to crisis situations. In 1998, the Bank also created the Special Financial Operations Department (SFO) to provide the team and the concentrated resources to respond to financial crises. While the SFO was disbanded, Management s t i l l retains the knowledge o f financial sector expertise within the Bank and has the ability to pull together strong teams on short notice when necessary. 11, Recommendations Recommendation 1. The Bank’s financial sector anchor should provide more guidance for Bank staffand client countries, in areas such as restructuring of banks (8 when, and how); asset management companies (8 when, how); privatization o f banks; promotion o f capital markets (8 when, and how, in conjunction with IFC on this); andfor topics related to the strengthening the legal, regulatory, and supervisory environment, a particular focus on implementation. I n addition, thefinancial sector network should become more pro-active in quality control o f financial sector components in multi-sector loans. 12. The OED review notes that Bank support has generally followed good practice and international norms. The review also notes the generally good quality and outcomes o f operations under the direct control o f the Financial Sector Board. According t o the review, however, there have been a number o f areas where the Bank’s approach may have lacked coherence, in terms o f differences in the process o f reforms (how), sequencing (when), and the selection o f specific reforms. Specific areas where a wider variation o f approaches may have been more apparent were in bank privatization, payments systems, deposit insurance schemes and capital market development. In this context, the report recommends that the financial sector anchor provide good practice notes o n a range o f topics, including those where there i s ongoing debate on various approaches t o reforms. There i s a large body o f literature based o n Bank research and policy work, as well as that o f other institutions, o n financial sector reform approaches, including the areas mentioned in the review, that are available to our clients and our staff. This knowledge i s evolving, as empirical work carried out by the Bank and others often challenges conventional wisdom. The availability o f such knowledge i s important, as policy makers, with Bank support, need to adapt known best practices to local conditions, including the capacity to implement reforms. 72 Management nevertheless recognizes the need for operational guidance to staff that will distill 72 OED recognizes in footnote 26 that i t s concerns about the variation o f policy approaches in a number o f areas does not mean that the Bank should prescribe a one-size-fits-all prescription for reforms, as there are large differences in initial local conditions, levels o f economic development, government commitment to reform, and institutional capacity to implement reforms; and these factors all need to be taken into account in supporting a successful sector reform program. 111 Attachment 1 in a comprehensive way best practice principles to reforming a particular policy and set o f institutions. Against this background, and in view o f OED’s recommendation, Management will build within the anchor program an operational and ongoing practice note series as an additional tool for knowledge sharing into the anchor’s work program. FSE i s also strengthening i t s staff training program. That training will include many o f these practical operational lessons. Within the Bank’s existing review processes, FSE will strengthen its review efforts on financial sector programs to ensure appropriate consistency o f financial sector reforms (without ignoring the need for customization), Management also notes that the planned update o f the Bank’s 2001 financial sector strategy will provide an opportunity for guidance to staff o n key priorities and approaches in financial sector reforms. 13. Multi-Sector Operations with Financial Components. The review concludes that outcomes o f financial components in multi-sector loans have lower outcome ratings o n average than adjustment loans done by units linked to the Financial Sector Board, although these outcomes are not out o f line with those o f other sectors included in multi-sector operations. As these results could not be explained by country characteristics and differences in reforms, the review notes that these findings may be a result o f a number o f factors, including the presence o f specialized financial sector staff in programs under the management o f finance network staff, the review process, and the quality o f supervision within the network. In the recent sector strategy implementation update (SSIU), Management highlighted the growing importance o f finance components in multi-sectoral operations managed by other Sector Boards, and the need for addressing the quality assurance processes o f these finance components. Within this context, the Financial Sector Network has been promoting greater partnerships with other networks o n thematic activities (notably, economic policy and rural finance) to encourage a better sharing o f technical expertise across networks. As part o f the Bank’s regular review process, FSE has also systematically reviewed multi-sector development policy and other operations, and plans to strengthen monitoring o f the outcomes of these components. Finally, FSE i s strengthening i t s Bank staff training program to reach out to non- specialists to raise awareness o f financial sector issues. Recommendation 2. The Bank should develop monitorable indicators to assess progress on objectives in the area ofprudential regulations and supervision for financial intermediaries. 