51571 Turmoil at Twenty R Recovery, and Rec Recession, R Cen Reform in Central and Euro Eastern Europe and the Soviet Union Former So Pradeep Mitra, Marcelo Selowsky, and Juan Zalduendo Turmoil at 20 Twenty Recession, Recovery, and Reform in Central and Eastern Europe and the Former Soviet Union Turmoil at 20 Twenty Recession, Recovery, and Reform in Central and Eastern Europe and the Former Soviet Union Pradeep Mitra, Marcelo Selowsky, and Juan Zalduendo © 2010 The International Bank for Reconstruction and Development/The World Bank 1818 H Street NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved. 1 2 3 4 13 12 11 10 Th is volume is a product of the staff of the International Bank for Reconstruction and Development/The World Bank. The fi ndings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Direc- tors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The Inter- national Bank for Reconstruction and Development/The World Bank encourages dis- semination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Inter- net: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Wash- ington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. ISBN: 978-0-8213-8113-7 eISBN: 978-0-8213-8114-4 DOI: 10.1596/978-0-8213-8113-7 Cover photo shows Riga, Latvia. © istock/Sergey Skleznev Cover design by Naylor Design, Inc. Library of Congress Cataloging-in-Publication Data has been requested. Table of contents Acknowledgments xiii Overview 1 1 Prelude to the crisis 25 Vulnerable . . . but with variation 29 Transition meets global finance 36 Would different macroeconomic policies have lessened vulnerability? 54 Opening the toolkit 60 Annex 1.1 Separating wheat from chaff--evidence of market differentiation from EMBI spreads 69 Annex 1.2 Finance in transition 70 2 How much adjustment? How much financing? 75 Different shocks for different countries 76 Sharing the burden: private and public, domestic and external 78 Crisis, adjustment, and financing in low-income and lower middle- income CIS countries 91 Of parents and offspring: understanding rollover risks in ECA 95 Three concluding arguments--three caveats 113 Annex 2.1 Description of the Bank for International Settlements (BIS) Dataset 115 3 Restructuring bank, corporate, and household debt 117 Financial systems need to be fi xed 118 For the lenders: bank restructuring 122 For the borrowers: corporate and household debt restructuring 127 Lessons on restructuring from previous banking and capital account crises 142 Lessons for strengthening bank regulation and supervision 148 v 4 Scaling up social safety nets 163 Existing social assistance programs 164 Safety nets: ready to be scaled up? 167 How important are safety nets in transferring income? 171 Cost of expanding means-tested programs 173 An opportunity for further reform 175 5 Prioritizing structural reform 179 Interpreting business environment surveys 180 Overview of results 183 Growth bottlenecks 185 The persistence of legacy in shaping the business environment 196 Annex 5.1 Conceptual framework 206 Annex 5.2 Moving away from the benchmark: firm characteristics and constraints 210 Annex 5.3 Tables 213 6 The day after 219 Bottlenecks in electricity--an agenda for reform 220 The education and skills agenda--making the grade 237 Bibliography 253 Boxes 1.1 Shades of vulnerability--a cluster analysis approach to classifying countries 35 1.2 Playing cat and mouse--staying ahead of regulation arbitrage in Southeastern Europe 52 1.3 Why foreign currency lending did not take off in the Czech Republic 66 2.1 Sticking together through thick and thin: the European Bank Coordination (Vienna) Initiative 80 2.2 From Bangkok to Budapest: ECA's adjustment compared with East Asia's 86 2.3 Countercyclical fiscal policy in financially integrated countries: Kazakhstan and the Russian Federation 88 2.4 Tajikistan's declining remittances can hurt the poor disproportionately 93 vi TABLE OF CONTENTS 2.5 Foreign ownership and funding sources 107 3.1 ECA's growth prospects--green shoots? Maybe. High growth rates? Unlikely 121 3.3 An agenda for modern banking sector institutions in ECA countries 151 3.4 Capital--what is it and why require it? 155 3.5 Taking the rough with the smooth--dynamic provisioning in Spain 158 4.1 Aiming high to serve the poor: Georgia's new Targeted Social Assistance Program 168 4.2 Enough bang for the buck? Safety nets in the Russian Federation 172 5.1 Transition economies converge in structure to market economies 198 5.2 Comparing constraints in transition (BEEPS 1999­2005) and nontransition (ICA) countries 203 6.1 Electricity tariff increases and poverty impacts 227 6.2 Ukraine norms for education facilities 249 Figures 1 Means-tested safety nets: targeting accuracy, coverage, and transfers to the poorest quintile 14 1.1 Income convergence with EU15, by country, 1999 and 2008 26 1.2 Trade integration, by region, 1994­2008 26 1.3 Financial integration, by region, 1994­2008 27 1.4 Labor integration, by country, 2007 and 2008 27 1.5 Capital flows in developing East Asia 29 1.6 Capital flows in Europe and Central Asia 30 1.7 Capital flows in Latin America and the Caribbean 30 1.8 Capital flows in other emerging economies 31 1.9 Real GDP growth, median values, 2000­08 31 1.10 Current account, net of FDI, median values, 2000­08 32 1.11 Inflation, median values, 2000­08 32 1.12 Fiscal balance, median values, 2000­08 33 1.13 External debt to GDP, median values, 2000­08 33 1.14 Ratio of short-term debt to foreign exchange, median values, 2000­08 34 1.15 Current account, net of FDI, median per group, 2000­08 38 1.16 Fiscal balance, median per group, 2000­08 38 TABLE OF CONTENTS vii 1.17 External debt to GDP, median per group, 2000­08 39 1.18 Short-term debt to foreign exchange reserves, median per group, 2000­08 39 1.19 Bank ownership patterns in Europe and Central Asia transition economies, 1997 and 2005 40 1.20 Banking crises in transition economies, 1990­2002 41 1.21 Average return on equity, parent banks and competition, 2004­08 42 1.22 Average return on assets, parent banks and competition, 2004­08 42 1.23 Private sector credit to GDP, median value, 2000­08 44 1.24 Loans to deposits, median value, 2000­08 44 1.25 Foreign exchange assets to liabilities, median value, 2000­08 45 1.26 Liabilities to equity, median value, 2000­08 46 1.27 Private sector credit developments in 2005­08: catch-up or excess? 47 1.28 Real housing price developments 49 1.29 Exchange rate regimes 55 1.30 Average policy outcome/stance: exchange rate flexibility, by group and period 56 1.31 Average policy outcome/stance: fiscal policy, by group and period 56 1.32 Average policy outcome/stance: monetary conditions, by group and period 57 1.33 Average policy outcome/stance: nominal exchange rate volatility, by group and period 57 1.34 Policy response to balance-of-payment pressures: exchange rate flexibility, by group and period 58 1.35 Policy response to balance-of-payment pressures: fiscal policy, by group and period 58 1.36 Policy response to balance-of-payment pressures: monetary policy conditions, by group and period 59 1.37 Policy response to changes in net foreign assets: sterilization of foreign exchange, by group and period 59 1.38 Median real GDP growth, by group and period 64 1.39 Median current account balance, by group and period 64 1.40 Change in real effective exchange rate, by group and period 65 1.41 Policy response to balance-of-payment pressures: real effective exchange rate, by group and period 65 viii TABLE OF CONTENTS 1.1.1 Country-specific components of EMBI spreads 70 1.2.1 Financial deepening in ECA's transition economies, median values, 1995­2007 71 1.2.2 Bank efficiency in ECA's transition economies, median values, 1995­2007 71 1.2.3 Private credit to GDP versus GDP per capita, ECA's transition economies versus other regions, 1995 and 2007 72 2.1 Mean private capital flows during the East Asian capital account crisis 78 2.2 Crisis, adjustment, and financing in financially integrated Europe and Central Asia economies 83 2.3 Crisis, adjustment, and financing in financially integrated oil-exporting Europe and Central Asia economies 84 2.4 Public and private sector balances in low-income and lower middle-income CIS countries, 2008­10 94 2.5 Banking sector credit--national and domestic sources 96 2.6 International claims, end-2008 99 2.7 Foreign claims, end-2008 100 2.8a Foreign exchange liabilities, Republic of Korea and financially integrated ECA countries 102 2.8b Foreign exchange liabilities, Republic of Korea and financially integrated ECA countries 102 2.9 Wholesale funding, by country, 2007­09 105 2.10 Resident retail deposits, by country, 2007­09 105 2.11 Parent bank funding, by country, 2007­09 106 2.12 Ratio of liquid assets to total assets, by country, March 2009 108 2.13 Loans with a maturity of five years or more, by country, 2007­09 108 2.14 Sectoral composition of loans, by country, March 2009 109 2.15 Currency composition, by country, March 2009 109 3.1a Average duration and change in output of recessions, by type 119 3.1b Average duration and change in output of recoveries, by type 120 3.2 Household debt, by country, 2008 133 3.3 Household debt, earlier EU members, 1995­2004 133 3.4 Composition of household debt, by country, end-2008 134 3.5 Foreign currency-denominated loans, by country, 2008 134 3.6 Mortgage loans with adjustable interest rates, by country, 2006 135 TABLE OF CONTENTS ix 3.7 Mortgage interest service and interest rate shock, by income quintile 136 3.8a Total debt service under shocks 137 3.8b Total debt service under shocks 138 3.9a Ratio of actual to minimum required capital-asset ratio 149 3.9b Supervisory agency can legally declare a bank insolvent 150 4.1 Spending on overall safety nets, by country, 2006­08 165 4.2 Coverage of overall safety nets, by country, various years 166 4.3 Targeting accuracy of overall safety-net benefits, by country, various years 167 4.4 How ready are ECA's safety nets for rapid crisis response? A typology of countries 169 4.5 Targeting accuracy of means-tested programs, by country, 2004­08 170 4.6 Coverage of means-tested programs, by country, 2004­08 171 5.1 Average business environment constraint: transition economies in 1999, 2002, 2005, 2008, and nontransition economies 181 5.2 Infrastructure composite bottlenecks, 1999­2008 187 5.3 Infrastructure bottlenecks--electricity, 1999­2008 187 5.4 Infrastructure bottlenecks--transport, 1999­2008 188 5.5 Access to land, 1999­2008 188 5.6 Skills bottlenecks, 1999­2008 189 5.7 Corruption, priority measure, 1999­2008 191 5.8 Crime/theft/disorder: priority measure, 1999­2008 192 5.9 Tax administration: priority measure, 1999­2008 192 5.10 Customs regulations: priority measure, 1999­2008 193 5.11 Financing expansion--internal finance 194 5.12 Financing expansion--external financing, constraint level 195 5.13 Financing expansion--external financing, constraint priority 195 5.14 Physical infrastructure: (composite) priority measure, 1999­2005 200 5.15 Skills: priority measure, 1999­2005 200 5.16 Legal environment: priority measure, 1999­2005 201 5.17 Labor regulation: priority measure, 1999­2005 202 5.1.1 Demand and supply of public goods--one country, two firms 207 5.1.2 Demand and supply of public goods 208 5.1.3 Two transition economies, two nontransition economies (and many firms) 209 6.1 Relative importance of constraints, by country groups 221 x TABLE OF CONTENTS 6.2 Investment commitments in electricity with private participation, 1993­2007 222 6.3 Evolution of average collection rates in ECA countries 223 6.4 Weighted average electricity tariffs, 2008 (US cents per kilowatt-hour, excluding taxes) 226 6.5 Present structure of ECA electricity markets 229 6.6 Worker education: percentage of firms considering it a "major" or "very severe" constraint in 2008 238 6.7 Worker education: change in the frequency of satisfied firms between 2005 and 2008 238 6.8a 2007 TIMSS--mean content scores in grade 4 math 240 6.8b 2007 TIMSS--mean cognitive scores in grade 4 math 240 6.8c 2007 TIMSS--mean content scores in grade 4 science 240 6.8d 2007 TIMSS--mean cognitive scores in grade 4 science 241 6.9a 2007 TIMSS--mean content scores in grade 8 math 241 6.9b 2007 TIMSS--mean cognitive scores in grade 8 math 241 6.9c 2007 TIMSS--mean content scores in grade 8 science 242 6.9d 2007 TIMSS--mean cognitive scores in grade 8 science 242 6.10a PISA math scores 242 6.10b PISA science scores 243 6.11a PISA reading scores--difference between quintile 5 and quintile 1 244 6.11b PISA math scores--difference between quintile 5 and quintile 1 244 6.12 Positive and negative shifts in employment participation in 1995­2006, by occupation 245 6.13 Kazakhstan survey of firms, 2008: importance of general competencies and technical skills 247 Map 1 Income convergence to the EU15 average income, 2000­09 3 Tables 1 Credit market characteristics in financially integrated countries 10 1.1 Savings­investments balance 37 1.2 Evolution of banking sector indicators, by country, 1999­2008 43 1.3 Growth and composition of credit to the private sector 50 1.2.1 Comparing ECA's transition economies with other regions 72 TABLE OF CONTENTS xi 2.1 Savings­investment balances, adjustment without official financing 82 2.2 Savings­investment balances, adjustment with official financing 90 2.3 Direct lending as a share of total national credit, by country, 2005­09 98 2.4 International claims, by country, 2000­09 101 2.5 Stability of funding sources 103 2.6 Index of home country concentration of parent bank exposure, international claims 111 2.7 Importance of lending in foreign exchange among parent-subsidiary banks 112 2.1.1 Comparison of information on claims in BIS data 2.1.2 Sectoral breakdown in BIS data 116 3.1 Countries with banking and currency crises and nonperforming loans as a share of total loans 126 3.2 Credit losses--extrapolating from past crisis events 128 3.3 Median nonfinancial corporate leverage, Europe and Central Asia countries and EU cohesion countries, 1999­2008, and comparator countries for years of crisis 129 3.4 Median nonfinancial corporate leverage, by country, 2008 130 3.5 Median interest coverage in nonfinancial firms, Europe and Central Asia countries and EU Cohesion countries, 1999­2008, and comparator countries for years of crisis 131 3.6 Median interest coverage ratio in nonfinancial firms, by country, 2008 132 4.1 Transfer to beneficiaries in the poorest quintile, by country, various years 174 4.2 Percentage of transfers through means-tested programs to overall safety net transfers, by country, various years 176 5.1 Levels and priorities of constraints on business in BEEPS 4 and other surveys 182 6.1 Total technical and commercial losses in ECA countries 225 6.2 Regulatory institutions in ECA countries 231 6.3 Implementation of the directive 2003/54/EC in the contracting parties to the treaty establishing the energy community, December 2008 235 6.4 Features of electricity systems in Central Asia 236 6.5 Russian Federation--wage premia for additional schooling 246 6.6 Students tutored by their own teachers 249 xii TABLE OF CONTENTS Acknowledgments We would like to thank the many authors who contributed background papers and notes for this study: Miroslav Beblavy, Thorsten Beck, Andres Bernas- coni, Eduardo Bitran, Mario Blejer, Robert Branda, Wendy Carlin, Zsolt Dar- vas, Santiago Fernandez de Lis, Timothy Frye, Charles Griffin, James Hanson, Valerie Herzberg, Robert Holzmann, Patrick Honohan, Ira Lieberman, Kathy Lindert, Anil Markandya, Mathilde Maurel, Martin Melecky, John Nellis, Trang van Nguyen, James Parks, Aleksandra Posarac, Mark Schaffer, Veli- mir Sonje, Naotaka Sugawara, Ramya Sundaram, Gyorgy Suranyi, Claudia Vasquez, and Michel Zarnowiecki. We owe Shigeo Katsu an enormous debt. He conceived the idea for this book while serving as vice president of the Europe and Central Asia region and was a strong intellectual force during the book's execution. He encour- aged us to take a broad historical view of the issues, and he provided constant support and guidance. Many colleagues shared their findings with us. We wish to thank Gavin Adlington, Stijn Claessens, George Clark, Itzhak Goldberg, Christos Kostopou- los, Arvo Kuddo, Patricio Marquez, Fernando Montes-Negret, Mamta Murthi, Ernesto Porta, Roberto Rocha, Pedro Rodriguez, Gevorg Sargsyan, Lars Son- dergaard, Victor Sulla, Erwin Tiongson, and Sarah Zekri. Our understanding of the region has benefited over the years from interactions with country econo- mists in the Europe and Central Asia region's Washington and country offices, a number of whom provided inputs to the study. We owe a special thank you to colleagues in the Europe and Central Asia Office of the Chief Economist: Indermit Gill, Elena Kantarovich, Rhodora Mendoza Paynor, Bryce Quillin, and Willem van Eeghen for their friendship and support. We are grateful to the central bank and financial sector authorities in Bul- garia, Croatia, Estonia, Hungary, Kazakhstan, Latvia, Lithuania, FYR Mace- donia, Poland, Romania, Serbia, and Ukraine for making available consoli- dated banking system balance sheet data, and to colleagues in ECA's private and financial sector unit for following up on these and other requests. Helpful comments on earlier drafts were received from Asad Alam, Tamar Manuelyan Atinc, Luca Barbone, Sudarshan Canagarajah, Indermit Gill, xiii Daniela Gressani, Roumeen Islam, Philippe Le Houérou, Fernando Montes- Negret, John Pollner, and Martin Raiser. We have also benefited from dis- cussions with participants in early brainstorming sessions in Almaty and Moscow. Research analysis was done by Lauren Elizabeth Clark and Th i Trang Linh Phu. Charito Hain and Rhodora Mendoza Paynor provided adminis- trative support. Bruce Ross-Larson and Allison Kerns edited the book, and Elaine Wilson was responsible for its design and layout. The World Bank's Office of the Publisher coordinated the design of the book, production, and printing. Pradeep Mitra Marcelo Selowsky Juan Zalduendo xiv TURMOIL AT TWENTY Overview The countries in the World Bank's Europe and Central Asia (ECA) region, among all emerging and developing economy regions, are forecast to experi- ence the deepest contraction as a result of the global economic recession of 2008­09. Th is is partly due to the region's deep integration into the global economy across many dimensions--trade, financial, and labor flows.1 · Trade integration in the transition (formerly centrally planned) ECA countries--measured by the sum of merchandise exports and imports as a share of GDP in purchasing power parity--rose from 20 percent in 1994 to around 50 percent in 2008, about 10­15 percentage points higher than in developing East Asia and Latin America. Turkey saw an increase from 10 percent to 30 percent over the same period. The averages mask substan- tial variation across subregions--the ratio ranged from a median value of around 35 percent in the South Caucasus, Central Asia and Moldova, where exports are generally intensive in natural resources and unskilled labor, to nearly 85 percent in the new member states of the European Union and Croatia, where exports are intensive in capital and skilled labor.2 · Financial openness in ECA's transition countries, as measured by the sum of foreign exchange assets and liabilities as a share of GDP, was 30 percent in 2008--twice that for developing East Asia, with the ratio being as high as 45 percent in the new member states of the European Union and Croa- tia. Turkey's financial openness, at 18 percent of GDP, is closer to that of developing East Asia. 1. The International Monetary Fund's World Economic Outlook (October 2009) projects GDP in Cen- tral and Eastern Europe and in the Commonwealth of Independent States (the former Soviet Union excluding the Baltic States) to contract by 5 percent and 6.7 percent, respectively, in 2009. In this book, Central and Eastern Europe comprises Albania, Bosnia and Herzegovina, Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Kosovo, Latvia, Lithuania, FYR Macedonia, Montenegro, Poland, Roma- nia, Serbia, the Slovak Republic and Slovenia. Turkey is part of the World Bank's ECA region but is not a transition country; thus, it is added selectively to the discussion. The Commonwealth of Independent States (CIS) includes Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, the Kyrgyz Republic, Mol- dova, the Russian Federation, Tajikistan, Turkmenistan, Ukraine, and Uzbekistan. Georgia is not part of the CIS but is included in the group because its economy shares many features of the other countries. 2. The evolution of trade openness over time is measured using GDP in purchasing power parity because of volatility in market exchange rates in the early years of transition. Comparisons in 2008 across developing country regions are substantially unaffected by this choice. 1 · Labor and remittance flows have followed a broadly biaxial pattern, being dependent on the European Union for the countries of Central and East- ern Europe and on the resource-rich countries of the former Soviet Union for the South Caucasus and many of the Central Asian republics. Indeed, the ratio of remittances to GDP for the poorest countries in the region is among the highest in the world, ranging, for example, between 30 and 35 percent for Moldova and Tajikistan in 2008. Helped by integration, GDP in the ECA region grew from 2000 to 2008 by two-thirds--an enviable growth rate averaging 6.5 percent a year. And the ECA region made remarkable progress in converging to Western European (EU15) income levels since the late 1990s. In fact, no other region converged as much as a group to the average income of EU15 countries even with the GDP declines cur- rently projected for 2009 (map 1). The decade since the Russian Federation finan- cial crisis saw 55 million people moving out of absolute poverty.3 Globalization lifted many boats, since developing East Asia grew at comparable rates; Latin America less so. Yet those regions have not been equally affected by the current global economic crisis. Interestingly, however, a number of advanced countries in Western Europe have been affected: for every Ukraine, Latvia, and Hungary (three of ECA's hardest hit countries), there is an Iceland, Ireland, and Spain. Before analyzing crisis and recovery, however, it is worth noting how far the transition economies of Europe and Central Asia have come since the fall of the Berlin wall two decades ago. Price liberalization--together with the disruption of the organizational arrangements that governed production and trade under cen- tral planning--made many enterprises inherited from the command economy unviable. These factors also resulted in transition recessions everywhere but par- ticularly deep and protracted ones in the countries of the former Soviet Union. Macroeconomic stabilization was needed to prevent price liberalization from converting the repressed inflation of centrally planned economies into hyperin- flation. And budget constraints were hardened to restructure or close enterprises and make possible the transfer of assets and labor to viable enterprises. Private sector development was sought to be accomplished through privatization, an enforceable set of property rights, and the entry of new private firms. Economic and institutional transformation this major created a difficult first decade of transition.4 But processes were set that enabled inflation to fall to single digits in many countries by 2006 and raised the share of the private sector in GDP to 80 percent in Central Europe and the Baltic states. Progress was also made in 3. The absolute poverty line used is $2.50 a day at 2005 Purchasing Power Parities. 4. These developments are reviewed in EBRD 1999, Havrylyshyn and Nsouli 2001, and World Bank 2002. 2 TURMOIL AT TWENTY MAP 1 Income convergence to the EU15 average income, 2000­09 (projection as of October 2009) Highest diverging quintile Average diverging/ converging quintile Highest converging quintile EU15 No data Note: The map reflects the change in PPP per capita incomes as a share of the EU15 average between 2000 and 2009. Dark blue countries have converged the most to the EU15 average income per capita over the past decade; those in gray have diverged the most to the EU15 average income per capita or, for countries higher than the EU15 average, that the EU15 is catching up. institutional reforms, whether measured by the EBRD transition indicators or by the World Bank's Doing Business, which has found Europe and Central Asia to be a leading reforming region. And further evidence of "institutional catch- up" comes from business surveys, which show that key elements of the business environment (such as competition and market structure, and finance and the structure of lending to firms) have converged toward those in developed market economies, particularly in the new member states of the European Union.5 This book, written on the eve of the 20th anniversary of the fall of the Ber- lin wall in 1989, addresses three questions that relate to recession, recovery, and reform, respectively, in ECA's transition countries. · Did the transition from a command to a market economy and the period when it took place, plant the seeds of vulnerability that made transition countries (the region excluding Turkey) more prone to crisis than develop- ing countries generally? · Did choices made on the road from plan to market shape the ability of affected countries to recover from the crisis? 5. Mitra 2008. OVERVIEW 3 · What structural reforms do transition countries need to undertake to address the most binding constraints to growth in a world where financial markets have become more discriminating and where capital flows to transition and developing countries are likely to be considerably lower than before the crisis? Plan of the book Chapter 1 of the book analyses how countries fell into recession and crisis, why not all of them were equally affected, and whether different policies could have positioned them better to face the crisis. Chapter 2 discusses rescue and stabilization and the role of international collective action. The next two chap- ters focus on policies for recovery--chapter 3 on restructuring bank, corporate and household debt and chapter 4 on scaling up social safety nets. Chapters 5 and 6 focus on reform, examining the binding constraints to growth and the policy agenda in the most important sectors identified by that analysis. Recession For the fi nancially integrated countries, the crisis has hit them primarily through two channels.6 · There was global deleveraging, triggered mainly by distress in home country financial markets, which together with the unwinding of real estate booms in some host countries, reduced the willingness by creditors to finance cur- rent account deficits. These deficits were, as a percentage of GDP in 2008, in double digits in such countries as Bulgaria, Latvia, Lithuania, Romania, and Serbia and in the high single digits in Estonia, Hungary, and Ukraine. · A recession-induced downturn in exports to Western Europe had a neg- ative impact on output and employment in small open economies such as the Czech and Slovak Republics, Estonia, and Hungary where exports accounted for between 70 to 80 percent of GDP in 2008. To a somewhat less extent, this was also the case in larger economies such as Poland and Roma- nia, where the corresponding share ranged between 30 and 40 percent. For the low-income and lower middle-income countries of the former Soviet Union, such as Armenia, Georgia, Kyrgyz Republic, Moldova, and Tajikistan, the reversal in capital inflows has been less important, but they have been hurt by lower exports and lower remittance flows. 6. A fi nancially integrated country is one where foreign claims of banks reporting to the Bank for International Settlements exceed 10 percent of the recipient country's GDP. For all but two of the fi nancially integrated countries, this implies foreign claims well in excess of $10 billion; for the two small economies for which this is not the case, however, foreign claims represent a very important share of their GDP. Details of the group are in chapter 1, and foreign claims are defi ned in chapter 2. 4 TURMOIL AT TWENTY · The sharp contraction in the Russian Federation's GDP, estimated at more than 7 percent in 2009, has depressed export demand in Armenia, the Kyr- gyz Republic, Moldova, and Tajikistan, for which it is the dominant export market. Furthermore, the marked deceleration in growth in Kazakhstan following the sudden stop in capital flows in 2007 has adversely affected the Kyrgyz Republic, which sends a significant share of its exports to that country. Export demand for Georgian goods, for which the CIS countries are not a major destination, is also projected to fall due to recessions in Turkey, the European Union, and the United States. Indeed, export earn- ings in Armenia, Georgia, and Tajikistan are projected to decline between one-quarter and one-third in 2009. · As noted, the low-income former Soviet Union includes countries that are among the most remittance-dependent in the world. Preliminary estimates suggest that remittances fell sharply--by more than 30 percent in some countries--during the first quarter of 2009, and they are not expected to recover before 2010 at the earliest. Transition shaped the nature of financial integration (and vulnerability) . . . The reason transition countries were badly hit by the fi nancial crisis, rather than developing countries more generally, can be sought in the nature of transition. The proximate reason for the crisis in the fi nancially integrated countries was the extraordinary growth of credit to the private sector, made possible in some ECA countries by large external inflows intermedi- ated by banks. The fastest credit growth rates between 2000 and 2008 was in the Baltic states, Bulgaria, Kazakhstan, Romania, the Russian Federa- tion, and Ukraine, countries that could be characterized as experiencing excessive credit growth (with growth of credit to the private sector above that of all developing nations even allowing for their initially shallow fi nan- cial sectors). They were all latecomers to the transition. Those in the for- mer Soviet Union--the Baltic states, Kazakhstan, the Russian Federation, and Ukraine--started in the early 1990s but underwent deep and protracted recessions. Bulgaria and Romania experienced double transition recessions due to macroeconomic crises in the mid-1990s that led to a second dip in GDP. Both sets of countries were therefore rapidly catching up--indeed, except Estonia, their fi nancial sectors were initially quite shallow. In con- trast, other transition countries such as Croatia, the Czech and Slovak Republics, Poland, and Serbia witnessed private sector credit growth in line with the experience of other developing nations--they are described in this book as experiencing convergent credit growth. OVERVIEW 5 . . . but vulnerability was also fed by unusually high global liquidity The growth of credit during the catch-up was the result of both demand and supply factors. On the demand side, the fast consumption catch-up by house- holds towards European living standards (particularly their adjustment in the stock of durable goods) drove the credit boom--from Hungary and Romania in the West through Lithuania and Ukraine and to the Russian Federation and Kazakhstan in the East. It was facilitated by their ability to borrow in for- eign currency, which had lower interest rates and typically longer maturities than those in local currency. And on the supply side, 2003­06 was a period of historically high global liquidity, with fierce competition in international banking and abundant supplies of credit to emerging market economies.7 For some countries, external fi nancing allowed loan-to-deposit ratios to become substantially greater than one, implying that external financing made up the difference between the credit extended by banks and what they were taking in from depositors. The sources of this external funding were multiple. For Kazakhstan, the Rus- sian Federation, and to lesser extent Ukraine, it was dominated by wholesale funding operations made possible by the appetite for private and public emerg- ing market paper. For the Baltic states, Bulgaria, and Croatia, it originated in the Western European parent banks of host country subsidiaries that had decided to expand operations in a low-debt and potentially fast-growing region. Indeed, as will be seen, some of the seemingly debt-creating flows had foreign direct investment (FDI)­like features and were on-lent at long maturities. Variability across the financially integrated ECA countries is more marked than is recognized Not all ECA countries, even among the fi nancially integrated, have been equally hit by the crisis--indeed, differences abound. The year-on-year fall in Latvia's GDP during 2009 is expected to be 18 percent, compared with the Czech Republic's 4 percent.8 Both are deeply integrated into the global trade and fi nancial system--and both are highly dependent on economic devel- opments in Western Europe. Fully 85 percent of the Czech banking sector is foreign-owned, compared with 65 percent of the Latvian banking sector. Acquiring those assets was a lucrative proposition for Western European banks: return to equity was higher for banks that had established a presence 7. Th is was a period when the annual growth of capital flows to emerging market economies exceeded the growth of the G7 economies by more than 10 percentage points. 8. The intent here is not so much to compare the Czech Republic and Latvia as to show them as emblematic of degrees of vulnerability. 6 TURMOIL AT TWENTY in Central and Eastern Europe than for those with mainly domestic market exposure in their home country. And yet the loan-to-deposit ratio at end-2008 was 280 percent in Latvia compared with 80 percent in the Czech Republic, implying that an extraordinarily large proportion of loan growth in Latvia was funded from abroad. Why did this not happen in the Czech Republic? Largely because early dis- inflation--and thus the credibility of monetary policy--and continued tight fiscal policy kept interest rates lower than those in the euro area. Despite real exchange rate appreciation due to productivity growth, borrowing in local currency was more attractive than borrowing internationally. This removed a source of potential instability. And yet, an ounce of prevention would have helped--in particular, fiscal policy should have played a stabilizing role Better policies in the boom years would have positioned countries better but not fully insulated them from the crisis. Countries with fi xed exchange rates, such as Latvia, had limited scope for monetary policy. And their choice of exchange rate regimes (though perhaps justified when introduced) together with the prospect of euro adoption, might have increased risk-taking in for- eign currency. The countries were not notably active in sterilizing foreign assets. That would have raised domestic interest rates, which, in a period of abundant global liquidity, would have stimulated even more capital inflows. To limit the growth of foreign borrowing, countries introduced prudential measures, such as marginal reserve requirements on bank foreign borrowing, higher risk weights for housing loans in calculating capital adequacy ratios of banks, and, more generally, higher capital adequacy ratios across the board. The measures would eventually be circumvented, either by going offshore or to less regulated parts of the financial sector, given the strong incentives for borrowing on both the demand and supply side. But putting some sand in the wheels of massive capital inflows was still worth doing. Indeed, for some coun- tries in Southeastern Europe, it reduced (but did not prevent) over-heating. The limits of monetary policy and prudential measures argue for a stronger role for fiscal policy, and not just in countries with fi xed exchange rates. With an open capital account and high global liquidity, fiscal policy--adjusted for the business cycle--should play a stabilizing role in the face of external imbal- ances, even when they are not of public origin. This would dampen demand and lead to some slowing of growth. But by creating domestic sources of finance for the private sector, it would reduce the attractiveness of borrowing abroad. The opposite policy stance was, however, observed in some financially OVERVIEW 7 integrated vulnerable countries.9 This is not to suggest that fiscal policy alone would have insulated countries--the necessary adjustment would have been too much to bear for a single policy instrument--but it could have made them less vulnerable to a reversal in market sentiment. In fact, those fi nancially integrated countries that pursued counter-cyclical policies more actively were more successful in reducing over-heating. It is more difficult to assign causal- ity to a single instrument, such as fiscal policy, in accomplishing this, since the more successful countries had for the most part already achieved disinflation and had credible monetary policies before the period of high global liquidity. It is striking however that the most vulnerable countries relaxed their fiscal stance--adjusted for the business cycle--just when this was least advisable. Recovery The crisis in advanced country fi nancial markets was transmitted in ECA through a less-than-full rollover of maturing external debt, a drying up of new external financing sources, and a slowdown in exports and remittances. Adjustment and financing trade-offs--in some ECA countries adjustment has been massive The scarcity of new external financing implies that the current account defi- cit had to shrink to whatever fi nancing was available--to zero, if there was none. Less-than-full rollover implies that the current account has to adjust further--perhaps even into surplus to finance maturing debt that is not rolled over. For many fi nancially integrated countries, this combined adjustment was significant, particularly when initial current accounts were in double dig- its and maturing debt was above 20 percent of GDP. The result has been a sharp improvement in the private sector balance (pri- vate savings less private investment) to accommodate both a worsened fiscal bal- ance and a large reduction in capital inflows. Current account deficits on the eve of the crisis reflected more an excess of private spending over income (as in the East Asia crisis 1997­98) than fiscal profligacy. So, the bulk of the adjustment has been a massive cut in private spending, making the private sector a net saver. Indeed, Latvia and Ukraine, two of ECA's worst hit countries, are now run- ning current account surpluses and experiencing marked declines in imports. Fiscal balances have worsened sharply, reflecting both a collapse in revenue (due to the downturn in activity) and an increase in spending (due in part 9. For example, public expenditures in real terms increased by more than 110 percent in Latvia over the period between 1998 and 2008, compared with an increase of slightly less than 40 percent in the Czech Republic. 8 TURMOIL AT TWENTY to the need to expand social programs). Hence, the improvement in the pri- vate sector balance as a share of GDP that has occurred between 2008 and 2009 in the hardest hit countries has been extraordinary, amounting to more than 25 percent in Latvia and 10 percent in Ukraine. Oil exporters among the financially integrated countries, such as Kazakhstan and the Russian Federa- tion, have had more room for maneuver since the stabilization funds saved for a rainy day could support a more expansive fiscal stance. But they have had more problems rolling over maturing debt and securing new money. Transition also partly shaped the crisis response and rollover risks Navigating the aftermath of the crisis has so far been easier for financially integrated countries with majority foreign-owned banking sectors. Foreign banks exposed countries to considerable risk, as did other foreign lenders. This is because domestic interest rates in excess of international lending rates, together with the expectation that borrowing countries would join the euro area at a fi xed rate for peggers and at an appreciated rate for floaters, strongly encouraged lending in foreign currencies. But once the crisis hit, countries with majority foreign-owned banking sectors, such as Romania and some of the Baltic states, have generally rolled over maturing external debt owed to par- ent banks, at least so far. Countries with majority domestic-owned banks, such as Kazakhstan and the Russian Federation, more reliant on volatile wholesale funding, have had considerably more difficulty. Reliance on wholesale funding is negligible in Romania and ranges between 5 and 20 percent of total fund- ing in the Baltic states. In contrast, half the funding for the banking sector in Kazakhstan comes from the wholesale market. Ukraine is a borderline case, with a mix of foreign banks with minority ownership in banking sector assets and a number of domestic banks with weak governance structures. Latvia, to a lesser extent, is also a borderline case: compared to the other Baltic states, it is more dependent on wholesale funding and nonresident deposits. In sum, the rollover of wholesale funding has proved difficult across the region. Table 1 classifies the fi nancially integrated ECA countries by available source of funding (the columns) and by the pace of growth of private sec- tor credit as defi ned earlier--as excessive or convergent (the rows). In addi- tion to wholesale and parent bank funding sources, the table includes resident deposit-taking operations, which have so far proven to be a source of stabil- ity. It is worth noting that crisis-hit countries that have recently sought offi- cial financing, such as Latvia, Romania and Ukraine, generally experienced excessive credit growth. The same is broadly true of the Russian Federation and Kazakhstan, which have tapped stabilization funds accumulated from oil OVERVIEW 9 TABLE 1 Credit market characteristics in financially integrated countries Bank funding Resident Wholesale Parent bank deposits Czech Rep. Croatia Slovak Rep. Convergenta Hungary Lower risk Credit growth Serbia Poland Turkey Macedonia, Kazakhstan Bulgaria Gray area FYR Russian Ukraine Lithuania Excessiveb Romania Higher risk Federation Latvia Estonia Medium Low stability Gray area High stability stability a. Countries experiencing credit growth rates that could be viewed as in line with those of coun- tries with similar initial ratios of private sector credit to GDP. b. Countries experiencing credit growth rates that could be considered as above average; that is, their credit growth rates are too high relative to their initial ratios of credit to GDP. revenue. The exception is Hungary, where the origin of macroeconomic imbal- ances, in contrast to much of the region, was more fiscal in nature. Secur- ing a high rollover has been easier for Romania and some of the Baltic states, which rely primarily on a combination of parent bank and deposit funding, but also for Hungary and Ukraine, where reliance on wholesale funding is borderline. For the Baltic states, this also reflects the long maturity of some of the loans that have been extended (for example, mortgages that account for almost 50 percent of the loan portfolio have an average maturity of slightly over 25 years), which in effect limits the liquidity that is available for with- drawal. What is clear, however, is that countries that rely more heavily on wholesale funding, such as Kazakhstan and the Russian Federation, have had considerable difficulty in rolling over maturing debt. At the other extreme, countries with convergent credit growth and predomi- nantly reliant on resident deposit funding, such as the Czech and Slovak Repub- lics and Poland, as well as Turkey, a nontransition country, have so far come through the crisis well. Indeed, countries relying primarily on such sources have had lower credit growth rates and have prevented a run on banks through coordinated action and deposit insurance schemes. Of course, the classification in table 1 broadly suggests the principal factors in play and does not fit every case, as is evident from the experience of Serbia, which has sought official sup- port after being hit by the crisis, even with convergent credit growth. A high rollover has been partly determined by the nature of financial integra- tion, itself the result of choices made in the early years of transition. Introducing 10 TURMOIL AT TWENTY foreign-owned banks while countries were moving from plan to market broke the symbiotic link between governments and state enterprises and newly priva- tized enterprises. Foreign bank ownership helped harden budget constraints and attain macroeconomic stability. Indeed, in most Central European coun- tries, systemic banking crises had been the trigger to sell banks to foreign finan- cial institutions. In turn, foreign banks engaged in restructuring and intro- duced better management practices. And at least till the onset of the crisis, they made possible a reduction in nonperforming loans and an increase in loan loss provisions that were accompanied by rising profitability. They also provided households with mortgage and consumer finance. And in the Czech Repub- lic, Hungary, and Poland, they provided credit to previously underserved small and medium enterprises. But here, too, there is variability. For example, foreign ownership of banking sector assets is 39 percent in Kazakhstan and 17 percent in the Russian Federation, but has a more dominant role among the new mem- ber states of the European Union and in the Western Balkans. Financial integration driven by a majority foreign-owned banking sector is, however, less of an option for ECA countries that do not have European aspirations. Among transition countries, majority foreign-owned banking sectors are generally found in the new member states of the European Union, accession and candidate countries, and the countries in the Western Balkans that have Stabilization and Association agreements with the European Union. Since this is less of an option for countries such as Kazakhstan, the Russian Federation, and Ukraine, their pattern of financial integration might be more akin to that in the financially integrated East Asian countries, with greater reli- ance on wholesale funding. In such cases, it becomes particularly important to calibrate the openness of the capital account to the strength of domestic finan- cial sector institutions and to ensure that monetary, fiscal, and prudential mea- sures limit the transmission of risks from world financial markets. Easing the pain in aggregate: collective action and official financing have so far been effective, but more of both are likely to be needed--and should continue for several years Official financing, more generous than in the East Asia crisis, has been comple- mented by collective action efforts. Reflecting lessons from earlier emerging mar- ket crises, the international financial institutions led by the International Mon- etary Fund and, for its member states, the European Union, have mounted an adequate response to the crisis. And the European Banking Coordination Initia- tive (the Vienna initiative) has attempted to ensure that Western European par- ent banks maintain their exposure to ECA countries and adequately capitalize OVERVIEW 11 their banks. The World Bank Group, the European Bank for Reconstruction and Development, and the European Investment Bank have put forward a joint initia- tive to support the recapitalization of banks and the provision of debt financing for banks across the region, including that for the subsidiaries of parent banks. The combination of generous official financing and policy reform in return for support led by the international financial institutions has allowed countries with liquidity problems to avoid imposing standstills on external payments, while providing creditors some assurance that policy reforms will help enlarge their capacity to service debt. By staying engaged over the multiyear period required for reforms to bear fruit, official finance provides countries an incentive to imple- ment reforms. These are encouraging examples of collective action. Collective action should continue until a robust world recovery is in place. A protracted recession and weak economic recovery cannot be ruled out. As discussed later in this overview, the need to reduce overleveraging in Western European parent banks and recognize their estimated losses could cause them to reduce exposure to ECA countries, in turn undermining the latter's abil- ity to navigate the crisis successfully. And deleveraging will be necessary and inevitable as banks move resources to countries with more favorable lending opportunities. Collective action can ensure that such deleveraging remains orderly and gradual. Address the human dimension of the crisis . . . The crisis has the potential to exact a high human toll, especially among the low-income and lower middle-income countries in the former Soviet Union. Since much of their external borrowing comes from official sources, these countries have not, for the most part, been directly affected by a reversal in inflows of foreign capital on account of global deleveraging. Instead, they have been hard hit by the global economic recession, which has led to a collapse of export demand and, for a number of them, of workers' remittances. The fall in remittances can create great hardship. In Tajikistan, the poorest country in the region, it is estimated that a 30 percent decline in remittances would cut household consumption in the poorest quintile by around 20 percent. The human toll is not confined to the poor countries. Preliminary informa- tion suggests that registered unemployment more than doubled in the Baltics while it increased by about 60 percent in Turkey and between 20 and 40 percent in the Czech and Slovak Republics, Romania, the Russian Federation, Slovenia, and Ukraine. Even recognizing that incentives to register depend inter alia on the generosity of benefits to be received, the data suggest that unemployment may be on the rise in many financially integrated ECA countries. Poverty is on 12 TURMOIL AT TWENTY the rise, too. In the Russian Federation, the numbers in poverty as a proportion of the population (using a national poverty line based on an officially defined subsistence level) increased by almost one-third between the last quarter of 2008 and the first quarter of 2009, implying 6 million more people in poverty. . . . and scale up safety nets (to avoid a humanitarian crisis) Social safety nets will have to be strengthened to deal with the human dimen- sion of the crisis. Most ECA countries--including, encouragingly, several low- income and lower-middle-income countries--have at least one well-targeted program where a high proportion of benefits reaches the poorest quintile of households and that could be scaled up in response to the crisis. Indeed, in many ECA countries the targeting compares favorably with that of specific programs in Latin America. Examples of well targeted programs include the Unified Monthly Benefit in the Kyrgyz Republic and the Ndhima Ekonomike Program in Albania, which deliver more than 45 percent of their benefits to the poorest quintile. And in the South Caucasus, the Family Benefit Program in Armenia and the Targeted Social Assistance Programs in Azerbaijan and Georgia deliver between 55 and 60 percent of their benefits to the poorest quintile. But their coverage--the share of the poorest quintile reached by these programs--is gen- erally less than 35 percent and could be expanded by consolidating other leg- acy "privileges." There are examples of good targeting in middle-income coun- tries, too. For example, 80 percent of the benefits of the Romanian GMI and the Ukraine Extreme Poverty program accrue to the poorest quintile of house- holds, but reach only between 15 and 25 percent of them. The generosity of means-tested safety nets is highly variable, ranging from modest in Albania, Bosnia, Kazakhstan and the Kyrgyz Republic to generous in Estonia, Georgia, and Kosovo (figure 1). Ten of the countries transfer at least one-fift h of post-transfer consumption of beneficiary households. Many countries in the region inherited social programs from central plan- ning with room for rationalization and targeting. A few ECA countries, such as Bosnia, Moldova, and the Russian Federation, have adequate spending but inadequate programs for channeling resources to the poor. Reforms can help consolidate programs, eliminate most untargeted privileges, and refocus design and eligibility criteria. Alternatively, new poverty-focused programs can be introduced (within available spending envelopes) while applying tar- geting tools and improving implementation arrangements. While social safety nets in many ECA countries have several strengths, there is no room for complacency. Indeed, preliminary data from a few coun- tries show significant declines in the number of beneficiaries between June OVERVIEW 13 FIGURE 1 Means-tested safety nets: targeting accuracy, coverage, and transfers to the poorest quintile Targeting accuracy (percent of total means-tested program benefits going to poorest quintile) 100 Ukraine XP Kazakhstan 17 Romania TSA 13 Bulgaria GMI 32 GMI 23 Estonia SB 40 Kosovo 75 Montenegro MOP Turkey CCT SA Croatia 43 SW 29 Georgia TSA Armenia FPB Turkey Green Card Azerbaijan 48 33 Hungary TSA 50 RSA 27 Poland Serbia Macedonia, FYR CA SFA 27 TSA 9 Lithuania SB 20 Kyrgyz Rep. 22 UMB 9 Albania NE 10 Russian Federation CA 25 Bosnia CPA 12 Transfers as a percentage of the receiving household's posttransfer consumption Programs for which transfer data are not available 0 0 10 20 30 40 50 60 Coverage (percent of households in the poorest quintile receiving the benefit) Note: The figure shows the targeting accuracy and coverage of means-tested safety net programs. The size of the bubbles is a measure of their generosity--the transfer received by households in the poor- est quintile as a share of the post-transfer consumption of all beneficiary households in that quintile. 2008 and June 2009--when more households have become vulnerable. The situation bears close monitoring. Official financing in the poorer ECA countries should be stepped up if expected private flows do not materialize While official financing has been adequate so far, larger and sustained official financing for the low-income and lower-middle-income countries of the for- mer Soviet Union will be required to support desirable social spending and to prevent the global recession from becoming a humanitarian crisis. For- eign direct investment, which fi nanced between 70 and 85 percent of cur- rent account deficits in Armenia, Georgia, Kyrgyz Republic, and Tajikistan in 2008, is expected to decline on average by more than half in 2009 to a point where it will fi nance roughly one-third of current account deficits. If even those levels of FDI do not materialize, however, as a result of delayed recovery of trade and exports, official financing will need to be stepped up. Prepare for a less exuberant future in an even more competitive external environment . . . Despite early evidence suggesting that the crisis has bottomed out, the growth outlook for the region is fraught with difficulty. Th ree factors support this 14 TURMOIL AT TWENTY concern. First, when compared with other types of recessions, those associ- ated with financial crises that are globally synchronized, like the current one, witness a deeper decline in GDP from the previous peak to the new trough and take longer to arrive at the new trough. And the time taken for GDP to recover to its previous peak is longer, and the gain in real GDP a year after the trough has been reached is weaker. This is because households increase sav- ings out of disposable income and firms repair their balance sheets after the excesses of the boom years, leading to deep declines in private consumption and investment. Indeed, IMF projections suggest that the contraction in GDP in the Baltic states, Bulgaria, and Hungary will continue in 2010 as well.10 Second, given that the leverage of Western European banks is exces- sively high, attaining more sustainable levels would require fresh capital on a scale that may constrain the ability of fiscally strapped Western European governments to support their banking systems. That outcome, if it materi- alizes, will entail substantial deleveraging, including from the ECA coun- tries. Th is will limit external credit as a source of fi nancing. If in addition exports do not expand rapidly because of a weak global recovery, the growth of imports will be restrained, limiting the expansion in private consumption and investment. Third, even after world economic growth restarts and the Western Euro- pean banks return to health, growth is unlikely to continue at the pace seen before the crisis as growth rates over the past decade were underpinned by abundant global liquidity. For the new member states of the European Union and the accession countries, this could imply inter alia a slower convergence to Western European living standards. . . . while addressing bank, corporate, and household indebtedness (to prevent the recovery from stalling) . . . Slow restructuring of banks could hold back the recovery of growth. Nonper- forming loans are signaling systemic distress among borrowers: in Latvia and Ukraine, for example, they account for between 15 and 25 percent of all loans. The proportion is higher in sectors that were booming during the years of rapid credit growth, such as construction. In this context, regulators have begun to triage banks into those that are viable and meet regulatory requirements, non- viable and insolvent, and viable but undercapitalized. Based on such assess- ments, they are also taking actions appropriate to each case--from liquidation and recapitalization, to sale and merger. These efforts should proceed swift ly 10. IMF 2009b. OVERVIEW 15 to avoid the earlier experience of transition countries in the 1990s: until banks were put on a strong footing, economic performance lagged. . . . and establishing enabling frameworks for debt restructuring--but committing public resources only under special circumstances and careful design To strengthen the recovery, household and corporate financial difficulties need to be addressed upfront. Most ECA countries have an insolvency framework that can deal with bankruptcy, reorganization, and liquidation. But fledgling judicial systems can be overwhelmed when a large number of firms need restructuring-- as in a systemic crisis. To expedite restructuring, governments should consider setting up a system of out-of-court voluntary workouts between debtors and creditors to expedite debt restructuring, just as the Czech Republic, Indonesia, Korea, Malaysia, Mexico, Thailand, and Turkey did in previous crises. While households drove a large share of the credit growth over the past decade, here too there is variability within ECA. On average, household debt accounts for a quarter of GDP in financially integrated transition countries, ranging from 10 percent of GDP in the Russian Federation to 50 percent in Estonia. This is not out of line with that in countries at similar stages of development. Governments may need to set up mortgage restructuring protocols--adapted to local circum- stances--to facilitate negotiations between debtors and creditors. These proto- cols put forward objectives and options on how to restructure loans and may require changes to the insolvency legislation to ensure that the legal framework gives debtors the incentive to negotiate with creditors in good faith. But beyond setting up enabling frameworks, governments should discourage regulatory for- bearance and resist calls to use public resources other than for the relief of the poorest households, which account for only a modest proportion of debt. Reform bank regulation and supervision . . . It is important for ECA countries to strengthen bank regulation and supervision to reflect the lessons learned from the present crisis. Better bank regulation and supervision alone would not have averted the crisis, owing to the strong demand and supply incentives for rapid credit growth. Yet, better regulation and super- vision might have placed countries in a stronger position to deal with its con- sequences. In fact, reform of bank regulation and supervision appears to have weakened after 2000: more specifically, a survey of bank regulators reported declines in the ratio of actual-to-required capital adequacy ratio and a fall in the number of countries responding that the supervisory agency had the authority to declare a bank insolvent. Unlike those in a number of industrial countries, 16 TURMOIL AT TWENTY banks in most ECA countries were neither exposed to toxic assets, though some parent banks were, nor part of a shadow banking system. This implies that improvements in regulation and supervision should be tailored to the problems facing ECA countries at their current stage of development and that are likely to arise as their systems of financial intermediation develop further. . . . by increasing capital requirements, Higher overall required ratios of capital to risk-weighted assets--larger than the minimum 8 percent under Basel I and Basel II--are desirable for financial institutions in countries that face volatile capital movements. This would pro- vide a first line of defense to protect against operational and market risks and foreign currency loans, all of which are important in the financially integrated ECA countries--even though it would err on the side of financial stability at the expense of some loss in financial intermediation. . . . strengthening supervision of foreign banks Cross-border banks play a major role in many ECA countries, and this raises important issues of supervision for home and host countries. While information sharing between home and host countries is desirable, there are also incentives to keep the supervisors apart. Host countries can ring-fence subsidiaries to pro- tect depositors and limit costs to the deposit guarantee system. And home coun- tries can centralize a bank's assets while keeping its liabilities decentralized. The European Union is considering proposals for a new financial architecture-- a European System of Financial Supervision that would develop a consistent and strengthened set of supervisory standards, coordinate the application of national supervisory standards across countries, and set up supervisory colleges for major cross-border financial firms. These would automatically apply in the new member states and likely be adopted by accession and candidate countries, thus covering the bulk of ECA countries where cross-border banks are impor- tant. But absent the ability of blocwide regulators to influence spending priori- ties of member states, the effectiveness of these arrangements is unclear. . . . and participating in a system of macro-prudential supervision Consideration needs to be given to assessing global systemic risk that arises from a common exposure of many financial institutions to the same risk factors. The European Union is considering setting up a European Systemic Risk Council comprising the European Central Bank and central banks of the member states to pool information relevant for financial stability and issue macro-prudential warnings for the European Union. Other financially integrated ECA countries OVERVIEW 17 without European aspirations may wish to consider how such micro- and macro- prudential supervisory arrangements could be adapted to their contexts. Finally, there is an extensive agenda for modernizing banking sector institutions. The World Bank-IMF Financial Stability Assessment Programs (FSAPs) and the Reviews of Standards and Codes (ROSCs) provide numerous recommendations on how to improve bank supervision in ECA countries. Reform Recession and immediate recovery have so far dominated the policy agenda, but, to remain competitive in a post-crisis world, countries now need to reinvigorate structural reforms in areas that constitute the tightest bottlenecks to growth. Despite an uncertain future, it is likely that capital flows will be considerably lower and will go to countries with the most attractive business environment. Greater discrimination by investors is already present in country spreads. High pre-crisis growth caused infrastructure and labor skills--the positive legacy of socialism--to emerge as binding constraints to growth and raised the cost of weak market economy institutions Which are the most important areas for action? On the eve of the crisis, firm managers in transition countries in an ECA-wide enterprise survey reported that, among all the elements of their business environment, infrastructure and labor skills were the most constraining in terms of their ability to oper- ate and expand their businesses.11 Other elements of the business environ- ment included taxation, labor regulation, customs administration, licens- ing, the rule of law, and finance. Because all of them, with the exception of fi nance, resemble public goods whose supply is common to all firms in the economy, firms' responses are a measure of the high costs imposed by short- ages of infrastructure and labor skills on the operating and growth of their businesses. The survey evidence highlights emerging shortfalls of investment in physical infrastructure, especially in the upper middle-income transition economies, and in education, especially in the low-income and lower middle- income-countries.12 And access to land for business expansion had become more problematic in many countries. 11. The analysis is based on the Business Environment and Enterprise Performance Survey (BEEPS), conducted by the World Bank and EBRD every three years since 1999. The responses in 2008 (BEEPS 4) correspond to 10,000 fi rms in 28 transition countries. They are contrasted against responses by 51,000 fi rms in 74 non-transition countries under the World Bank's Investment Climate Assessments. 12. Comparison of the constraints to growth across countries requires that the responses correct for the fact that samples of fi rms from different countries differ by characteristics such as firm size, eco- nomic sector, and nature of ownership, whether domestic or foreign. 18 TURMOIL AT TWENTY In addition, strong economic growth appears to have increased the cost of weak market economy institutions, especially the legal environment in the lower middle-income transition economies and corruption in the low- income and lower-middle-income transition economies. But other insti- tutions of the market economy--such as tax administration and customs regulation, which traditionally ranked high among concerns in transition economies--are seen by fi rms as less constraining for business and have fallen in importance in line with nontransition economies at similar per capita incomes. These bottlenecks are particularly important for exporters During a period of growth fueled by large capital inflows and increasingly oriented toward nontradables, exporting firms in ECA's transition countries reported their business environment as being more constraining compared with earlier years. Exporters in the upper middle-income transition coun- tries complain mainly about electricity and access to land. Those in the lower middle-income transition countries complain about corruption. And those in the low-income countries complain about tax administration. In contrast, exporting firms in nontransition economies at similar per capita incomes do not report the business environment on average to be more constraining than nonexporters. Transition economies have come to resemble nontransition economies in some respects The changing composition of constraints to growth shown in the 2008 survey allows broader comparisons between transition and nontransition economies after two decades of transition. First, the superior endowment of infrastructure and labor skills with which countries started transition was the product of the legacy of forced industrial- ization. Over-industrialization involved levels of physical infrastructure and, for the poorer transition economies, labor skills that were "excessive" when measured against a market economy benchmark. It is thus striking that in 2008 fi rms in transition economies reported higher levels of concern with infrastructure and labor skills, compared with nontransition economies, for the first time since transition began. Second, the richer transition economies--with a stronger collective mem- ory of market institutions--show greater sensitivity to weaknesses in market- related institutions such as the legal system; a variation by income that is missing in nontransition economies. However, progress in transition in the OVERVIEW 19 lower middle-income transition economies has steadily raised the sensitivity of firms to poorly functioning legal systems there as well. Third, the structure of finance for fi xed investment in transition economies had become similar to that in nontransition economies by 2008. While firms in transition economies had relied more on retained earnings to finance fi xed investment, the gap has been falling over time. And while transition firms relied as much or more on informal finance than nontransition firms a decade ago, they had become less reliant on informal finance than nontransition firms by 2008, reflecting a shift to the formal economy. Bank finance had replaced both informal and internal finance by 2008. The structural reform agenda Priorities for reform will vary across countries depending on the relative severity of different constraints. However, the survey evidence suggests that structural reforms in infrastructure and labor skills will be important, as will continuing reform of market economy institutions. In an environment with limited public resources, private investment will be crucial to opening infrastructure bottlenecks For electricity, the challenge is to have a business environment conducive to private investment and participation in the sector. The latest enterprise sur- vey shows that complaints about electricity as a constraint to growth in the richer transition economies rose above those in comparable nontransition countries. But those in the poorer transition economies rose to levels that are comparable to those in nontransition countries. Countries for which concerns about electricity were greater include Albania, the Czech Republic, Kosovo, and the Kyrgyz Republic. Because investment needs will be significant and public resources limited, attracting private investment is an option that ECA countries could exploit more. Although the situation varies by country, this will require commercially viable utilities, competitive trading regimes, and credible regulatory frameworks. There is also considerable potential for deep- ening regional cooperation and trade in power in Southeastern Europe and Central Asia. Also key is to strengthen the education and skills of the labor force For education and skills, strengths need to be maintained and emerging weak- nesses addressed. Firms in Estonia, Belarus, and to a somewhat lesser extent Kazakhstan and the Russian Federation, as well as Moldova and Tajikistan among the poorer countries, encountered more severe problems with skilled 20 TURMOIL AT TWENTY labor than did other countries. In response to the growing demand for skilled labor, there is a high, and in some cases growing, wage premium associated with tertiary education. Primary school-age students perform well in international tests, but this advantage needs to be consolidated. Vocational education, which has been historically important, is in need of reform. Indeed, evidence from the Rus- sian Federation suggests that the wage premium to vocational education has fallen sharply during the most recent decade. Yet vocational graduates do rel- atively well in employment relative to general education graduates in many countries in the region. While tertiary education has expanded in large part through the entry of private providers, the lag for strengthening licensing and accreditation systems and the lack of information for students to make informed choices means that the discipline of either the regulator or the market is absent. Governments need to unshackle institutions, especially of primary and secondary education, and allow them greater autonomy in the management of staff and resources. They also need to hold institutions more accountable for learning outcomes--by monitoring quality and making infor- mation public on quality and, for tertiary education, by strengthening licens- ing and accreditation. And they need to use financing more strategically-- for example, by allowing financing to follow students (based on capitation formu- lae) and by building in incentives for performance, as opposed to making allo- cations based on outdated norms that bear no relationship to quality. Conclusion This overview started with three questions, the answers to which may now be summarized. First, transition shaped the nature of vulnerability but was not its sole determinant. In particular, in some financially integrated countries that were late comers to the transition, excessive credit growth was fed on the demand side by households attempting to catch up to Western European living stan- dards. In that sense, transition shaped the nature of vulnerability and made transition more prone than developing countries to being hit by the global economic crisis. But that is only part of the story. Turning to the supply side, the ECA countries were integrating into the world economy at a time of his- torically high global liquidity, when fierce competition in international bank- ing provided an abundant supply of credit to emerging market economies. However, countries could have done more to manage risks: in particular, fiscal policy could have played a greater role in limiting vulnerability to a sudden change in market sentiment. OVERVIEW 21 Second, crisis-hit countries that had decided as part of the transition to privatize banks to foreign investors (in order to harden budget constraints and attain macroeconomic stability) have generally navigated the crisis more successfully over the past year. Foreign banks--and other foreign lenders-- exposed countries to greater risk. But crisis-hit countries with majority foreign-owned banking sectors have so far been able to roll over maturing external debt to parent banks more easily than those with majority domes- tic-owned banking sectors, which relied more heavily on wholesale funding. Indeed, the rollover of wholesale funding has been and continues to be diffi- cult across the region. Two caveats should be noted, however. One is that while parent bank fund- ing has proved to be stable so far, risks remain and whether it will continue to be stable depends on the health of Western European parent banks. Those banks could come under strain in the event of a prolonged recession and weak global recovery. In that case, collective action and generous official financing, both distinctive features of this crisis, will need to continue for some years. The other caveat is that fi nancial integration driven by a majority foreign owned banking sector and collective action under the aegis of the European Union and the International Monetary Fund to ensure continuation of expo- sure by parent banks may, however, be an aspect of European integration that is less of an option for ECA countries in the CIS that do not have European aspirations. For them, the pattern of integration may be more akin to that in the financially integrated East Asian countries. But the poorer, less financially integrated countries must not be forgotten. They have been hit hard as a result of their integration into product and labor markets, due to the downturn in world trade and a falloff in workers' remit- tances. In these countries, substantial official fi nancing will need to be sus- tained over a number of years in support of desirable social spending to pre- vent the human dimension of the crisis from becoming a humanitarian one. Third, high growth in the years preceding the crisis caused firms in transi- tion economies to encounter capacity constraints in infrastructure and labor skills--the positive legacy of socialism--for the fi rst time since transition began. At the same time, progress in transition caused firms and economies to outgrow the capacity of institutions to provide public goods to support mar- ket economies, as is evident from concerns about the legal and judicial system, labor regulation, and corruption. Looking ahead, an economic environment where growth is likely to be lower (as households and fi rms repair their balance sheets) and public resources more limited will reward countries that improve their business 22 TURMOIL AT TWENTY environment. Moreover, staying internationally competitive in a world of reduced capital flows, when financial markets have already become more dis- criminating, requires that governments undertake structural reforms in the areas that firms in their countries report as imposing the highest costs for their operation and expansion. For example, attracting private investment in infrastructure sectors such as electricity will be key, and this book highlights ways to do so efficiently without adding to the burden on public finances. At the same time, the state has the important task of continuing to strengthen institutions that provide essential public goods--from strong banking super- vision and regulation of tertiary education to a working judicial system. The record of the past 10 years shows that transition countries can accomplish much in building institutions of the market economy. As they did 20 years ago, there is every reason to believe that countries in the region will rise to the challenges facing them today. OVERVIEW 23 CHAPTER 1 Prelude to the crisis The Europe and Central Asia (ECA) region, among the most integrated of emerging and developing regions, has been hit hard by the global economic and financial crisis through all three channels of integration: financial (a sud- den stop in capital flows), goods and services (a sharp decline in exports), and labor (a slowdown in remittances).1 It is the hardest hit developing region partly because of the buildup of imbal- ances that accompanied financial integration. But not all ECA countries--and not even all financially integrated countries in the region--are equally vulnerable. Questions · Was integration into the global financial system desirable? · Does variation in country outcomes suggest that different policies could have cushioned the impact of the global economic and financial crisis? Findings · High rewards involve risks and integration is no exception; it brought the ECA countries above average income convergence to EU15 levels and mod- ern banking sector systems. · Unusually abundant global liquidity and strong demand for credit by house- holds attempting to catch up to Western living standards stretched the usual policy toolkit to its limits. But countries could have done more on the fiscal policy side to position themselves better when the crisis struck. Several coun- tries tried prudential measures, which lessen overheating pressures. These measures are worth pursuing even though they are eventually circumvented. Notes 1. The ECA region comprises Central and Eastern Europe, including Turkey (CEE), and the Commonwealth of Independent States (CIS). In the decade since the 1998 Russian fi nancial crisis, the ECA region has become well integrated into the world economy. In the quest to converge with EU15 per capita incomes, most ECA countries outperformed other develop- ing regions (figures 1.1­1.4). Some 55 million people were lifted out of absolute poverty between 1999 and 2006.1 1. The poverty line used is $2.50 a day in 2005 purchasing power parities. 25 FIGURE 1.1 Income convergence with EU15, by country, 1999 and 2008 Ratio to EU15 median value (income per capita in purchasing power parity terms) 100 1999 2008 75 50 25 0 Czech Rep. Slovak Rep. Albania Armenia Azerbaijan Belarus Bosnia and Herzegovina Bulgaria Croatia Estonia Georgia Latvia Lithuania Kazakhstan Kyrgyzstan Macedonia, FYR Moldova Poland Romania Russian Federation Serbia Slovenia Tajikistan Turkmenistan Ukraine Uzbekistan Latin America & Caribbean East Asia & Pacific South Asia Middle East & N. Africa Sub-Saharan Africa Hungary Turkey Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. FIGURE 1.2 Trade integration, by region, 1994­2008 Exports plus imports (percent of GDP in purchasing power parity terms, median values) 100 EU10+1 75 50 Western Latin America Balkans East Asia & Caribbean 25 Turkey Low-income and lower Middle-income CIS ECA middle-income CIS 0 1994 1996 1998 2000 2002 2004 2006 2008 Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. · Trade integration in ECA's transition countries (the ECA region less Turkey)-- measured by the sum of exports and imports as a share of GDP in purchasing power parity--increased sharply, particularly since 2002, to reach around 50 percent of GDP in 2008 compared with 35­40 percent in developing East Asia and Latin America. The corresponding share for Turkey rose to 30 percent.2 2. The country groups used to describe the transition economies of Europe and Central Asia are: EU10 and Croatia (Bulgaria, the Czech Republic, Croatia, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, the Slovak Republic, and Slovenia); the middle-income CIS (Belarus, Kazakhstan, the 26 TURMOIL AT TWENTY FIGURE 1.3 Financial integration, by region, 1994­2008 Foreign exchange assets plus liabilities (percent of GDP, median values) 50 EU10+1 and Croatia 40 Middle-income CIS 30 Western Developing Balkans East Asia ECA 20 Turkey 10 Latin America & Low-income and lower middle-income CIS Caribbean 0 1994 1996 1998 2000 2002 2004 2006 2008 Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. FIGURE 1.4 Labor integration, by country, 2007 and 2008 Remittances (percent of GDP) 50 2007 2008 40 30 20 10 0 Slovak Rep. Czech Rep. Tajikistan Turkey Moldova Kyrgyzstan Bosnia and Herzegovina Serbia Albania Armenia Georgia Bulgaria Romania Macedonia, FYR Azerbaijan Lithuania Ukraine Croatia Poland Estonia Latvia Hungary Belarus Slovenia Russian Federation Kazakhstan Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. · Financial openness in the new member states of the European Union plus Croatia--measured by the sum of foreign exchange assets and liabilities as a share of GDP--was three times the average for developing East Asia in 2008. Russian Federation, and Ukraine); Western Balkans (Albania, Bosnia and Herzegovina, Kosovo, FYR Macedonia, Montenegro, and Serbia); and the low-income and lower middle-income CIS (Armenia, Azerbaijan, Georgia, the Kyrgyz Republic, Moldova, Tajikistan, and Uzbekistan). Georgia is not part of the CIS but is included in this group because its economy shares many features of the other countries. PRELUDE TO THE CRISIS 27 · High oil prices and a raft of structural reforms in the Russian Federation in the four years after 1998 dramatically reduced poverty, particularly in the middle-income CIS countries. · Labor and remittance flows broadly followed a biaxial pattern, with Central and Eastern Europe depending on the European Union and with the South Caucasus and many Central Asian republics depending on the resource- rich CIS countries. So it is not surprising that the global economic and fi nancial crisis has affected the region through all three dimensions of integration--a sudden stop in capital flows, a decline in exports in line with the weakening of the world economy, and a brake on migration and remittances following the slowdown in Western Europe and the resource-rich CIS. Among all emerging and devel- oping regions, ECA is projected to experience the deepest downturn in 2009 (­5 percent for CEE and ­6.7 percent for the CIS) and the weakest recovery in 2010 (1.8 percent for CEE and 2.1 percent for the CIS). Indeed, the degree of integration--and now vulnerability--show how far ECA countries have come since the fall of the Berlin wall two decades ago. But not all ECA countries have been affected to the same degree through each channel. Some countries were more vulnerable to the change in market sentiment and the decline in global liquidity. Balance sheet vulnerabilities accu- mulated during many years of high private sector credit growth and the external imbalances associated with this growth. In Latvia, for example, current account deficits were more than 20 percent of GDP in 2006 and 2007, only to decline into the mid-teens in 2008. Countries in the Western Balkans also had dou- ble-digit external imbalances. It is thus not surprising that these differences are now being priced by markets in sovereign bond spreads--changing the order- ing of countries by risk and showing marked increases in volatility (annex 1.1). Ukraine's spread has increased sharply, but Hungary's has not changed much since the fall of 2008, when the crisis intensified in both countries. Still oth- ers are suffering a slowdown in exports, and for many the prospects of a rapid recovery are unclear. The effect on migration and remittances is likely to be acute, particularly among the poorer countries in the region. But some of them, such as Tajikistan and Moldova, have remittances to GDP ratios of between 30­35 percent, among the highest in the world. Early estimates suggest that the region is likely to see 11 million individuals fall into absolute poverty by 2010. With this as background, the chapter answers two questions. First, was integration--particularly into global financial systems--desirable in light of evidence showing how deeply the region is being affected by the global eco- nomic and fi nancial crisis? Focusing on fi nancial integration, the analysis 28 TURMOIL AT TWENTY highlights a distinctive feature of how ECA countries--Central and Eastern Europe (including Turkey) and the former Soviet Union--have become part of the world economy and the risks and rewards that such integration has brought. Second, does the significant variability in country outcomes, even among financially integrated countries, suggest that different policies could have positioned countries better when the crisis struck? Vulnerable . . . but with variation Compared with other emerging economy regions, all ECA countries received substantial capital inflows for much of the current decade (figures 1.5­1.8). The region received high levels of FDI flows and large amounts of debt- creating external financing, evidence of the substantial transformation of the region. Some of these flows originated in parent banks and corporates and were directed to their subsidiaries in host countries. As will be seen in chap- ter 2, some debt-creating flows can be as stable as FDI. The region's GDP growth rate outperformed all other regions until the cri- sis. The differences across ECA countries partly reflect progress in the tran- sition to market (figures 1.9­1.14). Countries where transition-based output recovery had largely been accomplished by the turn of the new century grew slower than latecomers to the transition process--the Balkans and the coun- tries of the former Soviet Union. In addition, the region as a group recently recorded a gradual pickup in inflation--in line with other developing regions. Accompanying the strong growth was a buildup of external imbalances, which cannot be explained by a decline in terms of trade or, for the most part, by weak fiscal positions (see figures 1.9­1.14). External developments mirror an excess of private investment over private savings. Even in Hungary, Romania, FIGURE 1.5 Capital flows in developing East Asia Median values Private Foreign direct Official Current Change in (percent of GDP) (excluding investment account foreign exchange FDI) (FDI) net reserves 10 0 ­10 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. PRELUDE TO THE CRISIS 29 FIGURE 1.6 Capital flows in Europe and Central Asia Median values Private Foreign direct Official Current Change in (percent of GDP) (excluding investment account foreign exchange FDI) (FDI) net reserves 10 0 ­10 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. FIGURE 1.7 Capital flows in Latin America and the Caribbean Median values Private Foreign direct Official Current Change in (percent of GDP) (excluding investment account foreign exchange FDI) (FDI) net reserves 10 0 ­10 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. and Ukraine, where fiscal positions are known to have been weaker, private sector imbalances were still important. While the real exchange rate appre- ciated in most of ECA, this was driven in some cases by higher productivity, reflecting integration with Western Europe in particular and global markets more generally. Unlike other emerging and developing regions, ECA saw no decline in external debt-to-GDP ratios (figure 1.13). And the ratio of short-term debt to foreign exchange reserves increased--a harbinger of things to come. There is, however, substantial variability across countries in the region. That makes it instructive to contrast ECA countries with other regions, and with each other. For this, the region could be divided into countries integrating rapidly into financial markets--the emerging markets--and the low-income and lower middle-income countries. But using only two groups would not do justice to the diversity of economic experience among the region's financially integrated middle-income countries. To make these differences more evident, 30 TURMOIL AT TWENTY FIGURE 1.8 Capital flows in other emerging economies Median values Private Foreign direct Official Current Change in (percent of GDP) (excluding investment account foreign exchange FDI) (FDI) net reserves 10 0 ­10 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. FIGURE 1.9 Real GDP growth, median values, 2000­08 Percent 12 Middle-income CIS Low-income and lower middle- income CIS ECA 9 6 Western Developing Balkans East Asia 3 EU10 + 1 Latin America & Caribbean Turkey, 2001: ­5.7% 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. a technique known as cluster analysis classifies countries into groups based on the sources of vulnerability that affect ECA countries. Such analysis allows differences across countries to be identified simultaneously in a number of dimensions of domestic sources of vulnerability (private sector credit growth, fiscal deficits, loan-to-deposit ratios) and external sources (short-term debt- to foreign exchange reserves, external debt-to-GDP ratios, and exchange rate regime), without setting ad-hoc thresholds (box 1.1). A classification of financially integrated ECA countries by the sources of vul- nerability identifies five distinct groups. Ordered from groups that have the most sources of vulnerability to those that have the fewest the groupings are as follows: PRELUDE TO THE CRISIS 31 FIGURE 1.10 Current account, net of FDI, median values, 2000­08 Percent of GDP 10 Middle-income CIS 5 Developing East Asia Latin America & Caribbean 0 Turkey ­5 Low-income ECA and lower EU10 + 1 middle-income CIS Western Balkans ­10 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. FIGURE 1.11 Inflation, median values, 2000­08 Percent 30 Turkey, 2000: 55.0% 2001: 54.2% 2002: 45.1% 20 Middle-income CIS ECA Latin Low-income and lower America & middle-income CIS 10 Caribbean EU10 + 1 Developing East Asia Western Balkans 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. · Group 1: Estonia and Latvia. · Group 2: Bulgaria, Croatia, Hungary, Kazakhstan, Lithuania, and Montenegro. · Group 3: FYR Macedonia, Poland, Romania and Serbia, and Ukraine. · Group 4: Czech and Slovak Republics and Turkey.3 3. Slovenia in all likelihood would also belong to this group, but not all the data required to implement the cluster analysis methodology were available for this country when the exercise was conducted. 32 TURMOIL AT TWENTY FIGURE 1.12 Fiscal balance, median values, 2000­08 Percent of GDP 6 Middle-income CIS 3 Low-income and lower ECA middle-income CIS Developing East Asia 0 ­3 Turkey, 2000: ­11.8% Latin EU10 + 1 America & 2001: ­17.6% Caribbean 2002: ­14.4% 2003: ­10.4% Western Balkans ­6 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. FIGURE 1.13 External debt to GDP, median values, 2000­08 Percent of GDP 80 EU10 + 1 60 ECA Western Balkans 40 Turkey Latin America & Caribbean Middle-income CIS Low-income and lower middle-income CIS Developing East Asia 20 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. · Group 5: Belarus and the Russian Federation. The remaining countries in ECA are aggregated into Group 6. It is also important to recognize that the discussion in the remainder of this book reflects average characteristics within a group, but might mask intra-group differences. Still, looking at developments at the level of country groups pro- vides policy guidance for countries that might face similar sources of vulner- ability in the future. PRELUDE TO THE CRISIS 33 FIGURE 1.14 Ratio of short-term debt to foreign exchange, median values, 2000­08 Percent 250 Middle-income CIS 200 Turkey EU10 + 1 150 Developing East Asia 100 Low-income and ECA lower middle-income 50 CIS Latin America & Caribbean Western Balkans 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. The magnitude of current account imbalances varied widely across coun- try groups (table 1.1 and figures 1.15­1.18). · Current account imbalances as a percent of GDP during the period 2003­ 05 in the fi nancially integrated country groups were in deficit amount- ing to double digits in Group 1 and fell monotonically across the country groups to a surplus of nearly 4 percent in Group 5. · Current account imbalances as a percent of GDP worsened in 2006­08 in all the fi nancially integrated country groups, but particularly sharply in Groups 1 and 2. · Most external imbalances originated in the private sector; particularly in Groups 1, 2 and 3. Indeed, savings­investment balances in table 1.1 also reveal the change in the source of vulnerability between 2003­05 and 2006­08. While public sector imbalances were important in 2003­05, they seem to have abated for the most part during the most recent 3-year period. In contrast, private sector imbalances were not only large, but increased further during 2006­08. External imbalances were accompanied by improvements in the fiscal balances (figure 1.16). Indeed, both domestic and external public debt ratios declined in most countries while total external debt increased in some coun- tries. External debt ratios tripled during the decade in Group 1 and increased quite sharply in Groups 2 and 3, in contrast with the rest of the develop- ing world. Perhaps more worrisome is the deterioration of foreign exchange 34 TURMOIL AT TWENTY BOX 1.1 Shades of vulnerability--a cluster analysis approach to classifying countries Classifying countries by underlying vulnerabilities is not straightforward. For instance, focusing on short-term debt-to-GDP might provide a ranking of coun- try vulnerabilities. But how does one decide which are the relevant thresholds to divide countries into groups? The typical approach is to defi ne them based on standard deviations--but this still is ad hoc. And a classification based only on one indicator (such as short-term debt-to-GDP) fails to compensate for large foreign currency holdings at the central bank or a flexible exchange rate--both of which should relieve market pressures but would not be captured by a clas- sification built only on short-term debt-to-GDP ratios. An alternative that reduces ad-hoc biases is to apply a data partition technique known as cluster analysis, which breaks the dataset into high (top) to low (bottom) risk groups, based on indices of data similarities (or dissimilarities). The iterative process begins by treating each country as a separate group and reduces the num- ber of groups by aggregating observations that are the most similar--a process known as hierarchical agglomeration. No thresholds are needed. Instead, groups are aggregated based on Euclidean distances within and across subgroups. When applied to short-term debt-to-GDP ratios, the data seem to be repre- sented by numerous groups (box figure 1). The dissimilarity is represented by the length of the horizontal lines--the short horizontal lines suggest that dis- similarity is low. Some of the countries ranked at the top also are not the ones expected to be vulnerable. For instance, the Czech Republic is significantly more vulnerable than Hungary and Ukraine. Box figure 2 goes a step further by carrying out cluster analysis on domestic (private sector credit growth, fiscal deficits, and loan-to-deposit ratios) and external (short-term debt-to-foreign exchange reserves, external debt-to-GDP ratios, and exchange rate regime--as classified in the IMF's AREAER) sources of vulnerability. BOX FIGURE 1 Cluster analysis Latvia Lithuania Estonia Czech Rep. Bulgaria Macedonia, FYR Romania Croatia Hungary Kazakhstan Belarus Slovak Rep. Poland Ukraine Montenegro Serbia Turkey Russian Federation 0.0 0.2 0.4 0.6 Dissimilarity measure Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. (continued) PRELUDE TO THE CRISIS 35 BOX 1.1 (CONTINUED) Shades of vulnerability--a cluster analysis approach to classifying countries The advantage of cluster analysis resides with its ability to classify countries on numerous indicators. The country groups now appear to be more represen- tative of our priors, including recent developments in ECA countries and with- out imposing ad-hoc thresholds. Specifically, we group the 18 most financially integrated ECA countries into the 5 most different groups. The horizontal lines are longer than is the case when applying only one indicator, suggesting that it is important to consider the variety of country circumstances when classifying countries into groups. BOX FIGURE 2 Cluster analysis Latvia Estonia Montenegro Bulgaria Croatia Hungary Lithuania Kazakhstan Ukraine Macedonia, FYR Romania Poland Serbia Czech Rep. Slovak Rep. Turkey Belarus Russian Federation 0 20 40 60 80 Dissimilarity measure Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. liquidity indicators: by 2008, all the financially integrated groups 1 through 5 had ratios above one, implying significant liquidity risks.4 Transition meets global finance Nowhere is ECA's integration more striking than in the role of foreign banks. The financial sector played a historically important role in the transition from plan to market (annex 1.2). After nearly two decades of transition, two-thirds of the countries in the region have half or more of their banking sector assets in foreign hands (figure 1.19). The increase in foreign ownership of banking sector assets has gone hand in hand with increasing financial openness. 4. The Guidotti-Greenspan rule suggests that the ratio of short-term debt on a remaining matu- rity (short-term debt and medium- and long-term debt falling due in the next 12 months) to foreign exchange reserves should be less than one. 36 TURMOIL AT TWENTY TABLE 1.1 Savings­investments balance (percent of GDP, median values) Latin Europe and Central Asia America Developing and the Other Group 1 Group 2 Group 3 Group 4 Group 5 Group 6 East Asia Caribbean regions 2003­05 National savings 21.9 17.0 16.9 19.0 27.7 21.9 27.1 19.0 23.1 Public 2.9 3.4 1.9 ­0.4 7.2 3.0 6.0 3.1 3.6 Private 19.0 13.5 15.0 19.4 20.5 18.8 21.1 15.9 19.5 National investment 33.0 24.5 21.3 25.5 23.8 25.8 24.7 19.4 22.9 Public 5.2 4.5 4.0 4.5 7.4 4.3 8.1 4.7 6.4 Private 27.8 20.0 17.3 21.0 16.4 21.5 16.6 14.7 16.5 Savings­ investment, public ­2.3 ­1.1 ­2.1 ­4.9 ­0.2 ­1.3 ­2.1 ­1.5 ­2.9 Savings- investment, private ­8.8 ­6.5 ­2.3 ­1.7 4.1 ­2.7 4.5 1.2 3.1 Current account ­11.1 ­6.8 ­5.0 ­5.3 3.8 ­3.6 0.2 ­0.5 1.0 Total capital flows 13.0 10.0 7.8 6.8 0.9 4.2 1.3 1.9 0.4 Change in foreign exchange reserves 1.9 3.2 2.8 1.6 4.7 0.6 1.5 1.3 1.4 2006­08 National savings 19.3 15.7 18.3 21.7 29.8 18.5 28.6 19.1 25.3 Public 3.6 3.6 2.0 2.8 10.9 3.6 6.5 4.7 6.9 Private 15.6 12.2 16.2 18.9 18.9 15.0 22.0 14.4 18.4 National investment 36.2 28.4 26.6 26.0 29.4 24.9 26.0 22.7 25.9 Public 6.7 4.5 4.4 6.6 10.4 3.2 9.4 5.5 7.0 Private 29.4 23.9 22.1 19.4 19.0 21.7 16.6 17.2 18.9 Savings­ investment, public ­3.1 ­0.9 ­2.4 ­3.8 0.5 0.3 ­2.9 ­0.8 ­0.2 Savings- investment, private ­13.8 ­11.7 ­5.9 ­0.5 ­0.1 ­6.7 5.4 ­2.8 ­0.4 Current account ­17.0 ­10.3 ­7.5 ­5.7 0.4 ­7.1 0.1 ­1.7 ­1.8 Total capital flows 18.1 14.0 9.7 5.0 3.6 6.2 1.6 2.9 3.1 Change in foreign exchange reserves 1.0 3.7 2.2 ­0.8 4.0 ­0.9 1.7 1.2 1.3 Note: Latin America and the Caribbean excludes island economies and International Development Association­ eligible countries. Source: IMF World Economic Outlook, Eurostat databases, and authors' calculations. PRELUDE TO THE CRISIS 37 FIGURE 1.15 Current account, net of FDI, median per group, 2000­08 Percent of GDP 10 Group 5 5 Group 4 Group 3 Group 6 0 Group 2 ­5 ­10 Group 1 ­15 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. FIGURE 1.16 Fiscal balance, median per group, 2000­08 Percent of GDP 5 Group 5 Group 1 Group 2 0 Group 3 ­5 Group 6 Group 4 ­10 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. Reaping the benefits . . . Moving to foreign-owned banking systems made possible the attainment of macroeconomic stability. Indeed, the symbiotic link between governments and state-owned enterprises and newly privatized enterprises was a threat to sta- bility. Perhaps the most important impact of foreign bank entry was severing relationships between politically connected enterprises and the banking system. 38 TURMOIL AT TWENTY FIGURE 1.17 External debt to GDP, median per group, 2000­08 Percent of GDP 120 Group 1 100 Group 2 80 60 Group 3 Group 4 40 Group 5 Group 6 20 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. FIGURE 1.18 Short-term debt to foreign exchange reserves, median per group, 2000­08 Percent 400 Group 1 300 Group 5 200 Group 2 100 Group 4 Group 3 Group 6 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: IMF World Economic Outlook and authors' calculations. Failure to do so had resulted in quasi-fiscal bailouts of incumbent state-owned enterprises and systemic banking crises during the 1990s, with banking distress reaching its peak in the mid-1990s (figure 1.20). Moreover, many foreign banks brought in long-term strategic goals and stabilized the banking systems. Econometric evidence shows that firms with ownership links to domestic banks or the government grow less, while stronger and younger firms benefit from the PRELUDE TO THE CRISIS 39 FIGURE 1.19 Bank ownership patterns in Europe and Central Asia transition economies, 1997 and 2005 1997 2005 Mixed Majority private domestic Majority Majority private foreign domestic Majority Majority foreign government Majority government Source: Barth, Caprio, and Levine 2008; EBRD (various years); and World Bank staff estimates. entry of foreign banks. While foreign banks initially focused on large and foreign- owned firms, improvements in the contractual and information framework in the host country made lending to small and medium firms more attractive as well.5 Foreign strategic investors were also crucial for the development of financial ser- vices: by bringing in new technology and skills from their home countries, foreign banks had an overall positive impact on their home countries and on their host countries' financial systems. More recently, foreign banks were instrumental in providing credit to households, both mortgages and consumer finance. Foreign banks strengthened ECA's banking sector. Restructuring and better risk management within the banking sector sharply reduced nonper- forming loans and maintained high capital adequacy ratios with significant changes between 1999 and 2007 (table 1.2). Provisions against loan losses gen- erally increased steadily until 2007. Profitability, measured by rates of return on equity and assets, was high till 2007 but declined in 2008 to ranges seen earlier in the decade, perhaps reflecting the gradual onset of the crisis. Even so, profitability appears high enough to accommodate the substantial addi- tional provisioning expected in 2009­10. Parent banks that entered Central and Eastern Europe have been more profitable than their domestic competitors at home. Indeed, banks with a presence in Central and Eastern Europe had higher rates of return-on-equity 5. De Haas and Naaborg 2005. 40 TURMOIL AT TWENTY FIGURE 1.20 Banking crises in transition economies, 1990­2002 Number of countries in crisis 20 Nonsystemic Systemic 15 10 5 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Note: Episodes of distress are classified as systemic if emergency measures were taken to assist the banking system (such as bank holidays, deposit freezes, guarantees to depositors or other creditors) or if large-scale nationalizations took place. Episodes were also classified as systemic if nonperforming assets reached 10 percent of total assets at the peak of the crisis or if the cost of rescue operations was more than 2 percent of GDP. Source: Honohan and Laeven 2005. over 2004­08 than banks that have only market exposure in their home coun- try (figures 1.21 and 1.22).6 Although the return on equity declines over the period in both groups, it does so more in the latter. Similar results hold for the rate of return-on-assets. Part of the reason for the higher rates of return for the internationally diversified banks is that markets perceived these as riskier, as reflected in more volatile stock prices than their domestic counterparts.7 . . . while taking on risks There was also a rapid increase across most groups in private sector credit-to- GDP ratios and in loan-to-deposit ratios (figures 1.23 and 1.24). Although the increase was most marked in Group 1 countries, the increase in other groups, such as Group 2 and 3, is also telling. But credit expansion in excess of the growth of deposits increased the reliance of banks on external borrowing and exposed countries to the risk of sudden stops and reversals in capital flows. Loan-to-deposit ratios increased across all groups, showing that credit to the 6. The estimates refer to the rates of return for a group of banks with exposures in the ECA region and a group of other Western European banks. The rate of return for each bank is weighted by its share of equity or assets in the group. 7. The comparison is between the beta coefficients of a bank with exposure in Central and Eastern Europe with that of its domestic counterpart; those for the former are significantly higher than those for the latter. PRELUDE TO THE CRISIS 41 FIGURE 1.21 Average return on equity, parent banks and competition, 2004­08 Percent 20 15 10 Parent banks 5 Competition 0 2004 2005 2006 2007 2008 Source: Andreeva and Branda 2009. FIGURE 1.22 Average return on assets, parent banks and competition, 2004­08 Percent 0.8 0.6 0.4 Parent banks 0.2 Competition 0.0 2004 2005 2006 2007 2008 Source: Andreeva and Branda 2009. private sector was not fi nanced through rising deposits (except for Group 4 where the increase was more modest). By late 2008, foreign currency loans constituted the majority of bank loans in many new member states of the European Union--over 80 percent in Esto- nia and Latvia and around 60 percent in Bulgaria, Hungary, Lithuania, and Romania. But there are important exceptions to the generally high share of foreign currency-denominated lending. It stood at 10 percent in the Czech Republic, 20 percent in the Slovak Republic, and 25 percent in Poland, also the countries with the lowest loan-to-deposit ratios.8 8. As will be seen in chapter 2, these are also the countries where the share of international claims in total lending by parent banks and their subsidiaries are the lowest; in other words, resident deposit- taking was the primary source of funding. 42 TURMOIL AT TWENTY TABLE 1.2 Evolution of banking sector indicators, by country, 1999­2008 Loan-loss provisions Capital adequacy (ratio of bank provisions Nonperforming loans (risk-weighted capital- for loan losses to Return on equity Return on assets (percent of total loans) asset ratio) nonperforming loans) (percent) (percent) Countries 1999 2004 2007 2008 1999 2004 2007 2008 2000 2004 2007 2008 1999 2004 2007 2008 1999 2004 2007 2008 Bosnia and Herzegovina 9.9a 3.5 3.0 3.1 26.3a 18.0 17.1 16.3 64.2 96.1 .. 37.8 ­5.8a 5.6 8.9 4.3 ­1.3a 0.6 0.9 0.4 Bulgaria 29.0 7.1 1.7 2.2 43.0 16.6 13.8 14.9 65.9 49.0 46.6 7.9 20.9 20.0 23.8 20.5 2.5 2.1 2.4 2.1 Croatia 11.8 4.5 3.1 3.2 20.6 14.1 16.4 14.2 79.9 60.3 51.5 48.4 4.8 16.6 10.9 10.1 0.7 1.4 1.6 1.6 Czech Rep. 22.0 4.1 2.7 3.2 13.6 12.6 11.5 12.3 46.8 69.4 56.4 56.9 ­4.3 23.4 24.5 21.7 ­0.3 1.3 1.3 1.2 Estonia 1.7 0.3 0.5 1.9 16.1 13.4 14.8 18.8 .. .. .. .. 9.2 13.8 30.2 13.8 1.4 1.6 2.6 1.2 Hungary 3.6 2.7 2.6 2.6 14.9 11.2 11.0 11.1 57.0 51.1 58.1 80.3 7.1 25.2 18.6 15.1 0.6 2.0 1.7 1.3 Kazakhstan .. .. .. .. .. .. 14.2 14.9 .. .. .. .. .. .. 18.4 9.7 .. .. 2.3 1.2 Latvia 6.0 1.1 0.4 2.2 16.0 11.7 11.1 11.8 74.1 99.1 129.8 92.6 11.2 21.4 24.2 4.6 1.0 1.7 2.0 0.3 Lithuania 12.5 2.3 0.8 1.1 17.4 12.3 10.9 12.9 34.6 21.6 73.0 .. 1.1 13.4 26.0 16.1 0.5 1.3 1.7 1.2 Macedonia, FYR 41.3 13.2 7.5 6.9 28.7 23.0 17.0 15.6 .. 76.2 94.1 .. 3.5 6.2 15.2 19.1 0.8 1.1 1.8 2.2 Poland 13.2 15.5 5.3 5.0 13.2 15.6 12.0 10.7 40.5 58.0 .. .. 12.9 17.6 24.8 20.6 0.9 1.4 1.9 1.5 Romania 35.4 8.1 9.7 9.8 17.9 18.8 12.7 11.9 .. 34.3 25.7 27.8 ­15.3 19.3 11.5 15.9 ­1.5 2.5 1.3 1.4 Russian Federation .. .. .. .. .. .. 15.5 14.5 .. .. .. .. .. .. 22.7 12.1 .. .. 3.0 1.6 Serbia 21.6b 22.8 3.8 5.3 25.6b 27.9 27.9 21.9 .. .. .. .. ­60.6b ­5.3 10.2 10.6 ­8.4b ­1.2 1.7 2.1 Slovak Rep. 23.7 5.4 2.8 3.2 29.5 19.0 13.2 11.1 75.1 89.1 93.3 91.4 ­36.5 11.9 19.6 14.1 ­2.3 1.0 1.1 1.0 PRELUDE TO THE CRISIS Slovenia 5.2 5.7 1.8 1.6 14.0 11.0 11.2 11.2 45.3 34.0 .. .. 7.8 14.2 16.3 8.1 0.8 1.1 1.4 0.7 Turkey 10.5 6.0 3.5 3.2 8.2 28.8 19.0 17.7 59.8 88.1 86.8 81.1 33.1 17.4 21.3 18.8 3.3 2.5 2.7 2.2 Ukraine .. 3.2 1.3 14.5 .. .. 13.9 13.6 .. .. 26.3 26.0 .. .. 12.7 10.9 .. 1.1 1.5 1.3 Average 16.5 6.8 3.2 4.3 20.3 16.9 14.6 14.2 58.5 63.6 67.4 55.0 ­0.7 14.7 18.9 13.7 ­0.1 1.3 1.8 1.4 Austria 1.7 1.5 1.7 .. 13.9 14.7 11.6 10.8 .. .. .. .. 6.9 9.3 8.2 .. 0.3 1.5 0.5 .. .. is not available. a. Data are for end-2000. b. Data are for end-2002. 43 Source: Central banks (Annual Reports or Financial Stability Reports), IMF World Economic Outlook, and authors' calculations. FIGURE 1.23 Private sector credit to GDP, median value, 2000­08 Ratio 1.25 East Asia 1.00 Group 1 0.75 Group 2 Group 4 0.50 Group 3 Group 5 0.25 Group 6 0.00 2000 2001 2002 2003 2004 2005 2006 2007 2008 Note: East Asia denotes the median values for Indonesia, Korea, Malaysia, and Thailand for 1990­ 98; this data is superimposed on 2000­08 in the charts for the different groups of ECA countries. Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. FIGURE 1.24 Loans to deposits, median value, 2000­08 Ratio 2.5 Group 1 2.0 Group 2 East Asia 1.5 Group 5 1.0 Group 3 Group 4 0.5 Group 6 0.0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Note: East Asia denotes the median values for Indonesia, Korea, Malaysia, and Thailand for 1990­ 98; this data is superimposed on 2000­08 in the charts for the different groups of ECA countries. Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. Together the charts have two implications. First, there has been increasing reliance on direct borrowing from parent banks and on wholesale funding abroad. It is also worth noting that the increase among the countries of East Asia hit by the emerging market crisis of 1997­98 was not as pronounced, in line with the more limited role of parent banks in the buildup of vulnerabili- ties. Second, while deposits are typically the cheapest source of funding, par- ent bank expansion strategies relied heavily on foreign financing sources that were on-lent in foreign exchange to unhedged borrowers. 44 TURMOIL AT TWENTY Foreign exchange liabilities and leverage (liabilities-to-equity) ratios of banks also rose among financially integrated countries. The foreign exchange exposure ratios show that foreign exchange liabilities increased across all groups, in some cases sharply, signaling growing vulnerability to a reversal in market sentiment (figure 1.25). And while leverage ratios are not high when compared with advanced economies, they increased steadily and rapidly over the past few years in the three groups (Groups 1 through 3) facing more sources of vulnerability (figure 1.26). Financial deepening or irrational exuberance? There is no doubt that credit to the private sector expanded very rapidly in the ECA countries, particularly since 2005. The growth of private sector credit has however been decelerating in the Baltic States since 2007 and, for the most part, elsewhere since mid-2008. But it is less clear how much of the acceleration reflects catch-up due to low initial indebtedness characteristic of late financial deepening and how much reflects excess growth. In particular, it is worth ask- ing if credit growth was excessive in the years leading up to the crisis. To answer that question, figure 1.27 defines excess credit growth relative to the average experience of all developing countries during 2005­08. The top two panels of figure 1.27 describe how each arrow in these charts should be interpreted. Where each arrow begins depicts private sector credit as a share of GDP in 2000 and the annual real private sector credit growth in 2001­04. The arrows end at the level of private sector credit as a share of GDP in 2004 and the annual real private sector credit growth in 2005­08. Whether credit FIGURE 1.25 Foreign exchange assets to liabilities, median value, 2000­08 Ratio 2.5 Group 5 2.0 Group 6 1.5 Group 4 Group 2 Group 3 1.0 Group 1 0.5 East Asia 0.0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Note: East Asia denotes the median values for Indonesia, Korea, Malaysia, and Thailand for 1990­ 98; this data is superimposed on 2000­08 in the charts for the different groups of ECA countries. Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. PRELUDE TO THE CRISIS 45 FIGURE 1.26 Liabilities to equity, median value, 2000­08 Ratio 10 Group 1 East Asia 8 Group 2 6 Group 4 Group 3 4 Group 6 Group 5 2 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Note: East Asia denotes the median values for Indonesia, Korea, Malaysia, and Thailand for 1990­ 98; this data is superimposed on 2000­08 in the charts for the different groups of ECA countries. Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. growth was convergent or excessive depends on where the arrows end relative to the confidence interval. For countries whose arrows begin below the regres- sion line, three relevant outcomes are possible. First, if the arrow ends within the confidence interval, these countries are considered to be catching up (con- vergence). Second, if the arrow exceeds the upper bound confidence interval, this indicates "excess" credit growth. Third, if arrows go toward the regression line but do not reach the lower bound, countries are said to be catching up but at a slow rate. The same logic applies to countries whose arrows start above the regression line--though in this case, excess growth relates to arrows going away from the origin of these graphs. Financially integrated countries in ECA are represented in the bottom panel of figure 1.27. The upward sloping arrows that end above the upper bound con- fidence interval are interpreted as cases of excess credit growth. This includes Belarus, Lithuania, Romania, the Russian Federation, and Ukraine: for all five the second subperiod's growth was higher than the first subperiod's--in excess of 30 percent a year except in the Russian Federation, where it was somewhat lower. Estonia already had the financial depth characteristic of Central Europe in 2000 and had far overtaken the latter by 2004 as a result of high credit growth, which continued with only slight moderation in the second subperiod. Bulgaria and Kazakhstan, however, went through a slow convergence. The Czech and Slovak Republics and Turkey experienced financial disintermediation between 2000 and 2004, but credit growth was positive between 2005 and 2008. FYR Macedonia's growth picked up in the second subperiod--a borderline case. But the other transition countries were in line with the average developing country. 46 TURMOIL AT TWENTY FIGURE 1.27 Private sector credit developments in 2005­08: catch-up or excess? Initial level in 2000 Initial level in 2000 (growth 2001­04) (growth 2001­04) Exceeding the parallel line Exceeding the upper bound of the interval = EXCESS (positive divergence) = EXCESS (positive divergence) Within the range of confidence interval Above the upper bound, = CONVERGENCE but below the parallel line = SLOW CATCH-UP Below the lower bound, Below the parallel line but above the parallel line Below the lower bound Within the range of = DIVERGENCE = SLOW CATCH-UP = DIVERGENCE confidence interval (negative) (negative) = CONVERGENCE Real private sector credit growth (percent) (arrows begin on growth rate for 2001­04 and end on growth rate for 2005­08) 50 Ukraine Romania Regression and 95% confidence Kazakhstan interval of the relationship for all Belarus Lithuania developing countries (initial level Russian Fed. in 2004; growth 2005­08) Latvia 30 Macedonia, FYR Bulgaria Estonia Serbia Turkey Czech Rep. Hungary Bosnia & Herz. 10 Slovak Rep. Poland ­10 0 25 50 75 Initial private sector credit as a share of GDP (arrows begin on initial level in 2000 and end on initial level in 2004) Note: The horizontal axis depicts the initial level of private sector credit as a share of GDP in either 2000 or 2004 in all three charts. The vertical axis depicts annual percentage changes of real pri- vate sector credit for four-year periods beginning in 2001 or in 2005. The downward sloping line reflects a negative relationship for developing countries (including nontransition countries) between the initial level of private sector credit as a share of GDP in 2004 and real private sector credit growth during 2005­08; also shown are the 95 percent confidence intervals for the relationship. Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. In sum, the evidence suggests that financially integrated transition coun- tries that experienced excess credit growth were those that had come late to the transition. Thus, using the share of private credit in GDP as a measure of fi nancial deepening, it is clear that most countries of the former Soviet Union--Belarus, Kazakhstan, the Russian Federation, and Ukraine, as well as Lithuania and (to less extent) Latvia--having come later in the transition, were financially much shallower in 2000 than countries in Central Europe, such as Croatia and Serbia. So too were Bulgaria and particularly Romania, both of which had encountered severe macroeconomic difficulties in the mid-1990s and undergone a second transition recession before a durable recovery took hold. This relative lack of financial depth continued to be true for the share of private credit in GDP even in 2004 for Belarus, Kazakhstan, Romania, the Russian Federation, and Ukraine in relation to Central Europe and Croatia. PRELUDE TO THE CRISIS 47 Indeed, Kazakhstan, the Russian Federation, and Ukraine had similar levels of financial depth in 2004. But that was no longer the case for Bulgaria and Latvia where financial depth, starting at CIS levels, had respectively become comparable to and exceeded that of Central and Eastern Europe by 2004. Bringing finance home The domestic credit expansion of 2005­08 in much of ECA was driven by the household sector. As large corporates turned to cheaper cross-border lend- ing, the share of households in bank loans to households and nonfi nancial corporates rose sharply after 2003, and in 2008 stood at more than 50 percent in the Czech Republic, Estonia, Poland, and Romania; somewhat less than 50 percent in Hungary, Latvia, and the Slovak Republic; and below 40 percent in Bulgaria and Ukraine (table 1.3). Indeed, the increase in the ratio over five years was by far the highest in Kazakhstan and the Russian Federation, both of which were financially less developed and where the share of households in private sector loans was around 10 percent in 2003. It was the least in Estonia and the Czech Republic, where households already accounted for more than 40 percent of private sector loans in 2003. With real credit growth in Kazakhstan and the Russian Federation at 30 per- cent or more a year in 2005­08, there was a steep rise in household lending, albeit from a very low base (figure 1.27 and table 1.3). Among households, mort- gages increased in importance in many countries, particularly Croatia, Latvia, and Poland and, as a share of household lending, stood at around 80 percent in Estonia and Latvia, 70 percent in the Czech Republic and Lithuania, 50 per- cent in Poland and just above 40 percent in Bulgaria and Croatia and about 30 percent in Ukraine in 2008 (table 1.3). But the share of mortgages in house- hold lending between 2003 and 2008 stayed broadly unchanged in the Slovak Republic and fell sharply in Hungary and, to much less extent in Lithuania. In Hungary, the Slovak Republic, and Slovenia, adjustable interest rate mortgages accounted for more than three-quarters of mortgage debt in 2006. Household indebtedness in the new member states of the European Union is however much lower than in the EU15. Within the corporate sector, whose share was falling, lending to previously underserved small and medium enterprises increased faster in the Czech Republic, Hungary, and Poland--a positive development. Not surprisingly, the increase in private sector credit was accompanied by increases in asset prices. There were marked increases in real housing prices-- beyond those in the Western European benchmark (figure 1.28). This is true of the countries that had excess credit growth during 2005­08 for which data on house prices are also available: Bulgaria, Estonia, Latvia, Lithuania, the 48 TURMOIL AT TWENTY FIGURE 1.28 Real housing price developments, 2005 = 100 Bulgaria Croatia Czech Republic 180 180 180 140 140 140 100 100 100 60 60 60 20 20 20 2000 2004 2008 2000 2004 2008 2000 2004 2008 Estonia Hungary Latvia 180 180 180 140 140 140 100 100 100 60 60 60 20 20 20 2000 2004 2008 2000 2004 2008 2000 2004 2008 Lithuania Poland Russian Federation 180 180 180 140 140 140 100 100 100 60 60 60 20 20 20 2000 2004 2008 2000 2004 2008 2000 2004 2008 Serbia* Slovak Republic Ukraine 180 180 180 140 140 140 100 100 100 60 60 60 20 20 20 2000 2004 2008 2000 2004 2008 2000 2004 2008 Country Benchmark mean Benchmark median *For Serbia the data end in June 2008. Source: Global Property Guide based on data from statistical offices and real estate companies. Benchmark countries are France, Germany, Ireland, Spain, and the United Kingdom. PRELUDE TO THE CRISIS 49 TABLE 1.3 Growth and composition of credit to the private sector (percent) Ratios of lending Average Average to households to Share of housing growth of growth of lending to households loans in total credit to credit to and corporations household lending households corporations Country 2003­08 2003­08 2003 2008 2003 2008 Bulgaria 41 57 27 37 .. 43 Croatia 15 11 53 57 31 43 Czech Rep. 26 12 41 56 65 70 Estonia 39 32 45 51 78 81 Hungary 21 7 35 49 64 51 Kazakhstan 71 34 12 30 .. .. Latvia 44 28 33 47 64 79 Lithuania 59 31 22 43 76 69 Macedonia, FYR 44 20 22 40 .. .. Poland 28 13 50 65 30 52 Romania .. 51 .. 21 Russian Federation 59 27 10 25 .. 28 Slovak Rep. 28 10 28 46 69 68 Turkey 45 24 27 44 27 33 Ukraine 84 47 25 39 25 31 .. is not available. Source: IMF International Financial Statistics, central banks, and Bank staff estimates. Russian Federation, and Ukraine. But the timing of the peaks differs--house prices started falling in 2007 in Estonia and Latvia, but only in 2008 in Lithu- ania. Here too, however, there is no tight association between credit growth and housing price increases, since there were rapid increases in the latter in Poland and the Slovak Republic, both countries with convergent credit growth.9 Financial integration clearly brought benefits to borrowers and lend- ers alike. Credit became widely available to countries that had rudimentary financial sectors at the beginning of their transition. Established corporates could work with cross-border banks directly. Countries with majority for- eign-owned banks saw credit extended to younger and less established firms. Households could borrow to finance housing and other durables, driving the 9. There are also significant structural breaks during the period, including in the role of private build- ers in a sector that was among the latest to be liberalized to market forces (both in prices and in the public sector as a supplier of housing). 50 TURMOIL AT TWENTY credit boom in many countries. On the supply side, as already noted, Western European banks with subsidiaries in CEE earned higher returns on equity and assets than those serving only home countries (see figures 1.21 and 1.22). In a broader institutional context, modernized lending practices and strengthened risk assessment procedures led to lower nonperforming loans and higher loan loss provisions as well as high capital adequacy ratios until 2007. But borrowing in foreign currency posed macroeconomic risks for coun- tries that did not belong to a reserve currency area. The incentives to do so depended on whether the differentials between domestic and international lending rates were narrowing, reflecting market expectations of the credibility of disinflation efforts and the exchange rate regime. And disinflation was far from achieved in a number of countries: in 2007, Bulgaria, Estonia, and Hun- gary had high single-digit inflation, and Latvia had double-digit inflation. In 2008, Croatia and Romania had high single-digit inflation and Bulgaria, Esto- nia, Latvia, Lithuania, and FYR Macedonia had double digit inflation. The interest rate spread in the Baltic states and Bulgaria, which have fi xed exchange rates, increased after 2006, while it increased from end-2007 in Cro- atia, Hungary, and Romania, where the first country had an exchange rate peg and the other two were facing fiscal difficulties. The expectation that these countries, either new EU member states or close to accession, would eventually adopt the euro, and that convergence implied trend real currency appreciation, made it extremely profitable to engage in borrowing and lend- ing in foreign currencies when the lending spread was high. Borrowers evi- dently perceived pegs to be credible in countries with fi xed exchange rates and expected nominal appreciation in countries with flexible exchange rates on account of an expected catch-up in living standards.10 Policymakers attempted to put "sand in the wheels" during the boom years with mixed success, illustrating the difficulty of restraining macroeconomic imbalances in small, open, and integrating economies. What needed to be done about the growth of credit, particularly the foreign exchange exposure, has been a subject of widespread debate since at least 2005, and many coun- tries in the region introduced prudential measures in an attempt to limit vul- nerabilities (box 1.2). By way of example, Estonia increased the risk weight for housing loans in calculating capital adequacy from 50 percent to 100 percent. Latvia relied on monetary reserves. Bulgaria and Romania introduced tighter loan-loss provisioning, more stringent rules for classifying claims and cal- 10. Th is appears still to be the case in the midst of the fi nancial crisis. Specifically, households in Hun- gary have largely not taken up the option--available after the onset of the crisis--of converting foreign currency mortgages into local currency. PRELUDE TO THE CRISIS 51 BOX 1.2 Playing cat and mouse--staying ahead of regulation arbitrage in Southeastern Europe High credit growth rates have been a major concern for countries in the region, in particular when accompanied by risk-taking that banking supervision authorities might regard as excessive, as when foreign exchange borrowing is undertaken by domestic entities that are not naturally hedged. But regulation is no easy task. As countries know only too well: regulate banks, and leasing activities will emerge, regulate "the system," and cross-border lending flour- ishes. The trick is to strike a balance and to keep adapting regulations as means to circumvent them are developed. But behavior to circumvent regulations occurs against a broader backdrop, where a country's macroeconomic policy mix and exogenous factors that affect global liquidity are at play. Despite these difficulties, many banking supervision authorities have actively introduced monetary and prudential regulations to slow, at least for a while, what is perceived as high risk or excessive lending. Croatia, FYR Macedonia, Montenegro, Romania, Serbia, and Slovenia have pursued countercyclical mon- etary and prudential policies to pour "sand in the wheels," and shared successes and failures. Examples of such monetary and prudential regulations include: · Increased risk weights for capital adequacy ratio calculations. When growth rates in certain business lines, such as credit card and overdraft credit lines, were considered too high, the corresponding risk weights were adjusted upward. For example, in FYR Macedonia the risk weights were increased from 100 to 125 percent for consumer lending. In Serbia the risk weights also varied between local and foreign currency loans. · Regulations on loan classification. Foreign currency loans are classified as higher risk; with exceptions allowed only with first class collateral, such as government bonds. The effectiveness of these exceptions depends on a well developed registry of collateralized assets, which cannot always be assumed. For instance, Latvia introduced such a registry only in early 2008, when banks realized for the fi rst time that some of the collateral they held had been pledged multiple times. · Indicative credit growth thresholds. Whenever credit growth rates exceeded a predefi ned threshold considered too high by the banking supervision authorities, a compulsory deposit equal to the excess above the threshold needed to be deposited with the central bank. FYR Macedonia introduced such a policy in the second quarter of 2008 for household credit, and the deposit was remunerated at a 1 percent rate. In Croatia, the deposit is remu- nerated at the same rate but in special central bank bills that are marketable. The regulation is intended not as a credit ceiling, but as a cost markup when- ever credit growth exceeds the specified threshold. In Croatia and Montene- gro, the ceiling applies to total credit as opposed to household credit alone. (continued) 52 TURMOIL AT TWENTY BOX 1.2 (CONTINUED) Playing cat and mouse--staying ahead of regulation arbitrage in Southeastern Europe In Croatia, the limits are also tied to the source of funding on which the banks depend, and in Montenegro, the ceiling growth rates are tied to the bank's own capital adequacy ratios. · Liquidity-asset ratios--typically distinguishing between loan currencies and maturities. In FYR Macedonia, banks are obliged to maintain floor thresh- olds for liquidity-asset ratios. The thresholds are applied to 30-day and 180- day maturities, distinguish between local currency and foreign exchange exposures, and allow only low-risk highly liquid assets to be used in the cal- culations. Banks that do not comply with the ratios are required to make up the shortfall in phases. Concentration in lending results in the imposition of additional liquidity requirements. · Minimum foreign exchange liquidity requirements. In Croatia the foreign exchange liquidity has a floor of 32 percent of the total to discourage foreign exchange liabilities as a source for funding credit operations. · Leverage ratio limits or special reserve requirements. Romania introduced limits on leverage ratios or special reserve requirements. · Varying reserve requirements. These have been set at different levels for local currency and foreign exchange deposits, aimed at increasing the interest rate and thus reducing credit demand. While supervision authorities develop rules to slow credit growth, banks have looked for ways to circumvent them as they are introduced. The typical credit growth control regulation is said to have a shelf life of one to two years; and even monetary regulation can be circumvented, as with capital controls more generally. One way to circumvent controls is through establishing leasing companies, which typically operate outside the reach of banking supervision authorities. When they are closely linked to bank activities, as in FYR Mace- donia, they can easily be brought within the purview of the regulatory authori- ties. But in the Baltic states, they have emerged as entirely separate entities that compete with banks for shorter maturity lending. In Latvia, leasing institutions make it more difficult for banks to restructure mortgage loans, as they attempt to coerce early repayment by threatening foreclosure. Since mortgage lending has a priority claim, this obliges banking institutions to foreclose on loans that they would otherwise have preferred to have restructured. culating capital, increased risk weights, and limits to loan-to-value and loan payments-to-net income ratios. In addition to such prudential measures, Cro- atia introduced direct limits to credit growth whenever a specified threshold was exceeded, marginal reserve requirements on banks' foreign borrowing (as Chile had in 1992), and specific additional capital requirements related to for- eign exchange exposure. PRELUDE TO THE CRISIS 53 Although it might be too early to judge their effectiveness, some of these measures had unintended consequences. For example, limits on credit growth in Croatia could be circumvented by corporates directly accessing cross-border banks or capital markets, and in the end the limits constrained lending to small and medium enterprises. The Chilean-type tax in Croatia made foreign borrowing very expensive for banks and created an incentive for banks to finance expansion through equity rather than debt. Even there, however, a better capitalized banking system could have been accomplished more directly. Would different macroeconomic policies have lessened vulnerability? The vulnerabilities of ECA countries cannot be delinked from their pre-cri- sis policy choices. As an example, exchange rate regimes were evenly distrib- uted between countries where their primary role was as a nominal anchor and those where external competitiveness was the main concern (figure 1.29). Not surprisingly, there is a concentration of less flexible regimes among countries that face more sources of vulnerability. And yet, the evidence on the role of exchange rate regimes is not conclusive. Indeed, it was only in early 2007 that overvaluation concerns emerged among small and open economies experi- encing externally financed consumption booms. These were rationalized as driven by catching-up effects and the role of financial integration in a context of low private indebtedness. But it is less clear that the buildup of vulnerabilities would have been avoided if countries had followed different policies. The magnitude of the capital inflows to these countries stretched the limits of the policy toolkits available to the authorities. More likely, the combination of initial conditions (degree of consumption catch-up effects, varying relative sizes of economies, progress in institutional reforms), exogenous factors (EU accession where applicable, high global liquidity), and policy choices made some countries more prone to balance of payments difficulties. So, even though differences in policy alone are unlikely to have averted the crisis, they might have placed countries in a better position to face the consequences when it struck. To that end, the differences in macroeconomic conditions across groups, as well as the policy reaction of the authorities to balance of payments pressure across countries, are compared. The analysis focuses on (1) exchange rates, as reflected in measures of exchange rate flexibility and volatility, (2) fiscal policy, as measured by changes in the fiscal position vis-ŕ-vis the cyclically- adjusted fiscal balance, and (3) monetary conditions, as measured by changes 54 TURMOIL AT TWENTY FIGURE 1.29 Exchange rate regimes 8 1995 Q4 7 2000 Q4 6 2008 Q1 5 4 3 2 1 0 Estonia Latvia Montenegro Bulgaria Lithuania Croatia Kazakhstan Ukraine Macedonia Poland Romania Serbia Czech Republic Slovak Republic Belarus Russia Hungary Turkey Group 1 Group 2 Group 3 Group 4 Group 5 Note: 1 no separate legal tender; 2 is a currency board; 3 conventional pegged arrangement; 3.5 conventional peg to a basket; 4 pegged exchange rate within horizontal bands; 5 equals crawling peg; 6 equals crawling band; 7 equals managed floating with no predetermined path; and 8 equals independently floating. Source: IMF's AREAER, Bubula and Ötker-Robe, and authors' calculations. in velocity (as a proxy for demand and supply factors) and the central bank' sterilization efforts (as a proxy for a key determinant of the money supply-- that is, central bank liquidity operations). The assessment distinguishes between: · The average macroeconomic conditions in each of the five groups of finan- cially integrated countries (figures 1.30­1.33) and across each of the policy instruments at the disposal of the authorities. · The policy reaction functions to balance of payments pressure, i.e., the sum of the current account balance and capital flows (official and private) (fig- ures 1.34­1.37)--and, in the case of the central bank's sterilization efforts, to changes in net foreign assets. Each policy reaction controls for other policies and initial conditions.11 11. The analysis of policy reaction functions is based on econometric estimations and includes a dummy on group membership that is interacted with balance of payments pressures (that is, the sum of total capital flows--private and official--and the current account balance) corresponding to the country. For the central bank's sterilization effort, the group dummy is interacted with changes in net foreign assets. Different sample periods (1999­2008 and 2005­08) are considered, as well as episodes with high global liquidity (2003­06) and country-specific periods when economic activity is above potential. The fi xed effect estimations include controls for other macroeconomic policies (lagged) as well as for country-specific initial conditions and economic characteristics, such as the output gap, inflationary pressures, external liquidity (short-term debt-to-reserves) and solvency (external debt-to- GDP) indicators, exchange rate regime (a dummy based on the IMF's AREAER), and time dummies. PRELUDE TO THE CRISIS 55 FIGURE 1.30 Average policy outcome/stance: exchange rate flexibility, by group and period Index (0, peg, to 1, floating) 0.75 1998­2008 High global liquidity High economic activity 2005­08 0.50 0.25 0.00 Group 1 Group 2 Group 3 Group 4 Group 5 Source: IMF World Economic Outlook, World Bank DEC, and authors' calculations. FIGURE 1.31 Average policy outcome/stance: fiscal policy, by group and period Change in the difference between the general government balance and the cyclically adjusted general government balance (percent of GDP) 0.4 1998­2008 High global liquidity High economic activity 2005­08 0.3 0.2 0.1 0.0 ­0.1 ­0.2 ­0.3 Group 1 Group 2 Group 3 Group 4 Group 5 Note: A positive value implies tight fiscal conditions; a negative value, loose fiscal conditions. Source: IMF World Economic Outlook, World Bank DEC, and authors' calculations. The analysis for each individual policy instrument as well as the full port- folio of macroeconomic policies is done only for the groups of financially inte- grated countries. As noted earlier, moving from Group 1 to 5 is associated with a decline in the number of sources of vulnerability. What matters in the end is whether 56 TURMOIL AT TWENTY FIGURE 1.32 Average policy outcome/stance: monetary conditions, by group and period Change in velocity 1998­2008 High global liquidity 2 High economic activity 2005­08 0 ­2 ­4 ­6 ­8 ­10 Group 1 Group 2 Group 3 Group 4 Group 5 Note: A positive value implies tight monetary conditions; a negative value, loose monetary conditions. Source: IMF World Economic Outlook, World Bank DEC, and authors' calculations. FIGURE 1.33 Average policy outcome/stance: nominal exchange rate volatility, by group and period Absolute value of change in exchange rate (percent) 0.5 1998­2008 High global liquidity High economic activity 2005­08 0.4 0.3 0.2 0.1 0.0 Group 1 Group 2 Group 3 Group 4 Group 5 Source: IMF World Economic Outlook, World Bank DEC, and authors' calculations. macroeconomic policies within each group were calibrated at the right level based on the external position affecting each group. Thus, the assessment by policy instrument is then complemented by a discussion covering the macro- economic policies as a whole: specifically, the links between economic perfor- mance, the real exchange rate, and macroeconomic imbalances. PRELUDE TO THE CRISIS 57 FIGURE 1.34 Policy response to balance-of-payment pressures: exchange rate flexibility, by group and period Effect on the exchange rate flexibility indexa 0.2 1998­2008 High global liquidity High economic activity 2005­08 0.1 0.0 ­0.1 ­0.2 Group 1 Group 2 Group 3 Group 4 Group 5 Note: The figure represents the point estimate derived from a fixed-effects ordinary least squares estimation that includes controls for lagged economic policies (fiscal and monetary policy, exchange rate flexibility, and sterilization) and for initial conditions (output gap, liquidity, solvency, and exchange rate regime) as well as dummy variables for time. a. A positive value implies more flexibility and a negative value implies less flexibility. Source: IMF World Economic Outlook, World Bank DEC, and authors' calculations. FIGURE 1.35 Policy response to balance-of-payment pressures: fiscal policy, by group and period Effect on the fiscal policy stancea 0.5 1998­2008 High global liquidity High economic activity 2005­08 0.4 0.3 0.2 0.1 0.0 ­0.1 Group 1 Group 2 Group 3 Group 4 Group 5 Note: The figure represents the point estimate derived from a fixed-effects ordinary least squares estimation that includes controls for lagged economic policies (fiscal and monetary policy, exchange rate flexibility, and sterilization) and for initial conditions (output gap, liquidity, solvency, and exchange rate regime) as well as dummy variables for time. a. A positive value implies fiscal tightening and a negative value implies loosening. Source: IMF World Economic Outlook, World Bank DEC, and authors' calculations. 58 TURMOIL AT TWENTY FIGURE 1.36 Policy response to balance-of-payment pressures: monetary policy conditions, by group and period Effect on the change in velocitya 3 1998­2008 High global liquidity High economic activity 2005­08 2 1 0 ­1 ­2 ­3 Group 1 Group 2 Group 3 Group 4 Group 5 Note: The figure represents the point estimate derived from a fixed-effects ordinary least squares estimation that includes controls for lagged economic policies (fiscal and monetary policy, exchange rate flexibility, and sterilization) and for initial conditions (output gap, liquidity, solvency, and exchange rate regime) as well as dummy variables for time. a. A positive value implies monetary tightening and a negative value implies loosening. Source: IMF World Economic Outlook, World Bank DEC, and authors' calculations. FIGURE 1.37 Policy response to changes in net foreign assets: sterilization of foreign exchange, by group and period Effect on change in net domestic assets of the central banka 0.5 0.0 ­0.5 ­1.0 ­1.5 1998­2008 High global liquidity High economic activity 2005­08 ­2.0 Group 1 Group 2 Group 3 Group 4 Group 5 Note: The figure represents the point estimate derived from a fixed-effects ordinary least squares estimation that includes controls for lagged economic policies (fiscal and monetary policy, exchange rate flexibility, and sterilization) and for initial conditions (output gap, liquidity, solvency, and exchange rate regime) as well as dummy variables for time. a. A value of zero implies no sterilization and a value of ­1 implies full sterilization. Source: IMF World Economic Outlook, World Bank DEC, and authors' calculations. PRELUDE TO THE CRISIS 59 Before discussing each policy instrument and their role in the buildup of vulnerabilities, it is useful to provide an example of how to interpret fig- ures 1.30­1.37. For example, Group 1 had loose fiscal policies in the period 2005­08 (figure 1.31). Yet, this group appears to have tightened fiscal policies- --though only marginally--when balance of payments pressures were positive (figure 1.35). In sum, while the policy reaction is appropriate, it appears that the overall fiscal policy stance was still too loose. It is also important to recog- nize that the discussion reflects average policy responses in a group but might mask intragroup differences. For instance, the two countries in Group 1 had different fiscal stances before the global financial crisis. Even so, it is useful to look at developments at the level of country groups to provide policy guidance for countries that might face similar sources of vulnerability in the future. Opening the toolkit Nominal exchange rate policy In general, countries in all five groups experienced lower exchange rate flex- ibility in 2005­08 than for the full period (figure 1.30).12 Countries also expe- rienced lower exchange rate flexibility during episodes of high global liquidity, 2003­06. This suggests that policymakers attempt to limit appreciation pres- sures originating outside the country, most likely to prevent losses in exter- nal competitiveness. Interestingly, countries appear more favorably disposed toward exchange rate flexibility when domestic activity picks up. In levels, this appears to mimic the exchange rate regime that dominates each group. In fact, the groups facing more sources of vulnerability have lower flexibility, and those with fewer sources of vulnerability have more volatile exchange rates. Still, volatility has declined overall. Country groups with fewer sources of vulnerability appear to have reduced exchange rate flexibility in response to balance of payments pressure (fig- ure 1.34). This was particularly true when global liquidity was high or when domestic economic activity was above potential (figure 1.33).13 Although there are differences across groups, they generally had lower exchange rate flexibil- 12. The index of exchange rate flexibility is defi ned as the absolute change in the exchange rate divided by this term plus the change--also absolute--in foreign exchange reserves. The index defi ned in this way is restricted to the closed interval [0, 1] with 0 a full peg and 1 a fully floating regime. Exchange rate volatility is calculated as the absolute value of the percentage change of the exchange rate over the previous 12 months (calculated using monthly nominal effective exchange rates). Volatility is defi ned as the 12-month average of the observations. 