14. Management welcomes this recommendation. FSE has undertaken a priority work program to develop financial indicators for operational use in the next few years. I t will go beyond prudential regulation and supervision to include indicators o f financial stability, efficiency, and access to financial services. FSE has a good starting point o n indicators for bank regulation and supervision, as it has a large database73with more than 200 variables for over 150 countries, and it i s updated every 3-4 years, and on the findings o f FSAPs o n bank supervision o n a wide number o f countries. This database has been widely used outside the Bank and has also provided the foundation for ground-breaking research on effectiveness o f regulatory approaches in banking (see forthcoming book “Rethinking Bank Regulation and Supervision: Till Angels Govern” by Barth, Caprio, and Levine). Going forward, FSE (without ignoring the need for customization), with strong support from network staff will continue to implement a work program o n financial sector indicators that it trusts will be helpful t o the Bank and to our clients. Recommendation 3. On support for countries prior to andfollowing crisis: (9 The Bank should develop a rating system, in partnership with other relevant institutions, for vulnerability to crisis, making use of readily available information that can be used to engage countries in crisis prevention measures and issues in crisis response. The Bank should also do a better job than in the past ofpresenting assessments more candidly in documents. ~~ 73 (http://worldbank.org/research/projects/ 112 Attachment 1 15. As mentioned above, after the East Asian crisis, the W o r l d Bank put in place the S T R M G that identifies and monitors countries that Management considers vulnerable to crisis, and flags the risks t o senior management o n a regular basis. The STRMG has representation from all regions and several central units, notably from Finance. In ranking vulnerability, the S T R M G appropriately uses a broader set o f indicators that include political, macroeconomic, finance and other indicators to determine vulnerability. In view o f OED’s recommendation, FSE plans to provide the STRMG a more systematic framework for assessing the vulnerabilities arising from the financial sector. This work will draw on research, FSAPs, the IMF’s financial soundness indicators (see below), and other analytical work. In addition, the Bank and the IMF use the results o f FSAPs to engage authorities in identifying sources o f In i t s review o f CASs and vulnerability in the financial sector and ways to decrease ~ l n e r a b i l i t y . ’ ~ development policy operations, FSE also systematically integrates findings from FSAPs, including recognition o f vulnerabilities in the financial sector. Management will continue to pursue these efforts. 16. IMFIndicators. The IMF, consistent with i t s mandate, i s currently working o n deepening i t s financial soundness indicators. Bank staff have been working with the IMF on the development o f these indicators. Bank Management will work o n ensuring that it maintains the partnership with the IMF in this area and will draw upon these indicators in improving assessments o f vulnerabilities in the financial sector. (ii) The Bank should make internal arrangements to respond better to crisis by developing i circumstances warrant, guidelines for dealing with crisis, which should include the possibility, f of lending liquidity support to countries experiencing a crisis without stipulating ambitious reforms (that may not be realized) as justiJcation for the loan. 17. The OED Review notes that the Bank has been better equipped to respond to the more recent crises in Russia, Argentina, and Turkey than it had been in the earlier crises. Staff members who have expertise in dealing with financial crisis are n o w present in both the anchor and the Regions. In addition, internal papers have been written and disseminated o n the lessons o f experience o n this subject. Having said this, Management recognizes the problems associated with maintaining an appropriate level o f knowledge in systemically important countries where there i s no ongoing financial sector program. One o f the roles o f the financial sector Vice Presidency i s to coordinate with Regional management to address these r i s k management concerns. 18. Reform Versus Liquidity Support. Supporting countries with a series o f development policy loans, perhaps starting with one that seeks only to supply liquidity and establish the framework o f future supports, i s one o f the options that the Bank can use in time o f crisis. As the review itself points out, this was the approach the Bank used in assistance to Korea. Going forward, Management will draw upon this approach as appropriate, in coordination with the IMF. Management wishes to highlight an important lesson o f experience in assisting crisis countries: the onset o f a crisis creates windows o f opportunity to address fundamental issues. The Bank’s response in a crisis situation will, therefore, require judgment o n h o w i t can balance i t s assistance to support realistic opportunities for reform while also providing urgent liquidity support. The framework and internal guidelines for dealing with crisis will be developed in conjunction with the ongoing update o f the financial sector strategy. 74 Ultimately, however, following up on FSAP recommendations depends o n the country’s ownership o f the reforms. 113 Attachment 1 (iii) Coordination with the IMF and other IFIs in crisis assistance needs to be improved, and at the outset of the crisis, the IFIs should reach quick agreement on division o f responsibilities. 19. Management has continued to work o n improved coordination with the IMF. The Review does not note the creation and ongoing operation o f the Financial Sector Liaison Committee to oversee joint Bank-IMF programs and the fact that Bank-Fund cooperation has, in fact, significantly improved over the past five years, partly because o f the FSAP program. On coordination in times o f crisis, Management i s aware that one o f the lessons o f support for crisis countries i s the importance o f I F I s working together as a team with agreed and assigned lead and secondary responsibilities for the reform program at a time o f crisis. Thus, Management considers sustaining these strong partnerships with the IMF and other I F I s very important t o enable joint programs and facilitate division o f responsibilities at critical times, including at the outset o f a crisis. 