13. The results for Group 5 are probably dominated by the policy stance of the Russian Federation during the years of high oil prices. 60 TURMOIL AT TWENTY ity when global liquidity was high. Note, moreover, that Group 4 was disposed toward exchange rate flexibility when output was above potential. Fiscal policy stance The fiscal stance was in general looser in the more vulnerable country groups, especially during the last four-year period (figure 1.31).14 Fiscal policy appears to have run counter to what policy advice would usually recommend, with more vulnerable countries, some with pegged exchange rates, also following looser fis- cal policies in 2005­08. By contrast, countries appear to have tightened fiscal policies relative to the cycle during the high global liquidity period (2003­06) and when domestic economic activity was above potential. Such procyclical fis- cal behavior in the period preceding the global financial crisis suggests that more could have been done by country authorities to reduce or contain overheating. In contrast to the average fiscal stance, policy responses in the face of balance of payments pressure are more nuanced. More vulnerable countries tightened fiscal policies (figure 1.35), though this adjustment might not have been sufficient. But it is not clear whether fiscal policy alone could have brought the macroeconomic framework onto a sustainable path, given the magnitude of capital inflows and their role in exacerbating external imbalances. While fiscal policy might have lessened vulnerabilities, it is also possible that the necessary adjustment would have been too much to bear with only one policy tool. Group 1 did not tighten policies to any significant extent during 2005­08. And while Group 4 tightened fiscal policy when faced with balance of payments pressure across most subsam- ples, Groups 3 and 5 were looser even during high global liquidity periods. Monetary conditions and central bank sterilization efforts The underlying monetary conditions reflect the rapid monetization of the ECA region.15 Indeed, all country groups experienced a decline in velocity, reflecting an increase in money demand as financial deepening intensified (figure 1.32). The decline was least striking in Group 4, the group that also experienced declines in financial disintermediation during part of the decade. 14. Fiscal stance is defi ned as the change in the deviation between the fiscal balance and the cyclically adjusted fiscal balance (a measure of fiscal impulse). An increase represents a fiscal tightening. 15. Monetary policy is defi ned as the change in nominal GDP over the change in broad money: a measure of velocity. Sterilization is defi ned as the change in net domestic assets of the central bank. The econometric estimation links changes in central bank net foreign assets to this measure of net domestic assets; a zero implies no sterilization, a ­1 implies full sterilization. The coefficient can also be positive if, for example, liquidity exceeds even what would have been justified on account of the change in net foreign assets alone. Both net domestic assets and net foreign assets are scaled by the average of reserve money in period t and period t­1. PRELUDE TO THE CRISIS 61 But the differences between country groups across periods and events are quite revealing. Almost all groups experienced a loosening of monetary con- ditions during the period of high global liquidity, except again Group 4, which appears to have followed rather tight monetary policy during the past decade. While these developments might in part reflect increased money demand, they also suggest that more could have been done to slow down economic activity. A caveat is in order. Increases in interest rates as a result of tighter monetary conditions could also further stimulate capital inflows. The evidence is more varied when focusing on episodes of high domestic economic activity or on the most recent four-year period (2005­08). Groups 1 and 5 have tighter mon- etary conditions when economic activity picks up than at other times. Indeed, relative to other periods and events, this is the case across all groups--and it might mirror the choice of exchange rate regime that reduces the effectiveness of monetary policy. Interestingly, there seems to be more variety during the last four-year period. Group 1 appears to have experienced monetary condi- tions, which, while still loose, were tighter than those in Groups 2, 3, and 5. There is also a variety of monetary policy responses when balance of payments pressures are positive (figure 1.36). Specifically, while Groups 1 through 3 (and, to some extent, Group 5) had loose monetary policy responses, the opposite was the case in the most recent four-year period among countries in Group 4. Sterilization efforts were limited in the more vulnerable country groups (Groups 1 and 2) but more actively pursued in the less vulnerable country (Groups 3 and 5), although less so in Group 4. For the most part, Groups 1 and 2 appear to have allowed external financing and its impact on net foreign assets to be fully monetized, while Groups 3 through 5 and, to a lesser extent, Group 4 more actively sterilized the effects of the balance of payments on the mon- etary accounts. Although these efforts weakened across all groups in the most recent four-year period, they were important during periods of high domes- tic economic activity and, in some cases, also during the high global liquidity period. The limited sterilization effort might also reflect the unusually high global liquidity that characterized the period, including the effect of EU acces- sion on capital mobility in a number of these countries. But a more active ster- ilization effort might have only furthered stimulate capital inflows through the interest differentials this might have created relative to the EU15 countries. An overall policy assessment The real exchange rate can be used as a summary statistic to assess macro- economic developments. While the real exchange rate is outside the control of 62 TURMOIL AT TWENTY country authorities, it is affected by their policy choices and the response of the economy to those policies. Moreover, macroeconomic policies are not formu- lated by the authorities in a vacuum. Instead, they are informed by domestic and external developments: from a country's economic growth to the under- lying inflationary pressures and buildup of external vulnerabilities. Here too the real exchange rate serves as a summary of the effects of economic policies. Group 1 experienced the largest current account imbalances with the small- est real exchange rate appreciation while Group 4 in contrast had moderate current account imbalances with significant real appreciation (figures 1.38­ 1.41). Thus, a tighter set of macroeconomic policies would have been needed to contain the deterioration in the external accounts in Group 1. In fact, fis- cal policy could have played a greater role in containing the excessively rapid economic growth underpinned by large and ultimately unsustainable capital inflows. This also appears to be the case during the episodes of high global liquidity and high economic activity. Indeed, since the choice of exchange rate regime restricted the options for monetary policy--and particularly the effec- tiveness of sterilization efforts owing to their impact on interest rate differen- tials and, consequently, on capital inflows--fiscal policy should have played a greater role. In contrast, the appreciation pressures in Group 4 in all likeli- hood reflect productivity gains, and thus explain why external imbalances did not accumulate too rapidly or become a source of vulnerability. Countries in this group appear to have kept a balance of fiscal and monetary policies that enabled sustainable growth, as the Czech Republic shows (box 1.3). The countries in Group 3 were successful on average but experienced pol- icy slippages in recent years. Specifically, external imbalances accumulated in the last four-year period--the average for the period as a whole is much lower than during 2005­08 (figure 1.39). Moreover, appreciation pressures picked up (figure 1.40). While these pressures could reflect productivity gains, exter- nal imbalances cast some doubt on such an interpretation. Indeed, examin- ing each policy instrument points to the conclusion that fiscal and monetary conditions could have been tightened more during the high global liquid- ity period. Still, this group appears to have reached the era of large capital inflows with a lag, and this might have helped avoid an even more difficult set of risks. In conclusion, while the process of financial integration against the back- drop of high global liquidity reinforced the upswing features of the business cycle, not all countries exposed themselves equally to the risks of a change in market sentiment. Domestic macroeconomic policies were tightened in some countries but, quite remarkably, were loosened further in others during the PRELUDE TO THE CRISIS 63 FIGURE 1.38 Median real GDP growth, by group and period Percent 1998­2008 High global liquidity 12 High economic activity 2005­08 10 8 6 4 2 0 Group 1 Group 2 Group 3 Group 4 Group 5 Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. FIGURE 1.39 Median current account balance, by group and period Percent of GDP 1998­2008 High global liquidity 10 High economic activity 2005­08 0 ­10 ­20 Group 1 Group 2 Group 3 Group 4 Group 5 Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. 64 TURMOIL AT TWENTY FIGURE 1.40 Change in real effective exchange rate, by group and period Percent 8 1998­2008 High global liquidity High economic activity 2005­08 6 4 2 0 ­2 ­4 ­6 Group 1 Group 2 Group 3 Group 4 Group 5 Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. FIGURE 1.41 Policy response to balance-of-payment pressures: real effective exchange rate, by group and period Effect on the change in the real exchange ratea 2 1998­2008 High global liquidity 12.79 High economic activity 2005­08 1 0 ­1 ­2 ­3 ­4 Group 1 Group 2 Group 3 Group 4 Group 5 Note: The figure represents the point estimate derived from a fixed-effects ordinary least squares estimation that includes controls for lagged economic policies (fiscal and monetary policy, exchange rate flexibility, and sterilization) and for initial conditions (output gap, liquidity, solvency, and exchange rate regime) as well as dummy variables for time. a. A positive value implies appreciation pressures and a negative value implies depreciation pressures. Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. PRELUDE TO THE CRISIS 65 BOX 1.3 Why foreign currency lending did not take off in the Czech Republic With a share of industrial production in GDP of about 30 percent--the larg- est in the European Union--the Czech economy has a large export-oriented manufacturing sector. Despite strong lobbying, however, the Czech National Bank (CNB) did not manipulate the exchange rate in favor of export indus- tries, emphasizing instead that a small economy and its businesses and labor unions needed to learn how to deal with external shocks. Further, the sustained appreciation of the koruna-euro exchange rate, due to a favorable productiv- ity growth differential vis-ŕ-vis the eurozone, helped contain overly optimistic expectations and an excessive credit boom in the corporate sector. The credibility earned by an independent Czech National Bank as a result of implementing conservative monetary policy within an inflation target- ing framework since 1998, together with substantial fiscal consolidation since 2004, led to a negative interest rate differential between the koruna and euro interest rates since early 2005 (box figure 1). Low interest rates, far from producing an undesirable credit boom, offered some advantages. The koruna, as a relatively low yielding currency, was not attractive for speculators. Its lower yield also removed any incentive for Czech households to borrow in foreign currency and take foreign exchange risk. Th e relatively low foreign currency denomination of the domestic deposit base permitted the economy to reap the benefits of currency depre- ciation without experiencing major balance sheet effects. Notwithstanding the absence of foreign currency borrowing by Czech households, mortgage lending grew substantially, with fi nancial penetration in this segment reach- ing significantly higher levels than in other EU10 countries except the Slovak Republic, and indeed comparable to Belgium, Denmark or Luxemburg. This BOX FIGURE 1 Koruna and euro interest rate, 2001­09 Percent 6 Three-month Prague Interbank Offered Rate 5 4 3 2 1 Three-month Euro Interbank Offered Rate 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 Source: Datastream. (continued) 66 TURMOIL AT TWENTY BOX 1.3 (CONTINUED) Why foreign currency lending did not take off in the Czech Republic produced housing bubbles, especially in large cities like Prague. Although this could lead to a significant deterioration of credit quality in the down- turn, it is worth noting that lending practices in the past were conserva- tive with respect to the average loan-to-value and the debt service-to-income ratios. Supported by low koruna interest rates and excess liquidity in the financial system, the CNB has withdrawn liquidity from the market. In sum, a combina- tion of confidence in the domestic currency due to credible disinflation, deep fi nancial penetration, and more patient households, as captured by modest growth of private consumption below most of its EU10 peers, helped establish a resident deposit base.1 The loan-to-deposit ratio stood at 78 percent by mid- 2008, the lowest in the EU10 (box figure 2). Moreover, the banking system, which had negligible exposure to toxic assets, was highly profitable. By April 2008, the return on equity stood at about 23 percent, the third highest after Poland and Bulgaria (25 percent). This helped sustain the modest external funding made available by parent banks to their Czech subsidiaries. The institutional arrangements governing financial regulation and supervi- sion and their links to monetary policy proved important in coping with the global fi nancial crisis. In 2005, the Czech Republic undertook a functional reorganization of fi nancial supervision, resulting in consolidated supervision of all segments of the fi nancial system.2 The capacity for a rapid supervisory response has been boosted by increased monitoring efforts, collection of daily information, and frequent communication between the CNB's senior manage- ment and the bank association.3 BOX FIGURE 2 Ratio of loans to deposits, by country, December 2008 Percent 300 200 100 0 p. p. y ine nia ia y n ia ia nd ria via tia ke ar tio an n en Re Re lga oa la t ra ng to ma r La ra Tu ov hu Po Uk Es Cr ak h Bu Hu de Ro ec Sl Lit ov Fe Cz Sl ian ss Ru Source: IMF International Financial Statistics and authors' calculations. (continued) PRELUDE TO THE CRISIS 67 BOX 1.3 (CONTINUED) Why foreign currency lending did not take off in the Czech Republic The efficiency of supervisory action was tested in the Société Générale scan- dal. The independence of bank subsidiaries also protected their financial stand- ing when problems arose within fi nancial groups such as Komercni Banka (Société Générale) or AIG Czech Republic. In February 2008, the mandate of the CNB was expanded to include con- sumer protection in fi nancial services, which effectively increased the powers of the supervisor in business conduct enforcement. This enhanced the CNB's fi nancial prudential supervision by enabling it to curb the accumulation of indirect credit risk in the household sector. And living through a crisis in 1997­ 98 made the Czech Republic more agile in undertaking reforms to mitigate financial crises. In sum, a combination of credible disinflation led by an independent central bank and fiscal discipline, underpinned by integrated supervision, shaped by an earlier crisis, can lower spreads in lending interest rates between domestic and foreign currency and make foreign currency borrowing unattractive. Notes 1. Of adults, 80 percent have a bank account in Croatia and the Slovak Republic and 97 percent in Slovenia. 2. Nevertheless, there are still some fi nancial sector institutions such as leasing compa- nies or nonbank credit institutions, that the Czech National Bank does not supervise, leaving the door open to some regulatory arbitrage. 3. The decision to collect daily data was implemented on the next day, and the col- lected data include cash requirements, money market and government bond market, fi nancial institutions (banks and other important fi nancial players), liquidity stance, exposures within group, deposit levels, and fulfi llment of regulatory limits. Source: Based on background work by Martin Melecky. period preceding the global crisis, which could have exacerbated their vul- nerabilities. But private sector imbalances also accumulated owing to other factors, such as integration into European production and fi nancing struc- tures as well as differences in initial conditions, such as lagging consumption of housing and low levels of private debt. Thus, while differences in policy alone would not have eliminated the emerging vulnerabilities given the mag- nitude of capital inflows during this period, better policies might have made the impact of the global financial crisis less severe. 68 TURMOIL AT TWENTY Annex 1.1 Separating wheat from chaff--evidence of market differentiation from EMBI spreads Financial markets appear to be differentiating countries, particularly after the collapse of Lehman Brothers. It suffices to compare the variety of changes in spreads across emerging market economies--country spreads have increased sharply, but some have done so more than others. A formal assessment requires decomposing spreads into global, group-specific, and country-spe- cific factors. A global factor model is well suited for this purpose; such models have looked at co-movements in output16 and stock returns.17 The global factor model is estimated using cross-sectional and constrained ordinary least squares, with daily data for forty EMEs for the period February 2008 to April 2009. Spreads for each period are regressed on a constant term (or global component) and a set of group-specific regional dummies; these two regressors are the common factors determining spreads. The residuals in the estimation are considered to be idiosyncratic and country-specific. The goal is to identify differences (if any) across countries. Note that country fun- damentals need not change for markets to differentiate. It suffices for them to be perceived differently after an event such as Lehman's collapse. Indeed, reduced appetite for risk could lead to a reassessment of pre-exist- ing country weaknesses, such as high current account imbalances, high pri- vate sector credit growth, excessive dependence on external fi nancing, and balance sheet factors--from solvency concerns (debt-to-GDP ratios) to indi- cators of foreign exchange liquidity (short-term debt-to-reserves). In other words, the risks were present, but investors are only now starting to charge a premium for "bad" characteristics. But investors might also be incorporat- ing in their pricing the real effects of the crisis. For instance, product and trade integration with Western Europe among new EU members might have led to a re-assessment of risk on account of the effects of the crisis on the real economy. What is the evidence? There is both an increase in the volatility of country- specific spreads among new EU member states and a change in the risk-order- ing of these country-specific spreads (annex figure 1.1.1). 16. Stockman 1998. 17. Brooks and Catăo 2000. PRELUDE TO THE CRISIS 69 ANNEX FIGURE 1.1.1 Country-specific components of EMBI spreads Basis points Basis points 600 3,000 400 2,000 200 1,000 0 0 ­200 ­400 ­1,000 Feb. 1, July 1, Dec. 1, Apr. 9, Feb. 1, July 1, Dec. 1, Apr. 9, 2008 2008 2008 2009 2008 2008 2008 2009 Bulgaria Czech Rep. Estonia Croatia Kazakhstan Hungary Latvia Lithuania Russian Federation Serbia Poland Romania Slovak Rep. Turkey Ukraine Source: Bloomberg and authors' calculations. Annex 1.2 Finance in transition Financial systems have come a long way since the start of the transition.18 Their deepening (annex figure 1.2.1) is captured by a steady increase in stan- dard measures of financial development, viz., liquid liabilities to GDP and pri- vate credit to GDP. Growing confidence in the financial system is illustrated by the lower ratio of base money outside the banking system. Intermediation efficiency has improved, with the loan-to-deposit ratio exceeding unity in 2006, as have traditional measures of efficiency, with net interest margins and overhead costs in 2007 falling to half their values of 1995 (annex figure 1.2.2). Financial systems however are smaller than that predicted by GDP per cap- ita, although the gap has become narrower between 1995 and 2007 (annex fig- ure 1.2.3). This is the case for such EU member states as Poland, the Slovak Republic, and Slovenia--where the ratio is less than 40 percent and credit to the private sector expanded less rapidly than in some other new member states-- and for such lower middle-income countries as Armenia and Georgia. While the new member states of the EU and Croatia have overhead costs to total assets similar to those of the EU15, their net interest margins are somewhat higher. 18. Th is annex is based on Beck 2009. 70 TURMOIL AT TWENTY ANNEX FIGURE 1.2.1 Financial deepening in ECA's transition economies, median values, 1995­2007 Percent 50 Ratio of currency outside banking system to base money 40 Ratio of liquid liabilities to GDP 30 20 Ratio of private credit to GDP 10 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Note: Private credit to GDP equals claims on nonfinancial private sector by deposit money banks and other financial institutions relative to GDP. Liquid Liabilities to GDP equals liquid liabilities of the financial system (currency plus demand and interest-bearing liabilities of banks and nonbank financial intermediaries) divided by GDP. Currency outside banking system equals currency outside banking system relative to base money. Source: Beck and Demirguc-Kunt 2009. ANNEX FIGURE 1.2.2 Bank efficiency in ECA's transition economies, median values, 1995­2007 Percent 0.08 0.06 Net interest margin 0.04 Overhead costs 0.02 0.00 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Note: Net interest margin is net interest revenue relative to total earning assets averaged over all banks for each country. Overhead costs are total operating costs relative to total assets averaged over all available bank data. Source: Beck and Demirguc-Kunt 2009. PRELUDE TO THE CRISIS 71 ANNEX FIGURE 1.2.3 Private credit to GDP versus GDP per capita, ECA's transition economies versus other regions, 1995 and 2007 Percent of GDP Percent of GDP 2 Europe and Central Asia 2 Europe and Central Asia All other regions All other regions 1 1 0 0 4 5 6 7 8 9 10 4 6 8 10 12 GDP per capita, in logs GDP per capita, in logs Source: Beck and Demirguc-Kunt 2009. ANNEX TABLE 1.2.1 Comparing ECA's transition economies with other regions (group medians) (percent) Stock GDP Liquid Private Bank Loan- market Stock Net per liabilities credit deposits deposit capitaliza- market Overhead interest capita to GDP to GDP to GDP ratio tion to GDP turnover costs margin EU10 + Croatia 5,935 49.4 52.0 42.8 107.3 32.4 15.4 2.5 3.6 Turkey 5,045 40.4 26.0 37.6 72.1 34.3 93.0 4.1 5.7 EU15 26,511 97.5 115.7 93.1 149.7 88.0 107.4 2.4 2.3 Southern EU Accession 1989 10,952 68.2 49.7 59.3 84.0 15.4 21.5 3.0 4.9 Western Balkans 2,037 51.5 32.7 40.3 98.1 42.0 9.0 3.7 3.9 Middle-income 2,246 49.0 36.1 42.3 80.3 41.9 14.1 3.7 5.6 Large CIS 2,284 35.2 39.0 27.1 134.7 55.0 15.6 4.8 5.8 Middle-income 2,246 49.0 36.1 42.3 80.3 41.9 14.1 3.7 5.6 Smaller CIS 1,218 19.8 12.3 10.5 129.5 3.0 5.1 4.9 6.3 Low-income and lower middle-income 1,435 43.0 35.5 37.6 80.4 40.7 17.0 3.5 5.4 Note: Data are for 2007, unless otherwise specified. Source: Beck and Demirguc-Kunt 2009. 72 TURMOIL AT TWENTY A comparison of the EU10 plus Croatia to Greece, Portugal and Spain in 1989--that is, a few years after those countries joined the European Union-- shows that the former group, despite having lower per capita incomes than the countries that joined the EU during its southern enlargement, enjoy fi nan- cial development comparable to those of the latter in 1989 (annex table 1.2.1). Countries of the Western Balkans have financial markets whose liquidity and efficiency is comparable to those of countries at similar levels of per capita income. Overhead costs are similar, whereas net interest margins are lower. The middle-income CIS has financial indicators similar to those of countries at similar per capita incomes, with the exception of the share of liquid liabili- ties in GDP, but overhead costs and net interest margins are higher. Among all country groups, Kazakhstan, the Russian Federation and Ukraine have rela- tively large stock markets compared with their banking systems. A striking exception to this pattern of broad comparability is provided by the low-income and lower middle-income CIS countries where both the liquidity and efficiency of financial markets are much lower than in countries at similar per capita incomes. The modest financial development of the low- income and lower middle-income CIS is due in part to bank restructuring having been postponed to the second decade of transition. Furthermore, small firms rely less on external finance for new investment in the Western Balkans than in the EU10 + 1 countries, and indeed much less than in the middle-income CIS, falling to low levels in the low income CIS. But the ratio in Armenia and the Kyrgyz Republic and, to less extent, Mol- dova is comparable to or higher than in the middle-income CIS and some of the Western Balkans. This is consistent with evidence from the three rounds of the Business Environment and Enterprise Performance Surveys (BEEPS) in 1999, 2002, and 2005. Firms in transition economies rely more on retained earnings than external finance for new investment, but this reflects a move away from informal sources of fi nance rather than a decline in the institu- tions of formal finance, with the countries that joined the European Union in 2004 (the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, the Slovak Republic, and Slovenia) relying almost as much on external finance as the cohesion countries: Greece, Ireland, Portugal, and Spain.19 19. Mitra 2008. PRELUDE TO THE CRISIS 73 CHAPTER 2 How much adjustment? How much financing? The more vulnerable among the financially integrated ECA countries face two big risks: that maturing external debt might not be rolled over, and that new money to finance large current account deficits might not be available. The global recession has also reduced exports and remittances for most ECA countries, with remittances being particularly important for some low income countries. Questions · What is the role of official fi nancing in country adjustment to the global financial and economic crisis? · Has the international response to the crisis in the fi nancially integrated countries been adequate, and does it need modification? · Does the crisis in the poorer CIS countries, affected mainly by the collapse in export demand and the decline in remittances, call for a stronger and sustained international policy response? · What is the exposure of parent banks to the fi nancially integrated ECA countries? And what are the determinants of rollovers by creditors? Findings · International collective action has so far managed to keep an orderly draw- down of excessive exposures. · For the fi nancially integrated countries, more official fi nancing is likely to be needed, particularly because it has a short repayment horizon, and the global recovery is expected to be weak and prolonged. Sustained official financing will be needed for the poorer CIS countries. · Parent-bank exposure to the financially integrated ECA countries, while large, varies significantly across countries. It is much less than is frequently cited. · Rollover risks depend on the liability and asset structures of banks in the region. Rollover rates for wholesale funding are low, but parent-bank fund- ing of their ECA subsidiaries--though declining--has so far been a stabiliz- ing force. In its quest for convergence toward Western European living standards, the stereotypical ECA country relied on debt-creating external financing. Indeed, in some cases, external imbalances as a percentage of GDP rose into double 75 digits. Not surprisingly, when the global crisis erupted in advanced country financial markets, the risk that many creditors would be unwilling to roll over their exposures to the ECA region, let alone continue to extend new financ- ing for consumption and investment, came to the fore. In an extreme, no new external financing and 100 percent rollover imply a balanced current account. A less-than-100 percent rollover would then require the current account to be in surplus by the amount of maturing debt that creditors do not roll over. Reality lies in between these two cases: there is less-than-100 percent rollover but there is also some new money. Official fi nancing can soften the pain of this adjustment, as will be seen later. It is thus not surprising that external adjustment is proving to be quite sharp in many ECA countries--despite the effectiveness so far of international coor- dination efforts. Also, as discussed in Chapter 1, ECA as an integrated region is also affected through other channels; from sharp declines in exports given the slowdown in the world economy to the effect of a decline in remittances. Different shocks for different countries The 2008 global crisis affected fi nancially integrated ECA countries in two ways--and through an additional third channel for the oil exporters among them. First, global deleveraging made creditors unwilling to continue financ- ing large current account deficits. But deleveraging was also due to the unwinding of real estate booms in some host countries. Second, the deep world recession led to a downturn in exports to West- ern Europe, translating almost immediately into lower output and employ- ment. This effect is particularly strong in small open economies--such as the Czech Republic, Estonia, Hungary, and the Slovak Republic, where exports in 2008 accounted for 70­80 percent of GDP--and somewhat less in larger econ- omies such as Poland and Romania (30­40 percent). Furthermore, economies in Central Europe and the Baltic states--which depend heavily on exports of motor vehicles and auto parts as well as engineering products such as machin- ery, electrical goods, and transport equipment--can expect to see a particu- larly severe downturn due to falling investment demand and a shortage of credit.1 In contrast to East Asia in 1997, the Russian Federation in 1998, and Latin America in the second half of the 1990s and the turn of the century, the current combination of a seizure in advanced country financial markets and the deepest global downturn since the Great Depression makes a quick export-driven recovery more challenging. 1. Landesmann 2009. 76 TURMOIL AT TWENTY Th ird, a sharp decline of around 40 percent in oil prices between 2008 and 2009 had an adverse impact on oil-exporting countries, because these account for nearly a fi ft h of the Russian Federation's exports and a third of Kazakhstan's. The first shock is what is typically referred to as a capital account crisis,2 marked by an abrupt and massive reversal of capital flows to a country and carrying the potential of a loss of confidence. Creditors head for the exits. The currency comes under intense pressure. And the banking system experiences a deterioration of its assets as borrowers face financial difficulties, a drying up of wholesale and interbank funding, and, as in the Russian Federation and Ukraine, a withdrawal of deposits. Countries with floating exchange rates have already seen major depreciations of their currencies. In relation to the euro, the Hungarian forint lost 20 percent of its value between October 2008 and February 2009, the Romanian leu 20 per- cent between August 2008 and February 2009, the Serbian dinar more than 20 percent between September and December 2008, and the Ukrainian hryvnia 35 percent between September 2008 and June 2009. A country with a currency peg, such as Latvia, lost a fifth of its reserves over the past year as the central bank defended the peg. Earlier capital account crises suggest the following: · In the crisis year (year t) private capital flows sharply reverse (figure 2.1). While foreign direct investment and portfolio flows tend to be more stable, the reversal in bank and corporate financing is substantial, lasting several years, particularly in East Asia. · The banking sector is distressed by worsening asset quality and the effect of currency depreciations on unhedged borrowers that contracted loans in foreign currency. The deterioration in bank balance sheets can be sub- stantial. For example, nonperforming loans as a proportion of the total peaked at about 20 percent in Thailand during the East Asian crisis and in Argentina in 2002; at 30 percent or slightly more in Indonesia, Korea, and Malaysia; and at around 40 percent in Ecuador and Uruguay. · In the year following the crisis, private investment plummets. Between 1997 and 1998, it fell from 32 percent of GDP to 16 percent in Malaysia and from 22 percent to 11 percent in Thailand--a reaction to overinvestment in the boom years and the decline in asset prices that follows the reversal of capital flows. Early data suggest that such a collapse in private investment has taken place in the Russian Federation and Ukraine. Similarly, the slowdown in growth in 2. For a discussion on what criteria need to be met to classify an event as a capital account crisis, see Ramakrishnan and Zalduendo 2006. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 77 FIGURE 2.1 Mean private capital flows during the East Asian capital account crisis Percent of GDP 10 Private capital flows, net 5 Direct investment, net 0 Portfolio investment, net ­5 Other private capital flows ­10 t­3 t­2 t­1 t t+1 t+2 t+3 t+4 t+5 Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. Estonia and Latvia brought the overheating in real estate markets to an abrupt end--pushing investment down more than 10 percentage points of GDP. The second shock is a collapse in export demand, requiring an even deeper contraction in imports, intensifying the negative impact on consumption. And in small open economies, where imports are large as a proportion of GDP, government revenues fall steeply. In Bulgaria and Latvia, for example, revenues from such indirect taxes as the value-added tax have fallen sharply. The third shock adds the decline in oil prices to the fall in export demand for oil exporters, with the loss divided between private and public income, depending on oil taxes.3 It can further compress imports over what is required to equilibrate the balance of payments in response to the other shocks.4 Sharing the burden: private and public, domestic and external Maturing debt in some ECA countries is high relative to foreign exchange reserves. Short-term private capital flows have financed a good part of the cur- rent account deficits in ECA. The ratio of maturing debt to the country's for- eign exchange reserves, an indication of liquidity pressures, is quite high: more than three times for Estonia and Latvia and 1.5 times in other ECA countries. 3. An average for Kazakhstan and the Russian Federation yields an export tax of roughly 25 percent. 4. The oil price decrease benefits oil importers, an effect that is not taken into account here. 78 TURMOIL AT TWENTY Compare that with a median of 1.8 times reserves in capital account crisis countries from East Asia in 1997 and 2.4 times elsewhere. A coordinated rollover of the bulk of external debt coming due can restore investor confidence in a capital account crisis. A sizable share of capital inflows to ECA countries is external finance from Western European banks, both to their subsidiaries in the ECA countries and directly to borrowers. So, gener- ous official financing to support policy reform, coordinated with full rollover of a major component of private debt, could assure the markets while allow- ing the least committed creditors an orderly exit. Indeed, the European Bank Coordination Initiative provides a forum for the parties to discuss these roll- over matters (box 2.1). A large official financing package would reduce the risk of a country's imposing a standstill on its payments, which would transform a temporary liquidity squeeze into a worse situation of violating its debt-service commitments. Reforms covering fiscal policy, bank restructuring and resolution, and bet- ter bank regulation and supervision, together with corporate and household debt restructuring (chapter 3), could ease concerns about solvency and per- suade creditors with claims on the country not to withdraw. The evidence that markets have become more discriminating in judging country risk (annex 1.1 in chapter 1) supports the presumption that a rollover is more likely to be forthcoming in the context of domestic policy reform.5 Indeed, generous offi- cial financing and a rollover of short-term debt by major creditors can increase the likelihood of policy reform, since the program can bridge the time for the reforms to take effect. How difficult could it get? Since the possible adjustment path for some country groups is meant only as an illustration, it is limited to two of the financially integrated country groups described in chapter 1--Groups 1 and 4--as well as the special case of two oil exporters, the Russian Federation and Kazakhstan (table 2.1). Group 1, Estonia and Latvia, ran an average current account deficit of more than 11 percent of GDP in 2008 and had external debt maturing in 2009 of nearly 70 percent of GDP. Group 4 had an average current account deficit of 5 percent of GDP in 2008 and significantly less external debt maturing in 2009-- 25 percent of GDP. Initial fiscal deficits were moderate for both groups at 2­3 percent of GDP. Thus, the current account deficit--the sum of the excess of spending over income of the private sector, or net private dissavings, and the 5. IMF Independent Evaluation Office 2003. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 79 BOX 2.1 Sticking together through thick and thin: the European Bank Coordination (Vienna) Initiative As the first IMF-supported programs among the new EU member states were put together in late 2008, it became clear that the sustainability of these pro- grams depended heavily on the ability to ensure that Western European banks would remain engaged in host countries. They were, after all, at the center of the buildup of vulnerabilities in transition countries. In the context of interna- tional fi nancing packages, the IMF, the EC, and other international fi nancial institutions (as well as the banking regulators of the home and host countries), engaged in discussions with parent banks of subsidiaries aimed at ensuring high rollover rates in the exposure of these banks in each of the countries requiring IMF financial support. Another goal was to ensure that these programs would not be perceived as a private sector bailout. How has the process worked in practice? So far the coordination has been useful and has provided a forum for discussion among the key parties. For Hungary, banks committed in October 2008 to support their subsidiaries and did so, case-by-case, in discussions with the central bank and banking super- vision authorities. But the IMF arrangements in Romania and Serbia aimed at specifying further these commitments. In late March 2009, in both countries, the main parent banks gave a general declaration on maintaining their overall exposure in the host country and on adequately capitalizing their banks fol- lowing stress tests carried out by the central bank. The aim was to ensure pre- emptive increases in capital given that nonperforming loans were expected to pick up in line with the slowdown in economic activity. Later in May, the nature of the voluntary commitment was discussed in Brussels for Romania, clarify- ing further how overall exposures are defi ned, how large pre-emptive capital increases should be, and the mechanisms to re-assess the adequacy of these commitments as economic activity slows. A letter similar to the one signed in March was signed for Hungary in mid-May. Although the initiative is justified given the nature of a crisis that has unfolded with differing timeframes across the ECA region, it has limitations as to what it can accomplish. First, the exposure decisions by parent banks can only to a degree be seen as a country-specific commitment. For the parent bank, the exposure decision has a regional dimension. In an extreme, a 100 percent rollover in the countries that first face external financing difficulties would in effect imply deleveraging (a likely scenario) in the less vulnerable countries (an unlikely scenario). Second, although the agreements with parent banks (even if legally non- binding) are nondiscriminatory (the treatment of domestically owned and foreign-owned banks is the same), central banks of the countries interested in reaching these agreements might at times provide fi nancial incentives (such (continued) 80 TURMOIL AT TWENTY BOX 2.1 (CONTINUED) Sticking together through thick and thin: the European Bank Coordination (Vienna) Initiative as different reserve requirements on new fi nancing flows). These incentives are in compliance with EU regulations, but over time could potentially distort resource allocation across the region. Third, a key aspect of the initiative is capitalizing banks (foreign or domes- tic) for what promises to be a period of increasing nonperforming loans. But the differences across the regulations and accounting practices in the region are pronounced. For instance, some countries follow international fi nancial accounting standards and others follow national accounting standards. Thus, different countries might end up with varying degrees of capital adequacy ratios. What matters more, however, is not the strength of the subsidiary but the fi nancial health of the bank as a group. Yet the standards for such com- parisons are far from agreed with several initiatives being followed--both at the country level and at the group level--and substantial differences of opinion exist as to how to assess the health of Europe's banking system. Notwithstanding these caveats, international coordination has proved a useful undertaking and, at least so far, the region has avoided the rapid de- leveraging observed in past capital account crises events. excess of spending over income of the public sector, or net public dissavings-- mainly reflected net private dissavings in Group 1, less so in Group 4. The con- tribution of net private dissavings to the current account deficit ranged from 60 to 75 percent in these groups, confirming that ECA's imbalances had pre- dominantly private origins. The oil exporters had an average current account surplus of 6 percent of GDP and maturing debt of 12 percent of GDP. Their average fiscal surplus was nearly 3 percent of GDP, and both had oil funds and large foreign exchange reserves. Without official financing, the partial rollover of maturing debt will require a sharp cut in private spending--and a sharp fiscal contraction. The adjust- ment brings the current account, net of what FDI can fi nance, into surplus by the amount needed to service that proportion of maturing debt not rolled over.6 Because FDI is the only new money assumed to be available, the cur- rent account deficit would have needed to shrink to this level of FDI if matur- ing debt had been fully rolled over. Even this would be a major adjustment for Group 1 countries because they were running double-digit current account deficits before the crisis. 6. Empirical evidence suggests that a reduction in the flow of credit by 1 percent of income increases private saving rates by 0.32 percentage points (Loayza, Schmidt-Hebbel, and Serven 2000). HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 81 TABLE 2.1 Savings­investment balances, adjustment without official financing (percent of GDP) Oil-rich middle-income CIS economies (Russian Item Group 1 Group 4 Federation and Kazakhstan) Macroeconomic data, 2008 Current account balance ­11.2 ­5.0 6.2 Public sector savings­ investment balance ­2.9 ­2.1 2.7 Private sector savings­ investment balance ­8.4 ­2.9 3.5 Foreign direct investment 4.4 3.2 4.6 Short-term debt, remaining maturitya 68.9 24.6 12.4 Assumptions for external adjustment scenarios Share private sector debt 0.9 0.9 0.9 Rollover rate 0.75 0.75 0.50 Foreign direct investment (half the 2008 value) 2.2 1.6 2.3 Scenario for 2009 without official financing Estimated current account balance 15.0 4.6 3.9 Estimated public sector savings­investment balanceb 1.7 0.6 0.6 Estimated private sector savings­investment balancec 13.3 3.9 3.3 Note: The table presents the assumptions underlying the adjustment scenarios. First, in line with the experience of previous capital account crisis cases, external private creditors are willing to roll over 75 percent of debts coming due for Groups 1 and 4 and 50 percent for oil producers. Although rollover rates in capital account crisis cases of a private origin are typically lower (65 percent during the East Asia crisis), the higher figure for Groups 1 and 4 is justified in recognition of the fact that the parent-subsidiary relationship is likely to be more conducive to a higher rollover rate in those ECA countries where this covers a substantial proportion of banking sector assets, an argument which will be developed later in the chapter. By the same token, the lower rollover rate for oil exporters reflects the lower share of banking sector assets owned by parent banks in these countries. Second, although there is some variation across groups, it is assumed as a sim- plification that 90 percent of maturing debt in all groups is owed by the private sector, which thus has to service 90 percent of what is not rolled over. Third, the ratio of foreign direct investment (FDI) to GDP, which remained stable at pre-crisis levels in earlier crises, is assumed to be halved-- to reflect the ongoing recession in originating countries. Fourth, FDI is also assumed to accrue entirely to the private sector, adding to its savings or financing projects that would have been undertaken in any event. The assumption has the effect of making FDI fully fungible and thus, per- haps optimistically, available to finance maturing debt. Fifth, because portfolio flows have typically hovered around zero in earlier capital account crises, they are assumed to be zero. a. Includes short-term debt falling due in 2009 and medium- and long-term debt falling due in the same year. b. Calculated as [1 ­ (rollover rate)] × maturing external debt (2009) × (1 ­ share of private sector debt). c. Calculated as [1 ­ (rollover rate)] × maturing external debt (2009) × (share of private sector debt) ­ foreign direct investment. Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. 82 TURMOIL AT TWENTY That is not all. A further adjustment is required to bring the current account balance into a surplus equal to that proportion of maturing debt not rolled over. Recall that maturing debt is nearly 70 percent of GDP for Group 1 in 2009, compared with about 25 percent for Group 4. So, the former requires a very large external adjustment. Group 4 also needs to adjust, though the magnitudes are lower. And because the bulk of debt repayment falls to the private sector, the change in net private savings is much larger than that in net public savings. Figures 2.2 and 2.3 illustrate these scenarios. Net private saving as a share of GDP (measured horizontally in the figures) and net public savings as a share of GDP (measured vertically) make up a given current account balance. An iso-current account balance (iso-CAB) line is a negatively sloped 45-degree line whose intersection with either the vertical or the horizontal axis reflects the total current account balance. Deterioration in either the private or the public balance needs to be compensated for by improvement in the other bal- ance to remain on the same iso-CAB line. For reference, the iso-CAB that passes through the origin connects all combinations of private sector and public sector savings that yield a current account balance of zero. All lines to the northeast of the reference line represent a current account surplus, and those to the southwest a current account deficit. Not surprisingly, the iso-CAB lines for Groups 1 and 4 before the crisis are all located in the southwest quad- rant (figure 2.2), because all the balances are negative. The iso-CAB line for FIGURE 2.2 Crisis, adjustment, and financing in financially integrated Europe and Central Asia economies (percent of GDP) Public savings­investment balance 4 Initial conditions, 2008 2009 simulation without official financing Current account Group 4 current 2 balance = 0 account balance = 4.5 (13.3, 1.7) Group 1 current (3.9, 0.6) account balance = 15.0 0 2009 simulation (­2.9, ­2.1) ­2 with official (­8.4, ­2.9) (6.7, ­4.0) (2.0, ­4.0) financing Group 1 current account Group 4 current Group 4 current Group 1 current balance = ­11.2 account balance = ­5.0 account balance = ­2.0 account balance = 2.7 ­4 ­12 ­9 ­6 ­3 0 3 6 13 15 17 Private savings­investment balance HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 83 FIGURE 2.3 Crisis, adjustment, and financing in financially integrated oil-exporting Europe and Central Asia economies (percent of GDP) Public savings­investment balance 4 (3.5, 2.7) Current account 2 balance = 6.2 (before oil price decline) (3.3, 0.6) 0 (­4.7, ­0.1) Current account ­2 Current account balance = 3.9 balance = ­4.8 (oil price adjusted) Current account balance = ­2.4 (1.6, ­4.0) ­4 ­6 ­3 0 3 6 Private savings­investment balance oil producers (figure 2.3) is in the northeast quadrant, reflecting positive net private and public savings. More precisely, the iso-CAB line for Group 1 is farther to the southwest, reflecting a current account deficit that is more than twice that for Group 4 (figure 2.2). Finally, the dominance of net private dissaving in this deficit is represented by the greater horizontal distance of the Group 1 equilib- rium in 2008 from the vertical axis compared with that from the horizon- tal axis. The iso-CAB lines for each group for 2009 are consistent with a partial rollover of maturing debt and no new money other than some FDI, which is assumed to flow entirely to the private sector and be available for fi nancing debt repayment. Both iso-CAB lines move to the northeast quadrant, but the shift for Group 1 is much larger. Indeed, the current account surplus is 15 per- cent of GDP for Group 1--a shift of more than 26 percentage points of GDP! In contrast, Group 4 undergoes a smaller, yet still large, shift of just less than 10 percentage points of GDP. Why? Because its initial current account deficit as a share of GDP was lower, requiring a smaller adjustment, and because less is needed to service the group's relatively modest maturing debt that is not rolled over. For oil producers, the developments in the iso-CAB line are a little more com- plicated owing to the parallel effects of changes in oil prices. The iso-CAB line moves to the southwest quadrant of figure 2.3, showing a shift of 11 percentage points of GDP because of the fall in oil prices, reflecting a worsening of both 84 TURMOIL AT TWENTY private and public net savings.7 But the need to service the part of the maturing debt that is not rolled over then requires a current account surplus of just less than 4 percent of GDP, a shift of some 9 percentage points of GDP. This change is smaller than in Group 4 because of the lower share of maturing debt in GDP. Government steps in . . . The compression of private spending in response to new money equaling only half of 2008 FDI and the partial rollover of maturing external debt is very large. Government policy can lessen the adjustment, but official action has other justifications. Paralleling past capital account crises, the current episode is accompanied by systemic bank and corporate distress (box 2.2). And for some ECA countries, the adverse impact on households may not allow gov- ernments to stand back. Banks are experiencing distress as their loan port- folio deteriorates on account of households and fi rms not being able to ser- vice debts. Banks will need to be recapitalized by private shareholders, and in some cases governments might need to step in. The fiscal costs are likely to be significant. Gross fiscal costs over the five years following the start of the capital account crisis in East Asia in 1997­98 were 57 percent of GDP in Indonesia, 31 percent in Korea, 16 percent in Malaysia, and 44 percent in Thailand. In Latin America, gross fiscal costs are estimated to have been 10 percent of GDP in Argentina, 22 percent in Ecua- dor, and 20 percent in Uruguay.8 The largest part of these costs is using public funds to recapitalize banks. Some of these costs were eventually recovered, but the recovery rate ranged from around a quarter in East Asia to between a third and a half in Latin America.9 Although these numbers provide a bench- mark, the recapitalization of subsidiaries will likely be by parent banks, as in Latvia and Romania, where the authorities are seeking such support. So, there could be a call on public funds in partnership with the private sector to recap- italize domestically owned banks that are deemed viable and are of systemic importance and where private shareholders cannot raise enough capital. 7. Energy exports in dollars for both countries are projected to decline by 44 percent from 2008 to 2009, slightly more than the projected decline in world prices (39 percent). In 2008, energy exports were 18 percent and 32 percent of GDP in the Russian Federation and Kazakhstan, respectively; thus the decline of energy exports as a percent of 2008 GDP is 8 percent in the Russian Federation and 14 percent in Kazakhstan. The average of these two figures (11 percent) yields the impact of the oil price decline on the current account balance. Th is resulting current account balance is ­4.8 per- cent of GDP--the 2008 actual current account balance of 6.2 percent of GDP minus the 11 percent adjustment. 8. Laeven and Valencia 2008. 9. The exception in East Asia was Malaysia, where the recovery rate was two-thirds; the exception in Latin America was Argentina, where nothing was recovered. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 85 BOX 2.2 From Bangkok to Budapest: ECA's adjustment compared with East Asia's Reflecting lessons from earlier emerging market crises, official financing in the affected ECA countries has been significantly more generous than in the East Asian crisis. The projected increase in net private savings and the swing from deficit to surplus in the current account in East Asia were broadly comparable to those in Group 1, the hardest hit ECA countries. The box figure shows net private savings and the fiscal deficit in 1997 through 2000 in Indonesia, Korea, Malaysia, and Thailand (mean values). The changes in net private savings and in the fiscal surplus were 15 percentage points of GDP and a worsening by more than 2 percentage points of GDP respectively, while the current account swing from deficit to surplus was 13 percentage points of GDP. This can be compared with a change in net private savings and the fiscal surplus of 15 percentage points of GDP and a worsening by over 1 percentage point of GDP respectively in Group 1 between 2008 and 2009, and a corresponding improvement in the current account deficit of just under 14 percentage points of GDP in the equi- librium with official fi nancing. The latter amounted to more than 12 percent- age points of GDP. Of course, the external adjustment would have been much sharper if external financing had not been available. A fi rst difference with the current crisis is the larger deterioration in East Asia's fiscal balance in the aftermath of the shock than that of Group 1, which East Asia could cover thanks to its stronger fiscal position--one of virtual BOX FIGURE 1 Public and private sector balances, developing East Asia economies and Europe and Central Asia Group 1 economies Percent of GDP 20 Foreign savings Public balance Private balance 10 0 ­10 East Asia Group 1 ­20 1997 1998 1999 2000 2008 2009 2009 (without (with official official financing) financing) Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. (continued) 86 TURMOIL AT TWENTY BOX 2.2 (CONTINUED) From Bangkok to Budapest: ECA's adjustment compared with East Asia's balance--going into the crisis. Even so, the fiscal deficit in East Asia was just above 2 percent of GDP in the year after the crisis. Current IMF-supported programs in Central and Eastern Europe incorporate official fi nancing in the range of 8 to 12 percent of GDP and have allowed more recently for a reces- sion-induced deterioration of fiscal balances through its effects on automatic stabilizers. A second major difference relates to the circumstances in the world econ- omy. The pattern of adjustment in East Asia in the five years after the crisis was one of declining current account surpluses from double-digit percentages of GDP as capital flows, especially bank and corporate fi nancing, which had turned negative for a few years, gradually resumed. This allowed net private savings to fall, also from double-digit percentages of GDP. Indeed, the resto- ration of confidence implied by declining current account surpluses occurred against a backdrop of rapidly expanding exports to the rest of the world. Grow- ing exports and declining current account surpluses allowed imports to expand pari passu with the recovery in private consumption and investment. Finally, fiscal imbalances, which were allowed to deteriorate by more than 2.5 percent of GDP as countries switched to expansionary demand policies to offset the recession after the shock, were reined in as the recovery took hold. Fiscal policy can also help countries adjust to both the capital account cri- sis and the trade contraction. First, the public sector has to service that part of non­rolled over maturing debt for which it is responsible. Second, as dis- cussed, fiscal space will need to be found to recapitalize domestically owned banks that are deemed viable but undercapitalized, jointly with sharehold- ers of those banks, with the injection of public funds being accompanied by government representation in decision making. Third, the recession triggers automatic stabilizers. It causes tax revenues to decline, for example due to severe import contraction, and public spending to increase on items such as unemployment benefits and targeted social safety nets. This additional deficit needs to be financed. Fourth, countries with exchange rate pegs will need to undertake a deeper fiscal consolidation, accompanied by incomes policy, to reduce real wages-- in order to bring about the relative price changes needed to switch resources toward tradables and demand toward nontradables. The few ECA countries that entered the crisis with strong fiscal positions have implemented countercyclical fiscal policy to alleviate the recession (box 2.3). Households that have contracted mortgage and other consumer debt in foreign exchange might not be able to service these debts as a result of currency depreciation, putting pressure on the HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 87 BOX 2.3 Countercyclical fiscal policy in financially integrated countries: Kazakhstan and the Russian Federation Taking advantage of the reserves accumulated during the period of high oil prices, Kazakhstan and the Russian Federation have quickly introduced some countercyclical fiscal measures to cushion the impact of the crisis. The composition and timing of the fiscal response in the Russian Federation is summarized in box table 1. The fiscal measures implemented in 2008 and announced by mid-2009 are about 2.6 percent and 4.1 percent of GDP, respec- tively. The initial focus was stabilizing the financial sector, particularly through injecting liquidity and recapitalizing banks. In 2009, the focus has moved to supporting the real economy with an emphasis on reducing the tax burden on large firms. So far the support to vulnerable households has been limited. The recent announcement by Rosstat that the number of people below the poverty line rose by one-third during the first quarter of 2009, however, raises the issue of whether an increasing share of the fiscal stimulus should not be devoted instead to protecting these households. The fiscal stimulus announced by Kazakhstan, about 90 percent of which is fi nanced by the National Oil Fund, is relatively large, amounting to about BOX TABLE 1 Summary of fiscal anti-crisis measures--introduced in 2008 and announced for 2009 Distribution of each policy measure as Billions of rubles percentage of total Total % Measure 2008 2009 Total of GDP 2008 2009 Total Strengthening the financial sector 785.0 625.0 1,410.0 3.3 72.1 34.1 48.2 Supporting the real economy 304.0 798.3 1,102.3 2.5 27.9 43.5 37.7 Protecting the vulnerable -- 111.5 111.5 0.3 -- 6.1 3.8 Transfers to regions -- 300.0 300.0 0.7 -- 16.4 10.3 Total 1,089.0 1,834.8 2,923.8 6.7 100.0 100.0 100.0 % of GDP 2.6 4.1 6.7 -- is not applicable. Note: Excludes quasi-fiscal and monetary measures, state guarantees in the amount of 300 billion rubles planned for 2009, and measures planned before the crisis, such as increasing the minimum wage and indexing pensions, as well as external crisis-related lending to CIS countries and Mongolia. Source: World Bank staff estimates, Government of the Russian Federation. (continued) 88 TURMOIL AT TWENTY BOX 2.3 (CONTINUED) Countercyclical fiscal policy in financially integrated countries: Kazakhstan and the Russian Federation 8 percent of GDP. Nearly 60 percent had been disbursed as of mid-2009. As in the Russian Federation the majority of the initially disbursed funds (about 60 percent) were directed to recapitalize the banks and stabilize the financial sec- tor. The remaining disbursements have been directed toward supporting small and medium enterprises, mortgage refinancing, agriculture, and construction. But should future allocations include some expansion of safety nets for vulner- able households instead? Although Kazakhstan's spending on a safety net as a share of GDP is low compared with other ECA countries, it has a well targeted means-tested social assistance program worth expanding. At present the pro- gram is quite small: though 70 percent of the beneficiaries are in the poorest quintile, only about 2 percent of households in that quintile are being covered (figures 4.5 and 4.6). Source: Bogetic et al. 2009 and World Bank 2009a. government to provide relief. This should be agreed to only after careful exam- ination, informed by appropriate burden sharing and accountability to limit moral hazard and avoid the diversion of scarce public money from better uses. . . . with a little help from its friends There are limits to accommodating the tax and public spending consequences of the crisis within an unchanged fiscal envelope, highlighting the need for official financing. Such financing from international and regional financial institutions will be required to support the new fiscal needs. It can also help reverse some of the currency depreciation following the outbreak of a capital account crisis. The case for official fi nancing will be even stronger if the fiscal conse- quences of crisis and recession are accommodated through reallocations away from lower priority public spending and through greater efficiency in pen- sions and health spending. This is especially important in countries such as Hungary, Ukraine, and Serbia, where public spending as a share of GDP is already very high in relation to per capita income--just below 50 percent in the first two countries and 45 percent in the third. Indeed, the crisis may pro- vide an opportunity to enact reforms that might have eluded the authorities in normal circumstances. Notwithstanding the short-run increase in spend- ing financed from official sources, the structural fiscal balance should be on a path that ensures debt sustainability especially where the burden of external public debt was high before the crisis. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 89 Indeed, official financing from supranational and international financial institutions--and the use of oil funds by oil exporters--can reduce the inten- sity of the private sector adjustment and provide room for a fiscal deficit. For instance, the official fi nancing in figure 2.2 (and resort to oil funds in fig- ure 2.3) illustrates what is required to halve the substantial net private savings in the equilibrium without official fi nancing and to increase fiscal space in order to allow the three country groups to have net public dissaving equal- ing 4 percent of GDP. It moves all the iso-CAB lines in a southwesterly direc- tion. Using Group 1 as an illustration, official financing is equal to maturing debt that is not rolled over (17.2 percent of GDP) less the sum of the current account balance (the sum of net private savings of 6.7 percent of GDP and net public dissaving of ­4 percent) and FDI of 2.2 percent of GDP (table 2.2). So, the required official financing is 12.4 percent of GDP. Financing needs in the other two groups are about 6 percent of GDP, with the additional fi nancing in the oil-exporting countries assumed to come from their oil funds. Financ- ing of 6 percent of GDP allows the current account balance to change from a TABLE 2.2 Savings­investment balances, adjustment with official financing (percent of GDP) Oil-rich middle-income CIS economies (Russian Item Group 1 Group 4 Federation and Kazakhstan) Scenario for 2009 with official financing (or use of foreign exchange reserves) and a reduction in the external adjustment a Estimated current account balance 2.7 ­2.0 ­2.4 Estimated public sector savings­investment balanceb ­4.0 ­4.0 ­4.0 Estimated private sector savings­investment balancec 6.7 2.0 1.6 Official financing (or foreign exchange reserve use) 12.4 6.6 6.3 Effect of an increase in rollover ratesd Extra financing 3.4 1.2 0.6 Effect on Reduced Reduced Reduced by private surplus by half by half a third a. The goals are to allow room for fiscal impulse and to reduce the private sector balance (surplus) by half. b. Set at -4 percent of GDP. c. Arbitrarily set at half the surplus level identified in the projections that assume no official financing. d. Scenario in which rollover is assumed to be raised to 0.80 for Groups 1 and 4, and 0.55 for Kazakhstan and the Russian Federation. Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. 90 TURMOIL AT TWENTY surplus of 3­4 percent of GDP to a deficit of 2­2.5 percent and the fiscal bal- ance to move from a modest surplus to a deficit of 4 percent. A higher rollover of maturing debt cushions the required reduction in the private sector's imbalance between spending and income. An increase in the rollover rate by 5 percentage points (from 75 to 80 percent for Groups 1 and 4 and from 55 to 60 percent for the oil producers), provided it accrues entirely to the private sector, makes possible a reduction of net private savings of between one-third to one-half from the equilibrium with official financing. Clearly, if maturing debt is large (as in ECA), even a small improvement in rollover rates provides much needed relief in adjustment (table 2.2). The weakness of the expected recovery from the global recession highlights the importance of sustaining generous official financing. Crisis support from the international financial institutions and the EU has so far been adequate. Parent banks have largely maintained their exposure. The risk that Western European parent banks recapitalized by home governments would focus lend- ing in their home countries has so far been averted. For their part, some hard- hit ECA countries have positive net private savings, while their fiscal balances have deteriorated. But continuing collective action will be important. Unless replenished by new lending, net transfers from international fi nancial institutions are expected to turn negative in three to four years. A sluggish recovery and high unemployment in Western Europe will test the willingness of home govern- ments to recapitalize banks as their large losses have to be recognized. And financial protectionism could put strains on cooperation between home and host country authorities of cross-border banks. Crisis, adjustment, and financing in low-income and lower middle-income CIS countries Because the external borrowing of the low-income and lower middle-income countries of the CIS (Armenia, Georgia, the Kyrgyz Republic, Moldova, and Tajikistan) comes mainly from official sources, they have for the most part not been directly affected by a reversal in capital inflows.10 Instead, they have been affected by the global economic recession, which has led to a collapse of export demand and, for a number of them, of workers' remittances as well. Foreign direct investment has also declined during 2009 to about half of its 2008 level. The sharp downturn in the Russian Federation has depressed export demand 10. Georgia is not part of the CIS but is included in the group because its economy shares many fea- tures of the other countries. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 91 in Armenia, the Kyrgyz Republic, Moldova, and Tajikistan. The marked decel- eration in growth in Kazakhstan following the sudden stop in capital flows in 2007 has hurt the Kyrgyz Republic. And export demand for Georgian goods, for which the CIS countries are not a major destination, is also projected to fall due to recessions in Turkey, the European Union, and the United States. The fall in workers' remittances from the Russian Federation, Kazakhstan, and the European Union has been significant. Preliminary estimates suggest that remittances had accounted for nearly 35 percent of GDP in Tajikistan, 30 percent in Moldova, and 25 percent in the Kyrgyz Republic in 2008--and fell in the first quarter of 2009 and even more during the course of the year. In Tajikistan it is estimated that a 30 percent decline would cut the consumption of households in the poorest quintile between 17 percent (rural) and 21 percent (urban) and increase headcount poverty from 53 percent to nearly 58 percent (box 2.4). Terms of trade changes have not been positive for countries that export products intensive in natural resources. For copper in Armenia and Geor- gia, and aluminum and cotton in Tajikistan, the price prospects are dim. And the difficulties facing Kazakhstani banks as a result of their inability to roll over external debt have the potential to spill into some Central Asian coun- tries, such as the Kyrgyz Republic, where subsidiaries of Kazakhstani banks account for more than a third of banking assets. Parent banks have main- tained funding of their Kyrgyz subsidiaries so far. Without official external financing, a trade-and-remittance shock, together with lower foreign direct investment, will require tight fiscal policy to lower the current account deficit. The effect of fiscal consolidation on output and employment is negative, but this can be partly mitigated by a depreciation of the currency. In the likely event of the depreciation not being passed through fully into domestic price increases during a recession, this facilitates the trans- fer of factors away from nontradables and toward exportables and import sub- stitutes. As with the financially integrated countries, it can also switch foreign demand toward the country's tradable goods, but this will be more limited for exports than import substitutes because of currency depreciations in com- petitor countries. The ability of the low-income and lower middle-income CIS countries to mitigate the contractionary impact of fiscal consolidation has been made more difficult by the Russian ruble's losing an estimated 15 percent of its value rela- tive to a dollar-euro basket since September 2008 and the Kazakhstani tenge's losing 20 percent of its value in February 2009. While several low-income and lower middle-income CIS countries have allowed significant depreciations of their currencies, the depreciation of the ruble and the inflation differentials 92 TURMOIL AT TWENTY BOX 2.4 Tajikistan's declining remittances can hurt the poor disproportionately The inflow of remittances to Tajikistan in 2007 amounted to $1.4 billion, or about 40 percent of the country's GDP, the highest share worldwide. The Rus- sian Federation is the principal country of origin for such remittances, account- ing for about 80 to 90 percent of officially recorded remittances. The box table shows that low-income households in Tajikistan rely heavily on migrant flows. Almost half of the families in the bottom quintile in the income distribution have an absent migrant abroad. The difference is very sharp relative to the other quintiles. The income dependency seems to be critical for low-income rural households. In the bottom quintile, remittances account for almost 80 per- cent of rural household consumption in cases where they have a migrant abroad. Although estimates of the magnitude of the decline in remittances during the crisis vary, a range of a 30­35 percent decline is commonly used. A 30 per- cent decline will result in a 25 percent decline in the consumption of the poor- est rural households and an 18 percent decline for the poorest urban house- holds and increase headcount poverty in Tajikistan from 53 to 58 percent. BOX TABLE 1 Share of households with migrants, by preremittance consumption quintile Remittances as a share of total Preremittance Share of households household consumption consumption with currently quintile absent migrant Urban Rural 1 46.5 55.9 78.8 2 7.5 35.9 66.5 3 4.9 36.2 65.1 4 4.6 45.0 63.9 5 6.3 44.6 60.1 Average 14.1 39.5 66.4 Note: Quintiles are calculated over yearly total household consumption net of remittances. World Bank estimates using TLSS 2007 data. The analysis uses an absolute poverty line equal to 139 somoni per month while the value of the PPP $2.15 poverty line used by the World Bank for international comparison is 120 somoni per month. Source: World Bank 2009b. with other trading partners have resulted in some appreciation of their real effective exchange rates. As with other groups of countries, GDP growth forecasts for the low- income and lower middle-income CIS countries continue to be revised down- ward. In general, the reduction in current account deficits projected for 2009 is modest in all countries except Georgia, which had a large current account deficit in 2008 for conflict-related reasons (figure 2.4). Official financing makes HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 93 FIGURE 2.4 Public and private sector balances in low-income and lower middle-income CIS countries, 2008­10 Percent of GDP 20 Foreign savings Public balance Private balance 10 0 ­10 ­20 2008 2009 2010 Source: IMF World Economic Outlook, World Bank World Development Indicators, and authors' calculations. it possible for countries to run larger fiscal deficits, which reflect in part the adverse impact, particularly on revenue, of the significant slowdown in growth. This implies that there will be less net private dissavings. Between 2008 and 2009, the current account deficit as a proportion of GDP is broadly unchanged in the Kyrgyz Republic and Moldova and might increase in Armenia. Official multilateral and bilateral financing from the Russian Federation is being used in both Armenia and the Kyrgyz Republic to fund higher fiscal deficits, reflecting in part the adverse impact of slower growth on revenue, but also increased spending on well targeted social safety nets in both countries. In Moldova, by contrast, the fiscal deficit is expected to balloon because of revenue shortfalls from trade taxes due to the severe import compression and to pre-election increases in wages and pensions, while net private dissavings fall by an offsetting amount. In Tajikistan, the current account deficit as a proportion of GDP is expected to increase, but the fiscal deficit will rise more, with a substantial increase in social spending however being accommodated in the face of revenue shortfalls through cuts in current spending and delays in capital outlays. Georgia is expected to witness a large decline from 2008 in the current account deficit as a proportion of GDP. There will be a reallocation to domestic social and infrastructure spending from other less urgent needs. Larger official financing will be required if expected levels of foreign direct investment do not materialize. The average current account deficit as a share 94 TURMOIL AT TWENTY of GDP for the five countries is expected to decline over the 2009­10 period, and remain at an average of about 12 percent of GDP (figure 2.4). Th is will be helped if the official fi nancing now covering about a third of such defi- cits can be maintained. Although the high historical values for foreign direct investment as a share of GDP are expected to be reduced in 2009, they are still expected to cover about a third of the current account deficits, at par with official financing. If that FDI does not materialize because of a delayed recov- ery in trade and exports, official financing will need to be stepped up. Most countries are increasing their fiscal deficits because of the fall in revenue and the expansion of social programs. So, private sector imbalances will need to be reduced further. Of parents and offspring: understanding rollover risks in ECA The assumption of a 75 percent rollover, though informed by earlier capital account crises, might not do justice to the particularities of ECA's financial integration. With rollover prospects depending heavily on the behavior of cross-border parent banks, the rollover risks might differ from those in past capital account crises. Although a few capital account crises linked to private sector imbalances, such as the sudden stops in East Asia, come closest to the current episode in ECA, the comparison cannot be taken very far. In East Asia, the capital flows were not linked to decisions by banks to develop a long-term engagement in the region, so the financing could not be considered to have had FDI-like fea- tures. Indeed, the resources flowing to East Asia in the mid-1990s were sim- ilar to wholesale funding and the syndicated loans seen in some transition countries over the past few years. But the bulk of bank financing going into ECA countries was loan financing from parent banks to their subsidiaries for an aggressive credit expansion, frequently in foreign currency. The resource flow may have also been triggered by profit opportunities in the region's ver- sion of the carry trade, and doing so through the extension of loans--in lieu of equity flows--had accounting and tax advantages. In sum, the resource flow reflects a strategic opening by Western European banks into new markets that were underdeveloped and that turned out to be quite profitable. Equally important, the resources were on-lent by the subsidiaries to corpo- rates and households, frequently at very long maturities. For instance, mort- gages in some Baltic countries have an average maturity of 26 years and are funded by the subsidiary through very short-term borrowing from parent banks. This maturity mismatch is inherent in financial intermediation. Once the long-term loans have been extended, there is little the subsidiary can do HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 95 FIGURE 2.5 Banking sector credit--national and domestic sources Foreign Equity and lending Subsidiary Additional funding from bank to subsidiary of foreign issuance of bonds, loans, Cr ndi os ng bank and deposit-taking funding le Foreign currency Local currency s- bo rd er Data covered by Bank for International Cross-border Settlements and International Financial lending is on-lent Statistics; foreign currency and local currency directly to firms breakdown available only from Bank for in foreign exchange International Settlements Local currency lending; most Cross- Domestic likely funded locally border credit lending market Data covered by International Financial Statistics only Total or Funding exclusively from Data covered by issuance of bonds, domestic banks national Bank for loans, and deposit-taking credit market International (that is, no links to Claims of Settlements only parent bank funds) Domestic bank to improve its liquidity position. In many ways, mortgage lending has a hand- cuffs nature and limits the drawdown of liquidity that is likely to occur.11 This is not to say that deleveraging is unlikely, but that the reversal of capital flows is probably not as easy as for investments by institutional investors in emerg- ing markets during previous capital account crisis events. This raises two questions. First, what is the exposure of parent banks in the region and their behavior so far? Second, what are the liability and asset features that might determine rollover behavior of parent banks and their subsidiaries? It's national not domestic: parent bank exposure To answer the first question it is necessary to understand the role of parent banks in credit markets. With rollover prospects depending heavily on the behavior of parent banks, it is necessary to cover not only the domestic credit market (akin to the residency concept of gross domestic product), but also the national credit market (akin to the concept of gross national product) 11. For a discussion see Herzberg, Sugawara, and Zalduendo 2009. 96 TURMOIL AT TWENTY (figure 2.5). A key foreign source of credit is direct or cross-border lending, which lies outside the monitoring orbit of monetary and banking supervision authorities in the recipient (host) country and could be a source of rollover risk in ECA countries at a time of global deleveraging.12 When parent banks commit to maintain exposures in an IMF-supported program, it is unclear how this extends to cross-border lending because these transactions are in response to decisions by the clients of parent banks that do not reside in the host country.13 But it is also possible that these are voluntary decisions by par- ent banks aimed, say, at circumventing regulations that might constrain lend- ing through their subsidiaries. Note the increasing share of cross-border lend- ing in total credit in ECA (table 2.3). But there are also significant differences across ECA: cross-border lending is 44 percent of total credit in Lithuania, but only 11 percent in Belarus and 12 percent in FYR Macedonia. In addition to direct or cross-border lending, parent bank groups also pro- vide credit through their subsidiaries to borrowers in the host country, in either foreign or local currency. The parent can fund these activities directly through the equity and credit lines it extends to its subsidiary or through deposit-taking operations carried out directly by the subsidiary in the host country. The sum of cross-border and foreign currency lending, referred to in the Bank for International Settlements (BIS) data as international claims, constitutes the foreign exchange exposure of the parent bank. The local-cur- rency credit extended by the in-country subsidiary, together with the inter- national claims of parent banks, are referred to by the BIS as foreign claims. (Annex 2.1 describes in greater detail the data covered in BIS international financial statistics.) Although credit flows described in figure 2.5 preclude a breakdown of the funding sources of each credit channel, it is reasonable to assume that most of the local currency lending is funded through deposits and wholesale funding that the subsidiary obtains locally.14 In that event, the foreign exchange risk 12. The discussion in this section combines the BIS dataset on fi nancial statistics and the IMF's data on banking systems; there are gaps in coverage that are fi lled in this section through putting forward simplifying assumptions. 13. An example makes this point clearer. A client of Bank X in Austria might request the extension of a credit to Firm Z in Hungary and, later on, instruct for this not to be renewed. In such a case, it is unlikely that the parent bank will maintain this exposure. Direct lending might also be extended at the initiative of the parent to avoid host country regulations. For example, Firm Z in the host country might have requested this cross-border lending and choose not to renew it as economic activity slows down; or, similarly, the parent bank might deliberately slow down the renewal of direct lending. Either way, this credit channel might retrench more rapidly as global deleveraging works its way through the fi nancial markets. 14. See Keller and Maslova 2009 for a discussion of this subject. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 97 TABLE 2.3 Direct lending as a share of total national credit, by country, 2005­09 (percent) Country 2005 Q1 Peak 2008 Q4 2009 Q1 Belarus 2.9 13.5 2008 Q1 11.1 11.4 Bulgaria 10.0 28.8 2008 Q1 26.7 27.2 Croatia 16.6 46.5 2008 Q1 35.8 36.3 Czech Rep. 25.3 31.9 2005 Q2 24.9 24.0 Estonia 25.0 57.2 2006 Q1 41.3 43.3 Hungary 26.4 38.0 2005 Q3 35.2 36.8 Kazakhstan 13.6 22.1 2006 Q4 15.5 15.8 Latvia 13.8 35.2 2008 Q1 26.4 26.6 Lithuania 21.2 46.8 2006 Q1 41.0 44.0 Macedonia, FYR 6.3 14.4 2008 Q4 14.4 11.9 Montenegro .. 44.3 2008 Q4 44.3 .. Poland 20.5 23.6 2005 Q3 17.8 18.5 Romania 22.5 46.2 2006 Q4 37.8 37.5 Russian Federation 13.0 20.6 2007 Q3 17.0 .. Serbia .. 47.8 2007 Q4 40.2 39.4 Slovak Rep. 22.3 33.3 2005 Q3 26.4 17.4 Turkey 13.7 19.1 2007 Q1 17.4 18.2 Ukraine 9.9 23.3 2008 Q1 21.9 21.2 Average 16.5 32.9 27.5 26.8 Median 15.2 32.6 26.4 25.3 .. is not available. Source: BIS and IFS, and authors' calculations. for the parent originating from local currency is limited.15 Conversely, what has been referred to above as international claims is more likely to be funded directly by the parent bank, although even this might not be fully correct because some foreign exchange lending might be funded by the subsidiary through deposit-taking operations in foreign exchange. If so, the direction of bias is clear: international claims would overestimate the foreign exchange exposure of the banking system to a home country. Figure 2.6 provides the absolute magnitudes of international claims as of end-2008 and as a share of GDP, as well as of the exposure to ECA countries.16 15. Credit operations might take place through two additional channels: claims of domestic banks, which are not linked to parent banks and are not monitored by the BIS, and direct lending that fi rms might access either through their parent corporate or in domestic and external capital markets. Th is category is not depicted in figure 2.5. 16. See Árvai, Driessen, and Ötker-Robe 2009 for a discussion of interlinkages and the exposure of parent banks. 98 TURMOIL AT TWENTY FIGURE 2.6 International claims, end-2008 Exposures of "home" country Exposures of "host" country Estonia (21.4) Austria Latvia (25.7) (98.5) Lithuania (23.9) Sweden Hungary (77.8) (58.1) Croatia (33.8) Montenegro (2.1) Greece Bulgaria (21.8) (26.4) Slovenia (17.8) Romania (60.0) Belgium Slovak Rep. (25.4) (36.7) Serbia (11.2) Bosnia and Herzegovina (3.4) Switzerland Poland (93.7) (30.0) Albania (2.0) Czech Rep. (30.0) Netherlands Ukraine (23.6) (45.1) Macedonia, FYR (1.2) Turkey (82.6) Germany Kazakhstan (11.0) (134.5) Russia (128.4) Moldova (0.4) Italy Azerbaijan (2.6) (70.8) Georgia (0.6) Belarus (2.3) France Armenia (0.3) (44.4) Uzbekistan (0.5) United Turkmenistan (0.2) % of GDPa % of GDPc Kingdom Tajikistan (0.1) % of claims on ECAb % of claims on ECAd (20.2) Kyrgyz Rep. (0.0) 0 10 20 30 40 0 50 100 150 a. Claims of parent banks in any home (originating) country as a percent of their home country GDP. b. Claims of parent banks in any home (originating) country as a percent of total claims of all par- ent banks in Western Europe on the ECA region. c. Claims of all home country parent banks in any host (recipient) country as a percent of this coun- try's GDP. d. Claims of all home country parent banks in any host (recipient) country as a percent of total claims of all parent banks in Western Europe on the ECA region. Note: Numbers in parentheses are billions of euros. Source: BIS and WEO data, and authors' calculations. Figure 2.7 provides equivalent information for the concept of foreign claims. As an example, it is worth highlighting that international claims by Austrian banks account for 35 percent of Austria's GDP. What is widely cited instead is that Austria's exposure is 70 percent of its GDP--a number that is twice as large and that reflects the foreign claims on Austria (figure 2.7). But this mea- sure of exposure is somewhat misleading because, as noted earlier, it includes lending in local currency by Austria's subsidiaries in ECA countries. As pre- viously explained, these operations are most likely funded by local deposit- taking operations in the host country. Likewise, it could be that some of the foreign currency lending is also funded locally. In other words, the figure of 35 percent could overestimate Austria's exposure as well. Similarly, while HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 99 FIGURE 2.7 Foreign claims, end-2008 Exposures of "home" country Exposures of "host" country Estonia (25.4) Austria Latvia (29.1) (199.5) Croatia (57.4) Belgium Hungary (110) (100.6) Slovak Rep. (66.5) Lithuania (30.6) Sweden Czech Rep. (126.1) (75.2) Bulgaria (30.1) Slovenia (29.4) Greece Montenegro (2.5) (42.1) Bosnia and Herzegovina (8.2) Romania (87) Netherlands Poland (198.8) (79.1) Serbia (16.5) Albania (4.2) Switzerland Ukraine (33.9) (37.1) Macedonia, FYR (1.8) Turkey (113.3) Italy Russia (159.9) (146.4) Kazakhstan (11.5) Moldova (0.4) Germany Azerbaijan (2.6) (159.6) Georgia (0.7) Belarus (2.9) France Armenia (0.4) (108.1) Uzbekistan (0.5) United Turkmenistan (0.2) % of GDP % of GDP States Tajikistan (0.1) % of claims on ECA % of claims on ECA (34.1) Kyrgyz Rep. (0.0) 0 10 20 30 40 50 60 70 0 50 100 150 200 Note: Numbers in parentheses are billions of euros. The legends in this chart follow those described in figure 2.6. Source: BIS and WEO data, and authors' calculations. international claims account for 135 percent of Estonia's GDP, foreign claims are larger and amount to 180 percent of Estonia's GDP. What has been so far the rollover experience in ECA countries? The evi- dence suggests that the exposure, measured by international claims, has declined (table 2.4). Foreign banks' positions were the highest at the end of 2008 in only six countries, having peaked earlier elsewhere. There are excep- tions; for example, Bulgaria peaked in March 2009. More generally, most countries peaked in early 2008, with a decline in exposure after the collapse of Lehman. The countries where parent banks had committed earlier to main- taining exposure in IMF-supported programs, as in Hungary and Latvia, had stable international claims at end-2008, but exposure in Ukraine and (less so) in Serbia, also a country participating in the Vienna initiative), has declined. Even in countries with IMF-supported programs, there was some decline in the first quarter of 2009. And yet perhaps the most striking result is the decline in the Russian Federation, Croatia, Turkey, Kazakhstan, and the 100 TURMOIL AT TWENTY TABLE 2.4 International claims, by country, 2000­09 (billions of euros) Change 2000 2008 2008 2008 2009 from Country Q4 Peak Q2 Q3 Q4 Q1 peak Albania 0.2 2.2 2009 Q1 1.9 2.0 2.0 2.2 0.0 Armenia 0.1 0.4 2009 Q1 0.2 0.2 0.3 0.4 0.0 Azerbaijan 0.3 2.9 2008 Q3 2.2 2.9 2.6 2.6 ­0.3 Belarus 0.3 2.3 2008 Q4 2.0 2.2 2.3 2.2 ­0.1 Bosnia and Herzegovina 0.2 4.7 2008 Q1 4.3 4.4 3.4 3.6 ­1.1 Bulgaria 1.5 22.2 2009 Q1 20.7 21.9 21.7 22.2 0.0 Croatia 7.8 42.1 2008 Q1 34.9 34.1 33.8 33.8 ­8.3 Czech Rep. 12.3 33.1 2008 Q2 33.1 32.8 29.8 28.6 ­4.6 Estonia 3.1 21.6 2008 Q3 21.0 21.6 18.7 19.3 ­2.4 Georgia 0.2 0.6 2008 Q4 0.6 0.6 0.6 0.6 0.0 Hungary 18.1 76.3 2008 Q4 67.6 73.3 76.3 74.2 ­2.0 Kazakhstan 0.8 15.0 2007 Q4 12.0 11.7 10.9 11.3 ­3.8 Kyrgyz Rep. 0.2 0.4 2007 Q2 0.0 0.1 0.0 0.0 ­0.4 Latvia 1.0 26.0 2008 Q3 25.0 26.0 22.4 21.4 ­4.7 Lithuania 2.1 23.3 2008 Q3 22.3 23.3 21.3 22.5 ­0.8 Macedonia, FYR 0.2 1.2 2008 Q4 0.9 1.1 1.2 1.1 0.0 Moldova 0.1 0.5 2008 Q3 0.4 0.5 0.4 0.4 0.0 Montenegro .. 2.2 2008 Q3 1.9 2.2 2.1 2.0 ­0.3 Poland 26.2 93.9 2008 Q4 81.0 88.6 93.9 88.5 ­5.4 Romania 3.3 62.4 2007 Q4 54.2 58.0 60.0 58.9 ­3.5 Russian Federation 42.6 143.6 2008 Q3 133.9 143.6 129.9 131.7 ­11.9 Serbia .. 11.7 2008 Q1 10.9 11.2 11.3 10.7 ­1.0 Slovak Rep. 4.1 25.6 2008 Q4 19.3 22.3 25.6 11.5a ­14.1 Slovenia 4.0 18.6 2008 Q1 18.4 18.4 17.6 17.7 ­0.9 Tajikistan 0.1 0.2 2001 Q2 0.1 0.1 0.0 0.0 ­0.1 Turkey 46.9 87.1 2008 Q3 81.3 87.1 81.7 84.7 ­2.4 Turkmenistan 1.7 1.7 2000 Q4 0.2 0.2 0.2 0.2 ­1.5 Ukraine 0.9 25.6 2008 Q3 24.3 25.6 24.0 22.7 ­2.9 Uzbekistan 1.9 2.0 2001 Q2 0.5 0.6 0.5 0.5 ­1.5 .. is not available. a. The Slovak Republic's decline in international claims might in part reflect the adoption of the euro on January 1, 2009. This means that euro-denominated local claims by subsidiaries were excluded from international claims and treated as local-currency-denominated claims. However, it is worth noting that foreign claims also recorded a decline. Source: BIS and authors' calculations. Czech and Slovak Republics. Yet, on a more positive note, the sharp decline in foreign exchange liabilities seen in South Korea has so far not occurred in ECA. Indeed, figures 2.8a and 2.8b present the profi le of foreign exchange HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 101 FIGURE 2.8A Foreign exchange liabilities, Republic of Korea and financially integrated ECA countries (monthly data; t for the peak of each country--during 1997 for South Korea, 2007 for Kazakhstan, and 2008 for other ECA countries) Index (100 = peak = t) Kazakhstan 100 Lithuania Serbia 90 Estonia 80 Turkey 70 Czech Rep. 60 Korea Bulgaria 50 t­10 t­8 t­6 t­4 t­2 t t+2 t+4 t+6 t+8 t+10 Note: The crisis year is represented by t, and t­i and t+i denote the ith month before or after the crisis. Source: IMF, International Financial Statistics. FIGURE 2.8B Foreign exchange liabilities, Republic of Korea and financially integrated ECA countries (monthly data; t for the peak of each country--during 1997 for South Korea and 2008 for other ECA countries) Index (100 = peak = t) 100 Poland Croatia Hungary 90 Romania 80 Latvia Ukraine 70 Russia 60 Korea 50 t­10 t­8 t­6 t­4 t­2 t t+2 t+4 t+6 t+8 t+10 Note: The crisis year is represented by t, and t­i and t+i denote the ith month before or after the crisis. Source: IMF, International Financial Statistics. 102 TURMOIL AT TWENTY liabilities centered on the peak for the East Asian countries and a select group of financially integrated countries. The evidence so far is that the declines in foreign exchange liabilities for South Korea were substantially sharper than those observed in most ECA countries. Time to redeem promises made in Vienna: determinants of rollover Rollover rates will dictate the economic adjustment of ECA countries but should not be seen as an aggregate concept. Instead, rollover rates depend on the eco- nomic prospects of debtors and on the factors affecting decisions of individual creditors. In the case of ECA countries, a significant share of rollover risk lies in the banking system, and the different funding sources for banks identified in table 2.5 all have different stability features. This section looks in greater detail at the liability structure of the banking system and the particularities of the asset structure of banks that might affect their rollover decisions. The decision by a parent bank to roll over credit lines to a subsidiary depends on the expected return or loss at the time of the decision, the uncer- tainty about future returns, and the willingness to bear this uncertainty. If the investment of the parent is a small share of total funding and the subsidiary is liquid, the parent may choose not to roll over because of the certainty that the subsidiary can accommodate this with little consequence. The alternative of waiting would expose the parent to uncertainty about future payoffs, espe- cially if conditions deteriorate and the liquidity and solvency of the subsidiary are called into question. If, by contrast, the subsidiary is not liquid but well capitalized, the parent bank may choose to roll over its credit lines in view of the bankruptcy costs that are likely to be incurred otherwise. But considerations of solvency and liquidity are not all. If the parent bank accounts for a large share not only of the subsidiary's funding but also of economy-wide funding, it is likely that withdrawal would lead to a country TABLE 2.5 Stability of funding sources Source Degree of stability Residents Retail deposits High/medium Wholesale funding Low Nonresidents Retail deposits Medium/low Wholesale funding Low Parent bank funding Medium Equity High HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 103 default. Given the systemic impact of the parent bank, its withdrawal will also affect recovery rates and have a substantive impact on macroeconomic performance. In contrast, the presence of many investors and the inherent maturity mis- match between bank assets and liabilities also affect wholesale funding and nonresident deposits during periods of turmoil. Investors fear who might exit first. Debt might not be rolled over because less informed investors interpret a collapse in a bank's share price as signaling that shareholders who control the bank--and thus should be knowledgeable--have lost confidence in the bank's earning prospects. Small and uninformed depositors are particularly prone to runs, though typically with a lag. While this has motivated deposit insurance schemes, retail runs may still occur if the schemes are not seen as credible. There can also be some special situations: for example, bank runs can occur when retail depositors wish to transfer their resources from domestically owned banks to foreign-owned banks perceived as more stable.17 Not all funding sources are equally stable . . . The relative stability of the sources of funding for banks can be assessed by looking at the evolution of their liability structures.18 The conclusions from such assessment are revealing. Shortages of liquidity brought on by the tur- moil in global fi nancial markets have substantially restricted the availabil- ity of wholesale funding. In Bulgaria, Croatia, Estonia, FYR Macedonia, and Lithuania, wholesale funding has declined (figure 2.9). In contrast, domestic deposits held up much better. Even though they declined in the fourth quar- ter of 2008 and the fi rst quarter of 2009, they were still higher at the end of the first quarter of 2009 than at the beginning of 2008 (figure 2.10).19 Inter- estingly, parent bank funding has held well, reflecting the success so far of international coordination efforts (figure 2.11). In fact, parent bank funding has been used to offset declines in other types of funding. And yet it is fair 17. Th is happened in Latvia in 2008, when deposits were transferred from domestically-owned banks to the foreign-owned subsidiaries of Swedish banks following announcements by the Swedish govern- ment of support for those banks. 18. The discussion is based on data that was requested from central bank and banking supervision authorities for 12 countries in the region. The data distinguish between alternative funding sources and the maturity, sectoral, and currency composition of the loan portfolio. The authors wish to thank these authorities for this information. 19. The higher stability of retail funding compared with wholesale funding in such cases echoes the case of Northern Rock in the United Kingdom, where wholesale funding rollover needs dried up one month before queues of depositors formed outside the bank's own branches. 104 TURMOIL AT TWENTY FIGURE 2.9 Wholesale funding, by country, 2007­09 Index (2008 Q1 = 100) 175 Poland 150 Kazakhstan 125 Latvia Bulgaria Hungary Macedonia, FYR 100 Croatia Estonia Romania Lithuania 75 2007 Q4 2008 Q1 2008 Q2 2008 Q3 2008 Q4 2009 Q1 Source: Central banks and authors' calculations. FIGURE 2.10 Resident retail deposits, by country, 2007­09 Index (2008 Q1 = 100) 140 Kazakhstan 130 120 Ukraine Poland Romania 110 Latvia Bulgaria Macedonia, FYR Croatia 100 Hungary Serbia Lithuania Estonia 90 2007 Q4 2008 Q1 2008 Q2 2008 Q3 2008 Q4 2009 Q1 Source: Central banks and authors' calculations. to say that the growth in parent funding in the past few years is no longer present. The exception is Kazakhstan, though here it has been declining since 2007 Q4 (figure 2.11). Note that foreign ownership and funding structures are quite independent of each other. Some parent banks rely heavily on domestic deposit-taking operation by their subsidiaries, even though the banking sys- tem is foreign owned (box 2.5). HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 105 FIGURE 2.11 Parent bank funding, by country, 2007­09 Index (2008 Q1 = 100) 200 Bulgaria 175 Hungary Poland 150 Macedonia, FYR Lithuania 125 Croatia Latvia 100 Estonia Kazakhstan 75 2007 Q4 2008 Q1 2008 Q2 2008 Q3 2008 Q4 2009 Q1 Source: Central banks and authors' calculations. . . . while some components of the loan portfolio fasten "golden handcuffs" on lenders The stability of parent bank funding and thus the prospect for a rollover depend also on the composition of assets on the balance sheets of subsidiaries. Indeed, it may have been optimal for some parent banks to secure their claims immediately, particularly where liquidity positions were strong, as in Serbia, where 35 percent of bank assets were held in cash, as opposed to in Estonia and Lithuania, where the corresponding ratio was 10 percent (figure 2.12). In this context, low liquidity buffers in Estonia and Lithuania may have encour- aged dominant parent banks to maintain funding in these countries. For instance, one-day funding of parent banks in Lithuania accounts for 25 per- cent of available cash of local banks. So, nonrenewals would strain an already circumscribed liquidity position. More generally, bank loan portfolios in ECA countries frequently com- prise nonmarketable loans with very long maturities. Of all loans, 60 per- cent have a contract maturity of more than five years in Lithuania and more than 70 percent in Estonia, reflecting the dominance of long-term mortgage lending. As noted earlier, the average maturity of mortgage loans is 26 years in some Baltic states, and these make up close to 50 percent of the loan port- folio. In contrast, loans with a maturity of more than five years account for only 30 percent of loans in Serbia and Ukraine (figure 2.13). Indeed, while credit to households accounts for about 50 percent of total assets in Estonia 106 TURMOIL AT TWENTY BOX 2.5 Foreign ownership and funding sources Ownership has little to do with the banking system's funding structure. An examination of the liability structure of the consolidated banking system in 12 fi nancially integrated ECA countries shows considerable variation. In Lithu- ania, where 90 percent of the banking sector is foreign owned, liabilities are dominated by parent bank credit lines and domestic deposits (box figures 1 and 2). In Estonia, where the banking sector is almost fully foreign owned, the dependence on parent bank funding is similar. But ownership does not necessarily determine the relative importance of dif- ferent funding sources. In FYR Macedonia, where 85 percent of the banking sec- tor is foreign owned, the liability side of the balance sheet is less diverse, and funding by parents is substantially less important. Croatia, where the banking sector is more than 90 percent foreign owned, presents a more mixed case--it has similar shares of parent and wholesale funding and a large resident deposit base. BOX FIGURE 1 Liability structure of banking sector balance sheets, by country, March 2009 Percent Equity Foreign depositors Wholesale Parent bank Local (nonbank) funding debt funding depositors 100 75 50 25 0 ria tia nia n a ia YR nd nia ia ine y ar sta tvi an rb ,F lga oa la ra ng to ma La Se kh hu Po nia Uk Cr Es Bu Hu Ro za Lit do Ka ce Ma Note: For Serbia and Ukraine there is no distinction between parent funding and wholesale funding in the data provided by the central bank; the figure combines both sources and refers to it as wholesale funding. Source: Central banks and authors' calculations. BOX FIGURE 2 Foreign ownership of banks, by country, 2007 Percent of total assets 100 75 50 25 0 Es p. Mo ulga . Ar ep. erb . d Alb ia rze nia Lit vina Ge ia Cr ia Ma Ro atia ni ia ec R nt ria Po ro Se d Hu bia rgy ia U ia za ne Sl tan Mo nia Ru Bela a ian s Ta ijan en n an y B ep Az Fed ss ru ar v rkm sta lan Cz a, FY Re n an g do n Ky Latv n eg ldo Ka krai ist or R zR to He a ma ng me e r s a o go ov en kh Tu jiki hu ak h ov Sl ce an ia sn Bo Source: EBRD 2007. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 107 FIGURE 2.12 Ratio of liquid assets to total assets, by country, March 2009 Percent 40 30 20 10 0 ia tia a n ia YR d ia ia e y a ar ni ta an in ar tv an rb ni oa ,F to ra ng hs La ua l lg Se Po m Cr Es Uk ia Hu Bu k Ro th za on Li Ka ed ac M Source: Central banks and authors' calculations. FIGURE 2.13 Loans with a maturity of five years or more, by country, 2007­09 Percent 80 Estonia Croatia Macedonia, FYR Hungary Latvia 60 Lithuania Poland Bulgaria Romania 40 Kazakhstan Ukraine Serbia 20 2007 Q4 2008 Q1 2008 Q2 2008 Q3 2008 Q4 2009 Q1 Source: Central banks and authors' calculations. and Lithuania, it accounts for a much lower share--about 35 percent--in FYR Macedonia, Kazakhstan, and Serbia (figure 2.14)--and in many coun- tries the vast majority of the lending was in foreign currency (figure 2.15). Th is illiquid asset position associated with long-term mortgage lending may 108 TURMOIL AT TWENTY FIGURE 2.14 Sectoral composition of loans, by country, March 2009 Percent Corporate sector Households Banking sector Public sector 100 75 50 25 0 ria tia ia an ia R nd ia ia e y ia ar in FY n tv an rb st an la a oa to ra ng La lg Se Po m , kh u Cr Es Uk ia Hu Bu Ro th za on Li Ka ed ac M Source: Central banks and authors' calculations. FIGURE 2.15 Currency composition, by country, March 2009 Percent Foreign exchange Domestic 100 75 50 25 0 a ia tia a an ia nd ia ia e y a R ar ni in ar tv an rb ni FY st la oa to ra ng La ua lg Se Po m kh Cr Es Uk a, Hu Bu Ro th za i on Li Ka ed ac M a. Data for foreign exchange include foreign exchange loans and loans in denars with foreign exchange adjustment clauses. Source: Central banks and authors' calculations. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 109 underpin the stability of debt exposures of large parent banks in some ECA countries. In sum, the potential for debt investors to realize assets at lit- tle or no cost is lower in the Baltic states than in Serbia and Ukraine. The long maturities in Estonia and Lithuania will encourage parent banks to maintain exposure, while the comfortable liquidity position in Serbia might provide an incentive to cut back on exposure in the knowledge that liquid and marketable assets are available to pay debt investors back without much distress. Does "home country" concentration matter? . . . The extent to which a banking system relies on parent banks from a few countries, or many, is also relevant in assessing the stability of parent fund- ing. The underlying assumption is that higher concentration in one home country is likely to result in more stable links, because withdrawing sup- port to the subsidiary could prove costly. Indeed, it is even likely in such cases that home country supervisors would look at home country exposure, as opposed to individual bank exposure: the Nordic countries' willingness to participate in the IMF-supported program in Latvia with substantial offi- cial fi nancing (1.8 billion, in addition to the support provided for Swedish bank recapitalizations) is consistent with this hypothesis. Moreover, if a bank stays involved, it can capture high profits once the economy returns to some degree of normalcy owing to its likely importance in market share. Indeed, though home country concentration is not equivalent to monopoly power, there is a correlation. How can the degree of home country concentration be assessed? Through an index of home country concentration of foreign banks' international claims. A large index number indicates high concentration and thus, poten- tially, higher funding stability (table 2.6).20 The Baltic states and some West- ern Balkan countries had high concentration scores. In contrast, Kazakhstan, the Russian Federation, Turkey, and Ukraine had low concentration scores, while countries in Central Europe such as the Czech Republic, Poland, and the Slovak Republic, together with Bulgaria, were in between. In general, the concentration scores for South Korea, Malaysia, Thailand, Philippines, and Indonesia are lower than those observed in some ECA countries, though by no means all. 20. It is akin to the Herfi ndahl index and is defi ned as the sum of the squares of the market shares. Th is is summed over all international claims of foreign banks with claims in a country. The index lies between zero and one. Increases in the index indicate a decrease in competition and an increase of market power originating in one country, while decreases indicate the opposite. 110 TURMOIL AT TWENTY TABLE 2.6 Index of home country concentration of parent bank exposure, international claims Country 2000 Q4 2008 Q4 2009 Q1 Europe and Central Asia Albania 0.35 0.28 0.35 Armenia 0.72 0.33 0.24 Azerbaijan 0.29 0.17 0.18 Belarus 0.50 0.29 0.28 Bosnia & Herzegovina 0.55 0.39 0.40 Bulgaria 0.14 0.16 0.16 Croatia 0.38 0.26 0.26 Czech Rep. 0.28 0.19 0.20 Estonia 0.76 0.83 0.84 Georgia 0.45 0.27 0.29 Hungary 0.14 0.21 0.21 Kazakhstan 0.15 0.11 0.10 Kyrgyz Rep. 0.27 0.25 0.24 Latvia 0.52 0.58 0.58 Lithuania 0.41 0.64 0.64 Macedonia, FYR 0.27 0.46 0.50 Moldova 0.26 0.47 0.44 Montenegro .. 0.59 0.56 Poland 0.17 0.16 0.15 Romania 0.21 0.18 0.18 Russian Federation 0.13 0.12 0.11 Serbia .. 0.18 0.18 Serbiaa 0.18 .. .. Slovak Rep. 0.14 0.19 0.22 Slovenia 0.31 0.28 0.27 Tajikistan 0.46 0.32 0.33 Turkey 0.13 0.12 0.12 Turkmenistan 0.28 0.71 0.70 Ukraine 0.17 0.14 0.15 Uzbekistan 0.24 0.24 0.24 Mean 0.32 0.31 0.31 Median 0.28 0.26 0.24 East Asia Peak date for claims Index Indonesia 1997 Q4 0.27 Korea, Rep. of 1997 Q2 0.21 Malaysia 1997 Q2 0.30 Philippines 1997 Q4 0.14 Thailand 1996 Q4 0.44 Mean 0.27 Median 0.27 .. is not available. a. Includes Montenegro. Source: BIS and authors' calculations. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 111 TABLE 2.7 Importance of lending in foreign exchange among parent-subsidiary banks (foreign exchange lending by parent-subsidiary bank as a percentage of total group lending) Country 2005 Q1 Peak 2008 Q4 2009 Q1 Belarus 1.00 1.00 2005 Q1 0.79 0.82 Bulgaria 0.68 0.76 2006 Q1 0.72 0.74 Croatia 0.65 0.65 2005 Q1 0.59 0.60 Czech Rep. 0.26 0.31 2006 Q2 0.25 0.24 Estonia 0.83 0.85 2006 Q1 0.82 0.85 Hungary 0.67 0.71 2009 Q1 0.70 0.71 Kazakhstan 0.89 0.96 2009 Q1 0.95 0.96 Latvia 0.79 0.90 2008 Q3 0.87 0.88 Lithuania 0.77 0.84 2005 Q4 0.76 0.79 Macedonia, FYR 0.77 0.77 2005 Q1 0.66 0.68 Montenegro .. 0.85 2008 Q4 0.85 0.78 Poland 0.41 0.53 2006 Q1 0.48 0.48 Romania 0.73 0.76 2005 Q2 0.69 0.68 Russian Federation 0.91 0.93 2006 Q1 0.79 0.82 Serbia .. 0.68 2008 Q4 0.68 0.67 Slovak Rep. 0.42 0.42 2005 Q1 0.38 0.23 Turkey 0.93 0.93 2005 Q1 0.73 0.74 Ukraine 0.89 0.89 2005 Q1 0.70 0.69 Average 0.72 0.76 0.69 0.69 Median 0.77 0.80 0.71 0.72 .. is not available. Source: BIS and IFS, and authors' calculations. . . . yes, but demand matters as well Home country concentration is a supply-side factor associated with maintain- ing exposure to host countries, but the demand for international claims in the host country is also important. This is revealed to some extent in a bank's foreign exchange lending. International claims as a share of total parent-sub- sidiary lending had remained broadly constant at the end of March 2009. On average, two-thirds of lending in the parent-subsidiary group was in foreign currency and one-third in local currency (table 2.7). The average masks substantial variation. International claims accounted for only 24 percent of total parent-subsidiary lending in the Czech Repub- lic and 23 percent in the Slovak Republic--but 83 percent in the Baltic states. As noted earlier, foreign ownership does not determine lending. The 112 TURMOIL AT TWENTY determinants must be sought instead in the relative attractiveness of borrow- ing in foreign exchange over borrowing in local currency from the perspective of the borrower (box 1.3). Three concluding arguments--three caveats Some deleveraging due to pressures in advanced country fi nancial markets is likely. It is also to a degree needed as rollover gives way to restructuring of bank, household, and corporate debt. But collective action involving regional and international fi nancial institutions, supranational authorities, parent banks and their subsidiaries, and home and host governments--and ECA's unique rollover determinants--can keep this process orderly. Although parent bank funding has for the most part continued, external financing is unlikely to recover its prominent role. Still, rollover in ECA is more probable than it was in previous capital account crises. These factors support this assessment. First, high home country concentration of parent bank exposure to the host country is likely to create incentives to remain engaged in some countries- --though not all ECA countries face the same circumstances. Indeed, where home concentration is low, the incentives to sustain exposure might not be as strong. Estonia and Kazakhstan illustrate opposite cases of the broadly posi- tive association between home country concentration and ease of rollover. But it must be recognized that there are exceptions, such as Ukraine, where home country concentration is low, but rollover has been reasonably secured. Second, the more illiquid and nonmarketable are the assets held by subsid- iary banks, the less likely is the withdrawal of those resources at short notice. Indeed, doing so is in many ways self-destructive, equivalent to a fire sale of assets (housing prices have already halved in some crisis countries). Thus, the commitment of parent banks also depends on their asset structure, particu- larly the maturity structure and sectoral composition of the loan portfolio. Third, it is likely that renewing external financing in wholesale funding, syndicated loans, or nonresident deposits will be more difficult--and has so far been the case. In fact, some banks are facing severe difficulties. This is a systemic threat only in some countries, though individual banks in many may falter. These three arguments, which support a gradual and orderly reduction of exposure, are subject to three caveats. First, subsidiaries are strong partners only to the extent their parents are financially healthy. A faster deleveraging in asset classes viewed as high risk could occur if parent banks face difficulties at home. Some ECA countries--though clearly not all--might be less attrac- tive than less vulnerable emerging markets around the world. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 113 Second, parent banks seem committed to gradually recapitalizing their subsidiaries, assuming that nonperforming loans accumulate gradually. But the foreign exchange exposure of subsidiaries to unhedged borrowers could accelerate the process of recognizing losses in the event of a currency crisis. In such cases, parent bank commitment to their subsidiaries is less clear, and little guidance can be offered from past banking crises. Third, collective action and the rapid response by the international com- munity has so far prevented contagion, but this also depends on continu- ing success in restoring countries to sustainable development paths. The risk remains that the adjustment is so large that it could weaken political resolve and popular support. This can be eased only by rollover rates that are higher than were anticipated at the end of 2008 when the crisis first struck. It may need to be complemented by large official financing flows--and for a longer period--to slow down and soften the pain of the adjustment. 114 TURMOIL AT TWENTY Annex 2.1 Description of the Bank for International Settlements (BIS) Dataset The banking statistics published by the BIS can be divided into two sets: loca- tional (tables 1 to 8 in the statistical annex of the BIS Quarterly Review) and consolidated (table 9 of the same annex) statistics on claims and liabilities. The locational dataset is based on the location of each banking unit. Although this classification is similar to that in national statistics, such as balance of payments, the data are on a gross basis. Therefore, the locational statistics contain the positions against their own affi liates in other countries. The con- solidated dataset is based on the nationality of the head office; in particular, it has all inter-bank positions (businesses between different units of the same bank group) on a net basis. These statistics show the risk exposures of bank- ing systems of countries where the head offices are located and are provided in two different ways: immediate borrower (contractual) basis (tables 9a and 9b of the dataset) and ultimate risk basis (tables 9c and 9d of the dataset). Annex table 2.1.1 summarizes the information on "claims" available in the BIS banking data. It should be noted that the sectoral breakdown is defined differently in each of the two sets of statistics, as shown in annex table 2.1.2. The difference between these stems from where the default risk reside. ANNEX TABLE 2.1.1 Comparison of information on claims in BIS data Consolidated (net) Locational Immediate Ultimate (gross) borrower risk Type of claims Foreign claims (FC = IC + LL) International claims (IC = XB + LF) Cross-border claims (XB) Local claims in foreign currency (LF) Local claims in local currency (LL) Composition of claims (IC) (FC) By currency By maturity By sector Starting date of quarterly series Dec. 1977 Dec. 1999* Mar. 2005 Number of reporting countries (Dec. 2008) 41 30 24 *Semiannual data available from December 1983 onwards. Source: BIS 2003, 2005, and 2009; Maechler and Ong 2009. HOW MUCH ADJUSTMENT? HOW MUCH FINANCING? 115 ANNEX TABLE 2.1.2 Sectoral breakdown in BIS data Claims on Banks Nonbanks Public sector Locational · Commercial banks · General · Monetary governments authorities · Public corporations · International · Nonbank private organizations sector Consolidated · Commercial banks · Public corporations · General · Nonbank private governments sector · Monetary authorities · International organizations Source: BIS 2008a,b. Numbers on immediate borrower basis capture claims to the country where the original risk lies. This is based on the residence (nationality) of the head office of reporting banks and their claims to nonresidents. Claims are on ulti- mate risk basis when it takes into account banks' own internal risk manage- ment systems and the difference in risk exposures brought by risk mitigation with, for instance, collateral and credit derivatives. Ultimate risk claims are the ones extended to the country where the final risk lies or the country where the guarantor of a claim resides. In other words, claims are allocated to the country where the counterparty that has the fi nal responsibility of repaying them is located. The difference between the two series is captured by net risk transfers (differences between inward and outward risk transfers). The posi- tive transfers indicate that banks' risk exposure to the country is increased and, therefore, the country is a holder of an ultimate responsibility for the repayment of the claim. The country risk exposures can be more appropriately captured by the ultimate risk basis, but the series is available only from 2005, and fewer countries report on this basis. 116 TURMOIL AT TWENTY CHAPTER 3 Restructuring bank, corporate, and household debt Nonperforming loans picked up during 2009 in many ECA countries--up to 20 percent of all loans--particularly construction and mortgage lending. With a few exceptions, the indebtedness of nonfi nancial corporates is not higher than that in comparable countries. A distinctive feature of ECA's crisis is household debt, which in some new member states of the EU and in Croatia has reached levels comparable to those of Ireland, Spain, and Portugal in the late 1990s. Household debt is concentrated in the upper-income quintiles. Questions · Why is it imperative that ECA countries recognize nonperforming loans without much delay? · What can be learned from past banking crises about the best way to restruc- ture corporate and household debt? · What are the main challenges for the future of finance in the region? Findings · The outlook for ECA's growth is weak. So, there is urgency to recognizing and restructuring nonperforming loans to ensure that the region's economic recovery is not held back by weak loan portfolios. · Credit losses (even in a worsening scenario) should be manageable and could range from 7 to 19 percent of country GDP. · Past crises offer guidance on how to conduct corporate and household debt restructuring. Governments should consider establishing voluntary out-of- court workout mechanisms to avoid overwhelming the judicial system with a large number of debt restructurings. · Capital adequacy ratios for transition and developing countries should take into account the greater volatility of the shocks affecting them--and thus err on the side of financial stability, even at the expense of some loss in financial intermediation. Financial institutions, corporations, and households are facing difficulties ser- vicing their debt contracts on the original schedules.1 Banks are seeing their 1. Laeven and Valencia 2008. 117 loan portfolio deteriorate as firms are hit by the collapse of demand and unem- ployment rises. Two of ECA's hardest hit countries (Latvia and Ukraine) had nonperforming loans (defined as substandard, doubtful, and lost) at around 20 percent of all loans at the end of the first quarter of 2009. Although the lower income countries of the CIS have not been directly affected by a sudden stop in capital flows, their banking sectors are also hurting: nonperforming loans stood at over 6 percent in the Kyrgyz Republic and nearly 14 percent in Georgia in early 2009. By comparison, nonperforming loans in earlier capi- tal account crises peaked at 32 percent of total loans in Indonesia, 35 percent in Korea, 30 percent in Malaysia, and 23 percent in Thailand during the East Asian crisis--and at 20 percent in Argentina, 40 percent in Ecuador, and 36 percent in Uruguay during their crisis earlier this century.2 Nonperforming loans are particularly high in trade and construction. This is due to the devaluation of the local currency and a plunging real estate mar- ket, and the impact of these developments on mortgage loans. For instance, 25 percent of all mortgages in Ukraine were in default in mid-2009, up from 1 percent in late 2008. In Latvia, the proportion of loans to companies that have real estate as their business and are more than 90 days overdue was nearly 30 percent in March 2009. On the liability side of the balance sheet, banks with high loan-to-deposit ratios have seen their funding dry up. Th is is particularly evident in the wholesale and interbank markets (chapter 2). And in most countries, banks have faced a withdrawal of resident deposits because of a loss of confidence. In Ukraine, for example, there was a deposit flight of 25 percent from hryvnia deposits and 18 percent of foreign exchange deposits between October 2008 and April 2009. In other countries, the shift has been from local to foreign exchange deposits. Financial systems need to be fixed As in previous capital account crises, restoring the financial system to health is important because both consumption and investment depend on credit. Policy responses to systemic banking crises typically distinguish between a containment phase (primarily related to the liability side of banks' balance sheet), where the priority is to restore confidence by depositors and investors in the banking system, and a resolution phase (primarily related to the asset side of banks' balance sheets) focusing on the financial restructuring of banks 2. The increases observed in past capital account crises reflect increased and widespread corporate distress, as well as the introduction of better loan classification standards for fi nancial institutions. 118 TURMOIL AT TWENTY and their borrowers. The distinction is not hard and fast, since steps to con- tain the crisis shape the landscape for formulating resolution policies. . . . particularly when growth is expected to be sluggish Restoring the financial system to health is more difficult when the profi le of recovery and growth is likely to be sluggish. The expectation of weak growth is based on three considerations. · First, recessions associated with fi nancial crises that are synchronized across the world, as they are today, have had declines in real GDP from the previous peak that are deeper (5 percent), take longer to arrive at the trough from the previous peak (almost 8 quarters), and require more time for real GDP to recover from the new trough to the previous peak (about 7 quarters). This is the case both when compared with recessions caused by other types of shocks (fiscal policy contractions, monetary policy tighten- ing, oil shocks) or those following financial crises that are not synchronized across the world (figures 3.1a and 3.1b). The reasons are simple. Net trade is much weaker and exports are thus less likely to be a source of recovery. In addition, credit growth and the increase in prices of assets, such as real estate and equities, during the expansion preceding a fi nancial crisis are higher than during other expansions, and household savings out of dis- posable income in the pre-crisis period is lower. These factors necessitate a large increase in savings rates after the crisis strikes, which in turn implies a much deeper decline in private consumption. Indeed, private consump- tion typically grows more slowly than during other recoveries, while pri- vate investment actually continues to decline even after passing the reces- sion's trough. FIGURE 3.1A Average duration and change in output of recessions, by type Duration (quarters) Output loss (percent from peak) All Financial crises Highly synchronized financial crises 0 1 2 3 4 5 6 7 8 Source: IMF World Economic Outlook, April 2009. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 119 FIGURE 3.1B Average duration and change in output of recoveries, by type Output gain after four quarters Time until recovery to previous peak (percent from trough) (quarters) All Financial crises Highly synchronized financial crises 0 1 2 3 4 5 6 7 8 Source: IMF World Economic Outlook, April 2009. · Second, writedowns remain to be fully recognized in the Western Euro- pean financial sector, and leverage ratios in the banking sector are high. The IMF estimates that financial institutions in Western Europe (including the United Kingdom) will face around $1.5 trillion in writedowns through 2010, but just around 40 percent had been recognized by the end of the second quarter of 2009.3 The leverage of Western European banks prior to the crisis was also too high. A reduction of leverage to reach a ratio of tangible common equity to tangible assets of 4 percent would require a capital injection of around $550 billion for Western European (including United Kingdom) banks.4 However, high fiscal deficits in Western Europe may constrain governments' ability to do so rapidly. · Th ird, growth in ECA was underpinned by abundant liquidity that is unlikely to continue at the pace in the pre-crisis years. This implies inter alia a possible slowing of convergence to Western European living standards. Since recognizing losses in Western European banks will require substan- tial deleveraging, which would typically include some retrenchment from subsidiaries, the growth of credit to some ECA countries is likely to be more limited--so too the excess of imports over exports. In this context, if financ- ing is not rapidly restored, this could compromise growth. There is room to run public imbalances without crowding out private activity--public debt is low and the private sector needs to adjust. But public imbalances cannot be maintained for long without raising sustainability issues (box 3.1). 3. IMF 2009c. 4. Tangible common equity is total equity less preferred shares and intangible assets; tangible assets are total assets less intangible assets. 120 TURMOIL AT TWENTY BOX 3.1 ECA's growth prospects--green shoots? Maybe. High growth rates? Unlikely How can one assess the region's growth prospects? Quantifying potential growth is always difficult, more so in a region where the role of unusually high external financing in growth is not well understood. Moreover, the full impact of this crisis on GDP in 2009 and 2010 is still playing out. Notwithstanding the caveats, a crude estimate of potential growth can be derived by using fi l- tered series, such as Hodrick-Prescott techniques. Th is is far from a theory of growth, but it highlights the unusual features of growth in the region over the past five years. Specifically, the box figure presents two estimates in addition to the actual GDP growth rate for 2008.1 The first estimate is based on growth rates interpolated from country-specific data for 1995­2007. The second esti- mate is based on growth rates interpolated from 1995­2002. Among the CEE- SEE group, there is a 3 percentage point difference between the interpolated growth rate derived using HP fi ltered series based on different sample periods. Among the CIS, this difference is even higher--slightly more than 4 percentage points. Clearly, the concern that growth rates might not rapidly return to past levels is valid. The potential effects on fiscal sustainability remains to be seen, but they could obviously be important as the revenue performance of recent years benefited from the unexpectedly high growth rates in real GDP. BOX FIGURE 1 Potential growth (in percent) Percent 15 Azerbaijan 10 5 0 ­5 Actual HP HP Actual HP HP 1995­2007 1995­2002 1995­2007 1995­2002 CEE­SEE CIS Note 1. The full distribution for CEE and SEE countries (including Turkey) and CIS coun- tries is represented through the use of box-and-whiskers charts. The whiskers rep- resent minimum and maximum values. The boxes represent the 25th and 75th per- centile. The line in the box is the median for the group. An observation beyond the whisker occurs when there are outliers in the data (more than 1.5 times the inter- quartile range away from the neighboring observation). Source: World Bank World Development Indicators and authors' calculations. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 121 Against this background, this chapter attempts to answer three questions. First, what can be learned from past banking crises about the best way to con- tain and resolve problem banks? Second, why is it important to recognize non- performing loans without much delay, and what is the best way to restructure corporate and household debts? Th ird, what are the more important aspects of bank regulation and supervision that need to be reformed to absorb the les- sons from the present crisis and make banking sectors in ECA countries less vulnerable to future crises? Issues include strengthening capital requirements and cross-border banking supervision. For the lenders: bank restructuring Containment: restoring confidence The containment phase is intended to restore public confidence in the bank- ing system and limit its adverse effects on the real sector. Numerous instru- ments are available to the authorities, most targeted at stabilizing the liability side of banks' balance sheets. These policy measures include: · Liquidity support in local currency. Liquidity support includes a reduction in reserve requirements, access to overdraft facilities, and the use of repos and reverse repos against broader types of collateral. But this needs to be done under closely monitored conditions to prevent recipient banks from shifting assets abroad. And liquidity support should not be extended to banks that are reportedly insolvent. Both have occurred in some middle-income CIS countries. For the poorer countries of the former Soviet Union, liquidity injections were put in place in Georgia and Tajikistan, reserve requirements were reduced in Georgia, and deposit insurance coverage was extended in the Kyrgyz Republic. Monetary policy will need to stand ready to steril- ize excess liquidity where liquidity support put pressure on the exchange rate, though the risk of currency depreciation has been reduced with global monetary easing and associated declines in world interest rates. · Liquidity support in foreign currency. The ability of the central bank to pro- vide liquidity support in foreign exchange is limited by the availability of reserves. Countries can benefit from temporary arrangements, such as a swap line to provide euro liquidity. For instance, this was made available to Estonia by the Swedish Riksbank, and to Poland by the IMF's approval of a flexible credit line, which is extended only to countries with a track record of sound macroeconomic management. Countries without access to such swap lines have opted for high-access IMF arrangements that require policy reform. 122 TURMOIL AT TWENTY · Government guarantees. Some countries have introduced guarantees for third-party funding of banks. For example, guarantees have been extended for interbank credits in Hungary and Latvia, bank-issued securities used to roll over or refinance domestic banks' funding needs in Hungary, and new debt issuance by banks in Slovenia. · Deposit insurance. Countries have raised maximum limits on bank depos- its covered by deposit insurance and increased deposit insurance premia-- the uncoordinated increase in deposit insurance in the eurozone in 2008 and its effects on the new member states of the European Union that do not belong to the eurozone were addressed later. The provision of liquidity support and deposit and government guarantees should be accompanied by intervention in banks deemed insolvent. This was the case, for example, with the second largest domestically owned bank in Latvia, where liquidity support was not enough to stop a bank run--and with 17 banks in Ukraine, where temporary administrators imposed a freeze on household deposits and a moratorium on repayment of liabilities to allay con- cerns about the banking system's soundness. Resolution: restructuring banks The resolution phase of a systemic banking crisis seeks to restore the normal functioning of the credit system and calls for the restructuring of fi nancial institutions. The response to a crisis requires that banks be recapitalized to protect depositors and taxpayers from losses arising from deteriorating asset quality. Bank supervisors must make a judgment about the viability of individual banks based on the best available, if typically incomplete, infor- mation and a view of its future prospects. Th is forms the basis for a triage depending on capital adequacy ratios and bank viability, with banks clas- sified as those that are viable and meet regulatory requirements, those that are viable but undercapitalized, and those that are nonviable and insolvent. Solvent and undercapitalized banks need to be capitalized on a timetable agreed with regulators. Unless market players are prepared to absorb the assets of fragile banks prior to bankruptcy, nonviable and insolvent banks need to be taken over by regulators (or "intervened") and a decision taken on their future. · If a bank is to be closed: Deposits need to be transferred to a healthy bank, and creditors should share in the losses based on existing banking and bankruptcy laws. · If a bank is to be kept open: The range of options includes recapitalizing the bank with public funds, selling it, possibly with some government RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 123 guarantee on asset values, and merging it with a healthy bank, possibly with some enhancement of the balance sheet. While recapitalization of private banks should be done using private funds, crisis situations might call for public capital. In such cases, the government should acquire preferred shares in return for representation on the board, and existing shareholders should suffer a dilution. For undercapitalized subsid- iaries of cross-border banks in ECA, the burden of recapitalization should rest with parent banks. For example, Romania has asked parent banks to pre- emptively recapitalize their subsidiaries. Countries have also used the crisis to give supervisors the broad author- ity to respond to systemic risks in the banking sector. Kazakhstan now has a banking resolution framework that allows regulators to intervene, with appropriate powers, in cases of bank distress. Latvia has sought improve- ments in the legal framework for bank resolution, including intervention in troubled banks. Hungary has strengthened bank regulation and supervisory powers to allow forward-looking actions to preempt systemic distress. It also seeks to renew the focus on onsite verification of banks' safety and soundness and requires onsite inspections of the largest banks to evaluate asset quality, loan loss provisions and reserves, collateral values, capital solvency and gov- ernance; to calculate required adjustments to capital and provisions; and to recommend corrective action. In Ukraine, legislation is being sought to allow revaluating shareholder cap- ital; transferring the assets and liabilities of a bank, whether before or after revocation of its license without the prior approval of creditors, including depositors; simplifying the grounds for introducing temporary administrators in problem banks; and giving the central bank the authority to charter a bridge bank, tasked with administering the assets and liabilities of failed banks. Similar actions are being taken in the low-income and lower middle-income countries of the former Soviet Union--with regular stress-testing of banks in Armenia and Georgia, increased provisioning in Georgia, requiring existing shareholders to inject capital in banks in the Kyrgyz Republic, with the author- ities taking equity stakes when needed. Bank supervision is being strengthened in Armenia, the Kyrgyz Republic, and Tajikistan. All the low-income and lower middle-income countries of the former Soviet Union have made sure that the supervisory authorities have necessary powers of intervention. Avoid regulatory forbearance Some previous episodes of systemic banking distress, such as Argentina 2001, Bulgaria 1996, Ecuador 1999, Indonesia 1997, Korea 1997, Malaysia 1997, 124 TURMOIL AT TWENTY Mexico 1994, the Russian Federation 1998, and Thailand 1997 have also seen regulatory forbearance. Specifically, to help banks recognize losses and allow corporate and household restructuring to go forward, the government might exercise forbearance either on loss recognition, which gives banks more time to reduce their capital to reflect losses, or on capital adequacy, which requires full provisioning but allows banks to operate for some time with less capital than prudential regulations require. But regulatory forbearance has risks. First, a fi nancial institution might use the period of forbearance to engage in risky lending in an effort to recover its capital position, increasing the costs of an eventual failure. So, forbearance should be allowed only for fi nancial institutions whose long- term viability seems reasonably assured, and progress toward capital ade- quacy should be time-bound and monitored. Second, some types of forbear- ance on loss recognition may encourage the overvaluation of restructured debt or converted equity and thus discourage follow-on operational restruc- turing. It could also discourage loss-averse fi nancial institutions from liqui- dating nonviable companies, selling to a strategic investor, or making forced sales of overvalued collateral. Th ird, forbearance on loss recognition may impede private recapitalization of banks since investors might be reluctant to invest in an institution with murky loan classifications and unclear pro- visioning rules. So, forbearance should focus on capital adequacy instead of loss recognition, be limited in applicability and duration, and be closely monitored. More important, postponing bank restructuring has little to recommend it, since the global recession is expected to be more protracted than its recent predecessors. The likelihood of capital inflows recovering to pre-crisis levels is low, so there will be greater reliance on domestic savings. If problem loans are not recognized early and addressed swift ly, this could discourage efficient financial intermediation and hold back the region's growth recovery. ECA's credit losses: substantial but manageable Given that the full impact of the crisis on asset quality is still unknown, past banking and currency crises offer a rough guide to assess underlying risks. The focus is on banking crises, accompanied by a currency crisis that had GDP declines exceeding 5 percent in the year following the onset of the crisis. In such cases, the nonperforming loans on average rise to 30 percent (table 3.1). These are assumed to be a proxy for the probability of default. In addition, recovery rates are assumed to be roughly 40 percent on mortgages, in line with the marked declines in housing prices, and 15 percent on loans to firms, which broadly RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 125 matches the average assumption by the Swedish Riksbank on the exposure of Swedish banks to the Baltic states.5 A preferable approach no doubt would be to calibrate the recovery rate by sector and country depending on country-specific bankruptcy resolution frameworks and other institutional characteristics that impact recovery rates, but such data are only available to banking supervision authorities of each country. The shares of households and firms in the total loan portfolio--a measure of exposure--are provided by a broad characterization of the consolidated banking sectors in ECA countries. Expected credit losses are the product of exposure, the probability of default, and the recovery rate. TABLE 3.1 Countries with banking and currency crises and nonperforming loans as a share of total loans Nonperforming Nonperforming Crisis loans (percent Crisis loans (percent Country year of total loans) Country year of total loans) Banking and currency crisis Banking crisis only Argentina 1980 9.0 Argentina 1995 17.0 Argentina 1989 27.0 Bolivia 1994 6.2 Argentina 2001 20.1 Colombia 1982 4.1 Brazil 1994 16.0 Colombia 1998 14.0 Bulgaria 1996 75.0 Croatia 1998 10.5 Chile 1981 35.6 Czech Rep. 1996 18.0 Dominican Finland 1991 13.0 Republic 2003 9.0 Japan 1997 35.0 Ecuador 1998 40.0 Latvia 1995 20.0 Estonia 1991 7.0 Lithuania 1995 32.2 Indonesia 1997 32.5 Nicaragua 2000 12.7 Jamaica 1996 28.9 Norway 1991 16.4 Korea, Rep. 1997 35.0 Paraguay 1995 8.1 Malaysia 1997 30.0 Sri Lanka 1989 35.0 Mexico 1994 18.9 Thailand 1997 33.0 Philippines 1997 20.0 Vietnam 1997 35.0 Russian Average 19.4 Federation 1998 40.0 Median 16.7 Sweden 1991 13.0 Turkey 2000 27.6 Ukraine 1998 62.4 Uruguay 2002 36.3 Venezuela 1994 24.0 Average 28.9 Source: Laeven and Valencia 2008. Median 27.6 5. Sveriges Riksbank 2009. 