20. Management Action Record. The Management Action Record provides more specific responses to OED’s recommendations. I t i s attached below. 114 Attachment 1 OED Review of Bank Assistance for Financial Sector Reform ManagementAction Record OED Recommendation Management Response The financial sector anchor should also be more pro- Management i s putting in place an operational practice active in quality control, especially for financial sector iote series as an additional tool for knowledge components o f multi-sector loans. The anchor should nanagement for financial sector support. Management also provide clear guidance for Bank staff and client will consider this action complete once this series i s countries on a range o f issues connected with financial firmly established, anticipated at the end o f FY06. sector reforms, including privatization o f banks, restructuring banks (if, when, how), use o f asset Within the Bank’s existing review processes, financial management companies; promotion o f capital markets; sector staff will strengthen their review o f the finance and other topics related to the legal, regulatory, and :omponents in multi-sector projects with a view to providing systemic solutions t o quality assurance. This will be done mainly by FS staff in the Regions, supported as necessary by anchor staff. FSE will use the Sector Strategy Implementation Update and the revised strategy to report o n progress; Management will consider this action complete when FSE reports that it i s a well-established practice. FSE i s strengthening i t s Bank staff training program and improving i t s outreach to staff in other networks. Management will consider this action complete after one year o f implementation o f the strengthened training program, the end o f FY06. The Bank should develop monitorable indicators to FSE, in collaboration with regional finance units, i s assess progress on objectives, especially in the area o f developing financial sector indicators for operational strengthening prudential regulations and supervision for use as a priority task in the next few years. Indicators financial intermediaries. will also be developed o n access to finance. Management will consider this action complete when these indicators are available as reported in the Sector Strategy Implementation Update. On support for countries prior t o and following crises, Drawing o n existing research in the Bank and o n the the Bank should develop a rating system, in partnership I M F ’ s financial sector soundness indicators, FSE will with other relevant institutions, for vulnerability to produce an operational note to provide a framework for crisis, making use o f readily available information, and assessing financial sector vulnerabilities. This should use the rating system to try t o engage countries framework will be used in support o f the broader in developing policies and measures for crisis S T R M G framework for assessing country risks. prevention and response. The Bank should also develop Management will consider this action complete when guidelines for providing assistance following crisis, and the framework i s available and in use, as reported in the should include the possibility, if circumstances warrant, Sector Strategy Implementation Update. o f lending liquidity support to countries experiencing crisis without stipulating ambitious reforms. Finally, as The Bank and the IMF will continue to use FSAPs to part o f crisis response. the Bank and other I F I s should engage authorities in identifving sources o f financial 115 Attachment 1 OED Recommendation Management Response reach quick agreement at the outset o f the crisis, o n the sector vulnerabilities and ways to decrease these risks. division o f responsibilities. Since this action i s ongoing, Management considers it complete. The fiamework and internal guidelines for dealing with financial sector support in crisis situations will be developed in conjunction with the ongoing update o f the Financial Sector Strategy. This action will be considered complete when these guidelines are cleared by senior management and available to staff. 116 Attachment 2 Chairman’s Summary Committee on Development Effectiveness OED Review of Bank Assistance for Financial Sector Reform and Draft Management Response to the OED Review of Bank Assistance for Financial Sector Reform (Meeting o f March 30,2005) 1. On March 30,2005, the Committee o n Development Effectiveness met to discuss the report entitled OED Review of Bank Assistancefor Financial Sector Reform and the Draft Management Response to the OED Review o f Bank Assistance for Financial Sector Reform. 2. Background. Between FY93 and FY03 the W o r l d Bank assistance for financial sector reforms (FSR) was supported by some US$56 billion in lending, or 24 percent o f the Bank’s total commitments. Most o f the lending was embedded in multi-sector loans. Over the period, lending for financial sector reform (FSR) declined, due mainly to the sharp drop in lines o f credit (LOC). CODE discussed OED review on LOC o n October 13,2004. The earlier 1998 OED review o f Bank support for financial sector reform presented several recommendations that management gave prominence in the financial sector strategy o f 200 1. 3. OED Finding and Recommendations. The OED review examined the Bank assistance to financial sector reform over the decade. The OED review found that (i) the objectives o f Bank assistance for FSR were generally consistent with good practice in terms o f reducing government ownership o f banks, improving prudential regulations and strengthening banking supervision; ( ii) consistency within a country and coherence o f the Bank’s approach to FSR across countries need improvements; and ( iii) there i s wide variation in Bank support in payments systems, deposit insurance schemes, and capital market development. The report also highlighted that outcomes o f financial sector programs - in terms o f financial depth and credit to the private sector had been weaker than had been anticipated, partly because o f continuing instability in the macro-economic situation, and partly because further reforms were needed. 