126 TURMOIL AT TWENTY The results of the analysis suggest that credit losses could, in a worsening scenario, be substantial but manageable. They vary from 7 percent of GDP in Belarus and Turkey to 21 percent in Estonia, with an average of some 13 percent for the financially integrated ECA countries (table 3.2). The variation across countries is largely accounted for by the size of the loan portfolio--that is, the share of credit in GDP. Note that despite sharp declines in real estate prices, this is somewhat compensated for by the better recovery rates for these loans given the collateral underlying mortgage lending--and indeed despite the sharp declines in real estate prices of the past year. Of course, the scenario could be more optimistic about recovery rates. For example, housing prices in many countries in the region have not declined as much, and banks might choose not to proceed immediately to sell these assets to avoid worsening the housing market. In a scenario with recovery rates in mortgages averaging 75 percent, credit losses would range from 6 to 16 percentage points of GDP. For the borrowers: corporate and household debt restructuring Corporate debt: how much of a problem? With a few exceptions, nonfinancial corporates in ECA are only moderately indebted. Indirect evidence comes from these facts: · Financial development (private credit to GDP) was still lagging economic development (GDP per capita)--but the gap has closed only recently rela- tive to 1995 (see annex 1.2). · Small and medium-size enterprises in ECA's transition countries (exclud- ing Turkey) relied more on retained earnings and informal finance than external fi nance to fund fi xed investment, than did developing market economies, a gap that closed for the richer transition economies only in 2008, on the eve of the crisis.6 · The growth of credit to nonfi nancial corporates was considerably lower than that to households in many financially integrated ECA countries (see table 1.3). Direct evidence comes from the evolution of corporate leverage--the ratio of total debt to total assets--for large nonfi nancial corporates (table 3.3). Although leverage increased sharply in Hungary and, to less extent, in Turkey in 2008, it was still about half the elevated levels in East Asia during its crisis in 6. The sample of fi rms is drawn from the World Bank's Investment Climate Assessments for 2000­08 for developing countries and the Business Environment and Enterprise Performance Surveys for 1999­2008 for the ECA transition countries (see chapter 5). RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 127 TABLE 3.2 Credit losses--extrapolating from past crisis events Losses (including Share of Outstanding nonperforming lending to private credit loans) Billions Billions of local of local House- currency Percent currency Percent Country holds Firms units of GDP Assumptions units of GDP Belarusa Nonperforming 0.25 0.75 37,159 29 loansb 8,632 7 Bulgaria 0.35 0.65 50 74 29.50 11 17 Croatia 0.50 0.50 222 65 48 14 Czech Rep. Loss recovery given 0.40 0.60 1,947 53 default, householdc 431 12 Estoniad 0.50 0.50 245 99 0.40 52 21 Hungary Loss recovery given 0.40 0.60 18,527 69 default, firmse 4,099 15 Kazakhstana 0.25 0.75 7,972 50 0.15 1,852 12 Latviad 0.50 0.50 15 90 3 19 Lithuaniad 0.45 0.55 70 63 15 14 Macedonia, FYR 0.40 0.60 175 44 39 10 Montenegro 0.40 0.60 3 81 1 18 Poland 0.40 0.60 633 50 140 11 Romania 0.40 0.60 194 38 43 9 Russian Federationa 0.30 0.70 17,102 41 3,910 9 Serbia 0.40 0.60 1,072 38 237 8 Turkeya 0.30 0.70 310 33 71 7 a Ukraine 0.30 0.70 700 74 160 17 Average 58 13 Median 53 12 a. Assumes a lower share of household lending; loans to corporates still dominate. b. Nonperforming loans are assumed to match the levels observed in the Laeven and Valencia database for cases with a currency crisis; in effect this is broadly equivalent to cases where the decline in GDP in period t+1 is at least 5 percent. c. Assumes loan-to-value ratios of 1 and a recovery rate of only 40 percent given the decline in housing prices. d. Assumes somewhat higher role of mortgage lending given developments in housing prices. e. The loss recovery given default is set at the average level observed during the Asian crisis. Source: IMF International Financial Statistics and authors' calculations. 1997­98 and was also generally lower than in Argentina (2001), Brazil (1998), Mexico (1995), and Turkey (2001) in the years of their crisis. Corporate lever- age is notably higher in Greece, Ireland, Portugal, and Spain (the EU cohesion countries), reflecting their deeper and more liquid financial markets. 128 TURMOIL AT TWENTY TABLE 3.3 Median nonfinancial corporate leverage, Europe and Central Asia countries and EU cohesion countries, 1999­2008, and comparator countries for years of crisis (percent) Number Country of firmsa 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Czech Rep. 31 23.9 18.0 12.8 10.3 8.9 10.1 10.8 9.7 8.9 11.6 Hungary 33 11.1 17.4 11.7 12.8 17.2 17.1 15.7 21.1 13.2 27.6 Poland 167 11.7 13.0 16.5 17.7 18.3 15.9 15.6 15.9 15.5 17.1 Turkey 182 25.5 27.5 26.2 25.1 19.7 16.7 16.7 19.2 16.5 21.7 Greece 268 20.2 20.4 26.8 28.0 29.5 29.2 31.9 30.2 32.6 34.4 Ireland 68 25.5 27.4 30.6 29.3 28.7 23.5 27.0 29.3 28.4 34.7 Portugal 64 29.3 32.4 35.6 37.1 35.2 35.2 34.8 38.9 40.8 46.3 Spain 163 20.4 23.2 24.0 26.0 27.7 28.1 31.9 33.0 33.3 34.1 Period (t is crisis year) t­3 t­2 t­1 t t+1 t+2 t+3 Korea, Rep. (1997) 442 44.1 45.1 46.1 50.5 44.6 33.4 31.1 Thailand (1997) 273 32.6 37.7 40.9 53.1 45.1 41.1 43.4 Indonesia (1997) 171 28.6 31.6 34.7 51.8 61.3 50.3 47.1 Argentina (2001) 78 33.4 30.6 31.2 27.7 31.4 25.8 23.1 Brazil (1998) 257 18.0 24.3 26.1 27.3 26.2 26.1 31.4 Mexico (1995) 82 28.7 29.9 31.1 31.9 28.3 28.7 28.0 Turkey (2001) 158 24.1 25.5 27.5 26.2 25.1 19.7 16.7 a. Average over period. Note: DataStream (WorldScope) includes companies which meet some of the following criteria: market capitalization equal to or greater than $100 million; company belongs to the FTSE ALL World, Dow Jones Global, MSCI World, MSCI EMF, S&P Global, or S&P/Citigroup; company has an ADR listed on the NYSE, ASE, or NASDAQ, or a sponsored ADR that trades over the counter; companies included in EASDAQ or EURO.NM. Source: DataStream (WorldScope). Data for other countries in the region (taken from the Bloomberg data- base, which has a wider country coverage) confirm this view (table 3.4). Cor- porate leverage in 2008 was among the lowest in Bulgaria, the Czech Repub- lic, Poland, and the Slovak Republic; intermediate in Romania, Turkey, and Ukraine; and among the highest in Croatia, Estonia, Latvia, Lithuania, and Slovenia. But even the countries with the highest leverage have a total debt to total assets ratio broadly similar to those in East Asia and somewhat less than in the cohesion countries in 2008. In particular, corporate leverage in the ECA countries is much lower than that in East Asia during its crisis in 1997­98. The comparison, which focuses on the largest firms, is meant to be suggestive, and the small sample size in ECA's smaller countries in particular RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 129 TABLE 3.4 Median nonfinancial corporate leverage, by country, 2008 (percent) Country Number of firms Corporate leverage Europe and Central Asia countries Bulgaria 142 16.4 Croatia 201 26.5 Czech Rep. 13 10.8 Estonia 14 26.2 Hungary 24 19.5 Latvia 23 25.6 Lithuania 34 29.4 Macedonia, FYR 30 18.6 Poland 247 14.8 Romania 151 18.5 Russian Federation 713 23.5 Slovak Rep. 11 13.7 Slovenia 41 31.9 Turkey 219 20.6 Ukraine 193 17.5 Other countries Korea, Rep. 116 27.8 Thailand 364 24.9 Indonesia 244 30.2 Argentina 78 21.2 Brazil 313 28.0 Mexico 83 23.5 Portugal 53 41.1 Ireland 44 26.9 Greece 255 33.5 Spain 115 27.5 Source: Bloomberg. should be recognized. But it should be placed alongside the indirect evidence cited earlier about ECA's financial shallowness, the importance of households rather than nonfi nancial corporates in rapid credit growth in many ECA countries, and the dominance of retained earnings as a source of financing for fi xed investment giving way only recently to bank financing in a large sample of small and medium enterprises from across the region (chapter 5). The sustainability of corporate financial structures during the year of crisis is of somewhat more concern in some countries. Some indication of the extent to which nonfinancial corporates have a sustainable financing structure is the 130 TURMOIL AT TWENTY TABLE 3.5 Median interest coverage in nonfinancial firms, Europe and Central Asia countries and EU Cohesion countries, 1999­2008, and comparator countries for years of crisis (percent) Number Country of firmsa 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Czech Rep. 33 1.7 4.0 3.8 4.6 6.1 8.1 6.6 11.7 14.1 22.3 Hungary 34 5.5 5.9 5.7 4.7 4.8 3.4 4.1 5.1 4.2 1.3 Poland 240 3.6 2.1 1.5 2.1 3.7 6.0 6.4 9.4 11.5 4.8 Turkey 194 1.7 2.3 1.2 1.8 3.2 3.1 3.7 3.3 4.5 2.2 Greece 281 6.0 5.8 3.9 3.1 2.9 3.2 2.9 3.1 3.0 1.7 Ireland 70 4.4 3.8 2.0 1.4 2.6 3.1 3.7 4.0 4.4 1.5 Portugal 65 4.1 2.9 2.4 1.8 2.0 3.1 3.2 3.0 2.5 1.3 Spain 166 7.2 5.8 4.5 3.8 5.0 5.8 5.9 4.6 3.5 2.3 Period (t is crisis year) t­3 t­2 t­1 t t+1 t+2 t+3 Korea, Rep. (1997) 436 1.4 1.4 1.2 1.1 1.1 1.7 1.8 Thailand (1997) 259 4.1 3.2 2.2 1.2 1.6 1.0 1.5 Indonesia (1997) 169 4.2 3.3 2.6 1.0 0.1 1.8 0.6 Argentina (2001) 468 3.3 1.9 1.4 1.2 0.1 1.6 2.2 Brazil (1998) 255 1.6 1.6 1.8 1.3 0.9 1.4 1.2 Mexico (1995) 73 3.2 2.7 1.1 1.8 3.1 3.1 1.7 Turkey (2001) 170 2.4 1.7 2.3 1.2 1.8 3.2 3.1 a. Average over period. Source: DataStream (WorldScope). interest coverage ratio--the ratio of EBIT (earnings before interest and tax) to total interest expense (table 3.5). It fell sharply in Hungary between 2007 and 2008 to reach a low of 1.3 in 2008, a figure comparable to the lows in East Asia during its crisis and in Turkey in 2001. For a wider set of countries, it is the lowest in Croatia, followed by Slovenia, Turkey, Latvia, and Hungary (table 3.6). The highest interest coverage ratios are for the Czech Republic, the Russian Federation, Estonia, Poland, Romania, and the Slovak Republic (table 3.6). The table also reports the proportion of firms that had interest coverage less than unity--that is, where EBIT did not cover interest costs. Household debt: the crisis hits home Much of the rapid expansion of credit in the years preceding the crisis was driven by the household sector.7 The ratio of household lending to corporations 7. Part of the material in this section draws on Tiongson et al. (2009). RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 131 TABLE 3.6 Median interest coverage ratio in nonfinancial firms, by country, 2008 (in percent; median values) Number With interest coverage Interest coverage Country of firms ratio less than 1 a ratio (percent) Europe and Central Asia countries Bulgaria 123 34 2.1 Croatia 204 52 0.8 Czech Rep. 16 6 33.3 Estonia 13 15 5.5 Hungary 14 50 1.5 Latvia 24 38 1.5 Lithuania 31 32 2.8 Macedonia, FYR 30 50 1.1 Poland 319 28 4.2 Romania 186 28 3.5 Russian Federation 705 19 5.5 Slovak Rep. 9 33 3.3 Slovenia 38 32 2.0 Turkey 214 45 1.3 Ukraine 46 41 2.0 Other countries Korea, Rep. 1,550 39 2.2 Thailand 397 33 3.5 Indonesia 243 20 2.7 Argentina 83 24 3.1 Brazil 157 27 2.6 Mexico 96 19 3.9 Portugal 49 37 1.5 Ireland 51 35 2.1 Greece 256 31 2.1 Spain 81 23 3.1 a. Proportion of firms with an interest coverage ratio less than 1. Source: Bloomberg. doubled in most countries between 2005 and 2008 (see table 1.3). And mort- gage lending as a share of lending to households increased sharply in some countries. Despite this growth, household indebtedness is still significantly lower than in the EU15 and reflected a pattern similar to that in the cohesion countries during their financial integration. · Household debt represents on average more than a quarter of GDP in the new member states of the European Union (EU10), but there is significant 132 TURMOIL AT TWENTY cross-country variation, with the number reaching more than 40 percent in some countries (figure 3.2). These ratios are below the average of about 65 percent of GDP among EU15 countries, and closer to those for Ireland, Italy, Portugal, and Spain in the late 1990s (figure 3.3). · As household financial positions have grown, there has been a shift toward housing loans or mortgages on the liability side of the balance sheet and an increasing share of equities and pension and mutual funds on the asset side. Still there is much variability. Housing loans accounted for the bulk of FIGURE 3.2 Household debt, by country, 2008 Percent of GDP 50 40 30 20 10 0 p. Sl p. ria tia nia Lit a ia Ro d a ia Ru Kaz us Fe tan n ine y nia y rke ar tvi Sl ani tio lan an en Re Re lar lga oa ra ng to me ian khs La ra Tu ov hu m Po Be Uk Cr Es h ak Bu Hu de Ar ec a ov Cz ss EU10+1 Middle-income CIS Other FIGURE 3.3 Household debt, earlier EU members, 1995­2004 Percent of GDP 60 Core countries 40 Converging countries 20 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Note: The chart displays averages for each country group. Core countries includes Austria, Bel- gium, Finland, France, Germany, Luxembourg, and the Netherlands. Converging countries include Ireland, Italy, Portugal, and Spain. Source: Gaspar and Fagan 2006. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 133 household credit in the Baltic states, the Czech Republic, Hungary, and the Slovak Republic, while the opposite was the case in Romania, the Russian Federation, Turkey, and Ukraine (figure 3.4). · A large share of household debt is denominated or indexed to foreign cur- rencies. This has exposed households to recent exchange rate depreciations to the extent that the currency composition of their assets, particularly FIGURE 3.4 Composition of household debt, by country, end-2008 Consumer credit Housing loans Other loans Bulgaria Czech Rep. Estonia Hungary EU10 Latvia Lithuania Poland Romania Slovak Rep. Slovenia Russian Federation CIS Turkey Ukraine 0 25 50 75 100 Percent Source: European Central Bank and other national central banks. FIGURE 3.5 Foreign currency-denominated loans, by country, 2008 Percent of bank loans to households 100 75 50 25 0 Es . Ar p. ria ec ia nia a ia Ro d a nia Ru Kaz us Fe tan n ine YR y y p rke ar tvi Sl ani tio lan t an Re Re lar ,F lga oa ra to ng me ian khs La ra Tu hu m Po nia Be Ma Uk Cr h ak Bu Hu de a Lit ov do Cz ce ss EU10+1 Middle-income CIS Other Note: Foreign currency­indexed loans are included for Croatia and FYR Macedonia. Source: MNB and other national central banks. 134 TURMOIL AT TWENTY labor income flows, leaves them unhedged. But again, there is considerable variation across countries (figure 3.5). · In some EU10 countries, mortgages with variable (adjustable) interest rates account for the largest share of lending, thus exposing households to inter- est rate shocks (figure 3.6). More detailed information on the characteristics of household debt across income quintiles, including the burden of servicing these debts, can be derived from household surveys. These surveys contain information on either inter- est debt service or total debt service (interest and principal payments com- bined). Some surveys cover only mortgage debt and some refer only to total debt (mortgage plus consumer debt). For example, the columns per income quintile in figure 3.7 report data for new EU member states on mortgage inter- est payments from the EU Survey of Income and Living Conditions (EU-SILC) for 2007. The lower lines show the share of income devoted to mortgage inter- est payments. In contrast, figure 3.8 is derived from household budget sur- veys issued by statistical offices and jointly reports principal and interest debt service. In particular, it refers to total household debt. Here, too, the columns reflect the percentage of households holding debt at each income quintile, and the lower line in each figure is the share of income devoted to servicing debts-- except that in this case this refers to both interest and principal payments. Two patterns to highlight from figures 3.7 and 3.8 are: · Share of indebted households. Only a fraction of households in each quin- tile hold debt, and that share expectedly increases in the higher quintiles. FIGURE 3.6 Mortgage loans with adjustable interest rates, by country, 2006 United Kingdom Portugal Other countries Netherlands Italy Ireland France Denmark Croatia Belgium Slovenia EU10 Slovakia Hungary 0 25 50 75 100 Percent of all housing loans Source: IMF, OECD, and national central banks. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 135 FIGURE 3.7 Mortgage interest service and interest rate shock, by income quintile (vulnerability defined by a 20 percent income threshold) Estonia Hungary 30 30 20 20 10 10 0 0 I II III IV V Total I II III IV V Total Latvia Lithuania 30 30 20 20 10 10 0 0 I II III IV V Total I II III IV V Total Percent of Percent of Percent of Interest service ratio, Interest service ratio, households vulnerable vulnerable percent of household percent of household with interest households, households, income, before the income, after the service before the shock after the shock shock shock Note: The quintile distribution of the household survey in Latvia and Lithuania has few households in the lowest quintile and thus provides unreliable results for the interest service ratio--it is thus excluded. Source: EU Survey of Income and Living Conditions. Among the new EU member countries (except in Hungary), fewer than 5 percent of households in the fi rst quintile appear to have a mortgage, suggesting that few low-income households would benefit from mortgage restructuring schemes that include use of public funds. The pattern for total debt in figure 3.8 suggests great variability across countries, with more than 40 percent of households holding some debt in Belarus and some 20 percent in Bosnia and Herzegovina, Hungary, and Poland. So, fi nancial deepening has so far been concentrated in relatively few households. · Burden as a share of income. Debt service (mortgage interest service in fig- ure 3.7 and total debt service in figure 3.8) as a share of income generally decreases with income. More important, debt service (total or mortgage only) is a small share of income. But there are some exceptions. Debt ser- vice is 5 percent of household income in the poorest quintile in Ukraine but 13 percent in Serbia, 25 percent in Moldova, and 30 percent in Bosnia 136 TURMOIL AT TWENTY FIGURE 3.8A Total debt service under shocks (vulnerability defined as a financial margin) Panel A: Exchange rate shock Belarus, 2008 Bosnia and Hungary, 2007 50 50 Herzegovina, 2007 50 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total Kazakhstan, 2007 Moldova, 2007 Poland, 2008 50 50 50 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total Romania, 2007 Serbia, 2008 Ukraine, 2007 50 50 50 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total Panel B: Unemployment shock Belarus, 2008 Bosnia and Hungary, 2007 50 50 Herzegovina, 2007 50 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total 50 Kazakhstan, 2007 50 Moldova, 2007 50 Poland, 2008 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total 50 Romania, 2007 50 Serbia, 2008 50 Ukraine, 2007 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total Percent of Percent of Percent of Debt service ratio, Debt service ratio, households vulnerable vulnerable percent of household percent of household with debt households, households, income, before the income, after the service before the shock after the shock shock shock Note: In the case of Bosnia and Herzegovina, the results of the simulation are the results of aban- doning the exchange rate peg. Source: Household Budget Surveys. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 137 FIGURE 3.8B Total debt service under shocks (vulnerability defined as a 30 percent income threshold) Panel A: Exchange rate shock Belarus, 2008 Bosnia and Hungary, 2007 50 50 Herzegovina, 2007 50 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total 50 Kazakhstan, 2007 50 Moldova, 2007 50 Poland, 2008 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total 50 Romania, 2007 50 Serbia, 2008 50 Ukraine, 2007 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total Panel B: Unemployment shock Belarus, 2008 Bosnia and Hungary, 2007 50 50 Herzegovina, 2007 50 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total 50 Kazakhstan, 2007 50 Moldova, 2007 50 Poland, 2008 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total 50 Romania, 2007 50 Serbia, 2008 50 Ukraine, 2007 25 25 25 0 0 0 I II III IV V Total I II III IV V Total I II III IV V Total Percent of Percent of Percent of Debt service ratio, Debt service ratio, households vulnerable vulnerable percent of household percent of household with debt households, households, income, before the income, after the service before the shock after the shock shock shock Note: In the case of Bosnia and Herzegovina, the results of the simulation are the result of aban- doning the exchange rate peg. Source: Household Budget Surveys. 138 TURMOIL AT TWENTY and Herzegovina. But only a small fraction of these households carry such a burden in Moldova (2 percent) and Bosnia and Herzegovina (10 percent). In contrast, mortgage interest is 7­13 percent of household income in the poorest quintile in the EU countries, but the fraction of households with mortgages is similarly small. Figures 3.7 and 3.8 also estimate the effects of economic shocks on households' capacity to service their debt obligations. The simulations affect household welfare and shed light on the probability of loan default. Three shocks are assessed.8 · An increase in 5 percentage points in the interest paid on mortgages with variable interest rates. This shock is more accurate when applied only to data on mortgage interest debt service, so it is limited to the EU-SILC (fig- ure 3.7). · A change in the foreign currency­local currency exchange rate of 35 per- cent. This is best applied to figure 3.8 because it affects both the interest and principal components of debt service. · The effect of unemployment on the combined income of working house- hold members. An increase of 5 percentage points in the unemployment rate is used for all countries. This shock has a limited impact on debt ser- vice capacity because it is applied to household members without taking into account that these households may or may not hold debt. Assessing the impact of economic shocks on a household's capacity to ser- vice its debt obligations depends on the threshold for vulnerability--and this choice is specific to each survey.9 An arbitrary mortgage interest debt burden threshold (interest debt service payments in percent of income) is used for the calculations based on the EU-SILC database. The threshold is based on work for advanced economies.10 A household is considered vulnerable at 20 percent debt service as a share of income, and this is applied without distinction across quintiles--clearly a limitation.11 8. In all three cases the shock is relative to the country-specific level in 2007. Because data on loan characteristics are not household-specific, random draws are needed to assess the impact of economic shocks on a household's debt service capacity. More precisely, 1,000 random draws are used to choose which loans are foreign currency­denominated or have variable interest rates, and which household members become unemployed. 9. The methodology on stress testing household debt holdings and on assessing the share of vulnerable households follows Holló 2007, ochowski and Zajczkowski 2008, Johansson and Persson 2006, and Vatne 2006. The analysis here extends work by Tiongson et al. (2009) by introducing income quintiles. 10. Johansson and Persson 2006, among others. 11. Th is is not justified because higher income households can devote higher shares of their income to housing services. The differences become clearer when comparing the results for the EU-SILC and the household budget survey in Hungary. Th resholds specific to income quintiles would be worth pur- suing further. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 139 In contrast, two different thresholds are applied when using household budget surveys: an arbitrary total debt burden threshold of 30 percent12 and a financial margin threshold given by Financial margin (FM) = disposable income (DI) ­ basic living costs (BLC) ­ total debt service expenditure (DSE).13 A household is considered at risk if its financial margin is negative. What are the findings? Both figures 3.7 and 3.8 show the impact of differ- ent shocks under different assumptions on the threshold values defining vul- nerability. The lowest area within each column is the percentage of households within each quintile considered vulnerable even before any economic shock simulation occurs, referred to as vulnerable at origination. The shaded area on top of this lower area reflects the households in each income quintile that become vulnerable after an economic shock is applied; and the remaining area of each column shows the households that are not vulnerable. So, vulnerable households after the simulation include those that were already at risk before the shock plus those that became vulnerable after the shock. Note that vulner- ability depends on the threshold that best applies to each type of household survey. Only a 20 percent income threshold for mortgage interest debt service is applied in figure 3.7, and both the financial margin and a 30 percent thresh- old on total debt service are applied in figure 3.8. The higher lines reflect the interest (or total) debt service ratio after the shocks. The results of the analysis of EU-SILC data (figure 3.7) suggest that an interest rate shock can expand the pool of households unable to service their mortgage debt obligations. Specifically, a severe 5-percentage point interest rate shock in Estonia increases the share of total vulnerable households or borrowers at risk from less than 2 percent of all households to 4 percent, and raises the average interest debt service (interest service ratio) from 10 percent of income to almost 20 percent. Although many of the vulnerable households are in higher income quintiles, the proportion of vulnerable households is smaller than in lower quintiles. Also, vulnerability is partly tied to the choice of threshold. Indeed, high-income households are more likely than poorer households to spend a larger share of their income on housing services. In 12. Th is follows work by Beer and Schürz 2007. 13. The basic living costs data reflect what is referred to as subsistence level income. Th is is adapted to household size following a classification used by EU-SILC where an equivalized household size is the sum of the fi rst adult plus 0.5 times the number of other adults in the household plus 0.3 times the number of children, where adults are defi ned as people aged 14 years and over, and children are those aged 13 years and younger. The BLC is monthly and per person, and is converted for each household j by BLCj = BLC * equivalized household size in household j * 12. 140 TURMOIL AT TWENTY contrast, Hungary appears to have a large share of vulnerable households in lower income quintiles, perhaps reflecting the more even distribution across all five quintiles of households with mortgage debt. Latvia and Lithuania are more mixed, though the lower quintiles are proportionally more vulnerable.14 More worrying is that low-income quintiles with such debt can be considered vulnerable even before the shock. For example, the few households in the first quintile are all vulnerable in Latvia and Lithuania, and about half of house- holds in the first quintile are vulnerable in Estonia and Hungary. But very few of these households are at risk due to an interest rate shock. Instead, they are risk borrowers at loan origination, suggesting that compensating government financial support might fail to restore repayment capacity. The analysis of household budget survey data in figure 3.8 provides addi- tional insights. First, the effect of exchange rate shocks suggests a more limited impact on household welfare when using a financial margin threshold--and, as would be expected, vulnerability declines with income (figure 3.8a). The financial margin remains positive given that the BLC is applied to all house- holds irrespective of their income quintile--and, by definition, BLC might be quite small, allowing households to avoid fi nancial distress as income rises. Second, as with mortgage debt, many lower quintile households are vulnerable before the shocks. Indeed, vulnerability appears to change little once shocks are imposed. This does not mean that households in the bottom quintile are unaffected. Figures 3.7 and 3.8 show only a change in household status from nonvulnerable to vulnerable so, even though few households become vulner- able in the lowest quintile, many of them were already vulnerable before the shocks are simulated. Finally, using a 30 percent arbitrary threshold suggests more households at risk among the higher income quintiles (figure 3.8b). Yet, very large economic shocks seem to affect proportionally more households in low-income quintiles. The premise that all households should be compensated for the increase in debt service burden arising from economic shocks is not justified by the dis- tribution of indebted households across quintiles. In fact, the evidence sug- gests that households have room to confront important economic shocks-- with one caveat. It might be worth developing a simulation for simultaneous shocks to incomes and liabilities--such as sharper increases in unemploy- ment and declines in nominal wages--to sharp depreciations of the exchange rate. Notwithstanding these caveats, it would be sensible to target eligibility of a government financial support program based on loan size and to households 14. As noted in the notes to figure 3.7, however, the data for the lowest-income quintile is not reliable. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 141 with incomes below a certain income threshold--and it would be best to keep income eligibility criteria simple and monitorable. Also, the cost of compen- sating households for their income losses is modest. For instance, the largest estimate of the household income losses in the poorest quintile in Central Europe is about 6 percent for the interest shock. Bearing in mind that 10 per- cent of that quintile's households would be affected, and using a share of that quintile's income in GDP of 2.5 percent, yields 0.015 percent of GDP as the necessary compensation for this subgroup of poor households. What have countries done to alleviate household debt distress? In Hun- gary, the authorities entered "gentlemen's agreements" with banks to con- vert foreign currency­denominated loans to households into local currency loans without penalty, capitalize the increase in mortgage payments arising from the conversion, and possibly extend the term of the loan for creditwor- thy borrowers. But the option has not been widely exercised because forint interest rates are substantially higher than euro interest rates. Hungary has introduced legislation to provide temporary state guarantees for mortgage payments of the unemployed and also to expand the mortgage debt servicing guarantee scheme for the unemployed to other debtors whose payment capac- ity has been impaired by the financial crisis because of either a reduction in income or an increase in debt service burden due to revaluation effects. In such cases, the lender would be asked to rephase the loan to temporarily lower the payment burden, and the government would guarantee the rephased por- tion of the loan, subject to restrictions. Romania has sought an agreement with commercial banks to facilitate the restructuring of debt contracted in foreign currency by adjusting the maturity and repayment schedule of the debt, including offering the option to voluntarily convert it into domestic cur- rency. In Latvia, a partial state guarantee for restructured mortgage loans is being considered under guidelines intended to relieve borrowers' debt service to a level commensurate with their capacity to pay. And banks participating in Serbia's financial sector support program have been asked to facilitate the vol- untary conversion of foreign currency and foreign currency­linked loans into local currency loans and work with the central bank to develop loan workout programs. Lessons on restructuring from previous banking and capital account crises The East Asia and Latin America crises saw a sharp increase in nonperform- ing loans, reflecting increased and widespread corporate distress as well as the introduction of better loan classification standards for banks and other 142 TURMOIL AT TWENTY financial institutions.15 In fact, it took considerable time for firms in East Asia to reduce corporate leverage and build more viable financial structures. Banks were weakly capitalized and had limited incentives to write off debts and could continue to carry loans at low provisioning requirements. Asset management companies were not much more effective. The experience from earlier crises suggests that these companies were useful for taking distressed debt off banks' books but not always at engaging in corporate restructuring. Commercial banks and asset management companies, lacking the skills to manage equity, opted for rolling over claims rather than reducing prin- cipal, converting debt to equity or proceeding with more active operational restructuring. So, high rollover rates, while creating breathing space when confidence is being rebuilt, can subsequently mask a reluctance to restructure among creditors as well as debtors. It is appropriate for fi nancial sector restructuring to precede corporate restructuring in order to arrive at an assessment of the ability of banks to absorb losses, but it is also important that corporate restructuring not be delayed. Banks have two options in dealing with nonperforming assets: to financially restructure the loans or to take a total or partial writeoff on these loans early on. The latter could also involve selling their claims to a specialized institution, such as an asset management company for more efficient manage- ment and resolution. The operational restructuring that follows, at least for corporate-related assets, depends on the management control that emerges from the transfer of assets. In some cases, the earlier involvement of outside investors can facilitate operational restructuring. The challenge in ECA is also unique in that bank distress is linked in part to household debt, which was not so in previous crises. Facilitating debt restructuring Countries typically have insolvency frameworks to deal with bankruptcy, reorganization, and liquidation, but fledgling judicial systems can be over- whelmed in a systemic banking crisis. To expedite debt restructuring, govern- ments in countries as diverse as the Czech Republic, Indonesia, Korea, Malay- sia, Mexico, Thailand, and Turkey have been active in setting up out-of-court voluntary workouts between debtors and creditors. While the workouts lie outside the formal insolvency framework, their success depends on the qual- ity of that framework. 15. The reader interested in the experience with debt restructuring is directed to work by Claessens (2005), Lieberman et al. (2005), and Lieberman and Mako (2009), who examine in particular the expe- rience during the East Asia crisis and more recent experiences. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 143 The "London approach" . . . In the variant directed at dealing with corporate restructuring, the so-called "London approach" is based on three broad principles. It seeks, first, to mini- mize losses to creditors from unavoidable company failures through well pre- pared workouts. Second, it aims to avoid unnecessary liquidation of viable companies, through reorganization and preservation of employment and pro- ductive capacity. Third, it finds ways to provide financial support to compa- nies deemed viable in cases where creditors cannot agree to the terms for con- cluding a workout. Armed with these principles, creditors cooperatively--likely in the form of a coordinating lead bank and a steering committee comprising those with the largest exposure to the borrower--agree to a limited standstill (30­90 days in Korea, 90 days in Malaysia, and a maximum of 90 days plus three 30 day extensions for a total of 180 days in Turkey). During this period, the debtor can continue to operate normally, and information about the debtor's assets, liabilities, and prospects can be collected and shared with all creditors, with a view to developing proposals that can resolve the debtor's financial difficul- ties. During the standstill, creditors refrain from taking steps to realize their claims against the debtor while the debtor refrains from taking actions that would adversely affect any creditor. Proposals to resolve the debtor's financial difficulties and the order of addressing creditors' claims are in accord with the law. Any new money needed in the event of a liquidity shortfall during the standstill may be provided by all existing lenders, by specific lenders with pri- ority arrangements, or through the release of asset disposal proceeds subject to priority considerations. A workout plan agreed to by a majority of creditors- --the thresholds for agreement can vary--is then implemented. In the event of disagreement, an arbitration committee set up as part of the workout resolves differences between the debtor and creditors, as well as disputes among credi- tors on the allocation of losses and risks. . . . adapted to mortgage restructuring The resolution of widespread household debt distress--which has a negative social impact and threatens to overwhelm the existing legal and institutional framework for individual restructuring--may call for governments to estab- lish a mortgage restructuring protocol. A model applied in industrial econo- mies is known as the UK preforeclosure protocol. The goal is to facilitate nego- tiations between creditors and debtors by setting clear objectives and options on how to restructure mortgage loans. Discussions between the parties may 144 TURMOIL AT TWENTY include extending the term of the mortgage, changing the type of mortgage, deferring the payment of interest due, or capitalizing arrears. Assessing the creditworthiness of borrowers should be left to banks since they are better placed to judge a household's repayment capacity. The protocol should set out minimum requirements for restructuring loans, define which restructuring methods should not be pursued, and provide guidelines for the regulatory treatment of restructured loans. Programs of support to mortgage holders (such as government guarantees) should be limited and well targeted, adequately assessing the incidence of sub- sidy. State guarantees should be looked at very carefully to avoid excessive costs and abuse and stress-tested to a variety of scenarios--for eligibility--to avoid unsustainable liabilities. In addition, any default should remain recoverable by the banks. In sum, government support instruments, should be a possible option where other solutions do not work and where households meet rigorous criteria. In particular, they should be informed by the depth of the beneficiaries' poverty and not distort the incentives of banks and borrowers to perform. The insolvency framework matters . . . The success of an out-of-court system of voluntary workouts depends on the ability of creditors to impose losses on debtors. Without the threat of a court- imposed loss under the country's insolvency laws, there is not enough incen- tive for corporate debtors to agree to such measures as asset sales, dilutions of equity, and diminution of management control. Indeed, an important require- ment for an out-of-court process is a credible threat of seizure of assets and liquidation under a normal insolvency or bankruptcy regime. Creditors can- not otherwise force debtors to participate in workouts in good faith. But a workable insolvency framework also needs to balance carrots and sticks. Debtors should feel encouraged to seek protection in court-supervised restructuring as an alternative to liquidation. The courts should also allow a mutually satisfactory restructuring plan--worked out between a debtor and a majority of its creditors--to be "crammed down" on a holdout minority of creditors and then converted into a court-supervised liquidation if interim milestones and reasonable deadlines are not met.16 Insolvency laws did not always have these features.17 The lack of adequate protection made court-supervised restructuring unattractive for creditors 16. Kawai, Lieberman, and Mako 2000. 17. A variant on the theme is a prepackaged bankruptcies ("prepacks"), workouts agreed to by a debtor and its creditors and then taken to the courts for approval, usually through an expedited procedure. They RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 145 during the East Asia crisis. But in countries where insolvency frameworks were functioning well, the out-of-court systems were stronger as well. The crisis and its aftermath in East Asia gave countries an opportunity to improve on their insolvency legislation, including the introduction of specialized bankruptcy courts. Informed by the experience in past systemic banking crisis, the authori- ties in Latvia and Romania have begun legislative and regulatory reforms aimed at removing obstacles to out-of-court corporate restructuring, allowing pre- packaged recovery and settlement agreements between debtors and creditors, and introducing further flexibility and easier access to insolvency proceedings. . . . and any approach should be adapted to local conditions Countries should look at many options in deciding the way forward. Whether voluntary out-of-court mechanisms are appropriate will vary from country to country, depending on the quality of the insolvency legislation, but also on the availability of specialized and skillful mediators who can broker deals between debtors and creditors. Indeed, countries that have used out-of-court workouts have adapted the London approach to suit local circumstances. Some industrial countries with greater human resource capacity have chosen to work instead with specialized creditor courts that are easier to implement within their existing legal frameworks. Both approaches require specialized skills and the institutional infrastructure of accounting, classification, and provisioning rules that provide incentives for restructuring. Incentives to restructure: banks Since most creditors are either domestic or foreign headquartered banks, corporate and household debt restructuring efforts in ECA are likely to be bank-led. Foreign-owned banks, by engaging in restructuring and better risk management, have been reducing the share of nonperforming loans in ECA, maintaining high capital adequacy ratios, and increasing provisioning against loan losses up to 2007, before the crisis (table 1.2). If that track record is main- tained, those banks can be expected to be adequately recapitalized in the light of the undertakings provided by parent banks during the European Bank- ing Coordination initiative--and to increase provisions in the light of crisis- induced increases in nonperforming loans. In such a situation, foreign-owned banks have every incentive to collect what they can on provisioned loans, since that adds to their profits. prevent dissenting creditors or minority shareholders from using the courts to sue over the outcome of a workout, as in some countries in East Asia during the 1990s. 146 TURMOIL AT TWENTY Foreign creditors have generally resisted participating in the voluntary workout between debtors and creditors in previous crises. In Turkey, foreign banks did not join the Istanbul process and were paid off instead. In Korea, the government negotiated a medium-term repayment agreement with for- eign bank creditors of Korean banks that had been intervened by the bank- ing supervision authorities. This is different from the situation in emerging Europe, where a substantial proportion of corporate and household debt is owed to in-country subsidiaries of foreign-owned banks--and, to less extent, directly to cross-border banks in home countries. It is to be expected is that subsidiaries would fall within the ambit of insol- vency laws in the host country and thus be covered by voluntary workout processes set up by the government. But the host government may not have the power to require cross-border banks to submit to in-country processes, whether court-supervised or voluntary. Since cross-border banks for the most part are from Western Europe, a supranational authority such as the Euro- pean Commission could consider establishing a framework that can facilitate the participation of cross-border banks, with a view to ensuring broad unifor- mity of treatment across member states and discouraging cross-border banks from shopping across borders to secure more favorable treatment. The incentives for restructuring are less clear for domestically owned banks. Those banks could delay recognizing losses from interest rate concessions or grace periods, debt equity conversions that leave banks with illiquid or low value shares, and sales of the noncore assets of borrowers. The incentive to do so is that recognizing losses could push their capital below regulatory require- ments. Recapitalizing banks will dilute existing shareholders and risk ceding management control to others or, if public funds are used, the government. Incentives to restructure: corporates and households It is not only creditors who need to be nudged by the government into taking losses. Debtors may also delay entering restructuring inasmuch as the insol- vency laws that govern a court-supervised process or backstop a voluntary out-of-court process might lead to asset sales, dilution of equity, and dimi- nution of management control. So, a country's insolvency laws, or a threat of their invocation, will be needed to force debtors to accept losses as well. As in some crisis-hit countries in East Asia, insolvency legislation may need to be strengthened. But the difficulty of strengthening laws to provide more cred- ible threats to debtors in the middle of an economic downturn should not be underestimated. For households, recognizing losses is complicated by the fact that this might also represent foreclosure on their home. Furthermore, RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 147 operational restructuring is not in the lexicon, highlighting the impor- tance (and difficulty) of the systemic banking crisis challenges facing ECA countries. Incentives to restructure: small and medium-size enterprises Much of the restructuring in previous crises took place outside voluntary workouts and formal in-court processes. Small and medium enterprises typi- cally lose access to finance during the credit crunch that follows a period of rapid credit growth and have to dispose of their assets, shed labor, or close down. Inasmuch as many low-income families depend on such enterprises, governments may wish to roll over credits, recapitalizing interest past due, extending grace periods for repayment, and injecting liquidity for work- ing capital. The financial costs of stabilizing small and medium enterprises are modest relative to the social and political costs of allowing thousands of them to fail. But restructuring small and medium enterprises case-by-case is not viable either, due to the high transaction costs relative to the value of the firm.. Capacity constraints in the number of bankruptcy judges, bank workout staff, and government employees limit the scope of a systemic out-of-court workout process. Apart from small and medium enterprises, corporate dis- tress may have to be segmented into: · Mid-cap and large companies, which produce the bulk of employment and exports and which can be helped to realize a supply response through the provision of working capital, and which generally form the core of a volun- tary workout program. · Very large firms that must be dealt with case-by-case to prevent a second- ary crisis spilling over into the banking sector. International workout teams will typically be needed to help major banks build capacity in this area. The specialty is also required in confronting house- hold debt restructuring where the complications are many and the political economy of a systemic banking crisis is more important. Lessons for strengthening bank regulation and supervision The ECA countries need to strengthen bank regulation and supervision to reflect the lessons learned from the crisis.18 Better bank regulation and super- vision would not have averted the crisis owing to the abundance of global liquidity and the buildup of external imbalances. But by arresting the 18. The material in this section was contributed by James Hanson. 148 TURMOIL AT TWENTY FIGURE 3.9A Ratio of actual to minimum required capital-asset ratio (median, risk-weighted) Ratio 2.00 ECA, all (17) 1.75 ECA, FI (11) Asia (4) 1.50 Latin America & Caribbean (13) 1.25 2000 2003 2007 Note: The number of countries within each group is in parentheses. ECA FI refers to the financially integrated countries discussed in chapter 1 that also responded to the Barth, Caprio, and Levine questionnaire. Source: Barth, Caprio, and Levine 2008. deterioration in lending standards, they might have put countries in a stron- ger position to deal with its consequences. Banks in most ECA countries were neither exposed to toxic assets, although some parent banks were, nor part of a shadow banking system. This implies that improvements in regulation and supervision should be tailored to the problems facing ECA countries at their current stage of development and likely to arise as their systems of financial intermediation develop further. While this agenda does not apply immedi- ately to many low-income and lower middle-income countries in the region, they have much to learn from early comers into transition. With progress in banking supervision and regulation issues until the early 2000s, it is surprising that reform efforts then weakened. ECA countries expe- rienced declines in the ratio of actual to required capital adequacy ratios (fig- ure 3.9a) and in the share of countries responding that the supervisory agency has legal authority to declare a bank insolvent (figure 3.9b). And many of them still face a basic agenda of supervisory and institutional improvement, which owes little to the crisis but nevertheless needs to be addressed (box 3.3 dis- cusses some of these institutional features). Strengthening capital requirements Higher capital requirements for financial institutions--an integral part of the proposed regulatory changes in industrial countries--would be desirable in RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 149 FIGURE 3.9B Supervisory agency can legally declare a bank insolvent (percent of countries that answered yes) Percent 100 Latin America & Caribbean (12) 75 ECA, all (18) 50 ECA, FI (13) Asia (5) 25 0 2000 2003 2007 Note: The number of countries within each group is in parentheses. ECA FI refers to the financially integrated countries discussed in chapter 1 that also responded to the Barth, Caprio, and Levine questionnaire. Source: Barth, Caprio, and Levine 2008. ECA countries.19 Higher overall required ratios of capital to risk-weighted assets--larger than the minimum 8 percent under Basel I and Basel II--have been recommended for countries that face volatile capital movements and corresponding volatile output and relative prices (see box 3.4 for definitions of capital). These requirements apply to foreign bank subsidiaries and domes- tic banks alike and thus do not conflict with EU requirements that specify a minimum capital adequacy ratio for member states. Protect against operational and capital risk. Capital to protect against operational risk and market risk is part of Basel II and would also be desirable for the ECA countries. Operational risk, such as losses from failures of systems and internal processes and from fraud, is an issue in most developing countries. Currently, market risk and capital requirements against it do not appear to be an important issue in most ECA countries. Even so, capital requirements against market risk may prove desirable as the financial system becomes more complex. Moreover, harmonizing capital requirements for market risk, if not already in effect, is prob- ably desirable in the near future, to prevent the risk of a shift of a foreign bank's 19. For example, de Larosičre, Recommendation 1, page 8; also in Financial Services Authority 2009, Recommendation 1, page 7. 150 TURMOIL AT TWENTY BOX 3.3 An agenda for modern banking sector institutions in ECA countries Although there is no painless route to institutional reform, transforming the ownership of banks has been important. And it is at times quite useful to have a policy anchor. For instance, accounting and auditing standards were driven in Central Europe by the EU accession process and by foreign banks' demand for information for basing credit decisions. Similarly, credit registries have had a positive effect on firm financing by increasing availability and lowering costs. The remainder of this box focuses on three sets of institutional reforms: building credit information systems, improving the framework for collateral and the asso- ciated legal framework, and strengthening accounting and auditing.1 The box also includes some summary findings of the many FSAPs and ROSCs in the region. Building credit information systems A good credit registry makes important contributions to financial development and credit access. It provides easy, prompt, reliable access to a potential borrow- er's credit history, both positive and negative. Th is reduces the cost of obtain- ing such information by an individual bank, especially for small borrowers, and thus reduces the costs of financial intermediation as well as nonperforming loans. A good credit registry also increases competition for creditworthy bor- rowers. And it creates important incentives that improve sustainable financial development--increasing the cost of default and allowing borrowers to build up a nontangible asset, a good credit record. All these reasons may explain why there is some evidence that credit registries give a better return than improving the contractual framework, for example,2 and that banks in transition coun- tries with credit registries face lower default risk.3 ECA countries initially lagged somewhat in the worldwide trend to set up credit registries, but in recent years there has been a rush to establish credit reg- istries across the region. In 2004, only 13 countries in the region had credit reg- istries; now all but two have one. Most registries include positive information (whether a borrower has repaid loans) and negative information (defaults). But the young life of many of these registries means that credit histories are short and that the coverage of the credit bureaus is low (box figure 1). These problems can be addressed by: broadening coverage by including nonfinancial payments (such as utilities and cell phones and including data from nonbank fi nancial intermediaries), allowing the existence of private credit bureaus (many ECA countries have done so), and strengthening the system's reliability by assigning each potential borrower a unique identifier number. Improving the collateral framework and the associated legal framework Collateral can be crucial in lending. Indeed, its defi nition and the quality of its execution--creditor rights--determine how well the fi nancial system func- tions. Firms, whether large or small, provide collateral. Similarly, mortgages (continued) RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 151 BOX 3.3 (CONTINUED) An agenda for modern banking sector institutions in ECA countries involve the collateral of the building and associated land as a guarantee of the loan--in other words, an incentive to pay and a guarantee for the lender. The problem with collateral begins with titles in the ECA countries. Land is the main collateral, but land titling is often weak, and cadastres still need reform in many ECA countries. Furthermore, land to which the state may still have title cannot be used as collateral. Many land registries are updated infrequently, and registries of other types of collateral are usually nonexistent. Registries for mov- able property would expand the types of assets that could be collateral. A second problem with collateral is the weakness and cost of the execution of collateral in a default. Creditor rights have improved substantially in many ECA countries, as measured by the laws and regulations, but some countries still have room for improvement (box figure 2). More important than the laws is how they work. An indicator of how laws work with respect to collateral is provided by the estimates of the cost of con- tract enforcement relative to the cost of a debt contract (box figure 3). Costs vary widely across the countries. It appears that lenders in some countries with high legal creditor rights nonetheless encounter high costs in enforcing a con- tract. These costs reflect the slow execution of collateral through the courts. A related issue is the unsatisfactory bankruptcy procedures in many coun- tries, frequently noted in FSAPs and ROSCs. Problems occur in the length of the process, which often allows the firm owners to strip the firm assets. There are also problems with the treatment of creditors, minority shareholders, and, in the case of banks, depositors. Improving accounting and auditing Financial statements are a second element of the information framework that needs to improve. They are critical in assessing creditworthiness of the larger potential borrowers, since small and medium borrowers generally have weak BOX FIGURE 1 Private credit bureau coverage, by country, 2008 Percent 100 75 50 25 0 Sl Pola . Ka Turk . U ep. za ey an ia rze tia ec na ak d Ro stan Ar nia Es ia ian Hu nia de r y Lit tion Bu nia G ria rgy gia be ne Al an erb ia Be jan ce La s nia ia M FYR nt ova Sl gro Ta enia p p u ov n rb Re Re n Az ban do tv Fe nga lar Uz krai ist He Croa Cz vi lga t Ky eor zR ma e me to a ai kis Mo old ra Se go , ov kh en hu jik h & Ma ss nia Ru s Bo Note: Percentage of adult population covered. Source: IFC and World Bank 2008. (continued) 152 TURMOIL AT TWENTY BOX 3.3 (CONTINUED) An agenda for modern banking sector institutions in ECA countries BOX FIGURE 2 Legal rights index, by country, 2008 Index 10 5 0 Uk ep. nia p. Es p. de ey an ia nt via ak o erb e Bu ijan Mo ria Po va Ro land Ar nia H nia ce yz y Se R Cr ia ec tia Ge nia Sl rgia rze nia za ina Lit stan ia be on Be an Ta arus Ma rg r Az rain ov gr Y an do Re rb Re an Ky unga Fe Turk ldo ,F ist lga Cz oa Uz rati t Mo Lat R Sl ene ma me to He ove Ka gov kis a o kh Alb hu l jik h ian & ss ia Ru sn Bo Note: This figure measures the rights of secured and unsecured creditors vis-ŕ-vis enter- prises. Higher numbers indicate stronger protection. Source: IFC and World Bank 2008. BOX FIGURE 3 Cost of contract enforcement, by country, 2008 Percent 50 25 0 G ep. Mo Ser . Bu ep. Ro stan Sl rkey & Al ine Ma rze nia ni a ec YR rgy gia nt bia ak o Lit aria Be nia be us za an Ta stan Ar nia Es ia Tu a erb ia Mo ijan La a ian C via de ia Hu tion Po r y d ep v do vin ov gr ni lan n Az ven Fe roat a ldo Uz lar Cz a, F Ka kist t Ky eor ra R zR R Sl ene ng He ba a ma me to a lg ra ce go i kh hu jik Uk h o ss ia Ru sn Bo Note: This measure indicates the cost of enforcing a contract through the court system, relative to the average amount of a debt contract. Source: IFC and World Bank 2008. accounting and usually need help to even bring it to reasonable levels. Financial statements are also critical in any development of capital markets. Yet another problem is that the accounting statements, like many other pieces of fi nancial information, do not identify the true controllers and owners of firms. Lack of this information complicates financial development in many ways, from raising risk on loans that are excessively concentrated in interlinked conglomerates, to reducing investor interest in funding parts of these firms. There has been international support for improving the accounting stan- dards, and many ECA countries report adopting the International Financial Reporting Standard (IFRS). But firms and banks often maintain their own sys- tems and have been slow to fully adopt this standard. Accounting and auditing are considered weak, and local auditing fi rms are often small and considered (continued) RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 153 BOX 3.3 (CONTINUED) An agenda for modern banking sector institutions in ECA countries unreliable by foreign investors. Some progress has been made through pres- sures from EU accession, at least in Central and Eastern Europe, and from for- eign banks, which require good information. Recommendations from World Bank and IMF FSAPs and ROSCs The World Bank-IMF Financial Stability Assessment Program (FSAPs) and the Reviews of Standards and Codes (ROSCs) provide numerous recommendations on how to improve bank supervision in ECA countries--and many of these are identified in the "de Larosičre" report as also being necessary for industrial countries in the EU. Among those recommendations: · Bank supervisors need greater legal powers to monitor compliance with rules. · Bank supervisors must have independence, legal protection and adequate safeguards against malfeasance. · Supervision departments need to be funded better. · Pay for bank supervisors has to be more competitive to avoid losing talent to the banking industry. · Technical capacity of bank supervisors to evaluate solvency and liquidity risks needs to be strengthened. · Evaluation of loans and liquidity needs should be carried out both statically as well as looking forward. · Bank supervision should include an evaluation of banks' internal risk and liquidity management systems. · An improved regulatory structure for dealing with weak banks (and their intervention and closure) is an important part of bank regulation and has a major role in limiting the future impact of crises. · Transparent, prompt, and monitorable corrective actions are needed-- should be applied automatically. · Corrective actions need to be time-bound requirements to force the owners to put in new capital, or to merge or close down such banks in order to pre- vent the problem from spreading. · Institutional responsibilities and triggers for intervening a bank need to be defined and provided with legal backing--in some countries the courts have intervened to reverse interventions. · Consolidated supervision needs strengthening--for example, to ensure that regulatory limits on single exposures by banks and exposures to bank own- ers can be enforced. · Improve the ability to identify bank and corporate owners--existing owner- ship structures are often opaque and preclude an assessment of fit and proper when issuing banking licenses. Notes 1. A recent survey of fi nance in the transition countries provides valuable analysis and information (Beck 2009). 2. Djankov, McLeish, and Shliefer 2007. 3. Haselmann and Wachtel 2007. Source: Based on background work by Thorsten Beck and James Hanson. 154 TURMOIL AT TWENTY BOX 3.4 Capital--what is it and why require it? Capital consists of owners' equity and retained earnings (Tier I) and general provisions, revaluations of assets, subordinated debt, hybrid equity-debt instru- ments, and other funding instruments (Tier II). Tier I must be at least as large as Tier II. Both the de Larosičre and Turner reports recommend a harmoniza- tion of instruments considered to be capital. Capital requirements on banks represent a buffer against shocks for deposi- tors and the deposit insurance agency. They limit owners' leverage and thus reduce risk. They also represent an incentive to owners to monitor activities as their own funds are at risk. Without such requirements, owners would be tempted to leverage massively, leaving depositors and deposit insurers at risk. Furthermore, the collapse of undercapitalized institutions would have systemic effects. Regulation and supervision attempt to ensure that banks, which are difficult institutions for investors to analyze, maintain the required capital. A bank's market access to loans and capital may depend on maintaining capital in excess of required levels. Under Basel II, total capital must be at least 8 percent of risk-weighted assets, plus, effectively, the estimated value at risk in the institution's proprietary mar- ket activities and the estimated operational risk, which cover the probability of loss from internal fraud, failures of systems, internal processes, and the like.1 The Basel II standard approach suggests risk weights for various types of assets; for example, 100 percent for standard loans, 150 percent for substandard loans, 0 for government debt.2 The ratio of capital to assets may be greater or less than the ratio of capital to RWA depending on market and operational risk. Notes 1. Specifically, the defi nition of risk-weighted assets is adjusted to require capital, in addition to the capital for risk-weighted assets, that is equal to the amounts of market and operational risks. 2. Basel II includes two other ways to calculate risk-weighted assets, but the experience of the crisis has led most observers to recommend a return to the standard approach. activities to a subsidiary in a country with lower requirements--international regulatory competition--a shift that might prove costly to the host country. Protect against foreign currency loans. Additional capital requirements, through higher risk weights, may be desirable for bank loans in foreign cur- rency (or indexed to it), which are important in many ECA countries. These loans are financed from the banks' substantial liabilities in foreign exchange, whether in large loans from parents of subsidiary banks, borrowing in the wholesale and interbank markets, or foreign currency deposits. Countries in Latin America and East Asia also have a large fraction of their loans in foreign RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 155 currency, But the funding source is more likely to be domestic deposits in for- eign currency.20 Current experience and that from earlier crises in Latin America and East Asia suggest that bank lending in foreign currencies is risky. Foreign currency loans imply that the borrowers take a currency risk and the banks a credit risk, assuming that the loans are funded by foreign currency liabilities. The borrowers accept the currency risk because the lower interest rates on foreign currency loans make loans more affordable and offer a better cash flow profile.21 Exchange rate risk was viewed as small either because the countries had hard pegs and an exit strategy (euro adoption) or because the countries were expected to experience an appreciation of the exchange rate as part of the convergence process.22 How- ever, a devaluation increases their debt service burden relative to their incomes because they are in local currency. Even if there is no devaluation, a stabilization program to maintain an exchange rate can generate significant unemployment and losses of income, making it difficult to cover the debt service. Given these risks on loans in foreign currency, it would be desirable to establish higher risk weights for them, implying that more bank capital would be required to make such loans. This policy would increase the buffer against these risks, force the banks' owners to have more at stake on such loans, and raise the cost and reduce the attractiveness to the borrower of such loans. Some ECA countries imposed such a risk weighting toward the end of the credit boom. In some cases, they exempted loans to borrowers with access to foreign exchange. Although this idea seems attractive, it is difficult to measure access to foreign exchange. In addition, a borrower's access may drop sud- denly, or the borrower may simply refuse to use this access to repay an obli- gation in a sharp devaluation, as in East Asia for exporters who sold foreign exchange forward. Another option to create a buffer against the risks of such loans, reduce their attractiveness, and raise their cost would be to increase required general provisions on such loans, which would imply an increase in Tier II capital. 20. In Latin America, depositors wished to hold foreign currency (FX) deposits as insurance against inflation. Beginning in the 1980s, foreign currency deposits were legalized because without them, deposits (and bank loans) would have been even lower than their comparatively low ratios to GDP; in East Asia, the foreign currency deposits can probably be explained as insurance against currency risk--the share of such deposits in East Asia is much smaller than in Latin America. 21. Less of the loan plus interest obligations needs to be repaid in the earlier years of the loan, though the present value of the payments is the same. 22. These risks are not eliminated by using a currency board to maintain the exchange rate, as Argen- tina suggests. Even Hong Kong suffered substantial shocks as a result of maintaining its currency board. 156 TURMOIL AT TWENTY . . . and parent bank funding in foreign currency. Although foreign currency inflows from parent banks to their subsidiaries present a risk, the use of capital requirements on such liabilities would be unusual. These inflows could be with- drawn in the event of problems in the parent bank or concerns that the subsidiary may prove unprofitable because of macroeconomic developments. 23 But capital is a buffer to protect the providers of funds. Instead, as suggested above, reserve requirements could be and have been used to limit such capital inflows, espe- cially if they are short term. A differential reserve requirement would reduce the interest rate paid on such deposits or loans and thereby reduce their attractive- ness and the risks to banks. Such policies for foreign currency loans and deposits would tend to slow overall credit growth only temporarily, until borrowers shift to less regulated parts of the financial system or go to offshore banks directly. Monitor liquidity. Consideration could also be given to introducing a "core funding ratio." Banks that rely heavily on volatile finance and market funding are risky in terms of liquidity, as the current crisis has indicated once again. One approach might be to adapt the proposals for the industrial countries, suggesting that regulators and supervisors should monitor liquidity carefully and consider defi ning a "core funding ratio" to ensure the sustainability of balance sheet growth.24 Such measures would be desirable in ECA countries that have large foreign inflows relative to GDP.25 Provisioning goes dynamic Another set of proposals involve making the capital adequacy regime counter- cyclical.26 Since a constant capital to risk-weighted asset ratio, whatever the state of the business cycle, could create a sudden credit crunch and add to problems in the event of a slowdown in the economy, the proposal is to vary the capital ade- quacy over the cycle. One approach would be to use dynamic provisioning, as in 23. Note that the incentive of taking currency exposure by lending in local currency and borrowing from abroad would be curbed by capital requirements against net foreign currency exposure. 24. G20 Communiqué, Tasking of Ministers and Experts; de Larosičre, Recommendation 1; and Financial Services Authority 2009, page 7. 25. Regulatory actions may substitute or complement monetary policy to limit credit volatility. Requiring higher ratios of capital to lending will limit bank lending. To some extent, such regula- tory actions have similar effects on credit growth as limits on capital inflows, and thus may be at least a temporary solution to the "trilemma" problem of the inconsistency between fi xed exchange rates, open capital accounts, and effective monetary policy. All the various proposals try to create barriers against the international flow of capital into banks; these are motivated by the desire to limit risk in banks, but do not involve barriers to capital flows. 26. G20 Communiqué: Tasking of Ministers and Experts; de Larosičre, Recommendation 1; and Financial Services Authority 2009, page 7. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 157 BOX 3.5 Taking the rough with the smooth--dynamic provisioning in Spain Dynamic provisions are a mechanism to accumulate reserves in the good times so that they can be spent in the bad times. In the standard system, bank provi- sions depend on contemporary bad loans: in the good times, when credit grows and there are few bad loans, provisions tend to be low, and this fuels further credit growth. In the bad times, credit declines and bad loans increase, so that banks have to make a greater provisioning effort, which ultimately feeds back into credit contraction. In contrast, the idea behind dynamic provisioning is very simple: to reduce the tendency of the fi nancial system to amplify credit fluctuations, it makes sense to smooth the provisioning effort along the cycle. Until recently, this sys- tem had been adopted only in Spain. The Bank of Spain introduced this mecha- nism in 2000, shortly after euro adoption. Having lost monetary policy and the exchange rate as macroeconomic tools, the central bank saw with concern how higher economic growth than in other Euro area economies induced high credit growth and further inflationary pressures. Unable to use the traditional instruments of monetary policy, the Bank of Spain used a "macro-prudential" tool to try to fight the overheating of the economy. The system, introduced in 2000, was initially based on three types of provi- sions: specific (a function of contemporary bad loans), generic (a function of the credit stock), and the new statistical provision (depending on credit growth). A fund was accumulated, with the intention of using it in the downturn. In 2004 the statistical provision was subsumed in the generic provision according to: Generic provisions = Credit + Credit ­ specific provisions where the parameters and vary with the riskiness of the assets: 0 2.5 and 0 1.64. The 2004 reform also introduced limits to the Fund between 33 percent and 125 percent of credit weighted by . These limits were a response to concerns that the accumulation of provisions could become excessive, and put Spanish banks at a competitive disadvantage in the single European market. How has the system worked compared to expectations? In a context of very strong growth and low risk aversion, dynamic provisioning hardly discouraged credit growth between 2000 and 2007. After the 2004 reform, most entities rap- idly reached the ceiling of the Fund, and in 2006 credit grew again above 25 percent. While the dynamic provisioning had little impact on credit growth, it helped create a buffer that was useful when the crisis hit in 2008--specifically, the ratio of provisions to bad loans in December 2007was above 200 percent in Spain, compared with an EU average of 70 percent. After the crisis started, the Spanish banking system used the funds to mitigate the impact of delinquencies on total provisions. The decrease in generic provisions fully compensated for (continued) 158 TURMOIL AT TWENTY BOX 3.5 (CONTINUED) Taking the rough with the smooth--dynamic provisioning in Spain the increase in specific provisions. For the period as a whole dynamic provisions did not eliminate the procyclical pattern of total provisions (box figure 1). In sum, the Spanish dynamic provisioning model did not prevent the credit and housing boom or the subsequent bust. But it allowed financial institutions to distribute the provisioning effort more evenly along the cycle. The system can certainly be improved as more information is accumulated in calibrating the economic cycle. But it is preferable to the standard model, which explains why the Financial Stability Board has opened a discussion on dynamic provi- sioning, including the possibility that capital requirements (Basel II) should also be designed more countercyclically. BOX FIGURE 1 Provisions over credit and GDP, 1999­2009 Percent Percent 5 6 Change in GDP (right axis) 4 4 3 2 Ratio of generic to credit (left axis) Ratio of total provision 2 to credit (left axis) 0 1 Ratio of specific to credit (left axis) ­2 0 ­4 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Note: Provisions are corrected for the impact of the new accounting regulation in 2004. Source: Fernandez de Lis and Herrero 2009; Fernandez de Lis, Pagés, and Saurina 2001; and Financial Stability Board 2009. Note 1. In late 2008, the Peruvian banking regulator adopted a similar system aimed at smoothing provisions along the cycle. The Peruvian system is based on GDP growth instead of credit growth and is thus systemwide as opposed to institution-specific. Spain (box 3.5). The required annual rate of general provisions would increase during periods of rapid credit growth, and the stock of such provisions--part of Tier II capital--would be run down during slowdowns. Countercyclical capital requirements might also be considered. In either case, care would be needed in defining the appropriate cycle for the measures to take effect. Strengthening capital requirements should be phased in carefully to avoid sudden declines in bank lending as the current crisis winds down. Parallelism with EU requirements will occur naturally for the new member states. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 159 Supervising foreign banks The dominant role of cross-border banks in many ECA countries raises important issues of supervision for both the host and home countries as well as parent banks. Host country supervision is important since, at the most basic level, bank supervisors must have legal powers, including sanctions, to ensure that banks comply with applicable rules. Moreover, subsidiary banks represent a potential obligation to the deposit insurance agencies and the gov- ernment budget of their host countries, in the event of failure of either the subsidiary or the parent bank. Even so, there is probably a tendency for many ECA host countries to rely on the effectiveness of supervision in the home country, in part because of the difficulties of supervising a bank belonging to a bank in another country. This is quite complicated, however, because supervi- sors of different countries may have different views on major issues. Improving the supervision of subsidiaries is difficult for several reasons.27 First, home country supervision depends on the quality of its consolidated supervision, including supervisor scrutiny of the subsidiaries, both of which may vary across states. Second, the subsidiaries may generate much of the income of banks that own them.28 Correspondingly, they represent potential problems if there is an economic deterioration in one or more of the subsid- iaries' host countries. Problems in the home country or with the subsidiar- ies could affect not only the home country but also other ECA host coun- tries, through removals of funds from the host countries by the parent bank or deposit shifts away from subsidiaries in the host countries. A typical approach to improving the supervision of international banks is to improve the exchange of information. Cross-border supervision has tradi- tionally focused on improving information-sharing through memoranda of understanding between supervisors. The World Bank-IMF FSAPs in various ECA countries have often noted the lack of such memoranda of understand- ing. But memoranda of understanding may not always lead to sharing critical information. If an institution is systemically important in both countries, countervail- ing incentives that will help supervision are at work. To some extent, incen- tives exist to share information and reach common solutions to problems. But there are also incentives keeping the supervisors apart--for host countries 27. See also de Larosičre, annex III. 28. For example, Raiffeisen Zentralbank Österreich AG (RZB) holds 70 percent of the shares of Raif- feisen International Bank Holding AG that operates fully consolidated bank subsidiaries in fi fteen transition countries. In 2008, according to its annual report, the non-Austrian subsidiaries accounted for about 75 percent of RZB's income. 160 TURMOIL AT TWENTY to "ring fence" subsidiaries to protect depositors and ultimately to limit the potential costs to the deposit guarantee system and the government budget, which may represent a legal obligation, and for home countries to centralize a bank's assets and keep liabilities decentralized. These issues will worsen as the situations of the parent and the subsidiary bank diverge. Both authorities may need to rely on the incentives and willingness of the other authority, which may not have the flexibility to provide a solution, because the home and host countries are not accountable to each other in an insolvency. The management of the Fortis-group serves as real example of this case.29 These cross-border issues are even more complicated for ECA countries outside the EU. The cost to the host countries' deposit guarantee scheme could be sub- stantial in the extreme case of a failure of a group with a large subsidiary. The legal obligations of consolidated groups to foreign subsidiaries vary even among EU member states. The group owner may simply choose to with- draw funds from or abandon a failing large subsidiary, again at a cost to the host country's deposit guarantee scheme. Th is option is complicated for the subsidiary owner by the risk that such a failure would lead to deposit runs against other subsidiaries and perhaps even in the home country. But the group might view this risk as less important than the cost of resolving the subsidiary. In any case, the cost of a failing bank subsidiary can be high, and improved national regulation and supervision of them will be impor- tant, particularly for non-EU member states, as well as participation in coor- dinated actions to the extent possible, as in the current crisis.30 Finally, the fiscal support for weak or failing global institutions has so far been orga- nized in the current crisis by the home country, and this support may be con- strained by both national and EU laws. The global character of the current crisis has led to numerous proposals for improving supervision across borders. The de Larosičre report has gone beyond simply recommending better sharing of information and proposed a new financial architecture for the EU, comprising: · A new EU body, the European Systemic Risk Council (ESRC), to pool information relevant for fi nancial stability and issue macro-prudential warnings for the EU and to deal with national supervisors. 29. See a discussion in de Larosičre, p. 73. 30. One example is the "gentlemen's agreement" of the European Banking Coordination Initiative (see box 2.1), which has signed up international banks to maintain their exposure in host countries and set up stress testing arrangements. Another is the EBRD, World Bank, and EIB initiative to lend to banks in the region, in addition to the various measures for greater international exchange of supervi- sory and macro-prudential information. These initiatives are in addition to the various proposals for greater international sharing of information. RESTRUCTURING BANK, CORPORATE, AND HOUSEHOLD DEBT 161 · A European System of Financial Supervision--for banks, nonbanks, and financial markets. It would also intensify efforts in training and coordinate the application of national supervisory standards. It would set up supervi- sory colleges for major cross-border financial firms. And it would develop a consistent, strengthened set of supervisory standards; and ensure that appropriate information flows to the ESRC. National supervisory authorities would remain responsible for day-to-day supervision, but the proposal would eventually entail adoption of binding supervisory standards, legally binding mediation between national supervi- sors, and licensing specific EU institutions such as credit rating agencies and trading infrastructures. But the issue is complicated because legally binding arbitration and burden sharing among the countries would have implications for fiscal expenditures. The EU is considering these proposals, and it remains to be seen how they will develop and apply to ECA countries. 162 TURMOIL AT TWENTY CHAPTER 4 Scaling up social safety nets The sharp output declines expected in most ECA countries in 2009 and the possibility of a slow recovery will have a significant impact on the poorest and most vulnerable households. Existing safety net programs, if they have a good track record of targeting benefits, can be expanded to reach these groups. Virtually all ECA countries operate a mix of safety nets today, but the spend- ing, coverage, and ability to target resources vary greatly. Questions · Does every country possess at least one well targeted program that could be scaled up to channel additional resources to the poor cost-effectively? · In which situations is the need simply to scale up the program fi nancially and in which is it critical to improve efficiency by consolidating programs and improving targeting? · How important is the income transfer of some well targeted programs to poor households? Findings · Of the 24 ECA countries examined, 15 have a well targeted program that could be scaled up. Encouragingly, the potential to build on an existing well targeted program is not limited to middle-income countries--several lower income countries are among the 15. · In some of the other countries, spending is adequate but effectiveness is compromised because of weak targeting and implementation. · The income transfer of some of the targeted means-tested programs is sig- nificant as reported by households in the poorest quintile. In several ECA countries--including some of the lower income ones--the transfer amounts to more than 30 percent of their consumption. The significant output declines expected in most ECA countries in 2009 and the possibility that output recovery may be a slow process will hit the poorer and most vulnerable. Reduction in real wages and employment, increases in food prices due to currency devaluations, and declines in social services as public revenues dwindle are some of the mechanisms likely to hurt the poor. Declines in workers' remittances in some of the poorer ECA countries will have 163 a disproportionate impact on the poor. For families already below the poverty line, these events will be particularly serious, especially for those with restricted coping mechanisms, such as the elderly and those with young children. Latest available data confi rm these developments. Registered number of unemployed workers, as reported by public employment offices provid- ing unemployment benefits, increased sharply in several countries between March 2008 and March 2009. The number more than doubled in the Baltic states. The increase was also significant in other countries: about 60 percent in Moldova and Turkey, and between 20 and 40 percent in the Czech and Slo- vak Republics, Romania, the Russian Federation, Slovenia, and Ukraine. Dif- ferences in these rates partly reflect incentives to register and generous ben- efits. Notwithstanding these caveats, the data suggest that unemployment is increasing in many ECA countries. Poverty is also on the rise: the Russian sta- tistical agency Rosstat reported in late August 2009 that the population below the national poverty line increased by almost one-third--from 13.4 percent of the population in the last quarter of 2008 to 17.4 percent during the first quar- ter of 2009--or, an additional 6 million individuals in poverty. Safety nets or social assistance programs (other than contributory pen- sion programs and contributory unemployment benefits programs) can be leveraged to reach the poor. But the ability to respond quickly depends criti- cally on the quality of the existing programs. A rapid response would focus on expanding existing safety nets that have a proven track record in target- ing social benefits to the poorest fift h of households. This chapter shows that most countries in the region have at least one well targeted safety net program that can be scaled up in times of crisis. Well targeted programs should be pro- tected and even expanded, either in coverage or by "topping-up" benefits.1 Existing social assistance programs Total spending on social assistance averages 1.7 percent of GDP in the region but there is substantial variation across countries, ranging from 0.5 percent of GDP in Tajikistan through 2.0 percent in Ukraine to around 4.0 percent in Croatia and Bosnia and Herzegovina (figure 4.1). Spending on benefits for war veterans is significant in the former Yugoslav republics. About half of ECA countries spend about the same as or more than Argentina and Brazil--middle-income countries in Latin America with the highest spending as a share of GDP--but most ECA countries are below the OECD average of 2.5 percent of GDP. 1. Th is chapter is based on the material developed in Lindert (in progress) and Nguyen, Sundaram, and Tesliuc (in progress). Data on comparator countries comes from Lindert et. al 2006. 164 TURMOIL AT TWENTY FIGURE 4.1 Spending on overall safety nets, by country, 2006­08 Percent of GDP 5 4 3 2 1 0 Tajikistan Turkey Azerbaijan Georgia Macedonia, FYR Kyrgyzstan Kazakhstan Kosovo Poland Bulgaria Montenegroa Lithuania Belarus Estonia Latvia Armenia Romania Albania Russian Federation Serbia Moldova Uzbekistan Ukraine Hungary Bosnia and Herzegovina Croatia Peru Colombia Chile Mexico Guatemala Brazil Argentina OECD Note: Spending data for Latin America are for 2000­04. Almost all ECA countries operate some mix of safety nets programs. The mix is similar to that of the OECD countries, with an emphasis on family allowances such as child allowances and birth grants, noncontributory pen- sions, heating and housing allowances, and targeted antipoverty programs. Some countries are still reforming such programs, such as monetizing and consolidating a range of "privileges", including so-called categorical benefits that target particular categories of individuals (such as war veterans, teachers, and judges). In some cases a proliferation of multiple programs fragments and duplicates benefits. Coverage--the share of households in the poorest quintile of the popula- tion reached by social assistance programs--also varies sharply across coun- tries, with more than half the countries having coverage rates between 40 and 60 percent (figure 4.2). Hungary and Romania have the highest coverage rates, followed by Belarus, Poland, and the Russian Federation, through moderate rates in Armenia, Azerbaijan, and Turkey, falling off in the low-income CIS countries, reaching only 30 percent of the poorest quintile in Georgia and the Kyrgyz Republic and barely 1 percent in Tajikistan. Coverage is not always correlated with spending. Bosnia and Herzegovina is among the highest spenders but has rather low coverage, because many benefits are "rights based" rather than "needs based." More than three-quarters of cash transfers are rights-based transfers. Safety nets in Croatia have a similar feature. Many of its programs cover war veterans, children's allowances, and other family benefits that are not targeted. But Croatia also operates a well targeted guaranteed SCALING UP SOCIAL SAFETY NETS 165 FIGURE 4.2 Coverage of overall safety nets, by country, various years Share of poorest quintile covered (percent) 100 75 50 25 0 Tajikistan (2007) Turkey (2005) Bosnia and Herzegovina (2007) Montenegro (2006) Georgia (2007) Kyrgyz Rep. (2006) Kosovo (2006/07) Bulgaria (2007) Albania (2005) Macedonia, FYR (2005) Uzbekistan (2003) Armenia (2006) Azerbaijan (2008) Croatia (2004) Kazakhstan (2007) Moldova (2007) Lithuania (2004) Ukraine (2006) Serbia (2007) Latvia (2007) Estonia (2004) Poland (2005) Belarus (2007) Russian Federation (2002) Romania (2004) Hungary (2004) Argentina (2004) Brazil (2004) Mexico (2002) Colombia (2004) Guatemala (2000) Chile (2003) Peru (2004) Note: Performance indicators for ECA countries are calculated using a standardized methodology so as to facilitate comparison across countries. This methodology ranks households into quintiles based on harmonized consumption aggregates (World Bank 2005) and pre-transfer consumption per capita. Azerbaijan, Bosnia and Herzegovina, Kyrgyz Republic, Moldova, the Russian Federation, Tajikistan, and Uzbekistan are yet to be updated using the standardized methodology. Since this book went to press, data in 2007 for Armenia have become available. Moreover, the standardized methods have since been applied to Bulgaria and Serbia. For Bulgaria, coverage of social assis- tance increases from 40 percent to 54 percent and targeting accuracy from 28 percent to 50 per- cent For Serbia, coverage of social assistance increases from 56.7 percent to 60.3 percent and targeting accuracy from 56 percent to 63 percent. These changes have been reflected in figure 1 in the overview. The indicators of the rest of the countries are being updated with the standardized methodology. minimum income program as its core poverty safety net. By contrast, Azerbaijan spends relatively little, but its new Targeted Social Assistance Program is quite effective in basing eligibility on income testing and home visits. The targeting accuracy of social benefits--the share of benefits going to the poorest quintile of households--varies across the region, but most ECA coun- tries perform better than comparable countries (figure 4.3). However, in sev- eral countries the poorest quintile receives less than a third of the safety net benefits. Note that the ranking of countries has changed relative to figure 4.2. For example, at 80 percent, the Russian Federation has relatively good cover- age, but only 30 percent of benefits go to the poorest quintile. The main reason for this gap is that the per child transfer out of the children allowance pro- gram, the main benefit received by the poorest quintile, is very low. In Georgia, by contrast, only about 30 percent of the poorest quintile receives benefits, but those benefits account for more than 50 percent of the total value transferred by the safety net system. Targeting accuracy is 166 TURMOIL AT TWENTY FIGURE 4.3 Targeting accuracy of overall safety-net benefits, by country, various years Share of benefits received by poorest quintile (percent) 100 75 50 25 0 Turkey (2005) Tajikistan (2007) Bosnia and Herzegovina (2007) Belarus (2007) Latvia (2007) Moldova (2007) Bulgaria (2007) Russian Federation (2002) Uzbekistan (2003) Kyrgyz Rep. (2006) Kazakhstan (2007) Estonia (2004) Ukraine (2006) Poland (2005) Hungary (2004) Romania (2004) Albania (2005) Azerbaijan (2008) Croatia (2004) Lithuania (2004) Macedonia (2005) Georgia (2007) Serbia (2007) Armenia (2006) Montenegro (2006) Kosovo (2006/07) Guatemala (2000) Mexico (2002) Chile (2003) Peru (2004) Argentina (2004) Colombia (2004) Brazil (2004) Note: See note to figure 4.2. significant and on par with several OECD countries. It is largely achieved by the new Targeted Social Assistance Program, implemented in 2006 after care- ful preparation (box 4.1). Safety nets: ready to be scaled up? Countries are classified in two categories by asking the question: "Does the country possess at least one well targeted program that could be cost effec- tively scaled up to channel resources to the poor in a crisis." Figure 4.4 classifies countries by income level and (in parentheses) the spe- cific program with the potential to be expanded. They are generally means- tested programs. Figures 4.5 and 4.6 show the targeting accuracy and cover- age of such means-tested programs. The classification in figure 4.4 is not based on the performance of the overall safety net but on the existence of at least one well targeted program with minimal coverage that could be scaled up to reach the poorest households. Many countries that have such a means-tested pro- gram nevertheless require significant reform of their overall safety net. Countries with at least one well targeted means-tested program Most countries operate at least one well targeted safety net that could be scaled up in times of crisis. In many ECA countries the targeting accuracy compares favorably with that in Latin American. Interestingly, the potential to build on such programs is not limited to middle-income countries; several lower income countries are in the group. SCALING UP SOCIAL SAFETY NETS 167 BOX 4.1 Aiming high to serve the poor: Georgia's new Targeted Social Assistance Program The TSA, launched in July 2006 after 18 months of intense preparation, has sev- eral features worth noting. · The team: The authorities appointed a team of national experts to develop a targeting mechanism, examining experiences in several countries, includ- ing Mexico, Chile, and Turkey. · Targeting method: A "proxy means test" (complementing data on income with information of possession of durables, quality of the dwelling, education level of the household, and so on) was selected because of the high informal- ity of the Georgian economy. The test was then finetuned through additional surveys. The formula accommodates differences in consumption across geo- graphical locations. It combines both objective (measurable) variables and subjective assessment of the household situation by social workers. · Administrative arrangements: The application process has two stages. First, households register and give their consent to further collection of informa- tion. The blank registration forms can be obtained at many points across Georgia. In the second phase, the Social Service Agency visits the house- hold to collect the additional information. The data are entered in the data- base and locked for changes. Further processing is done centrally, includ- ing information verification and cross-checks, as well the assignment of the proxy score. The entire process is automated to assure objectivity and mini- mize the risk of bribery. · Informing households: Each "scored" household is issued a card with the case number and score, allowing households to fi nd out immediately whether they qualify for assistance. · Program flexibility: The program is open-ended. Households can apply--at any time--if their score is below the cut-off point they automatically become included in the program. This poses a challenge for forecasting the required resources and calls for budgetary flexibility. Several low-income to lower middle-income countries have developed some elements of effective safety nets. Typically, these countries have a single well targeted program focused on poor families, such as the Unified Monthly Benefit in the Kyrgyz Republic, the Ndhima Ekonimike Program in Albania, the Targeted Social Assistance Programs in Azerbaijan and Georgia, the Fam- ily Benefit Program in Armenia, and the Last Resort Program in Ukraine. Depending on the country, further improvements could come from geo- graphic targeting with poverty maps, verifying income-testing with hybrid proxies, and strengthening implementation, registry administration, and oversight. Coverage is typically low, however--generally less than 35 percent 168 TURMOIL AT TWENTY FIGURE 4.4 How ready are ECA's safety nets for rapid crisis response? A typology of countries Countries with no programs Countries with at least one or with programs that have well targeted program, though very low coverage or weak often with low benefit values targeting accuracy Low-income and lower Low-income and lower middle-income countries middle-income countries Middle-income and upper middle-income countries Middle-income and upper middle-income countries (figure 4.6). It could be expanded with increased fiscal effort, possibly from consolidating other legacy "privileges" and improving outreach to the poor. Several middle-income countries also operate some well targeted means- tested programs, often as a complement to extensive social protection systems. More than 80 percent of the benefits of the Romanian GMI program accrue to the poorest quintile of households. Eligibility for these programs is usually determined by means testing (and in a few cases, proxy means testing), with targeting outcomes ranging from reasonable to strong. Except for the Turkey Green Card, which has both good coverage and targeting accuracy, cover- age of these specific programs is fairly low. The poor in these countries also tend to benefit from other universal or categorical social assistance programs. In many cases, countries are also implementing "second-generation enhance- ments" to their safety nets--linking beneficiaries to "activation" programs, such as job search assistance or professional training, and introducing "exit strategies," such as time limits, to discourage welfare dependency. Although the existence of at least one fairly well targeted program positions these countries for a faster crisis response, most are underfunded and have modest coverage. Special effort is needed to expand these programs during the SCALING UP SOCIAL SAFETY NETS 169 FIGURE 4.5 Targeting accuracy of means-tested programs, by country, 2004­08 Share of benefits received by poorest quintile (percent) 100 75 50 25 0 Romania (GMI) Ukraine (XP) Croatia (S. Welfare) Kosovo (SA) Turkey (CCT) Montenegro (FMS/MOP) Estonia (MT Benefits) Kazakhstan (TSA) Lithuania (S. Benefit) Serbia (MOP) Bulgaria (GMI) Hungary (Regular SA) Georgia (TSA) Turkey (Green Card) Armenia (FB Prog) Azerbaijan (TSA) Macedonia, FYR (SFA) Poland (SW benefits) Kyrgyzstan (UMB) Albania (NE Prog) Serbia (CA) Uzbekistan (SA) Russian Federation (CA) Bosnia and Herzegovina (CPA) Brazil (Bolsa Familia) Chile (Solidario) Jamaica (CCT) Mexico (Oportunidades) Argentina (HHH) United States (Food stamps) United States (TANF) Note: Since this book went to press, the standardized methods have been applied to Bulgaria and Serbia. For Bulgaria, coverage of GMI increases from 6 percent to 14 percent, and targeting accu- racy from 64 percent to 77 percent. For Serbia, targeting accuracy of the MOP program increases from 64 percent to 77 percent, while coverage remains basically the same. These changes have been reflected in figure 1 in the overview. The indicators of the rest of the countries are being updated with the standardized methodology. present crisis by raising benefits and expanding coverage by enhancing out- reach, such as proactive registration of potentially poor households, commu- nications campaigns, and social worker involvement. This is critical because preliminary data from a few countries show the opposite trend--coverage has declined during the last twelve months. Countries where spending is adequate but programs are not well targeted Several ECA countries allocate enough resources to social assistance pro- grams, but their ability to reduce poverty or cost-effectively channel resources to the poor is compromised by weak targeting and implementation. Fortu- nately, only a few countries fall into this category (figure 4.4). Moldova spends about 1.8 percent of GDP on about 15 untargeted programs. Bosnia spends 4 percent of GDP on untargeted categorical benefits for civilians and war vet- erans. And the Russian Federation spends 1.7 percent of GDP on categorical benefits and weakly means-tested subsidies--and suffers from implementa- tion and design challenges (box 4.2). Belarus may also fall into this category, spending 1.5 percent of GDP on social assistance. But it has recently intro- duced a targeted program, and an assessment of its effectiveness awaits house- hold survey data. In these cases, significant reforms are generally needed to: 170 TURMOIL AT TWENTY FIGURE 4.6 Coverage of means-tested programs, by country, 2004­08 Share of poorest quintile covered (percent) 50 40 30 20 10 0 Kazakhstan (TSA) Uzbekistan (SA for low income) Serbia (MOP) Bulgaria (GMI) Estonia (MT Benefits) Bosnia and Herzegovina (CPA) Hungary (Regular SA) Lithuania (S.Benefit) Ukraine (XP program) Croatia (S. Welfare) Montenegro (FMS/MOP) Azerbaijan (TSA) Poland (SW benefits) Georgia (TSA) Romania (GMI) Kyrgyz Rep. (UMB) Serbia (CA) Turkey (CCT) Albania (NE Prog) Kosovo (SA) Macedonia, FYR (SFA) Armenia (FB Prog) Russian Federation (Child allow) Turkey (Green Card) Chile (Solidario) Argentina (HHH) Mexico (Oportunidades) Note: See note to figure 4.2. · Consolidate programs and eliminate most untargeted privileges. · Refocus design and eligibility criteria, or introduce a new poverty-focused program within the available spending envelope. · Develop and apply targeting tools. · Improve implementation arrangements. · Strengthen oversight, monitoring, and evaluation. How important are safety nets in transferring income? How much of the consumption by the poorest quintile is accounted for by income transferred by safety nets, and what would it cost to increase such transfers? To answer this question, households are sorted into quintiles based on pretransfer consumption.2 In table 4.1 panel A, each cell contains the total transfer amount to beneficiary households in the poorest quintile divided by the total of post transfer consumption of beneficiary households in that quin- tile. Table 4.1 panel B shows the total transfer to beneficiary households in the poorest quintile divided by the total consumption of all households, includ- ing those not receiving the transfer. The first column of each panel shows the 2. Household surveys which include detailed modules on consumption and income, including income from various social protection programs are used in producing these tables. The detailed consumption data are used to produce harmonized consumption aggregates (using the same basket of goods) across the various countries so as to facilitate comparison. Households are ranked into quintiles based on pre-transfer consumption levels. SCALING UP SOCIAL SAFETY NETS 171 BOX 4.2 Enough bang for the buck? Safety nets in the Russian Federation The overall spending on non-contributory social assistance programs and sub- sidies in the Russian Federation stood at about 1.7 percent of GDP in 2007, mak- ing the Russian Federation a "moderate spender" in the region. Th is includes three types of programs: subsidies and cash compensation for privileged citi- zens (specific categories); income-tested programs (mainly housing and utility allowances, and child allowances); and institutionalized care. The Russian Fed- eration does not operate a "last-resort" poverty-targeted program at the federal level, other than means-tested housing, utility, and child allowances. Despite spending enough, the impact of the safety net on poverty reduction is low. The share of funds that actually reach the poor is substantially smaller in the Russian Federation than elsewhere in the region. There are two reasons for this. Most privileges are assigned to broad categories of the population and are not income-tested or needs-based--thus only 16 percent of payments of privi- lege benefits reached the poor in 2006. And even the income-tested programs have poor targeting--only 33 percent of child allowances and 30 percent of housing benefits reached the poorest quintile in 2006. The reasons for weak targeting include: · Program design flaws, such as very low child allowance benefits, which do not appear to justify the higher administrative costs associated with means- testing; · Delegation of delivery and fi nancing of income-tested programs to sub- national governments, which over-stretches the capacity of poorer regions that have larger share of the poor populations and fewer resources · Implementation challenges, particularly in the measurement of income and in sharing information among government bodies. Thus the ability of the Russian Federation's safety nets to protect the poor cost-effectively from the impact of the crisis is limited. Reforms to strengthen the safety net are needed, in program design, targeting, and implementation. The federal government might consider introducing a new federally funded pro- gram with a stronger targeting instrument to identify the truly poor. Develop- ing a targeting instrument--such as a Hybrid Means Testing tool--could hap- pen fairly quickly. A new program could be introduced 6­9 months later--say in early 2010--using budget reallocation options and regional implementation capacities. The program should include strong federal oversight, monitoring, and evaluation to ensure program effectiveness and minimize fraud and error. transfers received from overall social safety nets, while the second column includes only income transfers out of the targeted means-tested programs. The difference between panels A and B is important. Panel A shows the consumption impact only on households that receive the transfer, whereas 172 TURMOIL AT TWENTY panel B averages the total transfer across all households in the quintile--irre- spective of whether they received it or not (the denominator is larger). Figures in panel B are generally smaller, reflecting smaller household consumption in the beneficiary households or the fact that only a share of households in the quintile receives the benefit. This difference can be large: for instance, in Cro- atia, the average transfer value per beneficiary of its means-tested program is 28.8 percent (panel A). But it declines to 4.4 percent of consumption when the consumption of all households (both beneficiary and nonbeneficiary) in the poorest quintile is included (panel B). The reason is that Croatia's means- tested program covers only 15 percent of households in the poorest quintile. Equal declines are observed in the Baltic states, Hungary, and Serbia. There is considerable variation across countries in the transfer effect of the overall safety net (table 4.1, column 2). This variability can be seen even among the new member states of the European Union. In Hungary and Estonia, the safety net transfers per individual (relative to consumption) are significantly larger than in the other countries, such as Poland and Latvia. There are also strong differences in the former Soviet Union. In Georgia the transfers are the largest of all ECA countries (52 percent) and are twice the values in Ukraine and Kazakhstan. Means-tested programs account for an important share of all safety net programs. They appear to have a good impact when expressed per beneficiary- --which is exactly what these targeted programs are expected to do. In theory, they target the most vulnerable within the poorest quintile, as well as other vulnerable groups in other quintiles. There are 10 countries in the group where these programs transfer at least a fi ft h of the equivalent (posttransfer) consumption level of the household receiving the transfer.3 The highest transfer is again in Georgia, followed by Kosovo and Estonia. (For details on Georgia's program, see box 4.1). Cost of expanding means-tested programs How much would it cost to expand means-tested programs? This can be cal- culated by taking the ratio of the cost of such programs over the cost of overall safety nets and multiplying it by the cost of safety nets as a share of GDP (the latter figure derived from figure 4.1). Table 4.2 shows an approximation of the ratio of means-tested safety net transfers to total safety net transfers. It is an approximation because it does not incorporate administrative costs, which may be higher for the most targeted program. The ratio is small for the Bal- tic states and Hungary (means-tested transfers are less than one-fifteenth of 3. Th is is equivalent to one quarter of the pretransfer consumption. SCALING UP SOCIAL SAFETY NETS 173 TABLE 4.1 Transfer to beneficiaries in the poorest quintile, by country, various years (percent) Panel A Panel B Share of consumption of Share of consumption beneficiary households of all households in in poorest quintile poorest quintile Targeted Targeted Safety net program Safety net program (all social (means (all social (means Country Year assistance) tested) assistance) tested) Albania 2005 19.2 9.7 8.4 3.5 Armenia 2006 33.1 32.7 17.7 15.6 Bosnia and Herzegovinaa 2007 24.9 11.9 .. .. Croatia 2004 26.5 28.8 15.1 4.4 Estonia 2004 42.0 40.2 28.4 3.1 Georgia 2007 52.0 48.0 19.2 12.2 Hungary 2004 45.8 26.9 43.3 2.0 Kazakhstan 2007 20.6 13.2 11.5 0.4 Kosovo 2006­07 nab 43.4 nab 18.2 Kyrgyz Rep. 2006 10.1 9.4 3.3 2.4 Latvia 2007 15.6 3.7 9.8 0.2 Lithuania 2004 35.4 20.4 20.5 1.8 Macedonia, FYR 2005 29.2 26.9 13.4 11.5 Moldovac 2007 9.4 nad 5.2 nad e Montenegro 2006 47.9 .. 12.2 ..e Poland 2005 11.2 9.2 8.5 1.6 Romania 2004 33.1 32.1 30.5 6.9 Serbiac 2007 39.3 22.2 23.1 1.0 Turkey 2005 15.8 ..f ..f ..f Ukraine 2006 23.5 17.2 13.6 2.2 .. is not available. na is not applicable. a. These computations use per adult-equivalent national consumption aggregates. b. All social assistance in Kosovo is targeted. c. Calculations use national consumption aggregates rather than harmonized consumption aggregates. d. Moldova has no means-tested program. e. Although Montenegro has a means-tested program called MOP, the 2006 HBS did not ask about it. f. Turkey has two means-tested program, CCT and Green Card, but survey data do not distinguish them. Source: Nguyen, Sundaram, and Tesliuc (in progress). total safety net transfers in these countries), higher in Poland, Romania, and Ukraine (between a tenth and a fi ft h), and significantly higher in the low- income and lower middle-income countries of the Western Balkans and the former Soviet Union (between half and three-fourths ). 174 TURMOIL AT TWENTY The ratios in table 4.2, in conjunction with the magnitudes in figure 4.1, estimate what it would cost to double total spending on means-tested pro- grams as a share of GDP. In the lower income countries of the Western Bal- kans and the CIS, it would cost between 0.50 and 0.75 percent of GDP. In most of the new member states of the European Union and Ukraine, the cost ranges from 0.10 to 0.40 percent of GDP. An exception is Croatia, where doubling the last resort program would cost a further 0.77 percent of GDP; but those costs could be reduced if the expansion of the last-resort program were accompa- nied by significant reforms of the overall safety net. An opportunity for further reform Crises can present an opportunity for further reforms, as public pressures for an effective response can increase the political will for action. Reforms cover both policies that can be adopted in the immediate or short run, and actions which require more time, such as technical and institutional measures to strengthen safety net systems. Since this crisis is expected to be protracted, the second set of actions also merits consideration. Immediate reform measures could typically include the following: · Eliminating, reducing, or at most maintaining nominal values of untar- geted benefits. · Removing automatic indexation of benefits to wages. · More ambitiously, consolidating benefits into fewer well targeted schemes and introducing new targeted programs (such as conditional cash transfers or workfare), though their development and implementation will require more time. Measures to strengthen the overall safety net for the medium run require additional technical inputs: · Developing and introducing improved targeting and eligibility mecha- nisms, such as hybrid means testing.4 · Strengthening and improving household registries and management infor- mation systems, which typically involves technical, institutional, and IT investments, as well as capacity-building at central and local offices. 4. Targeting in times of crisis can be challenging. Family circumstances change more frequently, increasing the need for repeated recertification for eligibility. Participation in the informal economy generally expands, making measurement of "needs" (incomes) more difficult. Finally, policymakers face the dilemma of prioritizing vulnerable populations: the "old poor" versus the "new poor." While political pressures often come from those who face sudden income losses in times of crisis (some becoming "newly poor"), the chronic poor tend to suffer the most adverse consequences of shocks due to their low asset base and inability to smooth consumption, many with irreversible consequences for human capital (reduced schooling, child health and nutrition). SCALING UP SOCIAL SAFETY NETS 175 TABLE 4.2 Percentage of transfers through means-tested programs to overall safety net transfers, by country, various years Percent of means tested transfers Country Year in total safety net transfers Albania 2005 39.50 Armenia 2006 56.75 Croatia 2004 17.84 Estonia 2004 5.99 Georgia 2007 54.02 Hungary 2004 2.98 Kazakhstan 2007 1.83 Kosovo 2006­07 100.00a Kyrgyzstan 2006 59.56 Latvia 2007 2.68 Lithuania 2004 6.44 Macedonia, FYR 2005 77.38 Poland 2005 14.86 Romania 2004 11.27 Serbia 2007 3.61 Ukraine 2006 8.58 a. All social assistance in Kosovo is targeted. Source: Nguyen, Sundaram, and Tesliuc (in progress). · Strengthening payment, monitoring, and oversight and controls mechanisms. · Overhauling, rationalizing and consolidating the overall safety net in countries where it is weak--to improve targeting, effectiveness, efficiency, and administrative feasibility. Introducing new programs Countries without well targeted programs can introduce new ones. It takes 12 to 18 months to develop basic targeting, registry, and safety net systems--conso- nant with the expected duration of the downturn--but considerably longer for lower income countries with weaker institutional capabilities, where the need for them is likely to be particularly great. But there could still be value in using the crisis to introduce a new, well-targeted program, as Turkey did in 2001 with its conditional cash transfer scheme. Some programs to be considered include: · Direct cash transfers, which have low administrative costs and do not dis- tort prices. If they are targeted and have adequate coverage and generosity, 176 TURMOIL AT TWENTY these advantages make them an attractive option. Requirements include developing a targeting mechanism, building up a household registry (data collection, database management), identifying payment channels, and introducing basic monitoring, oversight, and control mechanisms. "Near cash" instruments such as food stamps are an alternative, but their admin- istrative costs tend to be significantly higher than cash transfers, as are those for in-kind transfers. · Conditional cash transfers (CCTs), which have additional administrative requirements as direct (unconditional) cash transfers but also require some sort of monitoring of education and or health conditionalities or both. As mentioned, Turkey introduced such a scheme, and it has been quite effec- tive. FYR Macedonia is developing a CCT focused on youths (secondary school) and young infants (health for those 0­18 months). CCTs may need to adapt conditionalities for greater relevance, given the region's human capital profi le--for example, by focusing on secondary-school enrollment, attendance and graduation or early childhood development. · Public works schemes, such as workfare and cash-for-work, could be intro- duced as crisis response mechanisms and might be especially relevant for countries facing an influx of returning migrant workers or a spike in unemployment. These programs can have lower net costs than direct trans- fer programs since they could improve key infrastructure. Self-targeting should ensure that the program benefits the poorest, by setting cash pay- ments (workfare wages) lower than average wages. Programs can encour- age female workers, particularly those entering the workforce temporarily to smooth household income, for example, by offering onsite child care and nearby employment. SCALING UP SOCIAL SAFETY NETS 177 CHAPTER 5 Prioritizing structural reform Capital flows to transition (and developing) countries are likely to be consider- ably lower than before the crisis. That makes it important for rescue and stabi- lization, which have dominated the policy agenda since the global economic crisis hit the region, to give way to structural reforms and make the business environment attractive to investors. Questions · What are the most important bottlenecks to growth? · What sectoral and institutional changes occurred in the ECA countries dur- ing the years of abundant capital flows? · Have institutions that underpin the business environment by providing public goods converged to those in nontransition economies at similar per capita incomes two decades after the fall of the Berlin wall? Findings · The three years preceding the crisis (2005­08) saw the socialist legacy of high endowments of infrastructure and labor skills disappear. Concerns about those inputs constraining fi rm expansion became greater than in nontransition economies. · The record of building market economy institutions has been mixed. Con- cerns about tax administration and customs regulation--traditionally higher than in nontransition economies--fell to the levels in nontransition economies. But concerns about the legal environment and corruption rose and are now higher than in nontransition economies. · The structure of financing for fi xed investment in transition economies has converged toward nontransition economy norms: the reliance on retained earnings has fallen and that on bank fi nance has risen. Th is occurred in parallel with a decline in the use of informal finance, reflecting a shift to the formal economy. Rescue and stabilization have dominated the policy agenda since the global economic crisis struck the region. While the precise contours of the future are uncertain, capital flows to transition and developing countries will likely be considerably lower due to the reduced appetite for risk and will go to coun- tries with the most attractive business environment. Indeed, evidence of 179 greater discrimination among countries is already in country spreads (annex 1.1). Policymakers thus need to reinvigorate structural reforms. They need to address constraints to growth that firms identified as the tightest on the eve of the crisis. But two decades since the fall of the Berlin wall, it is also instruc- tive to ask whether the business environment in the transition economies still retains elements of the socialist legacy--or whether it has converged to that in nontransition economies at similar per capita incomes. Interpreting business environment surveys How the business environment in ECA's transition countries evolved dur- ing the years leading up to the crisis and how it compares with nontransition countries after two decades of transition are based on the responses of 10,000 fi rms in 28 transition countries to the fourth Business Environment and Enterprise Performance Survey (BEEPS) by the World Bank and EBRD, for the most part in 2008, and around 51,000 firms in 74 nontransition countries to the World Bank's Investment Climate Assessments over the last decade.1 The other economies surveyed ranged from very poor countries with a GDP per capita of $250 or less to major industrialized countries with a GDP per capita above $15,000. While the surveys provide a rich description of how managers perceive the costs of their business environment, their interpretation requires a conceptual framework. To illustrate the approach most simply, figure 5.1 measures GDP per capita on the horizontal axis. The vertical axis measures how costly firms in a country report the impact of inadequacies in elements of the business environment on their ability to operate and expand their business. Those ele- ments include regulation, physical infrastructure, availability of skilled labor, macroeconomic conditions, and the rule of law, all of which resemble public goods. The level of the constraint is rated on a scale of 1 (not important) to 4 (severe). Since the supply of public goods is common to all firms in the econ- omy, the firm's response on its severity is a measure of the cost imposed by the constraint on the operation and growth of its business. The reading on the vertical axis is the average of firm responses for the country. It can generally be assumed that countries provide more public goods as they get richer. But firms in richer countries are also more demanding of those public goods. If the provision of public goods as countries get richer outruns the greater demands made upon them, which is usually the case, the 1. For a full presentation of the analysis summarized here, see Carlin and Schaffer (2009) and Carlin, Schaffer, and Seabright (2009). 180 TURMOIL AT TWENTY FIGURE 5.1 Average business environment constraint: transition economies in 1999, 2002, 2005, 2008, and nontransition economies Constraint level, conditional means 4 N N 3 N N N T T N N N N N T T TT T T T T N T TT T N T T T T T N T N T TT N N N N N T T T T T N N T TN T T T T TT T T T T N N T TT N N N T T N N N N T N T NTN T T T T N N TT N N N T N N N T N T N TN TTT T TT T T NT N NN T T T T T T N T N TTTTN T N 2 N T T N T N T N T TTN TT T N NN N T N T T TN N N N N T T N T N T T T N N T T T N T N N 1 4 5 6 7 8 9 10 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 cost of the business environment in a country as reported by firms will decline with per capita income, the situation in figure 5.1. But if the supply of public goods as countries get richer cannot keep pace with demand, the cost of the business environment will increase with per capita income Table 5.1 presents the average rating of constraints for the transition countries covered in BEEPS 4 and comparable figures for earlier waves of the BEEPS surveys in 1999, 2002, and 2005 and for surveys of firms in non- transition economies.2 The table reports two types of business indicators. · The first is the response to the question asking managers to evaluate the importance of each business environment constraint for the operation and growth of the firm. Answers are scored from 1 (minor) to 4 (very severe). This is referred to below as a measure of the "level" of the constraint. 2. When comparisons are made between transition and nontransition economies, the data for non- transition economies refer to the pooled surveys (Investment Climate Surveys and a few BEEPS sur- veys of nontransition economies) conducted over 1999 to 2008. Whereas the same transition countries have been surveyed four times (1999, 2002, 2005, and 2008) and the evolution of constraints over time can be investigated, this is not possible for nontransition countries, which were typically only surveyed once. Although there were concerns that the data from BEEPS 4 might be contaminated by the early effects of the fi nancial crisis, there is no evidence of this. While the average complaint level across all dimensions of the business environment rises in 2008 compared with 2005, it is close to the 1999­2005 average (table 5.1). However, the 2008 complaint level for problems relating to finance remains broadly similar to 2002 and 2005. Th is evidence from the fi nance question suggests that the responses from BEEPS 4 should be interpreted as "the eve of the fi nancial crisis" rather than "early in the fi nancial cri- sis." In two of the ECA surveys (Albania and Croatia), some fi rms were surveyed in 2007. PRIORITIZING STRUCTURAL REFORM 181 TABLE 5.1 Levels and priorities of constraints on business in BEEPS 4 and other surveys Level of constraint (conditional means) Priority of constraint (conditional means) Nontransition Nontransition economies BEEPS 1 BEEPS 2 BEEPS 3 BEEPS 4 economies BEEPS 1 BEEPS 2 BEEPS 3 BEEPS 4 2000­08 1999 2002 2005 2008 2000­08 1999 2002 2005 20008 Average (6) 2.24 2.51 2.16 2.13 2.35 Infrastructure 1.68 1.99 1.51 1.43 2.18 0.22 0.25 0.12 0.12 0.38 Telecoms 1.60 1.48 1.38 2.10 0.19 0.12 0.11 0.35 Electricity 2.38 1.67 1.58 2.51 0.45 0.20 0.18 0.49 Transport 1.94 1.45 1.44 2.01 0.30 0.13 0.12 0.31 Land access 1.80 1.51 1.57 2.05 0.24 0.15 0.17 0.32 Skills 2.22 1.91 1.97 2.54 0.39 0.30 0.32 0.50 Tax rates 2.70 3.55 2.80 2.78 2.96 0.59 0.91 0.71 0.70 0.69 Tax admin. 2.37 2.97 2.53 2.47 2.38 0.45 0.71 0.59 0.57 0.44 Finance 2.46 3.02 2.49 2.41 2.51 0.49 0.68 0.55 0.52 0.50 Labor regulation 2.01 1.84 1.73 1.87 1.91 0.29 0.19 0.19 0.25 0.24 Customs 1.83 2.00 1.99 1.85 1.82 0.25 0.28 0.33 0.28 0.24 Licenses 1.92 1.95 1.97 1.93 2.04 0.25 0.24 0.31 0.29 0.28 Legal 1.88 2.06 2.03 2.03 2.15 0.25 0.27 0.33 0.33 0.34 Corruption 2.48 2.42 2.20 2.10 2.53 0.47 0.41 0.39 0.36 0.51 Crime 2.18 2.31 1.85 1.72 2.27 0.36 0.35 0.23 0.19 0.38 · The second is a relative measure of the "priority" of a constraint, where the priority for a responding firm is defined as a value above the average of the firm's answers for the six constraints common to BEEPS 4 and most of the decade's surveys. To compare countries, it is necessary to control for the fact that the sam- ples of firms for different countries will vary by characteristics such as firm size, their sector, and the nature of ownership, whether domestic or foreign. For instance, if a country has a dominance of energy-intensive fi rms in the survey, the answers to electricity as an obstacle to doing business might be quite important--and more so than those from a country that lacks such eco- nomic sectors. To make the responses comparable, these differences are cor- rected for, and the results presented for a benchmark firm. The ratings in table 5.1 are those of the benchmark firm, which is a firm in manufacturing employing 30 persons,3 privately owned with no state-owned 3. In the full set of all surveyed fi rms, median employment was 28 persons and mean log employment was 33 persons. 182 TURMOIL AT TWENTY predecessor, with less than 10 percent foreign ownership, exporting less than 10 percent of its sales, and with no reported change in employment in the pre- vious three years. These ratings for the benchmark firm are referred to as con- ditional means because they are corrected for sample composition. Table 5.1 reports conditional means for constraint levels and priorities for individual elements of the business environment as well as the average business envi- ronment.4 Figure 5.1 presents conditional means for the level measure of the business environment. Annex 5.2 discusses how the evaluations of the busi- ness environment change for firms that vary in different dimensions from the benchmark firm, and annex 5.3 present estimates of those variations (called "marginal effects"). Overview of results The survey evidence from 2008 reveals three important changes in the growth process as it impinged on the constraints facing existing businesses. First, the results show a dramatic turnaround from 2005, the year of the previous sur- vey (BEEPS 3), in infrastructure and skilled labor constraints. The legacy of communism of high endowments of these inputs that characterized the first decade and a half of transition had disappeared by 2008. Second, strong eco- nomic growth appears to have increased the cost of weak market economy institutions, especially the legal environment and corruption. But other insti- tutions of the market economy--such as tax administration and customs reg- ulation, which have traditionally ranked high among the concerns in tran- sition economies--are seen as less constraining for business. Th ird, in the context of appreciating real exchange rates and growth oriented toward non- tradables during 2005­2008, the complaints of exporting firms, measured by the average across all aspects of the business environment, increase. The aver- age constraint reported by exporters rises from below average in 1999 to well above the average for nonexporting firms--and in 2008 is well above the level reported by firms in nontransition economies. The survey results thus provide a micro-based picture of important dimen- sions of the physical and institutional infrastructure during the phase of rapid growth. There are also some important variations according to a coun- try's level of development. Taking the business environment as a whole, the upper middle-income transition economies appear to experience a tighten- ing of constraints in 2008. The evidence highlights emerging shortfalls of 4. To permit comparability with the earlier BEEPS surveys, the average constraint is defi ned using a relatively narrow set of six indicators: tax rates, tax administration, labor regulation, licenses, corrup- tion, and crime. PRIORITIZING STRUCTURAL REFORM 183 investment in physical infrastructure, especially in the upper middle-income transition economies, and education, especially in the low-income and lower middle-income countries.5 In addition, despite improvements in some aspects of the institutional environment--as measured, for example, in the EBRD Transition Indicators6- --firms report that the costs of weak institutions had risen since 2005. Th is may also have resulted from the period of fast growth, as firms and economies outgrew the capacity of institutions to provide the necessary public goods. Th is is especially evident in the lower middle-income group of transition economies for labor regulation and the legal system--and for both the low- income and lower middle-income groups for corruption. A pattern of growth fueled by large capital inflows, with increasing emphasis on nontradables, is also suggested by the growing complaints from exporters in the higher income transition economies. The evidence is consistent with the hypothesis that strong private sector growth in transition economies before 2008 led firms to encounter capacity constraints in physical infrastructure and skilled labor for the first time since the transition began. It points to underinvestment in infrastructure and edu- cation over the transition. It is also consistent with the macroeconomic evi- dence presented in chapter 1 on rapidly increasing current account deficits, in some transition economies to extraordinarily high levels, and appreciating real exchange rates.7 The pattern of economic growth, especially in some upper middle-income countries in the years preceding the crisis, favored nontradable sectors. Exporting firms in the upper and lower middle-income groups of transition economies report higher costs of their business environment in 2008 than in earlier years and than in countries outside the transition. For the average business constraint measure, there is no tendency of exporting firms outside the transition to complain more than nonexporters.8 The biggest concern for exporters in the upper middle-income group of transition economies in 2008 5. The standard World Bank classification toward the beginning of the period under consideration-- July 2005--is used. Slovenia is a high-income country. The upper middle-income countries are Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, the Russian Federation, and the Slovak Republic. The lower middle-income countries are Albania, Armenia, Azerbaijan, Belarus, Bosnia- Herzegovina, FYR Macedonia, Georgia, Kazakhstan, Kosovo, Montenegro, Serbia, and Ukraine. The low-income countries are the Kyrgyz Republic, Moldova, Tajikistan, and Uzbekistan. 6. EBRD 2008. 7. See, for example, IMF 2009a, chapter 2. 8. It is natural (and reassuring) to see that exporting fi rms in and outside transition countries com- plain more than nonexporters about customs regulations. 184 TURMOIL AT TWENTY is infrastructure, especially electricity and access to land, where the effects are large on both the levels and the priority measures, though they also complain more than nonexporters about tax administration and labor regulation (lev- els measure only). In the lower middle-income group, exporters' heightened concerns are concentrated on corruption and in the low-income group on tax administration. The 2008 surveys also provide evidence of the convergence of fi nancing patterns for fi xed investment in transition economies toward those charac- teristic of nontransition economies. In particular, reliance on bank finance, which was below that in nontransition economies in 1999, increases, and the gap is almost closed by 2008. In summary, on the eve of the financial crisis, transition economies expe- rienced a pattern of development in which some benefits inherited from the planning era had finally disappeared, while other unfavorable aspects of the legacy persist, in part those that required new market institutions. The evolu- tion of institutional constraints is quite complex. In 2008, there is a sharp rise in the priority accorded to problems associated with corruption and crime. On the positive side, concerns about customs regulations and tax administra- tion converge with those of the nontransition economies in 2008. Growth bottlenecks Physical infrastructure and skills A striking fi nding in the 2008 survey is the increase in complaints about infrastructure and skilled labor. All the evidence is consistent with, or does not contradict, the hypothesis that higher reported constraints result from growth causing capacity constraints to be encountered or relative price effects, or both. The likely importance of domestic infrastructure rather than exter- nal price shocks, such as the increase in oil price in the years preceding the global crisis, is highlighted by the stronger correlation between power outages and electricity as a constraint in 2008 than in previous years in the transition economies. And unlike the transition economies, there is no increase in infra- structure complaints in 2008 in Turkey (which was also surveyed in 2008 as part of BEEPS 4), supporting the conclusion that this is a transition-specific phenomenon and not a reflection of world energy prices. Additional support comes from the fact that expanding firms complain more about electricity.9 9. The fact that higher complaints by expanding fi rms appear more strongly in the absolute than in the relative measures is probably explained by increased numbers of complaints at the lower end of the scale. PRIORITIZING STRUCTURAL REFORM 185 Whether measured by the level of complaints or by its priority among the set of business environment constraints, there is a clear upward shift in reported concerns--and thus the cost on the operation and growth of busi- nesses--about infrastructure and skills in transition economies in 2008.10 This shift is from levels that were below those reported by nontransition econ- omies from 1999­2008. The jump in 2008 puts most infrastructure complaints in the richer transition economies above their comparators, whereas in the poorer transition economies, the jump reduces the gap but still leaves them less constrained than the comparable nontransition economies. And while complaints about infrastructure generally decline as GDP per capita rises in nontransition countries (as seen by the downward sloping dotted lines in fig- ures 5.2­5.4), this is less clearly the case in the transition economies. In electricity and transport, the complaint levels rise from well below to close to or just above those in nontransition economies (figures 5.3 and 5.4). When the surveys are split into three country groups by income, complaints about electricity and transport in the upper middle-income transition econo- mies in 2008 rise significantly above those in nontransition economies. And in the low and lower middle-income country groups, complaints rise to the level of comparable nontransition economies in electricity but remain lower in transport. Infrastructure bottlenecks affect firms adjusting their level of employment more than fi rms not adjusting. The results in table 5.1 and figure 5.1 refer to the benchmark firm that has seen no change in employment in the previ- ous three years and is therefore a "nonadjusting" firm. Moving away from the benchmark and looking instead at firms that are identical, except that they have adjusted employment over the previous three years, reveals that com- plaints about electricity rise more in expanding fi rms in the poorest transi- tion countries on both the levels and the priority measures. In transport, there are sharp increases in the reported costs of constraints on both measures from both expanding and contracting fi rms relative to nonadjusting fi rms in the second lowest income group, and there is evidence of a similar effect for expanding firms in the lowest income group.11 There is some indication 10. Complaints about infrastructure in 1999 were higher in nontransition economies than in transi- tion economies because the legacy effects of the infrastructure endowments were not apparent until the phase of transition known as "disorganization", the disruption of the organizational arrangements governing production and trade under central planning, was over. For a discussion of disorganization, see Blanchard (1997). 11. The second and third columns of annex table 5.3.4 indicate the change in the severity of a busi- ness constraint for expanding and contracting fi rms relative to the benchmark (nonadjusting) fi rm in transition economies. 