4. OED recommended that the Bank should ( i) provide more guidance to Bank staff and client countries, in areas such as restructuring o f banks, asset management companies, privatization o f banks, promotion o f capital markets and for strengthening the legal, regulatory and supervisory environment; (ii) develop monitorable indicators to assess progress o n objectives in prudential regulations and supervision for financial intermediaries; and ( iii) develop a rating system for vulnerability t o crisis, make internal arrangements to respond better t o the crisis, and improve coordination with the IMF and other International Financial Institutions (IFIs) in crisis assistance. OED found in the current review that a prior management recommendation for the Bank to prepare guidelines for crisis situations has not been implemented and continue to remain valid. 5. Management Response. Management i s preparing an operational practice note series as an additional tool for knowledge management for financial sector support. The financial sector anchor unit (FSE) i s committed to (i)strengthen the review o f the finance components in multi- sector projects and provide systemic solutions to quality assurance, and the training program while improving outreach to other network staff; (ii) collaborate with regional finance u n i t s in developing financial indicators for operational use; and ( iii) produce an operational note in collaboration with the Fund and provide a framework for assessing financial sector 117 Attachment 2 vulnerabilities. Both the Bank and the Fund will continue using Financial Sector Assessment Programs (FSAPs) to engage authorities in identifying sources o f financial sector vulnerabilities. The framework and internal guidelines for dealing with financial sector support in crisis situations will be developed in conjunction with the ongoing update o f the financial sector strategy. 6. Overall Conclusions and Next Steps. Members welcomed the opportunity to discuss the report, which they praised for i t s high quality and candor and agreed with its recommendations. They also expressed their appreciation for management’s draft response (MR). They noted that the report and the MR together appeared to be a good basis for updating the financial sector strategy. Some members felt that the coverage o f the evaluation report could have been more complete, with the inclusion o f FSAPs and Report on the Observance o f Standards and Codes (ROSC) as well as o f IFC and MIGA activities. Suggestions were made for (i) highlighting further the critical role o f the broader macroeconomic situation, as well as structural and institutional factors in determining the outcomes o f Bank operations; and ( ii)deepening the analysis on country-wide impacts such as those related to investment and employment generation. 7. Members supported recommendations for improving Bank’s operational consistency and policy coherence as well as coordination with the IMF. They were concerned with OED’s opinion that management had not implemented a recommendation in a management document several year ago to establish clear guidelines for responding to crisis situations and requested management clarification. Management noted the establishment o f the short-term r i s k monitoring function in the Bank and a close working relationship on financial sector issues with the Fund in response to the financial crisis o f the 1990s. Members had some questions regarding ongoing work on indicators o f crisis vulnerability. Finally, members sought the views o f OED and management on policy implications going forward. The committee endorsed the OED report and the MR. The main issues raised during the meeting were the following: 8. Coverage of the Report. Many members and speakers noted the reports findings on the positive outcome o f the Bank’s assistance in financial sector reform. In discussing the findings on the weaker impact o f financial sector reforms on outcomes - such as financial sector depth and credit to the private sector some members felt that other factors such as the macroeconomic situation, political context, corporate governance could be important contributory factors. Some members also felt that a discussion o f FSAP’s and ROSC’s should have been included in the study and a review o f IFC and MIGA work related to the financial sector could have been includedhere as well. OED concurred that the macroeconomic situation could partially explain the weak impacts. OED also informed CODE that the review o f the FSAPprogram w i l l integrate thefindings from reports on standards and code, as well as the Bank’s E S K I n addition, it was mentioned that two evaluation brieJngs were being prepared on IFC equity investment in the banking sector and leasing. Some members highlighted that the evaluation o f financial sector initiatives should be linked to the judicial sector, enforcement o f contracts, accounting and auditing systems, corporate finance and corporate governance, and other aspects, as well as considering the role o f the private sector. OED responded that it w i l l conduct an evaluation on judicial reforms, which w i l l cover extra-judicial issues, such as out-of-court resolution of non- performingfinancial assets. A member felt that the report should have addressed the issues o f asymmetric information in client countries, promotion o f global and regional integration, impact on small economies, and investment promotion and employment creation. Another member regretted that there was n o specific MDG target on financial sector. Management found that financial sector issues relate to MDGs in areas such as promotion o f growth and support to private sector, and link to income distribution. 