186 TURMOIL AT TWENTY FIGURE 5.2 Infrastructure composite bottlenecks, 1999­2008 Constraint priority, conditional means 1.00 N T 0.75 T T N T T T TT 0.50 T T N T T T T TT T T T N T T T T TT N T N T T T N T T T N TT T T T T 0.25 T TT T T N N T N T T TT T N T T T T T T T N T T T N TT N N T T T T T T T T T TTN T T N N T T TT T TT T T TT T T T T T T T T T T T TT T T N T T T N T 0.00 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 FIGURE 5.3 Infrastructure bottlenecks--electricity, 1999­2008 Constraint priority, conditional means 1.00 N NN T N NN N N N N T N T N T 0.75 N NN T N N N T N N T N N N N T T N N N N T NN N N N TN T T T N N T T 0.50 N N N N T T NN N T T T N T T NN N N TN N N TT T T T N N T N N N NN T N T TT T T TT N N N 0.25 T N NT N T T T N N T T N NT T N NT N T TT T N T T TT TT T T T N N T T T T T T TT T T T T T T T T TT N 0.00 4 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition economies 2000­08 economies 2002 economies 2005 economies 2008 that the weight of the increased complaints about electricity is concentrated in manufacturing. Problems with transport affect service firms for the first time, taking their levels and priority scores above those for manufacturing firms. A similar pattern to electricity and transport appears for access to land (figure 5.5). Service firms go from reporting land access as less problematic than manufacturing firms to reporting as more problematic in all country PRIORITIZING STRUCTURAL REFORM 187 FIGURE 5.4 Infrastructure bottlenecks--transport, 1999­2008 Constraint priority, conditional means 1.00 0.75 N N NN N N 0.50 NN N N N N N T N T N N NN N T N N T N N T N NN N T N N T NN T N N TT T N N N N T N N NT N T T NN NT T TN T N T T N 0.25 NT T N T N NT T N T TN N T N N TN NNT N N T N N T N TNN N N N T NT T N NT T N TT N T T T T T N TT T T T T T T TT T N T T T T T T TT T T T T T TT T T T T T T 0.00 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition economies 2000­08 economies 2002 economies 2005 economies 2008 FIGURE 5.5 Access to land, 1999­2008 Constraint priority, conditional means 1.00 0.75 N N N T N 0.50 N T N T T NN N T T N N N N T N T NN N T T N N T N NN T NT TT N TN NT N T T N N N N T T N 0.25 T N TN N T N NT T T N T T T N N N N T T N TN N TNN TN T N N N T NT N N N N T N TT T NT N T T N N T T N NT T N T TT N N T N T T T T TNT TT T NTT NN N TT T N T TT T T T N T T T TT 0.00 T 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition economies 2000­08 economies 2002 economies 2005 economies 2008 groups. Exporters in the richer transition economies report a big increase in problems with access to land. Some country variations in infrastructure in 2008 are worth noting. Against the background of higher reported infrastructure constraints across transition economies as a whole, and controlling for income and firm char- acteristics, Czech firms report a bigger increase for electricity and transport 188 TURMOIL AT TWENTY in 2008. Among the upper middle-income country groups, Hungary is at the opposite end of the spectrum for transport and access to land, with low lev- els of complaints. In the Western Balkans, infrastructure problems appear greater in Albania and Kosovo. In the CIS, problems with electricity increased relatively more in the Kyrgyz Republic; Azerbaijan was at the opposite end of the spectrum, with lower reported problems with electricity and transport in 2008. Access to land was more problematic across the region in 2008, espe- cially in Azerbaijan, Belarus, Moldova, the Russian Federation, and Ukraine. Complaints in both the levels and priority measures about skilled labor rise in 2008, continuing the increases across successive BEEPS surveys and taking the levels in 2008 above those recorded in nontransition economies (figure 5.6). Both expanding and contracting firms' complaints about skill shortages rise in 2008--especially in the richer transition economies. Unlike infra- structure, however, labor skills provide an example where complaints increase much more sharply with per capita income before 2008 in the transition econ- omies than the nontransition economies. This implies that the demands for labor skills outrun their provision in the richer transition countries. Across all surveys and all elements of the business environment, fi rms in Estonia typically record lower constraints than firms elsewhere, control- ling for income and firm characteristics. But in 2008, Estonian firms record increases in skills constraints above the increases reported by firms in other transition economies. Hungary is at the other end of the scale, reporting that access to skilled labor was less problematic than in 2005. In the CIS, Belarus FIGURE 5.6 Skills bottlenecks, 1999­2008 Constraint priority, conditional means 1.00 0.75 T T T N T T T N T T N TT N T NN N T N N N N T TT 0.50 NN N N TN T N TT T N N N N T N T NN NT N T N TN N N N T NT NT T T N N TT T N N TN NT N TT T T N T TT N NT N T N NNNT N T NN N N TN N NT T TTT NT N T N N T N N N N T T T T 0.25 N N TT T N T T N N N T N N N N T N T N T T T TT T TT T T T 0.00 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition economies 2000­08 economies 2002 economies 2005 economies 2008 PRIORITIZING STRUCTURAL REFORM 189 and to somewhat less extent Kazakhstan and the Russian Federation, as well as Moldova and Tajikistan, encountered more severe problems with skilled labor than did other countries. The correlation between firm characteristics and the pattern of complaints supports the idea that capacity constraints in infrastructure and labor skills were increasing the costs of the external environment to firms. Higher com- plaint levels are reported by expanding firms for electricity and by expanding and contracting firms for land access and skills, and skill shortages increase with the size of fi rm. Differences between rural and urban areas are con- sistent with this picture: in the poorer transition economies, higher growth revealed physical infrastructure and skills constraints as more concentrated in the faster growing urban areas. In the richer transition economies, by con- trast, the physical infrastructure problems appear to have hit fairly uniformly across rural and urban areas, but skill shortages switched from being more problematic in rural areas to being more problematic in urban areas in 2008. Market institutions A characteristic of the transition economies in earlier surveys is the higher level of concern about the legal system than in nontransition economies (table 5.1). The concern rises with income in the transition economies, indi- cating that the demands on the legal system by the richer transition econo- mies outruns its provision, whereas the relationship is flat outside transition. Concern about this important dimension of market institutions continues to rise in the 2008 survey. The dynamic in the aggregate picture is driven by the lower middle-income transition countries, where the priority accorded to the legal system as a concern was much lower in 1999 than outside transition: it increases and is above the level in nontransition countries in each subsequent survey. This is consistent with the idea that progress in transition raises the sensitivity of firms to the costs of a poorly functioning legal system. Such costs continued to rise during the rapid growth up to 2008. The falling concern from 1999 to 2005 for corruption and crime are reversed in 2008 (table 5.1). The cost of corruption is evaluated at its highest in the mid- dle of the income distribution outside transition (figure 5.7). While such an inverse U-shaped relationship was less marked in the transition countries before 2008, it is clearly visible in 2008, and the rise in levels makes it virtu- ally indistinguishable from the relationship for the nontransition countries. Moving away from the benchmark, a look at variation by firm characteristics reveals that the increased concern about corruption is concentrated in new pri- vate firms. Concern about corruption among foreign-owned firms falls in 2008 190 TURMOIL AT TWENTY FIGURE 5.7 Corruption, priority measure, 1999­2008 Constraint priority, quadratic specification, conditional means 1.00 T NNT N N N NN N N 0.75 T N NT NN N N TN TN T T N N N T T T N T N N NN T N T N T N TT T T T NT N T N N N NN T N N T T T T 0.50 T T T T N T T TTT T T N N T NT T T T TT T T NT T T T N T T T T T N N N NT N NN T T N T N T NT T T N N N T T TT T T T N T T T T T N T T T N TT NT T T T NN N TT T 0.25 T T T N T N T N T T T T T TN N T T N NT T N T N N N N N 0.00 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 in the richer transition economies. The survey also shows an increase in the prevalence of bribes in 2008. The tendency for complaints about corruption to rise is more marked in the Russian Federation and Ukraine than elsewhere. Concern about crime shows a pattern similar to that for corruption (fig- ure 5.8). The level of concern was lower than in nontransition economies and falling until 2005. There is a sharp reversal in 2008, however, with the concern rising back above that in 1999 and higher than that reported for nontransition economies. Tax administration and customs regulations emerged as transition econ- omy problems in all the surveys until 2008, where on both the levels and the priority measures their importance falls into line with nontransition econo- mies at similar incomes (figures 5.9 and 5.10). The 2008 survey also records a fall in concern with customs regulations. Whereas convergence on this mea- sure to the nontransition norm had already taken place for the richer transi- tion economies by 2005, it is recorded only in 2008 for the two lower income groups. The most marked change is in the lower middle-income group, where the gap with nontransition economies disappears in both the levels and prior- ity measures. Although overall concerns with labor regulation remain broadly unchanged in transition economies from 2005 to 2008, interesting developments for dif- ferent types of firms suggest convergence toward market economy norms. In the richer transition countries, contracting firms record higher complaints PRIORITIZING STRUCTURAL REFORM 191 FIGURE 5.8 Crime/theft/disorder: priority measure, 1999­2008 Constraint priority, conditional means 1.00 T N N T 0.75 N T NN N N N N N T N N T N N N N T NN N T T 0.50 N N NT T T T TN T T N T T N N NT N T NN T N NN N T N T NNN N T TT N T TT T N TN N N N T T N NN N T N T T T N N T T TT T T N N NT N T N T TT T T TT T T T N T 0.25 N NT T T TT T T T T T T N N TT T N T T N T TT T T N N N N T T T T T N TT TT N T T T T NT T T TT N T TT N N N T T T T T N 0.00 T 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 FIGURE 5.9 Tax administration: priority measure, 1999­2008 Constraint priority, conditional means 1.00 T T T T T T T T T T N T N T T T 0.75 T T TN TT T T TT T T T N T T N N N T T T T T T TT T N T T T T NT T T T T T T T N N N N T NT N N N N N T T N N T N NT T T T N N T T T T N N 0.50 N N T T T N TT T T TTTN N N N N NN N N T T T T NT TTT T NN N N N NN T N T T T T N N N T TT N N N N N N N N TT NT T N N T N N N T T N N N N N N T NT N 0.25 N T T N T T 0.00 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 about labor regulation than nonadjusting firms on the levels measure in each survey but, for the first time in 2008 in these countries, concern with labor regulation increases for contracting fi rms on the relative measure, bringing them into line with nontransition economies. Labor regulation emerges as a concern for firms adjusting employment in 2008 for the first time in the lower middle income group, consistent with countries in this income group outside 192 TURMOIL AT TWENTY FIGURE 5.10 Customs regulations: priority measure, 1999­2008 Constraint priority, conditional means 1.00 0.75 T T T NN 0.50 T N N N N N T N T N T N T N TT T T TT T T T T T N N NNNN N TT T N T TT T TT T TN TT T T TT N T N T T N TT N N TTT TTTTTN N NT T T T 0.25 N N T N NNT NN T T T N TNT T NT T NT N T T T TT T T N N T N TN TT N N T N N N N N N N T T N T T TT N N TT T N T T N N N N N NN T T T N NT T T N NN TNN TN N T TT T N N T TT 0.00 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 transition and there is some evidence of this effect for contracting firms in the poorest transition countries. Financing Many richer transition economies benefited from financial deepening in the years before the fi nancial crisis, as foreign-owned banks took over a major- ity of the domestic banking system. Th is is reflected in interesting ways in the survey data from 2008. There is a clear convergence to the nontransition benchmark in the way fi xed investment is fi nanced in transition countries. For example, firms in transition economies rely more on internal funds for financing fi xed investments than do firms outside transition, but the gap has been falling over time. Parallel with the fall in dependence on internal finance is that in the use of informal finance. In 1999, transition firms relied as much or more on infor- mal fi nance as firms outside, but by 2008, reflecting the shift to the formal economy, firms in transition were less reliant on informal financing than the nontransition benchmark. Replacing internal sources and informal finance is an increase in bank fi nancing of investment. Whereas firms in transition relied considerably less on bank finance in 1999 than firms outside, the differ- ence had almost disappeared in 2008. Outside the transition, there is in effect no relationship between the reli- ance on internal funds and income per capita--the line is flat (figure 5.11). For PRIORITIZING STRUCTURAL REFORM 193 FIGURE 5.11 Financing expansion--internal finance Percent 100 T T T N T T T N T T T TT N T T N N T N T T T TT 75 NT T T N T TT N TT T T T N N N T N TN T T TT T T T T TT T T T NT N N N T T N T T N T T T T T T T T T TTT TT N N T T T T NT N N NT NT N N TT T NT NN T T T NT NN T TT T T N N T T 50 N N N T N N N NT T N N N N T N T T T T T T N 25 T N N N N 0 4 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 Note: The figure represents the percentage of financing for investment in fixed assets by the aver- age firm that comes from internal funds. transition economies in 1999, 2002, and 2005, there is a clear negative relation- ship: firms in the poorer transition economies are more reliant on internally generated funds for financing investment. This negative relationship disap- pears in the latest BEEPS round. Apparently low-income and middle-income transition economies have recently become less reliant on internal financing. Finance is different from the public good elements of the business environ- ment. Access to finance and its terms are firm-specific, making the interpreta- tion of the results more difficult. For example, moving away from the benchmark reveals strong firm-size effects associated with the finance constraints: as firm size rises, the relative priority of finance (as compared with the other business environment elements) declines. But this does not necessarily mean that poor access to finance hampers the growth of small firms. Firms may be small because they are poor prospects, and prudent banks choose not to lend to them. Unlike the earlier surveys, the 2008 survey asks only one question about finance as a constraint on business, covering both cost and access. As noted, concern with finance in levels terms remains unchanged in 2008 at the level of nontransition economies and falls on the relative measure, leaving it similar to the nontransition score (figures 5.12 and 5.13). In the upper and lower middle-income groups of transition countries, the priority accorded to finance falls to about the level outside the transition. But 194 TURMOIL AT TWENTY FIGURE 5.12 Financing expansion--external financing, constraint level Constraint level, conditional means 4 N N T T T T NT N N N T T NN N N T T N TT T NN N N 3 N N N N N N T TTN N T T NT T N T NT NT N T TT N T N T T TT T T T T N NT T N TT T T T T TN N N T T T TN T T T T TT T N T N T T T TT T N N TNNT N N T T N NT T N N T T TT NT N TT T T T NT N TN N T N NTN T T N N TN T N N T N T N 2 N T NN T TT N T N N T N N NN T NN T TT T N T T N 1 4 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 FIGURE 5.13 Financing expansion--external financing, constraint priority Constraint priority, conditional means 1.00 T N T N T N N N T T T N T 0.75 T N NN T N T T T T N N N N NN T N T T T T TT N NTN T T T T TT N N T N T T N NN N T T N N T T T TT T N T TT T T N TNN T T N T N N N N TT T T T T TTTT N 0.50 T T TT N T T T T TTTT TT T TT T N N T N T N N T NN T N TN T T N T NN T N N T T T N NTT T T N N N T T N N N N N N T N T N N T N N T N N NT T 0.25 N N T T 0.00 4 5 6 7 8 9 10 11 Log of GDP per capita (constant 2000 US$) Nontransition Transition Transition Transition Transition economies economies 1999 economies 2002 economies 2005 economies 2008 2000­08 in the poorest transition group, fi nancial constraints in 2008 are markedly lower than in the comparable nontransition economy group. There are three other interesting findings away from the benchmark: · Size of firm effects almost completely disappears. Unlike the nontransition economies and in earlier surveys, larger firms in transition economies in 2008 do not report significantly lower financial constraints than smaller firms. PRIORITIZING STRUCTURAL REFORM 195 · For the first time, state-owned firms complain more than new private firms (on both the absolute and the relative measures). · Hungary's relatively low reported constraint in 2008 on both the absolute and relative measures indicates that fi nance was not a major problem in that country. The evidence from the financing patterns and from the evaluation of the finance constraint is consistent. Credit was more freely available and on more favorable terms in 2008 than in earlier years, even for smaller fi rms and espe- cially for those in the private sector. This increased reliance on bank finance for investment is closer to the levels in nontransition economies. While the convergence of financing patterns in transition toward nontransition econ- omy norms may have reflected the specificities of the credit boom, the long- run maturation of the financial system away from informal finance and the reduced reliance of low-income and lower middle-income countries on inter- nal financing are part of a continuing trend seen in earlier surveys.12 But firms in transition newly dependent on bank credit may have become more vulner- able to financing constraints, at least until the global financial system returns to normal. The persistence of legacy in shaping the business environment What do differences in the business environment in transition and non- transition economies owe to the legacy of central planning? The broad changes in economic structure as economies develop are well documented.13 As an economy grows and aggregate income rises, low-productivity labor moves from agriculture into industry, with the share of employment in services changing relatively little. Later, the share of employment in industry stabi- lizes, and in the richest economies, it starts to fall as the share of employment in services rises. The explanation is simple: productivity growth in manufac- turing outstrips the demand for manufactured goods and workers move into labor-intensive services. The way central planning changed the structure of economies can be sum- marized as follows. · Central planners accelerated the process of moving labor out of agriculture and into industry as part of forced industrialization. 12. Mitra, Muravyev, and Schaffer 2009. 13. The pioneering work here was by Simon Kuznets and then by Hollis Chenery and his colleagues at the World Bank (Chenery et. al 1986). See Rowthorn and Ramaswamy 1998 for a theoretical analysis and Raiser et al. (2003) for an application to transition economies. 196 TURMOIL AT TWENTY · Planners emphasized industry over services, so employment in market-ori- ented services (trade, finance) was relatively low.14 · The many inefficiencies of central planning introduced a wedge between the degree of industrialization and aggregate productivity: the level of income in planned economies was lower than in market economies at a similar level of industrialization. The institutions of a market economy were either suppressed (in economies that had them when planning was introduced) or never developed (in econo- mies that industrialized under central planning). The sectoral structure of transition economies has converged to that of market economies at similar per capita incomes (box 5.1). How has the busi- ness environment evolved in the poorer transition countries that underwent forced industrialization under central planning and in the richer transition countries that had a stronger collective memory of market institutions? Industrialization in poorer transition countries The superior endowment of infrastructure and labor skills, as reflected in the business environment of transition economies before the 2008 survey, was the product of the legacy of forced industrialization. Figure 5.14 plots the reported cost of the infrastructure constraint against GDP per capita (proxying the level of development) on the horizontal axis in the left-hand panel and the share of nonagricultural employment in industry and services (proxying the level of industrialization) on the horizontal axis in the right hand panel. The results for the transition economies pool the BEEPS 1­3 surveys, for 1999, 2002, and 2005. This is because, compared with BEEPS 4, the first three rounds have a wider range of questions and allow more nuanced analysis. The results show that transition economies had better infrastructure (lower reported costs of constraints) at a given level of industrialization than did nontransition econo- mies. This is consistent with the idea that overindustrialization involved levels of development of physical infrastructure that were "excessive" when mea- sured against a market economy benchmark. Similarly, figure 5.15 plots the reported cost of access to skilled labor con- straint against GDP per capita and the level of industrialization in transition and nontransition economies. It shows skills shortages to be less problematic in transition: this is consistent with the persistence of the effects of forced industrialization in easing this constraint. Both panels of Figure 5.15 show that the upward-sloping income profi le for the skills constraint in transition 14. See, for example, Raiser et al. 2003. PRIORITIZING STRUCTURAL REFORM 197 BOX 5.1 Transition economies converge in structure to market economies Overindustrialization has been substantially corrected everywhere in the region, but the share of employment in industry is short of the market economy bench- mark in the low-income and lower middle-income CIS countries yet continues to exceed the benchmark in the EU10+1 (box figure 1). The share of employ- ment in industry fell during the period of income contraction and rose when income began to increase. Overall, it fell by between 5 percentage points in the Western Balkans and 15 percentage points in the low-income and lower middle- income CIS during 1990­2007. In 2007, the share of industrial employment in the EU10+1 exceeded the market economy benchmark by 5 percentage points, whereas the massive deindustrialization in the low-income and lower middle- income CIS countries left them with a share of employment that fell short of the market economy benchmark by some 5 percentage points. The middle-income CIS and the Western Balkans were virtually at the market economy benchmark. The share of employment in agriculture behaved quite differently across the subregions but had mostly reached the market economy benchmark by 2007 (box figure 2). It shrank in the EU10+1 and more modestly so in the middle- income CIS. But it rose sharply by nearly 15 percentage points in the low- income and lower middle-income CIS, much of the rise occurring during the severe transition recession, when subsistence agriculture was a safety net. It began to fall toward the market economy benchmark in the recovery. Agricul- ture in the Western Balkans followed a pattern similar to the low-income and lower middle-income CIS, tracing the benchmark as income fi rst went down and the share of employment in agriculture went up and then subsequently as income rose and the employment share went down. BOX FIGURE 1 Benchmarking industrial employment in Central and Eastern Europe and the former Soviet Union against market economies: evolution over the transition Share of employment 50 Middle-income 40 CIS 1990 1990 EU10+1 30 1990 2007 2007 1995 Benchmark 2007 20 Western Balkans 2007 10 Low-income and lower middle-income CIS 0 0 5,000 10,000 15,000 20,000 25,000 GDP per capita (PPP, 2005 international dollars) Note: For data reasons, Western Balkans comprises Albania, FYR Macedonia, and undivided Serbia and Montenegro. Source: Based on background work by Mark Schaffer. (continued) 198 TURMOIL AT TWENTY BOX 5.1 (CONTINUED) Transition economies converge in structure to market economies With the exception of a slight fall in the early years of transition in the low- income and lower middle-income CIS, the share of employment in services increased toward its market economy benchmark everywhere (box figure 3). It increased by nearly 15 percentage points between 1990 and 2007 in the EU10+1, the middle-income CIS, and the Western Balkans, with the increase driven by market services. BOX FIGURE 2 Benchmarking agricultural employment in Central and Eastern Europe and the former Soviet Union against market economies: evolution over the transition Share of employment 50 2007 1995 40 Western Balkans 2007 1990 30 1991 Low-income 1990 20 and lower 1990 middle-income 2007 2007 CIS 10 Middle-income EU10+1 Benchmark CIS 0 0 5,000 10,000 15,000 20,000 25,000 GDP per capita (PPP, 2005 international dollars) Note: For data reasons, Western Balkans comprises Albania, FYR Macedonia, and undivided Serbia and Montenegro. BOX FIGURE 3 Benchmarking service sector employment in Central and Eastern Europe and the former Soviet Union against market economies: evolution over the transition Share of employment 80 Benchmark Middle-income 60 Low-income CIS 2007 and lower 2007 middle-income 2007 EU10+1 CIS 2007 40 1990 1990 1990 1995 Western Balkans 20 0 0 5,000 10,000 15,000 20,000 25,000 GDP per capita (PPP, 2005 international dollars) Note: For data reasons, Western Balkans comprises Albania, FYR Macedonia, and undivided Serbia and Montenegro. PRIORITIZING STRUCTURAL REFORM 199 economies implies that the priority measure comes together for high-income transition and nontransition economies. Together, figures 5.14 and 5.15 make it clear that the poorer transition economies benefited from the infrastructure and education investments that accompanied forced industrialization. As has been noted, this advantage had eroded in relation to nontransition economies by 2008. FIGURE 5.14 Physical infrastructure: (composite) priority measure, 1999­2005 Constraint priority, conditional means Constraint priority, conditional means 1.00 1.00 0.75 NN T 0.75 N TN N N N N N T N N TN N T NN N N T 0.50 N NNN TT T 0.50 N N N NN N N TN T T N N NNN N NT N N T N NN N TN N TN N T NN N N N T T TN N NNTT T NNTNT NN TN T N N N N T NN NN NT N T T T N N TT T N N N TT N N NN T N T TN NN N N N T N N NT N T N N T N N N T T N T T N T N T N T NNN N NNN N NT T TN N T T N T N TN NN T TT T T N T T T TN T N T N NT N TTTTT N NT N 0.25 T T TNT TNN N TN T N 0.25 T T T NTN T T TT T T N N NTT TTT N T N TT T T T T N TT T N NNTT N N T T TT T T N T T T TT T T T T TN TT T T T TTTTTTT NT TT T T T T T TT T TT T NN T TNTT T T T T TT N T T N 0.00 0.00 4 6 8 10 0 25 50 75 100 Log of GDP per capita (constant 2000 US$) Nonagricultural L share Nontransition Transition economies 2000­08 economies 1999­2005 FIGURE 5.15 Skills: priority measure, 1999­2005 Constraint priority, conditional means Constraint priority, conditional means 1.00 1.00 0.75 0.75 T T N N N T N T NN N N N N N N N N TT TT N 0.50 N N N 0.50 N N NN T T TTN T N N N T TN T TT N N N N NN N T T N N N N N N T N TT T N N TT T N N T N T N T TT T T N T T NT T N TN T NT T T T T T T N T T T T N T N N NN N T T N N TN T NT NNN N NT T T N TT TTT N N T T TT TT TTT T T T T 0.25 T NT T T T T N TTTTT N N N 0.25 NT T TT T T T T N T N T T NT N T T NT T TT N T T T TT T T T T T T T T T 0.00 0.00 5 6 7 8 9 10 0 25 50 75 100 Log of GDP per capita (constant 2000 US$) Nonagricultural L share Nontransition Transition economies 2000­08 economies 1999­2005 200 TURMOIL AT TWENTY Collective memory of market institutions in richer transition countries Did transition economies with a stronger collective memory of market institu- tions show more sensitivity to weaknesses in market-related institutions? For the legal environment, there is no relationship with GDP per capita in the non- transition benchmark in the levels measure and a negative relationship in the priority measure. But in transition countries, for the priority measure, concern about the legal dimension rises with GDP per capita and concern is higher in transition than outside at high levels of income (figure 5.16). This supports the idea that the costs of inadequacies in the legal system are felt more by firms in the more advanced transition countries. And the sensitivity to those costs rose further in 2008 but, as noted earlier, this has been driven by the lower middle income transition countries as they have progressed in transition. In both transition and nontransition economies, the reported costs of labor regulation rise with the level of income and with industrialization. Firms in transition report lower constraints, especially at low levels of income (fig- ure 5.17). Complaints about labor regulation remained broadly unchanged between 2005 and 2008. Concerns about tax administration and customs reg- ulations do not show a rising profi le with income per capita or level of indus- trialization in the transition economies, but corruption displays a nonlinear relationship, as in figure 5.7. Summary statistics for the comparison between the transition economies averaged over 1999­2005 and the nontransition economies are in box 5.2. The FIGURE 5.16 Legal environment: priority measure, 1999­2005 Constraint priority, conditional means Constraint priority, conditional means 1.00 1.00 0.75 N 0.75 N T T T T T T N N T NT TT T NT TN T T N N T TT T N N T T TN TT 0.50 T TT TTT T 0.50 T TT T TT T T T T T NT T T T N T T TT T T T TTT N N T T NT T T N N N N N T T TT TT N NN NT T TN N N N TT T N N N N N N TT TTTTT NT T N T T NTT T T T T N T N TT N T T N TT T T T T TT T NNN N T TNNTT T T TT TN T TNN NN T N T T T N TT N NT N N TT N N NNT TN N N N T T N TN N TT 0.25 N T N NT N N T N N NTT 0.25 N N T NT T N NN N N N N TN N N NTT T TN N TT N N NN T N N N T N NN TTT TN T TN N T NN N N N N T TN NNN N N N N NT T T N T N N N N NN T T T TN 0.00 0.00 4 6 8 10 0 25 50 75 100 Log of GDP per capita (constant 2000 US$) Nonagricultural L share Nontransition Transition economies 2000­08 economies 1999­2005 PRIORITIZING STRUCTURAL REFORM 201 FIGURE 5.17 Labor regulation: priority measure, 1999­2005 Constraint priority, conditional means Constraint priority, conditional means 1.00 1.00 0.75 T 0.75 T N N N N T T N N 0.50 N N N N 0.50 N N N T TT N N N TT N N N N N T T TT T N TT T T TT T T T N NN T T T TT N NN T T N TTTT T T NT T N N T T TT T T T TT N N TT TN TT N T N NT TT N T T N N N T TT TT T 0.25 N NT NN TT T N NN NT T T T T N T T T T 0.25 N NN NT N T T N N T N T T TTT TT NT N NN N T T T T N T TT T T N T T TNT T N N N N N T T T N T T T N T T N T TT TT N T T T TT T T T T T T T TT T TT N T T T T T N N T T T T TT T T T T T T NT N T NT T T T T T T N N T T 0.00 0.00 5 6 7 8 9 10 0 25 50 75 100 Log of GDP per capita (constant 2000 US$) Nonagricultural L share Nontransition Transition economies 2000­08 economies 1999­2005 legacy of central planning is also to be found in the characteristics of firms: the size distribution was skewed toward very large firms, the service sector was underdeveloped and there was little private or foreign ownership. How this affects the comparison of the business environment between transition and nontransition economies is presented in annex 5.2. 202 TURMOIL AT TWENTY BOX 5.2 Comparing constraints in transition (BEEPS 1999­2005) and nontransition (ICA) countries Box table 1 shows both mean and conditional mean measures, the latter correct- ing for differences in the composition of the samples as explained in main body of the text. The differences between unconditional and conditional means are small for the transition economies but somewhat larger for the nontransition economies; as a result, the conditional mean average measure of the business environment in the first row of the table is about 5 percent lower. As the table also shows, the mean level for the average of the business environment con- straints is similar in transition and nontransition countries. And the top five ranked constraints (on both BEEPS and ICAs) are also similar in each group of countries: they are tax rates, macroeconomic environment, cost of fi nance, and policy uncertainty. Elements of physical infrastructure and access to land are low-ranked constraints throughout, with telecoms viewed as least problem- atic. The box table presents the means for the levels and priority measures of business environment constraints separately for nontransition and transition economies. BOX TABLE 1 Constraints on business--transition and nontransition economies Constraint level Constraint priority Mean Conditional mean Mean Conditional mean Non- Transition Non- Transition Non- Transition Non- Transition transition economies transition economies transition economies transition economies economies 1999­ economies 1999­ economies 1999­ economies 1999­ 2000­08 2005 2000­08 2005 2000­08 2005 2000­08 2005 Average (9)a 2.36 2.32 2.25 2.31 Infrastructure (composite) 2.03 1.64 1.96 1.69 0.28 0.16 0.26 0.18 Telecoms 1.74 1.48 1.64 1.46 0.21 0.13 0.18 0.13 Electricity 2.43 1.62 2.38 1.71 0.40 0.18 0.38 0.22 Transport 2.01 1.56 1.94 1.56 0.28 0.16 0.26 0.16 Land access 1.84 1.62 1.80 1.63 0.23 0.18 0.24 0.18 Skills 2.31 1.96 2.21 1.95 0.41 0.30 0.38 0.30 Tax rates 2.74 2.91 2.69 2.94 0.57 0.71 0.56 0.72 Tax administration 2.46 2.58 2.37 2.59 0.49 0.58 0.46 0.59 Finance (composite) 2.53 2.54 2.46 2.59 0.51 0.54 0.49 0.56 Access to finance 2.44 2.36 2.39 2.43 0.45 0.44 0.44 0.48 Cost of finance 2.61 2.69 2.52 2.73 0.53 0.58 0.50 0.60 (continued) PRIORITIZING STRUCTURAL REFORM 203 BOX 5.2 (CONTINUED) Comparing constraints in transition (BEEPS 1999­2005) and nontransition (ICA) countries Constraint level Constraint priority Mean Conditional mean Mean Conditional mean Non- Transition Non- Transition Non- Transition Non- Transition transition economies transition economies transition economies transition economies economies 1999­ economies 1999­ economies 1999­ economies 1999­ 2000­08 2005 2000­08 2005 2000­08 2005 2000­08 2005 Labor regulation 2.08 1.85 2.00 1.86 0.35 0.23 0.32 0.24 Customs 2.03 2.01 1.83 1.96 0.33 0.33 0.26 0.32 Licenses 2.02 1.99 1.91 1.95 0.25 0.30 0.22 0.28 Legal 1.95 2.11 1.88 2.07 0.28 0.33 0.26 0.33 Corruption 2.57 2.27 2.47 2.25 0.47 0.41 0.43 0.39 Crime 2.25 2.04 2.17 1.97 0.36 0.29 0.35 0.26 Policy uncertainty 2.60 2.76 2.46 2.74 0.52 0.64 0.47 0.63 Macro- economic instability 2.75 2.69 2.65 2.68 0.57 0.59 0.53 0.59 Power outages (y/n) 0.66 0.47 0.64 0.46 Water supply outages (y/n) 0.22 0.18 0.20 0.17 Bribes (y/n) 0.34 0.45 0.29 0.43 Unreported sales (y/n) 0.21 0.16 0.23 0.16 Don't report all sales (y/n) 0.50 0.46 0.53 0.46 Tax gifts (y/n) 0.23 0.41 0.22 0.44 Contract gifts (y/n) 0.31 0.28 0.28 0.26 Security costs (y/n) 0.68 0.62 0.61 0.55 Crime losses (y/n) 0.22 0.24 0.18 0.20 a. The average of the business environment is taken over tax rates, tax administration, labor regulation, licenses, financial access, policy uncertainty, corruption, crime, and anti- competitive practices. Note: Annex tables 5.3.1 and 5.3.2 summarize the statistical tests of a linear relationship between income per capita and constraint levels and constraint priorities (box table 1). The tests reported are of the significance of the slope of the income/constraint relationship, and of difference between the averages constraints for transition and nontransition econo- mies at low and high levels of GDP per capita. There is a statistically significant (at the 5 percent level) linear relationship for the nontransition benchmark in either constraint level of constraint priority for the following dimensions of the business environment: physical infrastructure, land access, tax rates and administration, customs regulation, labor regula- tion, finance access and cost, crime and policy uncertainty. There are significant relation- ships with income in the transition but not outside for labor skills, licensing, and macro- economic instability. There is also a significant nonlinear relationship for corruption. Source: BEEPS surveys for 1999, 2002, and 2005. (continued) 204 TURMOIL AT TWENTY BOX 5.2 (CONTINUED) Comparing constraints in transition (BEEPS 1999­2005) and nontransition (ICA) countries There are interesting differences between nontransition and transition countries. For example, electricity is viewed as much more problematic in non- transition countries. More generally, the scores across the infrastructure con- straints and skills are much lower in transition economies: these elements of the business environment clearly are considerably less problematic for the tran- sition countries. Corruption is in the top five constraints on the levels measures in non- transition countries but below this in transition countries. By contrast, transi- tion countries complain more about the legal environment and attach a higher priority to it than do nontransition countries. For example, controlling for firm characteristics, the priority measure indicates that about one-third of firms in transition view the legal system as a constraint more severe than the average, as compared with one-quarter of firms in nontransition countries. Quantitative measures of the experience of firms regarding their environ- ment tend to confi rm the information in the evaluative questions (the lower panel of the box table). For example, consistent with the lower level of com- plaints about physical infrastructure in transition economies, power and water outages are much less prevalent. The prevalence of bribes is higher in transition economies. Underreporting to the tax authorities is somewhat more common outside than in the transition, and the opposite is so for gifts to tax officials. PRIORITIZING STRUCTURAL REFORM 205 Annex 5.1 Conceptual framework Comparable fi rm-level surveys in a large number of countries around the world can provide a rich description of how managers perceive the costs of their business environment. Interpreting the survey results requires a con- ceptual framework. The responses of firms to questions about the quality of the business environment could be interpreted as estimations by managers of the cost imposed on the firm by inadequacies of an aspect of the business environment such as regulation, physical infrastructure, availability of skilled labor, macroeconomic conditions and the rule of law--all of which resemble public goods.15 The response of managers to questions about the impact of such elements on their ability to operate and to expand their business can, in these cases, be interpreted as referring to the cost of the constraint to the firm, where the supply of the public good is common to all firms in the economy. Firms may take actions to mitigate the costs imposed by their environment, such as installing their own generator if the electricity grid is unreliable. But the faulty grid imposes a cost, and this should be reflected in the evaluation of the burden--in this instance--of "electricity" as compared with other dimen- sions of the business environment. Although different firms may face effectively the same supply of the pub- lic good, how they report the importance of this part of their environment will depend on the characteristics of the firm. For example, rapidly growing, highly productive firms would likely suffer higher costs from a poor quality electricity supply infrastructure than stagnating or downsizing firms that are operating with significant spare capacity. Annex figure 5.1.1 illustrates this: firm output or productivity Q is on the horizontal axis, and the reported cost of the constraint P is on the vertical axis. D' represents the demand of a high productivity firm for one of these public goods, and D the demand of a low productivity firm. The upward sloping S curve is the same for all firms in the country. It captures the notion that as firms become more productive, the sup- ply of the public good becomes less adequate and the shortfalls more costly to the firm. Analysis of the responses by firms to a survey from an individual coun- try can shed light on these constraints. In effect, they identify the factors that move the demand curve from D to D' along the common S curve. Measures of fi rm growth and productivity are naturally of particular interest. Thus if 15. The analysis uses the approach described in Carlin, Schaffer, and Seabright (2007), who refer to these costs as shadow prices. 206 TURMOIL AT TWENTY ANNEX FIGURE 5.1.1 Demand and supply of public goods--one country, two firms P (reported cost of constraint) S (of public good) BC' BC D' (of public good; high productivity firm) D (of public good; low productivity firm) Q expanding firms provide higher rankings for a particular public good-type element, this would be consistent with the idea that the weakness of the busi- ness environment along that dimension may be a constraint on firm growth. Firm size is also of particular interest, because policy attention is often focused on the special needs of small firms. If firms of different size perceive differ- ent relative costs of elements of the business environment, the relevant aspect of the business environment would seem a likely candidate for special policy measures directed at small firms. The business environment: mostly a public good? This analysis works because all the firms in the country share the same busi- ness environment, even though they have different needs and thus experience differences in the costs imposed on them by that common environment. In other words, they all face the same S curve (annex figure 5.1.1). The within- country variation identifies the impact of business constraints on firm growth by moving the D curve. Within-country variation from surveys conducted in different countries can be combined to obtain more efficient estimates of these impacts.16 And by splitting the surveys between transition economies and nontransition economies, it is possible to compare how growing firms (or small and medium enterprises, or whatever the focus of interest) in transi- tion economies and nontransition economies differ in their evaluations of the impacts of business environment constraints. 16. This can be done using fi xed-effects estimation techniques. Details are in Carlin and Schaffer (2009). PRIORITIZING STRUCTURAL REFORM 207 The quality of the business environment also varies systematically across countries. Richer countries typically have better developed economic insti- tutions, physical infrastructure, and human capital. Indeed, they are richer in large part because of the higher quality of their business environments. By allowing the construction of aggregate indicators of the business environ- ment, these surveys can also help understand the relationship between eco- nomic development and the provision of public goods that comprise the busi- ness environment, and how the transition and the inheritance of decades of central planning change this relationship (annex figure 5.1.2). Annex figure 5.1.2 represents the results of a business environment survey in one country and in two. The individual downward sloping lines represent the demand curves of individual firms for one element of the business envi- ronment. The average constraint reported by firms in the country is denoted by BC. Panel B of the same figure shows how the data from many surveys can be combined to derive a relationship between the cost of a business environ- ment constraint and the level of income or productivity in two different coun- tries. The SPoor line in Panel B represents an element of the business environ- ment in a low income country. The demand curves in a high-income country will be shifted to the right, but typically so will the provision of the public ANNEX FIGURE 5.1.2 Demand and supply of public goods Panel A: One country, many firms Panel B: Two countries, many firms P (reported cost of constraint) P (reported cost of constraint) SPoor S SRich BCPoor BC BCRich Q Q 208 TURMOIL AT TWENTY good as represented by SRich. Indeed, firms in high-income countries make more demands of the electricity grid, but the quality of the electricity supply is also higher in these countries. Panel B also presents the constraint-income locus, sloped downwards; in the electricity example, this implies that the provision of this aspect of the economic infrastructure in high-income countries more than compensates for the greater demands firms in these countries place on it. It is the relation- ship shown as a dotted blue line that appears in figure 5.1 in the chapter. But in other cases the locus could be flat or upward sloping; an example of the lat- ter discussed in the chapter is labor market regulation. This framework is readily extended to comparisons of transition and non- transition economies. Annex figure 5.1.3 illustrates a case where there is, on average, poorer provision of an aspect of the business environment in tran- sition economies (gray) than in nontransition economies (blue), but the gap is declining as income rises and disappears entirely at high incomes. How the constraint-income locus in transition countries varies systematically from what is observed in market economies can reveal much about both the lasting inheritance of central planning and the progress of transition economies in building market economies. ANNEX FIGURE 5.1.3 Two transition economies, two nontransition economies (and many firms) P (reported cost of constraint) SPoorTE SPoor SRichTE SRich PRIORITIZING STRUCTURAL REFORM 209 Annex 5.2 Moving away from the benchmark: firm characteristics and constraints Transition entailed the large-scale reallocation of resources from old to new enterprises, the creation of new patterns of ownership, and a major sectoral redistribution of activities. The survey data can identify how firm characteris- tics are related to the reported constraints on growth, how these compare with nontransition economies, and how they have changed over time in transition. The focus here is on firm size, the comparison between firms adjusting employ- ment levels and those that do not, sector, ownership, and exporter status. The marginal effects as a result of deviating from the benchmark appear in annex tables 5.3.3 and 5.3.4. The results here and in the tables refer to the transition economies over the first three rounds of the BEEPS, for 1999­2005. Firm size A number of firm-size effects are common to transition and nontransition economies: the increasing importance of labor regulation, skills, customs, and the legal system and, until 2005, the declining importance of the finance con- straints as firm size rises (annex tables 5.3.3 and 5.3.4). Firms expanding and contracting employment The survey asks managers whether their current employment level is too low, about right or too high. Other things equal, firms content with their existing employment level are likely to report lower costs of constraints than firms that need to adjust to achieve their desired size. This is consistent with the find- ing that where expanding or contracting firms place different weights on the constraints than nonadjusting firms, it is almost always a higher ranking. In interpreting the results for adjusting firms, note that a much smaller propor- tion of firms in transition report being satisfied with their current employ- ment level (45 percent) than do those outside transition (60 percent). More- over, a substantially higher proportion of expanding firms in transition than those outside report that their employment level is still too low. This reflects the immaturity of the firm size distribution in transition. Adjusting firms report a higher priority to problems with accessing skilled labor in both transition and non transition economies. Along other dimen- sions of the business environment the patterns are different inside and outside transition (annex tables 5.3.3 and 5.3.4). In both transition and nontransition economies, firms adjusting their lev- els of employment complain more about finance than do nonadjusting firms. 210 TURMOIL AT TWENTY This is an example of the need to interpret the data carefully: access to finance and its cost are likely to reflect the quality of the firm (as well as the quality and resources of the banking system). Since firms with stable employment in transition are more likely to report that their size is suboptimal than those outside transition, the fact that stable and expanding firms report less concern about access to finance than contracting firms suggests that finance is being allocated more efficiently rather than less. The complaints of contracting firms may well reflect that they do not have good projects to finance. In the case of finance, as mentioned in the chapter, the level of complaints cannot be inter- preted as a signal of policy priority. Sectoral effects Sectoral effects are in line with expectations.17 In both transition and non- transition economies, service sector fi rms place less importance on infra- structure, finance, labor regulation, and customs constraints. In addition, in transition, they are less concerned with access to skilled labor than are man- ufacturing fi rms. A pattern that also holds in both sets of countries is the greater relative priority to licensing by firms in construction than those in manufacturing. This accords with the higher reported relevance of bribes in construction. Ownership In transition economies, privatized and state fi rms generally evaluate the business environment as less constraining than do new private firms. Both privatized and state firms in transition economies rank tax administration, customs, and corruption as less problematic than do new private sector firms, and separately rate a range of constraints as significantly less constraining than the new private sector benchmark. The only exception is infrastructure, which state-owned firms fi nd more constraining than either new private or privatized firms.18 Foreign-owned and exporting firms Foreign-owned firms in transition report much lower concerns with access to and cost of finance than domestically owned firms and greater relative concern 17. The standard errors associated with estimated coefficients for the sectoral effects in the non- transition economies are much larger than in the transition economies, reflecting a stronger represen- tation of nonmanufacturing fi rms in the transition economy sample. 18. There are not enough state-owned and privatized fi rms in the nontransition economy sample to be able to estimate ownership effects with any precision. PRIORITIZING STRUCTURAL REFORM 211 about customs regulation (annex table 5.3.4). In transition economies, unlike in nontransition economies, foreign firms do not complain more about physi- cal infrastructure (which was less problematic in transition until recently) but attach greater importance to macroeconomic constraints. Exporting firms in transition share the higher priority of foreign-owned firms in relation to cus- toms, licensing, and macro constraints. Exporting firms outside transition attach a higher priority to labor regulation and the legal environment than nonexporters. 212 TURMOIL AT TWENTY Annex 5.3 Tables ANNEX TABLE 5.3.1 Country conditional means of constraint levels and quantitative indicators vs. income per capita Slope of income/ Conditional means constraint locus evaluated at: Low income High income Non- transition Transition Non- Non- economy economy Diff transition Transition transition Transition slope slope sig? economy economy Difference economy economy Difference Average (9) ­0.120* ­0.074 2.46 2.46 ­0.002 2.02 2.19 0.171 Infrastructure (composite) ­0.201* ­0.105* Yes 2.37 1.94 ­0.431* 1.63 1.55 ­0.078 Telecoms ­0.092* ­0.052 1.88 1.57 ­0.312* 1.54 1.37 ­0.165 Electricity ­0.336* ­0.159* Yes 3.01 2.05 ­0.956* 1.76 1.46 ­0.303* Transport ­0.149* ­0.038 Yes 2.23 1.63 ­0.602* 1.68 1.49 ­0.191* Land access ­0.113* ­0.066* 2.03 1.75 ­0.281* 1.61 1.51 ­0.107 Skills 0.029 0.100* 2.13 1.74 ­0.389* 2.24 2.11 ­0.129 Tax rates ­0.079* ­0.067 2.79 3.09 0.304 2.49 2.84 0.349 Tax administration ­0.079* ­0.156* 2.45 2.89 0.432* 2.16 2.31 0.148 Finance (composite) ­0.189* ­0.113 2.84 2.81 ­0.029 2.14 2.39 0.253 Access to finance ­0.169* ­0.124* 2.73 2.65 ­0.072 2.10 2.20 0.096 Cost of finance ­0.202* ­0.095 2.98 2.95 ­0.030 2.23 2.59 0.363* Labor regulation 0.113* 0.177* 1.74 1.50 ­0.234* 2.15 2.16 0.004 Customs ­0.104* ­0.163* 2.03 2.28 0.253* 1.65 1.68 0.034 Licenses 0.000 ­0.120* Yes 1.89 2.15 0.258* 1.89 1.71 ­0.185 Legal ­0.027 0.010 1.89 2.06 0.167 1.79 2.10 0.305 Corruption ­0.092 ­0.180* 2.59 2.62 0.029 2.25 1.95 ­0.299 Crime ­0.084 ­0.105 2.35 2.21 ­0.138 2.04 1.82 ­0.216 Policy uncertainty ­0.123* ­0.097 2.68 2.92 0.244 2.22 2.56 0.340 Macro- economic instability ­0.071 ­0.190* 2.77 3.02 0.253 2.51 2.32 ­0.185 Power outages (y/n) ­0.099* ­0.080* 0.86 0.63 ­0.228* 0.49 0.33 ­0.159* Water supply outages (y/n) ­0.061* ­0.062* 0.40 0.30 ­0.108* 0.18 0.06 ­0.114* (continued) PRIORITIZING STRUCTURAL REFORM 213 ANNEX TABLE 5.3.1 (CONTINUED) Country conditional means of constraint levels and quantitative indicators vs. income per capita Slope of income/ Conditional means constraint locus evaluated at: Low income High income Non- transition Transition Non- Non- economy economy Diff transition Transition transition Transition slope slope sig? economy economy Difference economy economy Difference Bribes (y/n) ­0.082* ­0.073* 0.44 0.68 0.239* 0.14 0.41 0.272* Don't report all sales (y/n) ­0.046* ­0.043* 0.63 0.59 ­0.040 0.46 0.43 ­0.031 Tax gifts (y/n) ­0.039* ­0.133* Yes 0.25 0.69 0.440* 0.11 0.20 0.092 Contract gifts (y/n) ­0.057* 0.008 Yes 0.41 0.37 ­0.042 0.20 0.40 0.198* Security costs (y/n) ­0.015 ­0.027* 0.70 0.71 0.008 0.64 0.61 ­0.037 Crime losses (y/n) ­0.075* 0.052* Yes 0.51 0.12 ­0.391* 0.23 0.31 0.081 * indicates statistical significance at the 5% level; tests adjusted for clustering on country. Note: Column 3 reports the statistical significance of the nontransition economy and transition economy slope coefficients at the 5 percent level. Low income is defined as log GDP per capita = 5.6, about $270; high income is defined as log GDP per capita = 9.3, about $11,000. ANNEX TABLE 5.3.2 Country conditional means of constraint priorities versus income per capita Slope of income/ Conditional means constraint locus evaluated at: Low income High income Non- transition Transition Non- Non- economy economy Diff transition Transition transition Transition slope slope sig? economy economy Difference economy economy Difference Average (9) ­0.120* ­0.074 2.46 2.46 ­0.002 2.02 2.19 0.171 Infrastructure (composite) ­0.201* ­0.105* Yes 2.37 1.94 ­0.431* 1.63 1.55 ­0.078 Telecoms ­0.092* ­0.052 1.88 1.57 ­0.312* 1.54 1.37 ­0.165 Electricity ­0.336* ­0.159* Yes 3.01 2.05 ­0.956* 1.76 1.46 ­0.303* Transport ­0.149* ­0.038 Yes 2.23 1.63 ­0.602* 1.68 1.49 ­0.191* Land access ­0.113* ­0.066* 2.03 1.75 ­0.281* 1.61 1.51 ­0.107 Skills 0.029 0.100* 2.13 1.74 ­0.389* 2.24 2.11 ­0.129 Tax rates ­0.079* ­0.067 2.79 3.09 0.304 2.49 2.84 0.349 Tax administration ­0.079* ­0.156* 2.45 2.89 0.432* 2.16 2.31 0.148 Finance (composite) ­0.189* ­0.113 2.84 2.81 ­0.029 2.14 2.39 0.253 214 TURMOIL AT TWENTY ANNEX TABLE 5.3.2 (CONTINUED) Country conditional means of constraint priorities versus income per capita Slope of income/ Conditional means constraint locus evaluated at: Low income High income Non- transition Transition Non- Non- economy economy Diff transition Transition transition Transition slope slope sig? economy economy Difference economy economy Difference Infrastructure (composite) ­0.060* ­0.034* 0.40 0.26 ­0.135* 0.18 0.14 ­0.038 Telecoms ­0.044* ­0.018 0.28 0.17 ­0.112* 0.12 0.10 ­0.019 Electricity ­0.105* ­0.055* Yes 0.60 0.34 ­0.264* 0.21 0.13 ­0.075* Transport ­0.034* ­0.011 0.34 0.18 ­0.155* 0.21 0.14 ­0.069* Land access ­0.034* ­0.017 0.32 0.21 ­0.111* 0.20 0.15 ­0.049 Skills 0.007 0.059* Yes 0.37 0.18 ­0.185* 0.39 0.40 0.009 Tax rates ­0.021 ­0.019 0.60 0.77 0.172* 0.52 0.70 0.182* Tax administration ­0.027 ­0.056* 0.49 0.69 0.197* 0.40 0.49 0.090 Finance (composite) ­0.046* ­0.033* 0.61 0.63 0.021 0.44 0.51 0.069 Access to finance ­0.041* ­0.036* 0.53 0.55 0.012 0.38 0.41 0.032 Cost of finance ­0.053* ­0.024 0.64 0.67 0.023 0.45 0.58 0.129* Labor regulation 0.060* 0.082* 0.18 0.08 ­0.104* 0.41 0.38 ­0.022 Customs ­0.033* ­0.053* 0.34 0.42 0.076* 0.22 0.22 0.002 Licenses 0.020 ­0.031 Yes 0.18 0.33 0.153* 0.25 0.22 ­0.034 Legal ­0.031* 0.032 Yes 0.33 0.26 ­0.066 0.22 0.38 0.167* Corruption ­0.092* ­0.055* 0.60 0.51 ­0.090 0.26 0.31 0.044 Crime ­0.057* ­0.025 0.48 0.32 ­0.154 0.27 0.23 ­0.033 Policy uncertainty ­0.027 ­0.015 0.52 0.65 0.138* 0.42 0.60 0.182* Macro- economic Instability ­0.023 ­0.061* 0.59 0.71 0.118 0.50 0.48 ­0.022 * indicates statistical significance at the 5% level; tests adjusted for clustering on country. Note: Column reports the statistical significance of the nontransition economy and transition economy slope coefficients at the 5 percent level. Low income is defined as log GDP per capita = 5.6, about $270; high income is defined as log GDP per capita = 9.3, about $11,000. PRIORITIZING STRUCTURAL REFORM 215 ANNEX TABLE 5.3.3 Priority and quantitative measures of business environment constraints for nontransition economies Size Expand- Con- Con- State- (log L) ing tracting Services struction Privatized owned Foreign Exporter Infrastructure (composite) 0.010* 0.005 ­0.010 ­0.016 ­0.003 0.010 0.017 0.039* 0.028* Telecoms 0.006* 0.017* ­0.010 0.049 0.030 0.017 0.007 0.032* 0.023 Electricity 0.005 0.006 ­0.008 ­0.038* ­0.020 ­0.003 ­0.005 0.014 0.018 Transport 0.015* 0.019* 0.003 ­0.056* ­0.026* ­0.009 0.001 0.030* 0.030* Land access ­0.002 0.024* 0.003 ­0.065* 0.001 ­0.058* ­0.068 ­0.023 ­0.027* Skills 0.021* 0.054* 0.045* ­0.020 0.008 0.032 0.028 ­0.031* 0.000 Tax rates ­0.001 0.025* 0.034* ­0.009 ­0.008 ­0.021 ­0.098* ­0.021 ­0.007 Tax administration 0.001 0.012 0.027* 0.009 0.035* ­0.042* ­0.077* 0.023 0.030* Finance (composite) ­0.021* 0.054* 0.076* ­0.050 0.017 0.040 0.020 ­0.118* 0.011 Access to finance ­0.025* 0.060* 0.072* ­0.067* ­0.002 0.025 ­0.006 ­0.114* 0.007 Cost of finance ­0.009* 0.047* 0.071* ­0.030 0.007 0.014 0.008 ­0.099* 0.007 Labor regulation 0.030* 0.006 0.040* ­0.029* 0.009 0.040 0.028 0.009 0.024* Customs 0.028* 0.034* 0.019* ­0.033* ­0.079* ­0.047 ­0.041 0.098* 0.115* Licenses 0.010* 0.038* 0.014 0.049 0.014 ­0.030 ­0.046* 0.012 0.001 Legal 0.015* 0.020 0.038* 0.002 0.008 0.015 ­0.041 ­0.006 0.029* Corruption 0.001 0.035* 0.024* 0.029 0.049* ­0.005 ­0.046 0.007 0.008 Crime 0.002 0.003 ­0.013 0.060* 0.023 ­0.006 ­0.036 0.000 ­0.014 Policy uncertainty 0.014* 0.042* 0.065* 0.002 0.000 0.048 ­0.053 0.011 0.010 Macro- economic instability 0.011* 0.039* 0.063* 0.004 ­0.027 0.044 ­0.053 ­0.011 0.038* Power outages (y/n) 0.007 0.034* 0.022* ­0.048* ­0.097* ­0.047* ­0.020 ­0.001 0.002 Water supply outages (y/n) ­0.007* 0.014* 0.002 ­0.008 ­0.010 ­0.026 ­0.013 ­0.004 ­0.006 Bribes (y/n) 0.005 0.047* 0.066* 0.013 0.084* ­0.026 ­0.103 ­0.014 0.011 Don't report all sales (y/n) ­0.034* 0.023* 0.006 ­0.040 0.006 ­0.046 ­0.127* ­0.106* ­0.028 Tax gifts (y/n) 0.004 0.027* 0.031* ­0.015 0.036 ­0.028 ­0.047 ­0.031* ­0.002 Contract gifts (y/n) ­0.006 0.033* 0.044* 0.024 0.038 0.003 0.014 ­0.022 ­0.023* Security costs (y/n) 0.055* 0.025* 0.032* 0.057* 0.021 0.012 ­0.022 0.015 0.024* Crime losses (y/n) 0.028* 0.022* 0.024* 0.048* 0.076* ­0.009 ­0.030 ­0.005 ­0.019 * indicates statistical significance at the 5% level; tests adjusted for clustering on country. Note: Coefficients are marginal effects relative to the omitted category, except for size (log employ- ment), which is an elasticity. 216 TURMOIL AT TWENTY ANNEX TABLE 5.3.4 Priority and quantitative measures of business environment constraints for transition economies Size Expand- Con- Con- State- (log L) ing tracting Services struction Privatized owned Foreign Exporter Infrastructure (composite) ­0.005* 0.002 ­0.014 ­0.029* ­0.042* 0.000 0.036* 0.006 ­0.011 Telecoms ­0.010* 0.004 ­0.012 ­0.007 ­0.019 0.006 0.029* 0.016 0.005 Electricity ­0.005* 0.002 ­0.012 ­0.049* ­0.072* ­0.008 0.027* ­0.002 ­0.017 Transport 0.002 0.011 ­0.017 ­0.017 ­0.017* ­0.017 ­0.006 0.014 0.021* Land Access 0.006* 0.026* ­0.007 ­0.016 0.040* ­0.046* ­0.028 0.011 ­0.007 Skills 0.024* 0.049* 0.034* ­0.046* ­0.018 ­0.018 0.015 0.017 0.004 Tax rates ­0.001 0.008 0.028* ­0.020 0.013 ­0.001 ­0.095* ­0.017 ­0.008 Tax administration ­0.002 0.015 0.008 ­0.015 0.021 ­0.031* ­0.075* 0.009 0.028* Finance (composite) ­0.019* 0.023* 0.053* ­0.076* ­0.024 0.002 0.015 ­0.113* 0.018 Access to finance ­0.020* 0.008 0.027 ­0.069* ­0.010 ­0.017 0.018 ­0.105* 0.028* Cost of finance ­0.007* 0.023* 0.053* ­0.060* ­0.019 0.016 ­0.035* ­0.086* 0.009 Labor regulation 0.023* 0.016 0.006 ­0.031* ­0.016 ­0.028* ­0.012 0.015 0.013 Customs 0.018* 0.030* 0.006 ­0.028* ­0.088* ­0.062* ­0.134* 0.111* 0.144* Licenses 0.006* 0.023* 0.015 0.005 0.029 ­0.024 ­0.103* 0.034* 0.028* Legal 0.018* 0.000 0.022* ­0.003 0.016 ­0.022 ­0.014 0.015 0.016 Corruption 0.000 0.026* 0.032* 0.002 0.048* ­0.038* ­0.081* 0.010 0.014 Crime ­0.003 0.007 0.006 0.055* 0.024* 0.003 0.004 ­0.038* ­0.033* Policy uncertainty 0.007 0.007 0.040* ­0.009 ­0.015 ­0.003 ­0.022 0.011 0.030* Macro- economic instability 0.003 0.005 0.024* ­0.021 ­0.029 ­0.012 ­0.043* 0.032* 0.059* Power outages (y/n) ­0.005 0.058* 0.054* ­0.015 ­0.097* ­0.043* ­0.060* ­0.024 ­0.009 Water supply outages (y/n) ­0.003 0.032* 0.034* 0.017 ­0.018 ­0.028* ­0.029* ­0.014 ­0.012 Bribes (y/n) 0.002 0.046* 0.032* 0.020 0.105* ­0.052* ­0.201* ­0.021* 0.031* Don't report all sales (y/n) ­0.032* 0.068* 0.031* ­0.004 0.030 ­0.065* ­0.158* ­0.080* 0.002 Tax gifts (y/n) ­0.004 0.031* 0.011 ­0.002 0.018 ­0.069* ­0.181* ­0.029 0.008 Contract gifts (y/n) 0.003 0.038* 0.034* 0.000 0.163* ­0.062* ­0.149* ­0.013 0.017 Security costs (y/n) 0.065* 0.044* 0.058* 0.043* 0.016 0.035* ­0.006 0.036* 0.028* Crime losses (y/n) 0.044* 0.023* 0.055* 0.057* 0.105* ­0.026* ­0.023 ­0.044* ­0.039* * indicates statistical significance at the 5% level; tests adjusted for clustering on country. Note: Coefficients are marginal effects relative to the omitted category, except for size (log employ- ment), which is an elasticity. PRIORITIZING STRUCTURAL REFORM 217 CHAPTER 6 The day after ECA's transition countries were well endowed with public infrastructure and workers' education and skills, but these were increasingly perceived by firms as bottlenecks to growth during the boom years 2005­2008. The economic down- turn has temporarily released these bottlenecks, but they will become a priority once recovery begins. Questions · What factors could facilitate greater private participation in electricity? · What education and labor skills may have to be strengthened in some ECA countries? Findings · Key to releasing bottlenecks in electricity is improving the environment for private sector involvement. Th is requires improving fi nancial viability, facilitating change in the power market structure, and introducing credible regulation. · Also important is maximizing the benefits from regional energy trading, both in the Western Balkans and in Central Asia. · Young students in ECA do well on international tests. Improving compul- sory education will require better incentives for teachers and greater control by schools over staffing decisions. · Specializing too young (vocational and technical colleges in lieu of general secondary education) might be detrimental and needs to be reconsidered. The wage premium to tertiary education is high. Countries have success- fully used the private sector to expand tertiary education--thus comple- menting scarce public resources. Just as the high growth phase on the eve of the crisis revealed infrastructure and skill constraints on the horizon, the recession is likely to loosen them-- but they are also likely to reappear once the economic recovery starts. This is especially likely in countries where fiscal restraint imposed during the reces- sion restricts public investment--in other words, in most ECA countries. Reforms in electricity and education are the main subjects of this chapter, which discusses how to increase private investment. 219 It is worth recalling the evidence on constraints to growth as seen by firms on the eve of the crisis (chapter 5). The metric is whether a constraint is seen as above the average of all constraints comprising the business environment in that country (figure 6.1). This relative concept helps to the extent there may be a systematic variation between country tendencies to report dissatisfaction across the board--for example, reflecting particular periods of instability or economic downturns. Note that workers' education is the second highest con- straint identified by firms, with electricity and corruption at broadly the same level. Courts and access to land are also important bottlenecks.1 Against this background, the chapter poses two questions. First, what could facilitate greater private sector participation in electricity--an impor- tant part of infrastructure? Second, what education and labor skills have to be strengthened in some ECA countries?2 Bottlenecks in electricity--an agenda for reform The investment needs to meet overall ECA electricity demands could be sub- stantial. An estimate by the World Bank based on annual growth of 4.5 per- cent has put the investment needs of the sector at about $1.5 trillion over the next two decades--or 1 to 1.5 percent of ECA's annual GDP. This would absorb a large share of the region's public sector investments, which now average 5.5 percent of the region's GDP. These projections also assume big improvements in energy efficiency and conservation: a decline in ECA's energy intensity from the current 0.70 tons of oil equivalent per $1,000 of GDP to 0.35 tons of oil equivalent in 2030--marked, but still far above the 0.13 value for the EU. Of course the current global economic crisis has slowed down the demand for energy and created some uncertainty on when the electricity sector might become a bottleneck to growth. Moreover, structural factors behind ECA's investment needs are also behind these estimates; for example, about 80 per- cent of the region's electricity generation capacity comes from power plants built before 1980. Since it will be difficult for the public sector to mobilize the required investment, it is imperative to create conditions for private involvement. So far, private participation in the electricity sector in ECA has been low. The private sector invested only $36 billion (in 2007 prices; or 1.1 percent of the 1. As is frequently the case with enterprise surveys, tax rates are perceived as the severest constraint, though it is less clear how to interpret these answers since fi rms are typically not asked about the ben- efits of government expenditures fi nanced by those taxes. 2. In addition to some of the fi ndings discussed in this chapter, the World Bank is working on two flagship reports: one on energy needs and one on the skill challenges facing the region. 220 TURMOIL AT TWENTY FIGURE 6.1 Relative importance of constraints, by country groups Number of countries 30 Middle-income CIS Low-income and lower middle-income CIS Western Balkans EU10+1 and Turkey 20 10 0 ns rm ng rt ion rts d m ce r ion or d es ty de lan rkf ate po ici co at tio an pe si u its ce at pt or Co ctr ns d en xr le wo uc ula fin str to ru dis Te Tra Ele an lic Ta ed r ini ss Co reg to d ss ce dm ely an ss ine ur Ac ce at xa eft bo s qu Ac Bu th Ta La de e, Ina im Cr Source: BEEPS 2008. region's GDP) in the electricity sector in ECA between 1998 and 2007 (figure 6.2).3 Compare that with private participation of $89 billion in Latin America (2.5 percent of the region's GDP), $50 billion in East Asia and the Pacific (1.2 percent), and $33 billion in South Asia (2.3 percent). Although ECA's share of total private investment commitments in electricity projects rose from 10 percent in 1998-2002 to 22 percent in 2003-07, most were in a handful of countries. The Russian Federation, the countries of the European Union, and Turkey accounted for 86 percent of total private activity. Indeed, the privatiza- tion of two wholesale generation companies in the Russian Federation in 2007 accounted for 38 percent of total private investment in ECA in 2003-07. The economic downturn provides an opportunity for countries to accel- erate reforms so that private investments can materialize before capacity becomes a bottleneck. But attracting direct private fi nancing requires sub- stantial efforts to improve the business environment. This section focuses on what might catalyze private investment--how to put in place a business envi- ronment that assures commercial viability. Economies of scale also play a role: 3. Investment data are from the Private Participation in Infrastructure (PPI) database, which includes projects that reached fi nancial closure between 1993 and 2007. While the data here are in real terms (in 2007 dollars), those in the PPI database are in current dollars. For more information, see www.ppi. worldbank.org. THE DAY AFTER 221 FIGURE 6.2 Investment commitments in electricity with private participation, 1993­2007 2007 $ billions 150 Europe and Central Asia Middle East and Africa South Asia East Asia and Pacific Latin America and Caribbean 120 90 Others 14% 17% Russian 60 EU10+ Federation Turkey 55% 31% 38% 30 Two generation projects in 2007 ($9.8 billion) 0 1993­97 1998­2002 2003­07 Source: World Bank and PPIAF, PPI Database. many countries in the region are small, and resource endowments are varied. So, deeper regional cooperation is integral to a sustainable solution to elec- tricity needs. Improving the business environment for private investment Significant private investment can be attracted if electricity utilities are allowed to operate in an environment that fosters commercial viability. This requires ensuring that electricity utilities function on a fi nancially sustain- able basis, reforming the power market structure to move toward more com- petitive trading, and putting in place a credible and predictable regulatory framework. Improving financial viability. The financial viability of firms depends on pay- ment discipline, technical losses, and tariff structure. Strengthening payment discipline is the fi rst line of action. It requires improving collection rates, defined as the ratio between cash electricity pay- ments received and total electricity billed to users, and reducing commercial losses stemming from unbilled consumption (such as the use of artificial con- sumption norms for users who are not metered, as well as theft in the form of illegal connections and meter tampering). Improving collection rates depends on the willingness of governments to introduce legislation enabling utilities to enforce payment. Public users 222 TURMOIL AT TWENTY (government departments and agencies) must have adequate budget alloca- tions for paying their service bills and be subject to the discipline of disconnec- tion for nonpayment. Progress in ECA on payment discipline has been mixed. · Countries in Central and Eastern European adopted the necessary legisla- tion early on. By the mid-1990s, the average collection rate in Poland, Bul- garia, Hungary, and Lithuania was about 90 percent. Since then, the situ- ation in the new EU member states has improved further: collection rates are close to Western European standards--100 percent. · Collection rates have also improved in many CIS countries (figure 6.3). The Russian Federation's parliament amended the necessary laws in the late 1990s, and RAO UES--the main electricity utility--improved collections on its distribution companies. Collection rates increased from 50 percent in the 1990s to 95 percent in the years before the crisis. · Collection remains a challenge in low-income and lower middle-income countries such as Albania, Azerbaijan, Kosovo, the Kyrgyz Republic, FYR Macedonia, and Uzbekistan. Weak governance poses a significant bar- rier. Critical to progress and to avoiding reversals is to strengthen the legal framework that allows utilities to disconnect federal, provincial, and municipal agencies and facilities. Financial viability is also hampered when consumption norms are used in lieu of metered systems. In many CIS countries, enterprise consumption is FIGURE 6.3 Evolution of average collection rates in ECA countries Average collection rates (percent) 1995­99 100 2000­05 2006­08 75 50 25 0 EU 10 + Croatia Western Middle-income Russian Low-income and and Turkey Balkans CIS Federation lower middle- income CIS Note: The middle-income CIS group does not include the Russian Federation, which is shown separately. Source: EBRD Transition Report 2008. THE DAY AFTER 223 billed according to industry norms, and about 10­15 percent of the electricity is sold to households without metering, also on the basis of household norms. Overall, the actual consumption of the household is higher than the norm, which most of the time underestimates real consumption.4 Other sources of commercial losses include theft of electricity through illegal connections, tam- pering with meters, and corruption. Theft in excess of 20 percent is reported in Georgia, Kyrgyz Republic, and Kosovo.5 Technical losses remain high, particularly in the CIS and the Western Balkans, even when compared with Central and Eastern European coun- tries.6 In centrally planned economies, technical design was optimized on the basis of artificially low administered prices of fuel, energy, and mate- rials; thus, efficiency gains were not valued. Improvements will materialize over the long term as old assets are retired and new ones replaced them. In the short to medium term, however, technical losses can be reduced by reha- bilitating the transmission and distribution systems. Investments to reduce technical losses in distribution networks have been undertaken in Georgia, Kazakhstan, Moldova, and the Slovak Republic, with notable improvement as a result. While reducing technical losses can be an important source of revenue, it also requires investments. Incentives to reduce technical losses will depend critically on progress made in payment discipline. It is worth noting that countries in Central Europe have brought down technical losses to about 10 percent--compared with losses of less than 8 percent in the EU15 in 2007. But there is variability across the region (table 6.1). Setting and maintaining tariffs that fully recover costs are central to ensur- ing the financial viability of electricity utilities. Tariffs need to recover the cost of inputs and operating and maintenance costs (short-run marginal cost) and contribute to the funds required for the capital investment needed to sustain the sector (long-run marginal cost). During the socialist period, electricity tariffs levels were far below the long-term supply cost, and the price structure was further distorted by cross-subsidies from industry and business to resi- dential customers. This subsidy pattern has been largely maintained during the transition, particularly in the CIS countries. 4. In Tajikistan, for example, a pilot study showed that household consumption tended to be 2.65 times the norm (Sharma 2005). 5. World Bank 2006, 2008a. 6. It is not clear exactly how much power goes into distribution networks and thus whether losses occur in transmission or distribution networks. Nor is it clear exactly how much power is consumed by end users--and thus whether losses in distribution are technical or commercial. 224 TURMOIL AT TWENTY TABLE 6.1 Total technical and commercial losses in ECA countries (percent) Total losses Country <8 EU-15, Slovak Republic, Czech Republic, Slovenia 9­11 Lithuania, Poland, Romania, Bulgaria, Estonia, Hungary, Russian Federation 12­14 Belarus, Ukraine, Armenia, Azerbaijan, Georgia, Turkey, Turkmenistan ~17 Croatia, Serbia, Tajikistan, Bosnia and Herzegovina, Latvia 20­30 Kazakhstan, Uzbekistan, Kyrgyz Republic, Tajikistan, FYR Macedonia, Montenegro > 35 Moldova, Albania, Kosovo Note: The table consolidates technical losses in both transmission and distribution networks and commercial losses. In practice, distinguishing between these measures is difficult given the absence of metering in many countries. Source: Kazakhstan, Tajikistan, Kyrgyz Republic, Uzbekistan, and Kosovo: World Bank staff; Monte- negro: IEA 2005; all other countries: World Bank World Development Database 2006. A rough indicative range for the long-run marginal cost of generation is 6.5­7.5 cents per kilowatt-hour.7 Based on a gas fired combined-cycle power plant, this does not include transmission or distribution charges.8 It is likely that this type of plant will be the marginal plant (the plant added to meet incremental demand) and so should form the basis for tariff setting, particu- larly with capital largely imported. What is the evidence on electricity tariffs? The new EU member states and some Western Balkan countries appear to cover long-run generation costs in 2008 (figure 6.4). The utilities in these countries have attracted private inves- tors. With some exceptions, such as Albania, Kosovo, and FYR Macedonia, the Western Balkans seem to be moving in that direction as well. The picture is very different in most CIS countries. In Armenia, Azerbaijan, Ukraine, and the Russian Federation, the average residential tariff is close to 5 cents per kilowatt-hour, with residential tariffs as low as 2 cents per kilowatt-hour in the Kyrgyz Republic (box 6.1). These tariffs are not adequate to cover the long- run marginal costs of electricity generation. In other CIS countries, such as Moldova and Georgia, tariffs seem to be closer to this objective. In most countries across the region, the nonresidential tariff is equal to or above that for residential consumers. The burden of tariff increases from a 7. Further work is required to establish the long-run marginal cost in the region, as it is specific to each (national or regional) power network, and depends on factors such as technology, fuel prices, and availability as well as transmission and distribution costs. 