118 Attachment 2 9. Country-Wide Impact. While recognizing the challenges o f reducing the govemment’s role in financial intermediation, several members highlighted issues o f sequencing; governments’ short-term needs to finance the huge costs o f reform including restructuring o f staff, branches and portfolio cleanup; and country specific context. Other situations, for example, where state-owned banks support state-owned enterprises, could have been covered in the report as well. In this regard, a few members recommended addressing the issues o f sequencing or prioritization o f reforms in the financial sector strategy update. A member sought clarifications on the difference o f outcomes between f i r s t and second generation reforms in the financial sector, particularly o n the Bank’s role to improve developmental impact o f the second phase o f reforms taking into account the OED’s recommendations. Management indicated that second generation reform i s the most critical issue shaping the financial sector strategy update because o f their multi-sector dimension, cross-sectoral nature and high visibility, as w e l l as need for strong country ownership. 10. Lending and Non-Lending Instruments. One member noted the important lesson regarding the need for selectivity in identifying technical assistance (TA) opportunities and country ownership, which emerged from the outcomes o f adjustment loans accompanied by TA loans; o n the one hand, in countries with limited institutional capacity, adjustment loans accompanied by TA loans had better outcomes than adjustment loans without TA. On the other hand, in countries with better institutional capacity adjustment, loans accompanied by TA loans had significantly worse outcomes than adjustment loans without TA loans. While preferring more emphasis o n lending programs, another member wondered whether new lending instruments could be developed that could benefit from ESW and could support financial sector reforms in an innovative way. Management recognized that there i s increasing need for advisory services. However, the Bank’s lending instruments t o promote financial reforms are somewhat limited in a non-crisis context. Management felt that a more strategic coordination with IFC would be desirable, given the latter’s flexibility, knowledge o f the private sector, and range o f financial and advisory instruments. 11. Coordination and Coherence. M a n y members and speakers suggested improving the Bank’s operational consistency and policy coherence; strengthening coordination with the IMF based o n each institution’s comparative advantage; and broadening coordination with other IFIs. Management noted the concerns related to consistency in designing early reform packages, which were attributed to inadequate assessment o f the market conditions such as competitiveness of thefinancial sewices industry, and lack o f collateral laws or insolvency regimes. Regarding coordination with the I M F , management highlighted that the Bank concentrates in areas where it has greater advantages such as in TA and capacity building for bank supervisors, without competing with the IMF or Bank for International Settlement. Speakers recognized the importance o f country ownership and accountability as basis for support, and in implementation o f reform initiatives. A few members sought more information o n the set o f indicators being developed to better track progress in the financial sector, and o n the Bank’s work with the lMF to strengthen the IMF financial indicators. Management responded that an extensive database was built covering about 200 variables offinancial regulation and supervision in 150 developed and developing countries. This database is available to outside sources. Management also shared that the Bank (Finance and PSD) in collaboration with outside partners (IMF, UN) was developing indicators to assess progress in areas like outreach offinancial services, depth and breadth o f the financial system, and diversijkation. These indicators w i l l be used in the context of the FSAP program. 12. Bank Support to Countries Facing Financial Crisis. Members requested management to address the need for expanding guidance (i) to respond to crisis situations as previously recommended by a review commissioned by management; and ( ii)to provide liquidity support, 119 Attachment 2 which was not fully addressed in the MR although they recognized the difficulties in developing universal guidelines. Supplementing M R regarding this matter, management indicated that Bank’s experts have been identijled within the regions, FSE, and other networks, particularly PREM, who are prepared to respond to financial crisis situations. Management also commented that lessons learned in past crisis were applied in dealing with recent cases in Turkey and Argentina. Moreover, the Bank has developed a system for monitoring country risk, it has improved coordination with the IMF and other IFIs, and it has redesigned the lending instrument such as the development policy lending. One member noted the challenges in evaluating the outcomes o f the Bank assistance to crisis countries because the implementation o f reforms and the full recovery o f the financial sector, especially credit to private sector, require time. Management acknowledged that there is a time lagfor increasing credit to the private sector and increased outreach offinancial services that may be attributed to a wide range o f factors from weak institutional capacity to macroeconomic policies. Other speakers highlighted the importance o f Bank policy dialogue with the countries in non-crisis but potentially vulnerable situations. Regarding the recommendation to develop a rating system for vulnerability to crisis, a member observed that there were enough analytical tools, including those at the Fund in addition to the Bank. Chander Mohan Vasudev Chairman