8. A gas price of $250­300 per 1,000 cubic meters is equivalent to a crude oil market price of between $75 and $100 per barrel. Th is estimate is based on the typical indexation formula for long-term imports of gas from the Russian Federation to be delivered at Ukraine's western border. See World Bank 2008b. THE DAY AFTER 225 FIGURE 6.4 Weighted average electricity tariffs, 2008 (US cents per kilowatt-hour, excluding taxes) Nonresidential tariff, US cents per kilowatt hour 25 Montenegro 20 Slovak Rep. Hungary Bosnia and 15 Herzegovina Poland UNMIK Kosovo Turkey Lithuania Croatia Latvia 10 FYR Macedonia Albania Bulgaria Moldova Romania Serbia Georgia Ukraine Azerbaijan Estonia 5 Armenia Kazakhstan Russian Federation Kyrgyz Rep. 0 0 5 10 15 20 25 Residential tariff, US cents per kilowatt hour Source: Energy Regulators Region Association, Tariff database 2009. pure efficiency point of view should be borne by residential consumers given that the cost of supplying nonresidential consumers is lower than the cost of supplying residential consumers--technical losses associated with supply- ing large consumers are low (figure 6.4). Some rebalancing of tariffs to reduce cross subsidies has taken place, and the average ratio of residential to nonresi- dential prices is 0.97. But more progress has been achieved in Central Europe than in other countries such as Armenia. Appropriate sequencing of reforms is critical in the presence of weak pay- ments discipline and low tariffs. Compared with Poland and Hungary, which focused on improving their electricity laws and the operation of their utilities before privatization, Kazakhstan, Georgia, Moldova, and Ukraine privatized some of their electricity assets without first addressing payments discipline, with mixed result on efficiency.9 More recently, the Kyrgyz Republic engaged in sector unbundling, albeit in the context of extensive use of barter and off- sets in lieu of cash for the settlement of transactions. To date, tangible results of such an ambitious program have not materialized, and the level of service to the population has not improved. One of the main lessons is that power generation should not be privatized before distribution, particularly when there are severe payments problems. 9. World Bank 2003. 226 TURMOIL AT TWENTY BOX 6.1 Electricity tariff increases and poverty impacts A key challenge is to ensure that tariff reforms are accompanied by social safety nets to protect the most vulnerable from price shocks and secure social and political support. There are two ways to mitigate the impact of higher electricity tariffs: block tariff structures and targeted subsidies. Under a block tariff structure, a basic amount of power is provided below cost, with higher tariffs for consumption beyond the specified threshold. Block tariffs are attractive because they are simple and can be designed to be revenue neutral (they do not require budget financing). But they do not exclusively tar- get the poor, they might even unjustly punish poor families (if they use elec- tricity for heating without any evident alternatives), and the systems can be circumvented by consumers that have numerous meters. From an energy con- servation point of view, the higher marginal tariff of the highest block has the benefit of discouraging inefficient consumption through price signals. Under targeted subsidies, an earmarked transfer is provided (such as a voucher for power consumption or a cash transfer specifically for poor con- sumers). Subsidies are attractive because they target the poor and do not distort resource allocation, but they require good targeting and have a fiscal impact. Transfers to compensate the bottom quintile of the population as a result of adjustments in tariffs are not that large. The calculations in box table 1 are based on household consumption of utilities, which includes not only electric- ity consumption but also charges for district heating and water consumption. The estimated figures therefore overestimate the impact of changes in electric- ity tariffs alone, unless district heating charges--the other main item in utility expenditure--are also adjusted in the process. As an example, an adjustment in tariffs of 50 percent in Armenia, Azerbai- jan, Kazakhstan, the Russian Federation, and Ukraine--where tariffs are in the BOX TABLE 1 Percentage of GDP needed to compensate the bottom quintile Tariff increase 50 percent 70 percent Armenia 0.10 Azerbaijan 0.12 Kazakhstan 0.11 Ukraine 0.29 Russian Federation 0.11 Kyrgyz Republic 0.28 0.39 Note: Utilities include both direct consumption of electricity by households as well as dis- trict heating charges. (continued) THE DAY AFTER 227 BOX 6.1 (CONTINUED) Electricity tariff increases and poverty impacts range of 5 cents per kilowatt-hour--will need a transfer ranging from 0.11 per- cent of GDP in the first four countries and a much larger amount of about 0.28 percent of GDP in Ukraine. The reason for the higher figure for Ukraine is a higher share spent on utilities but particularly a higher share of income of the poorest quintile in total GDP. In the Kyrgyz Republic, the transfer would have to be 0.28 and 0.39 percent of GDP for tariff increases of 50 and 70 percent respec- tively; these figures also influenced by the high share of income of the poorest quintile in total GDP. Adding the second quintile to the compensation scheme increases the cost of the transfer by about 2.5 times in most countries. To help place these amounts in perspective, it is worth comparing them with what coun- tries spend on safety nets, ranging between 1 and 2 percent of GDP (chapter 4). Reforming the power market structure and trading regimes. An enabling envi- ronment for private participation also requires unbundling the energy sector to create more transparency and competition. The traditional prescription is to transform vertically integrated utilities into separately managed companies. The application of such prescriptions would lead to full vertical unbundling (separation of the state-owned power company into generation, transmission, and distribution entities) and horizontal unbundling of electricity supply (separating power generation companies with a view to deepening competi- tion, and separating distribution companies to support liberalization). The varied experience in the developing world suggests that power supply reform needs to be adapted to country-specific conditions. The relevant fac- tors to decisions on vertical and horizontal unbundling are power system size, resource endowment, and country institutional capacity to manage complex trading mechanisms.10 In medium to large power markets with strong institu- tional capacity, full horizontal and vertical unbundling is generally preferred. For small markets, horizontal unbundling into small entities generally would not make sense. But some vertical unbundling would increase the transpar- ency of operations and facilitate regional trade. Figure 6.5 describes progress toward sector unbundling in the region. Motivated by the requirements set by EU Directives,11 countries that recently joined the EU are among those that have unbundled the most. These 10. ESMAP 2006. 11. The EU Directive on the creation of an integrated electricity market (EU 2003) focuses on break- ing up vertically integrated supply chains to allow competition in the power market, regulated third- party access to the power network, coexistence of regulated and competitive markets side by side, and allowing consumers to choose suppliers. 228 TURMOIL AT TWENTY FIGURE 6.5 Present structure of ECA electricity markets Retail competition: End users can legally choose their power EU10, supplier (full market opening) with TPA to the transmission and Croatia, distribution networks and competitive wholesale market. Western Balkans, Wholesale competition TPA: Distribution retailers and large consumers Turkey, Vertical unbundling buy directly from generation entities as an organized power exchange or Kazakhstan, bilaterally. They transmit electricity under open access arrangements Moldova, (third party access, TPA) with the transmission and distribution entities. Ukraine Single buyer: Purchasing agency which has the exclusive right to purchase and sell Armenia, Azerbaijan power in a wholesale electricity market. Vertically integrated monopoly + independent power producers Tajikistan (IPPs) that sell power to it Vertically integrated Uzbekistan, monopoly Turkmenistan, Belarus Horizontal unbundling Source: Based on Besant-Jones 2006 and Bank staff. countries have unbundled their transmission entities, implemented third- party access to their networks, and opened, to varying degree, their retail market to competition.12 But progress has also taken place elsewhere in the region. · The three biggest generators control more than 70 percent of generation capacity in the Czech Republic, Estonia, Latvia, Lithuania, the Slovak Republic, and Slovenia. Concentration in the retail market is even higher. Further market restructuring is expected with the implementation of the proposed third legislative energy market package.13 · In Turkey, substantial progress in sector restructuring, in line with relevant EU legislation, has enabled private participation. In 2008, four regional distribution companies have been privatized (with total receipts from the privatization estimated at about $2.4 billion). Bidding for the privatization 12. Third-party access is to grant open access for parties other than a vertically integrated power utility to use its transmission and distribution network for a fee. It is the most tangible way of creating a real level playing field between the incumbent and new project developers to reach customers directly. 13. The proposal for the third direction concerning common rules for the internal market in electric- ity and gas (EU 2007) includes the following measures: effective separation of supply and production activities from network operation; further harmonization of the powers and enhanced independence of the national energy regulators; establishment of an independent mechanism for cooperation among national regulators; creation of a mechanism for transmission system operators to improve the coor- dination of networks operation and grid security, cross-border trade and grid operation; and greater transparency in energy market operations. THE DAY AFTER 229 of the next three distribution companies is under way and expected to be completed in 2009. · The Western Balkans have also adopted some relevant EU legislation in the sector and advanced fairly quickly in vertical unbundling. They have unbundled transmission systems, implemented third-party access to net- works, and partially opened their retail market to competition. But the small size of these markets limits the competition that can be expected in each market. Thus, the development of competitive electricity markets will come from cross-border trade. · Some CIS countries have also taken important steps, such as establishing a wholesale electricity market, as in Kazakhstan, Moldova, and Ukraine. Credible and predictable regulatory framework. A credible and predictable regu- latory framework is also necessary in any strategy to attract private investment. When such a framework is in place, utilities and new investors can predict with some confidence that tariffs will generate enough revenue to cover cost. Other- wise, private financing would be attracted only on costly terms, which would result in high rates of return sought by investors, take-or-pay contracts as used in independent power producers, or investors insisting on sovereign or third party guarantees. Effective regulation, together with competitive markets, can encourage cost reductions and mitigate some of the adverse affordability impacts of tariff increases. Some of the options and factors to be tackled include: · Establishment of an independent, sector-specific regulatory authority. Independence is required to ensure that regulatory decisions critical to investors--from awarding licenses to setting tariffs--are based on tech- nical factors and relatively immune to political and bureaucratic interfer- ence. There are various criteria for assessing whether a regulator is inde- pendent. Four dimensions are crucial: the independent regulator should be separate from the government, appointed for a fi xed term, funded by the regulated industry (through, for example, license fees), and have the power to approve tariffs without requiring approval from government officials. Table 6.2 shows the extent of progress along these dimensions. · Regulatory reform has progressed significantly in the region over the past five years because of both widespread privatization and EU acces- sion.14 Indeed, a totally independent regulator (the first four columns of table 6.2) is now present in 10 ECA countries (Albania, Bosnia, Georgia, 14. The Power Directive of the European Union requires that regulatory bodies be established, and that they have the authority at least to fi x or approve tariff methodologies for power transmission and distribution, and ideally to set tariff levels (EU 2003). 230 TURMOIL AT TWENTY TABLE 6.2 Regulatory institutions in ECA countries Fixed-term Full tariff- Separate appoint- Industry setting Trans- regulator ment funding power parency Redress South East Europe Albania Bosnia & Herzegovina Croatia Macedonia, FYR Montenegro Serbia EU countries Bulgaria Czech Republic Estonia Hungary Latvia Lithuania Poland Romania Slovak Republic Slovenia Black Sea and Belarus Armenia Belarus Georgia Moldova Ukraine Turkey Central Asia Azerbaijan Kazakhstan Kyrgyz Republic Tajikistan Uzbekistan Turkmenistan Latvia, FYR Macedonia, Moldova, Montenegro, Romania, Slovenia, and Turkey) compared with only three in 2003.15 Countries such as Alba- nia, Romania, and FYR Macedonia have made significant progress in 15. A survey carried out in 2003 by EBRD and the World Bank found that only three countries (Lat- via, Moldova, and Slovenia) demonstrated total regulatory independence as defi ned above (Kennedy 2003). THE DAY AFTER 231 conjunction with the privatization of their distribution companies. In the Russian Federation and in Central Asia, with the exception of Tajiki- stan and Turkmenistan, separate regulatory bodies have been set up, even though they cannot yet be regarded as fully independent. · To the extent possible, an independent regulator should retain control over tariff setting, but practice in the region has been mixed. In Bulgaria, Croa- tia, the Czech Republic, Hungary, and Serbia, tariffs are recommended by the regulator but must be approved by the government (see table 6.2). In Bulgaria and Romania, the tariff methodology must be approved by the government. Such arrangements can limit the economic and technical information on which tariffs decisions should be based. Provisions for the government to approve tariff setting might be appropriate for an interim period while a regulator establishes a track record. For example, the staff of newly formed regulatory office often come from the power companies, and the risk of capture is thus real. This would warrant greater government involvement because an effective regulator needs to be independent of both political and industry interference. But as the regulator establishes its cred- ibility as an independent agent, tariff setting should be separated from gov- ernment decisionmaking. · Although an independent regulator must be isolated from the political pro- cess, regulatory discretion opens the door to regulatory interference. Regu- latory discretion can be mitigated by making the regulators more account- able for their actions. Th is can include substantive reporting and audit obligations, a high level of transparency in regulatory decisions and their justification, and appeal rights for parties believing that their interests have been harmed by regulators' decisions not following the law. Independent regulators in almost all countries in ECA are now legally obliged to pres- ent an annual report on their activities to the legislature or the executive branch. The publication of rules, regulations, and decisions is now com- mon in most countries. Transparency has significantly improved recently as the consultation process, an element in improving accountability, plays an increasing role in regulatory decision-making. · Regulated companies in nearly all countries must have recourse to the judi- cial system to seek redress. The local judiciary is unlikely to have the req- uisite expertise and may not be perceived by regulated companies or inves- tors as impartial (see table 6.2). Recourse to national courts is enforced in Azerbaijan, Georgia, Kazakhstan, Montenegro, Poland and Roma- nia. Additional procedures of redress through international arbitration have proved useful, particularly where privatization has taken place, as in 232 TURMOIL AT TWENTY Kazakhstan and Moldova.16 In EU member states, potential recourse to European courts may also provide adequate investor security. · It will take time before the new regulatory system becomes fully func- tioning and credible--and the lack of track record may still inhibit private investment. This is the situation in some countries in the region. Regula- tory institutions are sometimes too new to develop a good track record, and the new regulator lacks the experience with private sector operators. In such cases, governments may wish to consider temporary risk-mit- igation instruments to facilitate the mobilization of commercial debt or equity. Some of the risks that private investors are not willing to take can be transferred, at a price, to third-party official institutions, such as multi- lateral or bilateral agencies, that are capable of taking such risks. These have proved useful to facilitate a smooth transition to a credible regulatory framework while a track record is being developed. For example, Romania and Albania successfully privatized their distribution companies and sig- nificantly diminished the cost of capital by using the World Bank's partial risk guarantee. Deepening regional cooperation and trade Expanding power trade in the ECA region is an untapped option that could yield significant economic savings and boost security of supply. Trading can be either short- or long-term. Short-term trade allows countries to change dis- patch patterns opportunistically. Long-term trade, by contrast, affects both dispatch and system expansion and involves firm commitments to interna- tional trade. Almost all ECA countries exchange electricity on a short-term basis. Some are now considering deeper trade as they develop their institu- tional capacity and as confidence strengthens in cross country arrangements for regional cooperation. The main drivers of long-term trade in the region are large mismatches between electricity demand and resource endowments. For example, central Asian countries could exploit coordination between thermal and hydropower production between the summer and winter seasons. Developing and inter- connecting a largely thermal power system in Uzbekistan and Kazakhstan with the largely hydropower systems of Tajikistan and the Kyrgyz Repub- lic would allow energy banking. Th is is also an important option for small countries rich in coal (such as Kosovo) where power markets are too small to 16. Private investors in Kazakhstan and Moldova initially sought recourse to international arbitra- tion, as allowed under their privatization contract, to solve disputes related to tariff-setting; in the end, resolution was reached without arbitration. THE DAY AFTER 233 exploit substantial economies of scale. Increased cooperation and trade would secure access to export markets and export revenues to support their energy export-led growth. Further economies are possible where utilities face differ- ent system load shapes or experience peak loads at different times of the day or year. In this situation, the output of peaking plants can be shared between countries. When networks are interconnected, as opposed to operating inde- pendently, power supply reliability can be achieved with a lower reserve mar- gin in generation capacity. Two untapped electricity trade arrangements opportunities exist in the ECA region: the South East Europe regional integrated electricity market, and elec- tricity projects based on seasonal and endowment features in Central Asia. South East Europe regional electricity market. Recognizing the potential gains from increased energy trade, the EC and the countries of Southeastern Europe subscribed in 2006 to the Energy Community Treaty.17 The Treaty contains a commitment to market reforms and the operation of an integrated regional electricity and gas market, compatible with the internal energy market of the EU.18 This helps offset the disadvantages of small market size, which together with a dominant national generator tend to create isolated markets with weak competition. The Treaty also establishes a mechanism for cooperation and dialogue among governments, regulatory authorities, industry and interna- tional donors. Network unbundling and third-party access are generally well advanced (table 6.3). Even so, an integrated regional electricity market still faces major obsta- cles. One of the most critical areas is market opening. Except in Romania and Bulgaria, most countries need to set a calendar and take concrete measures to open the market and allow competition in both household and nonhousehold electricity markets. In most countries, wholesale market activity remains low--the result of national markets dominated by a single, generally state- owned, generator that supplies at regulated rates to tariff customers. Another problem is that cross-border trade is hampered by the lack of harmonization of congestion management and cross-border capacity allocation mechanisms. 17. The Energy Community Treaty created the Energy Community of South East Europe with the following parties: the EC and the Western Balkans (Albania, Bosnia and Herzegovina, Croatia, FYR Macedonia, Montenegro, Serbia and Kosovo). Romania and Bulgaria joined the EU in January 2007 and are classified as "participants." 18. The Generation Investment Study (EU 2004) estimated that operating the South East Europe power system as a single, fully interconnected network would reduce investment requirements and save approximately 3 billion. 234 TURMOIL AT TWENTY TABLE 6.3 Implementation of the directive 2003/54/EC in the contracting parties to the treaty establishing the energy community, December 2008 Bosnia and Mace- Herze- donia, Monte- UNMIK Albania govina Croatia FYR negro Serbia Kosovo Public service and customer protection Monitoring security of supply Technical rules (grid code, distribution code) Authorization for new generation capacity Unbundling provisions Third-party access to networks Electricity market opening Cross-border trade All provisions Some Some Bottlenecks to available provisions provisions progress available missing Source: Energy Community Secretariat 2009. Although the possibility of secondary trading of cross-border capacities exists in a few cases, these markets do not function well. Developing energy trade in Central Asia. Kazakhstan, Kyrgyz Republic, Tajiki- stan, and Uzbekistan, possess complementary natural resources. For example, hydropower exported from the Kyrgyz Republic and Tajikistan during sum- mer months is cheap relative to thermal power generated in Kazakhstan and Uzbekistan. In contrast, the economic costs of hydropower generation in the Kyrgyz Republic and Tajikistan during the winter are substantial. The Kyrgyz Republic and Tajikistan could meet their energy needs far more efficiently using coal and gas imported from Kazakhstan and Uzbekistan. But the inabil- ity of the Kyrgyz Republic and Tajikistan to access these fuels and the lack of their own thermal generation capacity are causing them to use and develop hydropower to avoid endemic power shortages in the winter. Unfortunately, THE DAY AFTER 235 TABLE 6.4 Features of electricity systems in Central Asia Regulatory Cash collection Tariff codes and (percent of System losses (percent of institutional billing) (percent) cost recovery) structure Kazakhstan ~85 ~20 100 Reasonable Kyrgyz Republic 94 35 ~55 Inadequate Tajikistan 85 19 ~58 Inadequate Uzbekistan ~70 23 90 .. .. is not available. Source: World Bank staff. water released in winter can cause flooding and compromise the availability of water to meet irrigation needs in downstream countries in the summer. The potential economic savings from committing to long-term framework agreements on electricity and water exchange are large. But in part because of each government's post-independence apprehension of being dependent on outside sources for energy, such agreements have not been put in place. National policies so far have favored self-sufficiency and substitution even at the expense of trade and rationalization of energy use. The failure to move to new agreements has left energy trade in a complex structure of state-to-state barter arrangements that are neither efficient nor able to meet the changing needs of each country. 19 As noted before, countries exhibit large sector losses, non-billing, and non-payment. Electricity tariffs are below cost-recovery in the Kyrgyz Republic and Tajikistan (table 6.4), and this has led to fi nancial and physical deterioration of the sector and excessive demand for power by consumers. In addition, efforts to restructure electricity supply by utilities have been limited. The Central Asian countries, supported by an alliance of major multilateral development institutions, have announced initiatives to encourage energy trade but with limited impact so far. The Central Asia Regional Economic Cooperation program, initiated in 1997, may hold some promise. It focuses 19. In 1998, upstream (Kyrgyz Republic and Tajikistan) and downstream (Kazakhstan, Uzbekistan) riparian countries concluded an agreement on the water and energy resources of the Syr Darya River. It sought to compensate the upstream states for water storage services through the purchase of surplus summer electricity from the Kyrgyz Republic and the supply of fossil fuels needed for Kyrgyz winter needs by the downstream states. The agreement weakened when Uzbekistan's interest in importing electricity in the summer declined and when it insisted on cash for its gas rather than electricity and the benefits of water storage and seasonal release. The agreement is no longer functional. 236 TURMOIL AT TWENTY on key investment projects, such as improvements in power transmission sys- tems, and provides technical assistance to promote discussions on power sec- tor reform and analysis of electricity export potential, and is also developing the Water-Energy initiative. Azerbaijan, China, Kazakhstan, Kyrgyz Republic, Mongolia, Tajikistan, and Uzbekistan signed a memorandum of understand- ing establishing the CAREC Members Electricity Regulators Forum in 2005. The Forum is designed to help members capitalize on their shared experiences as they seek to reform their power sectors and identify possibilities for energy trade. In another initiative, Tajikistan and Kyrgyz Republic, with Afghanistan and Pakistan, are developing a Central Asia South Asia Regional Electricity Market. The objective is to exploit and monetize Central Asia's hydroelectric power surplus during the summer. The education and skills agenda--making the grade For the fi rst time since BEEPS started a decade ago, fi rms are identifying workers' education and skills as a major impediment to their growth pros- pects. Until 2005, this was not an area typically identified by firms as a con- straint, though the gap between transition countries and nontransition coun- tries had largely disappeared among the high-income transition countries. As mentioned in Chapter 5, firms were surveyed in 2008--at the eve of the pres- ent crisis--and had very different perceptions about labor as an impediment to growth. About 30 percent of firms (unweighted average) considered education and skills to be a major or severe constraint (figure 6.6). Most dissatisfaction is among the middle-income CIS countries--more than 40 percent of firms are dissatisfied. In contrast, there is less dissatisfaction in the Western Balkans and significant variability in the EU10 plus Turkey. The share of firms assessing the availability of the "right" workers as either no obstacle or a minor obstacle has declined in the last three years (figure 6.7). Negative values mean that a smaller share of firms were satisfied in 2008 than in 2005. For example, if the share of satisfied firms in a country declined from 0.50 to 0.25 between 2005 and 2008, the country would be in the -0.2 to -0.3 range of the figure. Except for one country, all magnitudes are negative. The worst deterioration has taken place in the middle-income CIS countries. But even in a large number of new EU members and other CIS countries, the reduction in satisfied firms is between 0.2 and 0.3. What attributes are firms valuing when they refer to educated labor and skills? Are they valuing specific technical knowledge, or are they valuing more general attributes associated with years of schooling, such as adaptability, THE DAY AFTER 237 FIGURE 6.6 Worker education: percentage of firms considering it a "major" or "very severe" constraint in 2008 Number of countries 8 Middle-income CIS Low-income and lower middle-income CIS Western Balkans EU10+1 and Turkey 6 4 2 0 % 0% 0% 0% 0% 0% 0% 0% 0% 10 <2 <3 <4 <5 <6 <7 <8 <9 x< x x x x x x x x % % % % % % % % 10 20 30 40 50 60 70 80 FIGURE 6.7 Worker education: change in the frequency of satisfied firms between 2005 and 2008 Number of countries 10 Mean = ­0.2 Low-income and lower middle-income CIS Middle-income CIS Western Balkans 8 EU10+1 and Turkey 6 4 2 0 5 4 3 2 1 <0 .1 .2 .3 .4 5 5 0. 0. 0. 0. 0. 0. 0. 0 0 <0 <0 x ­ <­ <­ <­ <­ x< x< x< x x< x x .1 0 1 4 x x x x ­0 2 3 .5 .4 .3 .2 0. 0. 0. 0. ­0 ­0 ­0 ­0 overall analytical thinking, and communication skills? Will simply adding years of schooling be the solution? Or does the type of education matter (such as general or vocational education)? Is tertiary education crucial, and, if so, in which fields? How important is the curricular content as a substitute for sim- ply more years of schooling? 238 TURMOIL AT TWENTY Evidence from students tests--youngsters do well internationally Young students in ECA countries generally do better in international tests than those with similar incomes. The recently published 2007 Trends in Inter- national Mathematics and Science Study (TIMSS) is the fourth of a series of international comparisons of young students at grades 4 and 8. For the first time, this study measures separately the knowledge of the curricula (content tests) from applications of that knowledge to real life problems (cognitive tests) (figures 6.8 and 6.9). Generally, ECA countries perform very well at grade 4 relative to countries with similar per capita incomes. These patterns hold for both mathematics and science and for both mastery of curricula and applied knowledge. These TIMSS results at grade 4 also echo results from the Progress in International Reading Study, another international assessment measuring students' reading abilities. Some country patterns emerge from figures 6.8 and 6.9.: · Students from the CIS and the Baltic states perform particularly well. · ECA countries perform better than their peers in math, but in science, Georgia, Turkey, and Romania are slightly below the line. · Students in the Russian Federation still perform quite well in grade 8--there is no clear signal of a deterioration of their relative position between grades 4 and 8.20 The TIMSS results can be compared with results from the 2006 OECD Program for International Student Assessment (PISA). These tests are taken at a later age and measure more general competencies of students and their abil- ity to apply knowledge (capturing education imparted at grades 9 and 10). For example, the PISA mathematics test asks students to apply their mathematical knowledge to solve problems set in various real world contexts. The compari- son of PISA scores of ECA students is relative to peers in countries with simi- lar per capita incomes. Again, the majority of students in ECA perform well relative to peers (figure 6.10). The relative position of ECA students remains rather invariant, either using TIMMS or PISA. Two additional countries are in the PISA sample: Azerbaijan and Kyrgyz Republic. Azerbaijan does extremely well in mathematics--far above the comparators. In science, however, its posi- tion is below the comparators. 20. The 1997 Report on Education by the Public Chamber of the Russian Federation points out that although Russian elementary school graduates do better than the world average in knowledge in math and natural science, they do start to lag behind their peers at grades 5­9 (lower secondary) in their ability to apply their knowledge (Public Chamber of the Russian Federation Report 2007). A more recent study on the Russian Federation's education system also advances the notion that Russian schools may place greater emphasis on acquisition of encyclopedic knowledge over problem solving, particularly as education evolves from primary to upper secondary schools (Kapelyushnikov 2008). THE DAY AFTER 239 FIGURE 6.8A 2007 TIMSS--mean content scores in grade 4 math Mean scores 80 R-squared = 0.56 Kazakhstan Russian Federation 60 Latvia Lithuania Armenia Hungary Ukraine Slovak Slovenia Czech 40 Georgia Rep. Rep. 20 0 7 8 9 10 11 Log of GDP per capita (PPP) FIGURE 6.8B 2007 TIMSS--mean cognitive scores in grade 4 math Mean scores 80 R-squared = 0.55 Kazakhstan Russian Federation 60 Latvia Lithuania Armenia Hungary Ukraine Slovak Slovenia 40 Georgia Rep. Czech Rep. 20 0 7 8 9 10 11 Log of GDP per capita (PPP) FIGURE 6.8C 2007 TIMSS--mean content scores in grade 4 science Mean scores 80 R-squared = 0.66 Russian Federation Latvia Hungary 60 Kazakhstan Armenia Lithuania Slovenia Ukraine Slovak Czech 40 Georgia Rep. Rep. 20 0 7 8 9 10 11 Log of GDP per capita (PPP) 240 TURMOIL AT TWENTY FIGURE 6.8D 2007 TIMSS--mean cognitive scores in grade 4 science Mean scores 80 R-squared = 0.66 Russian Federation Latvia Kazakhstan Hungary 60 Armenia Lithuania Slovenia Ukraine Slovak Czech 40 Rep. Rep. Georgia 20 0 7 8 9 10 11 Log of GDP per capita (PPP) FIGURE 6.9A 2007 TIMSS--mean content scores in grade 8 math Mean scores 80 R-squared = 0.45 60 Russian Lithuania Bosnia and Herzegovina Federation Hungary Czech Rep. Serbia 40 Armenia Bulgaria Ukraine Romania Slovenia Georgia Turkey 20 0 7 8 9 10 11 Log of GDP per capita (PPP) FIGURE 6.9B 2007 TIMSS--mean cognitive scores in grade 8 math Mean scores 80 R-squared = 0.46 Russian Federation Lithuania 60 Bosnia and Hungary Herzegovina Bulgaria Czech Rep. Serbia Armenia 40 Ukraine Slovenia Georgia Romania Turkey 20 0 7 8 9 10 11 Log of GDP per capita (PPP) THE DAY AFTER 241 FIGURE 6.9C 2007 TIMSS--mean content scores in grade 8 science Mean scores 60 R-squared = 0.44 Russian Federation Hungary Bosnia and Slovenia 50 Herzegovina Czech Rep. Lithuania Armenia Bulgaria Serbia 40 Ukraine Romania Turkey 30 Georgia 20 7 8 9 10 11 Log of GDP per capita (PPP) FIGURE 6.9D 2007 TIMSS--mean cognitive scores in grade 8 science Mean scores 60 R-squared = 0.46 Russian Hungary Federation Slovenia 50 Bosnia and Herzegovina Czech Rep. Lithuania Armenia Bulgaria Serbia 40 Ukraine Romania Turkey 30 Georgia 20 7 8 9 10 11 Log of GDP per capita (PPP) FIGURE 6.10A PISA math scores Mean scores 550 R-squared = 0.69 Czech Rep. Estonia Lithuania 500 Russian Federation Azerbaijan Croatia 450 Serbia Latvia Turkey Slovenia Romania Slovak Hungary 400 Rep. Montenegro Bulgaria 350 Kyrgyz Rep. 300 7 8 9 10 11 Log of GDP per capita (PPP) 242 TURMOIL AT TWENTY FIGURE 6.10B PISA science scores Mean scores 550 R-squared = 0.71 Czech Rep. Estonia Slovenia 500 Slovak Rep. Croatia Hungary Russian Federation 450 Serbia Turkey Lithuania Latvia Bulgaria 400 Azerbaijan Romania 350 Montenegro Kyrgyz Rep. 300 7 8 9 10 11 Log of GDP per capita (PPP) Interestingly, the PISA study allows computation of an index of economic, social, and cultural status (ESCS) comprising parents' education, occupational status, and possessions in the home. It allows classifying students in quintiles according to the level of the ESCS index, thus allowing comparisons on how students in one quintile perform against those in another quintile. Implicitly, it assesses how well the educational system has compensated for such ESCS differences. The main conclusion is that differences in test scores across socioeconomic groups are similar to those observed in many OECD countries. Figures 6.11a and 6.11b show for several countries the difference in students' scores between the highest and lowest ESCS quintiles. Countries are ranked according to the magnitude of such differences. Positive values indicate that students in the highest ESCS quintile performed better than those in the lowest quintile. Negative values mean students in the lowest ESCS quintile performed bet- ter than students in the highest quintile. The figures show ECA countries to be roughly in the middle of the distribution, at par with many OECD coun- tries. For most ECA countries, the (positive) differences are lower than those observed in the United States, France, Germany, and Belgium. Thus, at least on an international basis, it is not obvious that ECA countries participating in these programs have a distinctive distributive issue regarding the perfor- mance of students with different socioeconomic status. This depends on how the ESCS index is measured, of course, but the results provide a benchmark that may encourage further work in this area. Evidence from labor markets--changing structure of occupations and wages Most ECA countries have experienced strong changes in the structure of output as results of market liberalization and more open trade regimes. Job THE DAY AFTER 243 FIGURE 6.11A PISA reading scores--difference between quintile 5 and quintile 1 Points of PISA reading scale 120 ECA country Non-ECA country 80 40 0 ­40 Slovak Rep. Iceland Qatar Montenegro Finland Sweden* Slovenia* Macao, China* Denmark* Taiwan, China Greece* Japan Switzerland Estonia Australia Azerbaijan Austria United Kingdom Spain Korea Canada Hong Kong, China Ireland Netherlands Latvia Russian Federation Liechtenstein Serbia Romania Poland France Kyrgyzstan Lithuania Luxembourg New Zealand Belgium Bulgaria Tunisia Jordan Indonesia Thailand Mexico Portugal Chile Colombia Brazil Argentina Norway Italy Hungary Germany Turkey Uruguay Note: Differences are statistically significant unless indicated by *. FIGURE 6.11B PISA math scores--difference between quintile 5 and quintile 1 Points of PISA math scale 120 ECA country Non-ECA country 80 40 0 ­40 Slovak Rep. Czech Rep. Iceland Norway* Azerbaijan* Qatar* Montenegro Macao, China Slovenia Finland Australia Canada Taiwan, China Sweden Japan Switzerland Denmark Estonia Russia Ireland United Kingdom Netherlands Greece Hong Kong, China Spain Austria Kyrgyzstan Serbia Liechtenstein Luxembourg Latvia Poland Korea New Zealand France Lithuania Jordan Romania Belgium Indonesia Bulgaria United States Tunisia Thailand Mexico Portugal Colombia Chile Argentina Brazil Italy Germany Hungary Turkey Uruguay Note: Differences are statistically significant unless indicated by *. growth has been greatest in the service sector, with strong reallocations within manufacturing as a result of more open trade. New firms have been a significant force for job creation, while job contractions have taken place in large state owned and newly privatized firms.21 How have labor markets responded to these developments--and what do these changes suggest in terms of education and training systems? Figure 6.12 shows the number of countries where the percentage participation of differ- ent occupations increased or declined during the last decade. The definition 21. Mitra 2008. 244 TURMOIL AT TWENTY FIGURE 6.12 Positive and negative shifts in employment participation in 1995­2006, by occupation Number of countries 20 Positive Negative 10 0 10 20 Service Professionals Technicians Unskilled Skilled Clerks Agriculture and sales and managers manual manual of occupations is quite broad and follows the one-digit ISCO/ILO classifica- tion.22 Four patterns are worth highlighting: · Occupations in the service sector and retail have expanded--areas that were held back under central planning, as noted in chapter 5. · Managers and professionals have also expanded, in all likelihood, linked to the transformation of the enterprise sector.23 · Agricultural employment has shrunk. · For a mix of middle level and manual labor occupations, the picture is more diverse. In about half the countries, these occupations have expanded, while in the other half they have decreased. These are occupations where changes are very country-specific and sensitive to finer adjustments in the subsec- toral structure of the economy, as in the structure of manufactured exports. Have wage premia to skills embodying different levels of schooling (or dif- ferent types of schooling) changed? Wage premia and rates of returns have increased since the start of transition--particularly in the earlier periods of liberalizing the wage setting mechanism.24 By the early 2000s, rates of return to an average year of extra schooling (across levels) was comparable to that of developed economies. They varied significantly, from 4­6 percent in Roma- nia, Slovak Republic, Tajikistan, and Ukraine to 8­11 percent in Hungary, 22. The discussion in this section is based on work by Rutkowski (2009). 23. Professionals are individuals with a degree or training in tertiary education. 24. Rutkowski 2009. THE DAY AFTER 245 TABLE 6.5 Russian Federation--wage premia for additional schooling (percent) 2003 2007 Complete tertiary over: Incomplete tertiary 14% 47% Technical college 36% 56% Vocational schools 63% 63% General secondary 63% 57% Technical college over: Vocational schools 5% 5% General secondary 20% 0% Vocational schools over: Primary education 54% 25% General secondary over: Primary education 36% 30% Source: Kapelyushnikov 2007. Kazakhstan, and Poland. In the Russian Federation, they increased sharply from about 4 percent in the early 1990s to about 9 percent in the early 2000s. Comparing studies across countries and over time requires some caution because they may use different concepts of earnings (including some returns to capital), types of schooling, and different estimation techniques. A major shortcoming in this area is the absence of more recent wage data that would allow an assessment of how wage premia has evolved during the last decade. However, a study based on two of the most representative nation- wide surveys in the Russian Federation provides information on wage devel- opments.25 Table 6.5 shows how the Russian wage premium has evolved between 2003 and 2007, a period when GDP growth was sustained at about 7 percent a year: · Just completing tertiary education has a strong premium, increasing sharply between 2003 and 2007 from 14 percent to 47 percent. Perhaps completing tertiary and the associated certification is signaling other ele- ments of student's personalities and competencies (beyond pure additional training) that are highly valued by employers. This has also been found in other labor market studies in developing countries, attesting to the strong screening effect that employers may give to the certification effect of com- pleting tertiary education. 25. NOBUS 2003 and OZPP 2007. 246 TURMOIL AT TWENTY FIGURE 6.13 Kazakhstan survey of firms, 2008: importance of general competencies and technical skills Ability to work independently Time management skills Literacy Russian language skills Communication skills Analytical thinking Problem solving skills Customer relations skills Writing skills Mathematics (calculation) skills Computer skills Telephone conversation skills Kazakh language skills Management skills English language skills 0 25 50 75 100 Percent Note: The figure shows the percentage of firms reporting a specific skill to be important or very important for specialist/skilled workers. · The premium to technical college was small, and relative to general sec- ondary education, it disappeared completely in 2007. · The premium to vocational school over primary education was reduced by half--from 54 to 25 percent. · The premium to general secondary education over primary is roughly constant. The importance of general competencies more than narrow technical training is confi rmed by a recent survey of entrepreneurs in Kazakhstan. Kazakhstani firms highly value general competencies--such as the ability to work independently, time management, literacy, and language and communi- cations skills (figure 6.13). Math and computer skills were singled out less fre- quently.26 On the other hand, unemployment indicators across ECA provide a mixed picture in this area. In 2006, about half of ECA countries reported higher unemployment rates for workers with vocational education, and half reported higher unemployment for general secondary graduates.27 What are the implications of this evidence on labor markets--while acknowledging its fragmentary nature and its reliance on only a few coun- tries? First, there is an increase in demand for individuals with completed 26. Ivaschenko 2008. 27. Mertaugh (forthcoming). THE DAY AFTER 247 degrees in tertiary education, but it is not as obvious what attributes firms are valuing or how they are enhanced by tertiary education--this requires additional research. Second, the lower premia on specific technical degrees and the fact that management occupations have increased suggest that general competencies that may be associated with completing tertiary education are valued more than specific technical knowledge. Third, the low returns to mid- dle technical education observed in the Russian Federation, the mixed picture of the expansion of technicians as occupations, the increased share of ser- vices, the survey in Kazakhstan--all point to the importance of a secondary education that imparts a minimum set of general competencies. This is con- sistent with the discussions in ECA countries on the role of vocational edu- cation and on avoiding situations where students enter a narrow career path too early. Finally, the ability to adapt, retrain and absorb on-the-job training may be increasingly critical as the composition of output and employment changes. Education has more impact when it allows workers to cross sectors and occupations. Issues to focus on As mentioned earlier, the review of issues in education is based on incomplete evidence. ECA has a large variability of country situations calling for careful country taxonomies--which are absent here. But some areas appear common to most countries. Compulsory education (primary and secondary): the critical role of teachers. Most students in early grades seem to be performing well in international tests, and the information available reveals no major differences in perfor- mance between children of different socioeconomic status as measured by PISA data. It is important for these gains to be preserved--and accelerated in countries that are lagging. Assuring that minorities or more disadvantaged children in poorer rural areas have access to high quality schools remains a challenge. A specific problem, particularly but not exclusively in the CIS countries, is the coexistence of three developments. First, student-to-teacher ratios in the CIS countries are lower than other countries. For example, in Ukraine the student-to-teacher ratio is 9.5­60 percent of the OECD average. And these ratios are expected to fall with declining enrollment ratios and shrink- ing school age cohorts. Second, there is evidence that the teaching profession has become less attractive: the average age of teachers is rising rapidly, teach- ers are working fewer hours, and wages are low relative to other professions. 248 TURMOIL AT TWENTY TABLE 6.6 Students tutored by their own teachers Country Percent of students Tajikistan 51 Kazakhstan 40 Kyrgyzstan 39 Lithuania, Slovak Rep., Bosnia and <20.0 Herzegovina, Georgia, Ukraine Croatia, Poland <10.0 Source: Silova, Budiene, and Bray 2006. BOX 6.2 Ukraine norms for education facilities · 1 deputy school director per 11 class groups. · 1 managing director, if the village has more than 600 people (regardless the total number of children in the school). · 1 pedagogue per 8 class groups (regardless of the number of children in the school). · 1 extracurricular activities group leader per 16 class groups (regardless of the number of children in the school). · 1 cleaner per 500 square meters (0.5 per 250 square meters). · 1 coat room attendant per 200 coat spots in school. · 1 yard keeper per 1.5 hectares in each school. Third, payments by households to obtain additional tutoring from their own teachers seem to have become important (table 6.6). The combination of excessive numbers of teachers, fewer hours taught at school, and parents having to purchase additional tutoring services is consis- tent with poor incentives for teachers. But aggregate spending in education does not seem to be the problem--in fact, spending in many countries has increased, but these resources have not been targeted to increasing the remu- neration of highly qualified teachers. Spending on other inputs arise from centrally determined norms that prevent reallocation. For example, overman- ning and rigid input mixes characterize the education system in Ukraine.28 Specifically, the input mix is set centrally according to specific norms (in rigid ratios), and many norms are set per unit of physical facility, independent of demand and enrollment (box 6.2). The result is a high ratio of nonteaching 28. World Bank 2008c. THE DAY AFTER 249 staff to teaching staff--ranging from 0.7 in secondary schools to 1.5 in pre- schools. Such coefficients--added to the already low student-teacher ratio-- increase the cost per child.29 Specializing too young and too narrowly? In a large number of ECA countries, vocational education gives lower secondary graduates (usually age 15) the choice of moving to this stream rather than continuing into upper second- ary education. Th is choice allows students to complete grades 10 and 11 in vocational schools. Historically, this type of education has been important in the ECA countries. In fact, about 15 percent of the labor force in the Russian Federation in the mid-2000s had vocational education as the highest level of school attained. The figure for upper secondary graduates in the labor force was basically the same--about 16 percent. Vocational school enrollments have fluctuated widely among countries and across periods. During the first decade of transition (1989 to 1999), enroll- ments declined sharply across all countries, from 60 to 40 percent of total enrollment at the upper secondary level--the rest being general upper sec- ondary enrolment. During the next decade, vocational enrollments stabilized in most countries. This was a natural reaction by students to the first decade of transition, given that vocational schools were quite specialized and histori- cally tuned to the technological needs of the past industrial structure. And this is consistent with the sharp decline in premia still observed in the Rus- sian Federation. Countries are making vocational education less narrow and more adapt- able to changing circumstances. Poland has postponed by a year the time for graduates to move from lower secondary to vocational education, allow- ing students to absorb more general knowledge and avoiding excessive early specialization. In several countries vocational education is treated as a default option for weak-performing students--who are redirected into vocational schools rather than fi nishing upper secondary education. The performance of these students may be the result of having attended poor quality elementary schools and of their families' socioeconomic back- ground. To reduce the implicit income-based selection into vocational schools, it is critical to even out the quality of primary schools and lower secondary schools. 29. Simulations show that reforms that simultaneously address low student-teacher ratios, reduce non-teaching personnel, and allow flexibility of norms (including consolidation of facilities) could save 1 percent of GDP in Ukraine. These resources could then be used to increase the wages of teachers willing to go through training and certification and work normal hours at schools. 250 TURMOIL AT TWENTY Tertiary education. The most salient feature of education in ECA countries has been the massive enrollment increase in tertiary education. Some experts believe this is one of the fastest expansions of higher education in recent his- tory.30 And this happened across countries of very different development lev- els and economic structure. During the last decade, enrollment has more than doubled in Hungary, Kazakhstan, and Romania, and increased by more than 75 percent in Armenia, Kyrgyz Republic, Lithuania, the Slovak Republic, Tajikistan, and the Russian Federation. In some cases, the additional enroll- ment exceeded the flow of graduating students from secondary education-- signaling the enrollment in tertiary education of individuals already in the labor force. This massive increase in enrollment was fi nanced by the implicit liberal- ization of private financing. In some countries, private institutions emerged, and in others, public institutions simply started charging tuition and mobi- lizing private contributions. In countries as diverse as Georgia, Kazakhstan, and Poland, at least a third of students are enrolled in private schools. That figure is now one-fift h in Azerbaijan, Estonia, Latvia, and Romania.31 In many ECA countries, public institutions have been given significant latitude to raise funds through tuition--and are charging fees to at least 50 percent of their students.32 Much of the expansion in private schools was a response to a sharp increase in demand in specific areas not requiring expensive investments, (such as law, business, accounting, languages, economics, and management). In many countries, the authorities have provided tertiary schools with greater autonomy in their choice of personnel, programs, and tuition charges. Most progress has taken place in the new member states of the European Union and less in the CIS, with the exception of Kazakhstan. A recent study summarizes trends in competition, state regulation, and governance of insti- tutions.33 Poland and the Czech Republic seem to have progressed signifi- cantly in both competition and autonomy. No change is reported for Bulgaria, Croatia, and Slovenia. In most other ECA countries, improvements have been more modest. Tertiary education can adapt to changing labor market conditions if the system is more demand-driven. Allowing the private sector to enter the sector 30. Usher 2009. 31. For a comparative discussion of the role of the private sector in higher education in the Baltic countries and Belarus, see Vanags and Hansen 2005. 32. Usher 2009. 33. CHEPS 2006. THE DAY AFTER 251 and help mobilize resources and know-how could help--particularly in peri- ods of fiscal stringency. Public resources could flow to public institutions on the basis of per-student enrolled formulas, and these institutions could have more autonomy in curriculum changes and hiring and firing. Loans or grants for lower income students who are accepted under a competitive process are an important complement. Many countries have experimented with various options, particularly Hungary, Kazakhstan, Latvia, and Turkey. Many governments are concerned about allowing private institutions unless a good system of accreditation and monitoring of quality is in place. There is some concern about the transparency and objectivity of existing accreditation systems in several ECA countries. These are genuine concerns, but some good practice models are emerging. One is Chile, where new private institutions need to "earn" their right to full autonomy through a 10-year pro- bation period of monitoring and self-evaluation.34 34. Since its creation in 1990, the new regulatory regime granted autonomy to 26 universities and 13 professional institutes for successfully completing their period of supervision while closing down 16 private universities and 21 professional institutes for poor academic performance, fi nancial dis- tress, or poor governance. Thus as many institutions were closed as created. 252 TURMOIL AT TWENTY Bibliography Andreeva, Anastasia, and Robert Branda. 2009. Rates of Return on Assets and Equity for Private Banking Sector in Selected Countries of Western and Eastern Europe. Background Paper, World Bank, Washington, DC. Arvai, Zsofia, Karl Driessen, and Inci Ötker-Robe. 2009. Regional Financial Interlink- ages and Financial Contagion within Europe. IMF Working Paper 09/6, Interna- tional Monetary Fund, Washington, DC. Barth, James, Gerard Caprio, and Ross Levine. 2008. Bank Regulations Are Changing: For Better or Worse? Policy Research Working Paper 4646, World Bank, Washington, DC. Beblavy, Miroslav. 2009. European Union Influence on Institutional Lock-In in the Countries in Transition. Background Paper, World Bank, Washington, DC. Beck, Thorsten. 2009. Finance in Transition--A Long Road with More Challenges Ahead. Background Paper, World Bank, Washington, DC. Beck, Thorsten, and Asli Demirgüç-Kunt. 2009. Financial Institutions and Markets across Countries and over Time--Data and Analysis. World Bank, Washington, DC, mimeo. Besant-Jones, John. 2006. Reforming Power Markets in Developing Countries: What Have We Learned? Energy and Mining Sector Board Discussion Paper 19, World Bank, Washington, DC. Blanchard, Olivier, and Michael Kremer. 1997. Disorganization. Quarterly Journal of Economics CXII (4). Bogetic, Zeljko, Karlis Smits, Sergei Ulatov, Alisher Mirzoev, Marco Hernandez, and Stepan Titov. 2009. Russian Economic Report 18. Europe and Central Asia Region, World Bank, Washington, DC. Brooks, Robin, and Luis Catăo. 2000. The New Economy and Global Stock Returns. IMF Working Paper 00/216, International Monetary Fund, Washington, DC. Bubula, Andrea, and Inci Ötker-Robe. 2002. The Evolution of Exchange Rate Regimes Since 1990: Evidence From De Facto Policies. IMF Working Paper 02/155, Interna- tional Monetary Fund, Washington, DC. Carlin, Wendy, and Mark Schaffer. 2009. Public Infrastructure Constraints on Growth in the Transition Economies: The Legacy of Communism and Evidence from Enter- prise Surveys 1999­2008. Background Paper, World Bank, Washington, DC. Carlin, Wendy, Mark E. Schaffer, and Paul Seabright. 2007. Where Are the Real Bottle- necks? Evidence from 20,000 Firms in 60 Countries about the Shadow Costs of Con- straints to Firm Performance. IZA Discussion Paper 3059, Institute for the Study of Labor, Bonn, Germany. 253 ------. 2009. A Framework for Cross-Country Comparisons of Public Infrastructure Constraints on Firm Growth. Working Paper, mimeo. Center for Education Policy Studies. 2006. The Extent and Impact of Higher Educa- tion Governance Reform across Europe. Final Report to the Directorate-General for Education and Culture of the European Commission. Center for Higher Education Policy Studies (CHEPS), Enschede. Chenery, Hollis, Sherman Robinson, and Moshe Syrquin. 1986. Industrialization and Growth: A Comparative Study. New York: Oxford University Press. Claessens, Stijn. 2005. Policy Approaches to Corporate Restructuring Around the World: What Worked, What Failed? In Corporate Restructuring: International Best Practices, eds. Michael Pomerleano and William Shaw. Washington, DC: World Bank. Communication from the Commission to the Council and the European Parliament. 2007. Common Position of the Council on the Adoption of a Directive of the Euro- pean Parliament and of the Council Concerning Common Rules for the Internal Market in Electricity and Repealing Directive 2003/54/EC. EU Communication 2007/0195, Commission of the European Communities, Brussels. ------. 2009. Report on Progress in Creating the Internal Gas and Electricity Market. EU Communication 2009/115, Commission of the European Communities, Brussels. Darvas, Zsolt. 2009. The role of the European Union in Supporting Crisis-hit Countries in Central and Eastern Europe. Background Paper, World Bank, Washington, DC. De Haas, Ralph, and Ilko Naaborg. 2005. Does Foreign Bank Entry Reduce Small Firms' Access to Credit? Evidence from European Transition Economies. DNB Working Paper 050/2005, De Nederlandsche Bank NV, Amsterdam. De Larosičre, Jacques. 2009. Report of the High-Level Group on Financial Supervision in the EU. Brussels: European Commission. Djankov, Simeon, Caralee McLiesh, and Andrei Shleifer. 2007. Private Credit in 129 Countries. Journal of Financial Economics 84 (2): 299­329. EBRD (European Bank for Reconstruction and Development). 1999. Transition Report: Ten Years of Transition. London EBRD ------. 2008. Transition Report: Growth in Transition. London: EBRD. Enoch, Charles, and Inci Otker-Robe. 2007. Rapid Credit Growth in Central and Eastern Europe: Endless Boom or Early Warning. New York: Palgrave Macmillan. European Central Bank. 2009. Financial Stability Review. Frankfurt am Main, Germany: European Central Bank. Fagan, Gabriel, and Vitor Gaspar. 2007. Adjusting to the Euro. ECB Working Paper Series 716, European Central Bank, Frankfurt am Main, Germany. Fernández de Lis, Santiago, and Alicia García Herrero. 2009. The Spanish Approach: Dynamic Provisioning and Other Tools. In Towards a New Framework for Financial Stability, eds. David Mayes, Robert Pringle, and Michael Taylor. London: Central Banking Publications. Fernandez de Lis, Santiago, Jorge Martinez Pagas, and Jesus Saurina. 2001. Credit Growth, Problem Loans and Credit Risk Provisioning in Spain. In Marrying the 254 TURMOIL AT TWENTY Macro and Micro Prudential Dimensions of Financial Stability. BIS Papers 1, Bank for International Settlements, Basel, Switzerland. Financial Services Authority. 2009. The Turner Review: A Regulatory Response to the Global Banking Crisis. London: Financial Services Authority. Frye, Timothy. 2009. The Politics of Economic Policy Making and Economic Crisis. Background Paper, World Bank, Washington, DC. Hansen, Morten, and Alf Vanags. 2005. The Private Sector in Higher Education in the Baltics: Permanent Feature or Transition Phenomenon? Project RRC IV­086. Baltic International Centre for Economic Policy Studies, Riga, Latvia. Haselmann, Rainer, and Paul Wachtel. 2007. Risk Taking by Banks in the Transition Countries. Comparative Economic Studies (49): 411­29. Havrylyshyn, Oleh, and Saleh Nsouli. 2001. A Decade of Transition: Acvhievements and Challenges. International Monetary Fund, Washington, DC. Herzberg, Valerie, Naotaka Sugawara, and Juan Zalduendo. 2009. Golden Hand-Cuffs and Deleveraging. Europe and Central Asia Region, World Bank, Washington, DC, mimeo. Hoelscher, David, and Marc Quintyn. 2003. Managing Systemic Banking Crises. Occa- sional Paper 224, International Monetary Fund, Washington, DC. Holló, Dániel. 2007. Household Indebtedness and Financial Stability: Reasons to Be Afraid? MNB Bulletin. Budapest: Magyar Nemzeti Bank. Honohan, Patrick, and Luc Laeven. 2005. Systemic Financial Crises: Containment and Resolution. New York: Cambridge University Press. IFC (International Finance Corporation) and World Bank. 2008. Doing Business 2009. Washington, DC: International Finance Corporation. IMF (International Monetary Fund). 2009a. World Economic Outlook: Crisis and Recov- ery. Washington, DC: International Monetary Fund. ------. 2009b. World Economic Outlook: Sustaining the Recovery. Washington, DC: Inter- national Monetary Fund. ------. 2009c. Global Financial Stability Report: Responding to the Financial Crisis and Measuring Systemic Risks. Washington, DC: International Monetary Fund. Independent Evaluation Office of the IMF. 2003. The IMF and Recent Capital Account Crises: Indonesia, Korea, and Brazil. Washington, DC: International Monetary Fund. Ivaschenko, Oleksiy. 2008. The Survey of Skills/Labor Demand and Job Vacancies in Kazakhstan. PowerPoint presentation, March 11. European and Central Asia Region, World Bank, Washington, DC. Johansson, Martin, and Mattias Persson. 2006. Swedish Households' Indebtedness and Ability to Pay: A Household Level Study. Paper presented at the IFC Conference on Measuring the Financial Position of the Household Sector, Basel, Switzerland. Kapelyushnikov, Rostislav. 2008. Russia's Human Capital: An Assessment. Moscow: Cen- ter for Labor Market Studies, Higher School of Economics. Kawai, Masahiro, Ira Lieberman, and William Mako. 2000. Financial Stabilization and Initial Restructuring of East Asian Corporations: Approaches, Results and Lessons. BIBLIOGRAPHY 255 In Managing Financial and Corporate Distress: Lessons from Asia, eds. Charles Adams, Robert E. Litan, and Michael Pomerleano. Washington, DC: The Brookings Institution. Keller, Christian, and Svitlana Maslova. 2009. Western Europe's Banks in Emerging Europe. Barclays Capital Research Note. London: Barclays Bank, PLC. Kennedy, David. 2003. Power Sector Regulatory Reform in Transition Economies: Prog- ress and Lessons Learned. EBRD Working Paper 78, European Bank for Reconstruc- tion and Development, London. Laeven, Luc, and Fabien Valencia. 2008. Systemic Banking Crises: A New Database. IMF Working Paper WP/08/224, International Monetary Fund, Washington, DC. Landesmann, Michael. 2009. Trade Relations and the Economic Crisis. World Bank EU10 Quarterly Report. Washington, DC: World Bank. Lieberman, Ira, Mario Gobbo, William Mako, and Ruth Neyens. 2005. Recent Interna- tional Experience in the Use of Voluntary Workouts under Distressed Conditions. In Corporate Restructuring: International Best Practices, eds. Michael Pomerleano and William Shaw. Washington, DC: World Bank. Lieberman, Ira, and William Mako. 2009. An International Comparison of Recent Expe- riences in Corporate Restructuring Workouts during Financial Crises. Mimeo. Lindert, Kathy. 2009. Social Protection and Economic Shocks in ECA: The Social Side of the Global Crisis. Stocktaking note. Europe and Central Asia Region, World Bank, Washington, DC. Lindert, Kathy, Emmanuel Skoufias, and Joseph Shapiro. 2006. Redistributing Income to the Poor and the Rich: Public Transfers in Latin America and the Caribbean. Social Protection Discussion Paper 0605, World Bank, Washington, DC. Loayza, Norman, Klaus Schmidt-Hebbel, and Luis Serven. 2000. What Drives Private Saving around the World? Policy Research Working Paper Series 2309, World Bank, Washington, DC. Markandya, Anil. 2009. The Environment and Natural Resources. Background Paper, World Bank, Washington, DC. Mertaugh, Michael. 2008. Making Secondary Education Systems in ECA More Respon- sive to Changing Skill Needs. Background Paper for Closing the Skill Gap Study of the ECA Region of the World Bank (Forthcoming). Mitra, Pradeep. 2008. Innovation, Inclusion and Integration: From Transition to Con- vergence in Eastern Europe and the Former Soviet Union. Washington, DC: World Bank. Mitra, Pradeep, Alexander Muravyev, and Mark Schaffer. 2009. Convergence in Insti- tutions and Market Outcomes: Cross-Country and Time-Series Evidence from the Business Environment and Enterprise Performance Surveys in Transition Econo- mies. Policy Research Working Paper 4819, World Bank, Washington, DC. And IZA Discussion Paper 3863, Institute for the Study of Labor, Bonn, Germany. Nellis, John, and Ira Lieberman. 2009. Privatization in Transition States: Lessons for Latecomers. Background Paper, World Bank, Washington, DC. 256 TURMOIL AT TWENTY Nguyen, Trang Van, Ramya Sundaram, and Emil Tesliuc. In progress. The Redistributive Impact of Social Protection Systems in Europe and Central Asia. Draft. Europe and Central Asia Region, World Bank, Washington, DC. Official Journal of the European Union. 2003. Directive 2003/54/EC of the European Parliament and of the Council. European Union, Brussels. Parks, James. 2009. Turkey: Instability to Crisis and Recovery. Background Paper, World Bank, Washington, DC. Pistor, Katharina, Martin Raiser, and Stanislav Gelfer. 2000. Law and Finance in Transi- tion Economies. Economics of Transition 8: 325­38. Pricewaterhouse Coopers (PwC) Consortium. 2004. The South Eastern Europe Power Generation Investment Study. The European Union's CARDS Programme for the Balkan Region. Brussels: European Commission. Ramakrishnan, Uma, and Juan Zalduendo. 2006. The Role of IMF Support in Crisis Pre- vention. IMF Working Paper 06/75, International Monetary Fund, Washington, DC. Rowthorn, Bob, and Ramana Ramaswamy. 1998. Growth, Trade, and Deindustrialization. IMF Working Paper No. 98/60, International Monetary Fund, Washington, DC. Rutkowski, Jan. 2009. Demand for Skills in ECA: Main Trends and Patterns. Human Development Unit, Europe and Central Asia Region, World Bank, Washington, DC. Silova, Iveta, Virginia Budiene, and Mark Bray. 2006. The Growing Phenomenon of Pri- vate Tutoring. In Education in a Hidden Marketplace: Monitoring of Private Tutoring, eds. Iveta Silova, Virginia Budiene, and Mark Bray.New York: Open Society. Sharma, Raghuveer. 2005. Tajik Energy Utilities Reform Review. Europe and Central Asia Region, World Bank, Washington, DC. Sonje, Velimir. 2009. Credit Growth in CEE: Real Effects of International Integration and Crisis. Background Paper to be presented at the Second ZSEM Economic Con- ference, November 6. Zagreb, Croatia. Stockman, Alan C. 1989. Sectoral and National Aggregate Disturbances to Industrial Output in Seven European Countries. NBER Working Paper 2313, National Bureau of Economic Research, Cambridge, Mass. Sugawara, Naotaka, and Juan Zalduendo. 2009. Public Monies and Mortgage Restructuring--The Need for Targeted Interventions. Europe and Central Asia Region, World Bank, Washington, DC, mimeo. Sveriges Riksbank. 2009. Financial Stability Report. Stockholm: Sveriges Riksbank. Tiongson, Erwin, Anna Gueorguieva, Victoria Levin, Kalanidhi Subbarao, Naotaka Sugawara, Victor Sulla, and Ashley Taylor. 2009. Europe and Central Asia: The Crisis Hits Home. Stress Testing Households in Europe and Central Asia. Report No. 50158. Washington, DC: Europe and Central Asia Region, World Bank. Usher, Alex. 2009. Financing Higher Education in the ECA Region. Background Paper for Closing the Skill Gap Study of the ECA region of the World Bank (forthcoming). Vatne, Bjřrn. 2006. Financial Margins in Norwegian Households--An Analysis of Micro Data for the Period 1987­2003. Paper presented at the IFC Conference on Measuring the Financial Position of the Household Sector, August 30­31, Basel, Switzerland. BIBLIOGRAPHY 257 World Bank. 2002. Transition: The First Ten Years: Analysis and Lessons for Eastern Europe and the Former Soviet Union. Washington, DC: World Bank. ------. 2003. Private Participation in the Power Sector in Europe and Central Asia. Wash- ington, DC: World Bank. ------. 2005. Growth, Poverty, and Inequality: Eastern Europe and the Former Soviet Union. Washington, DC: World Bank. ------. 2006. Infrastructure in Europe and Central Asia Region. Washington, DC: World Bank. ------. 2008a. Developing a Sustainable Energy Sector for Kosovo. Washington, DC: World Bank. ------. 2008b. South East Europe: Regional Gasification Study. Washington, DC: World Bank. ------. 2008c. Ukraine: Improving Intergovernmental Fiscal Relations and Public Health and Education Expenditures. Washington, DC: Europe and Central Asia Region, World Bank. ------. 2009a. Kazakhstan Economic Report No. 2. Washington, DC: World Bank. ------. 2009b. Republic of Tajikistan: Poverty Assessment. Washington, DC: Europe and Central Asia Region, World Bank. ochowski, Dawid, and Slawomir Zajczkowski. 2006. The Distribution and Dispersion of Debt Burden Ratios among Households in Poland and Its Implications for Finan- cial Stability. MPRA Paper 692, University Library of Munich, Germany, mimeo. 258 TURMOIL AT TWENTY ECO-AUDIT Environmental Benefits Statement The World Bank is committed to preserving Saved: endangered forests and natural resources. The · 9 trees Office of the Publisher has chosen to print · 3 million Btu of Turmoil at Twenty on recycled paper with total energy 30 percent postconsumer fiber in accordance · 867 lb. of net with the recommended standards for paper greenhouse gases usage set by the Green Press Initiative, a · 4,176 gal. of waste nonprofit program supporting publishers water in using fiber that is not sourced from · 254 lb. of solid waste endangered forests. For more information, visit www.greenpressinitiative.org. 259 The transition economies of Central and Eastern Europe and the former Soviet Union, among all emerging- and developing-economy regions, have been hardest hit by the global economic crisis of 2008-09. This is partly due to the region's deep integration into the global economy across many dimensions--trade, financial, and labor flows. Attempts by countries that came later to the transition to catch up rapidly to Western European living standards at a time when global liquidity was unusually abundant, together with some policy weaknesses, made them vulnerable to reversals in market sentiment. Written on the eve of the twentieth anniversary of the fall of the Berlin Wall, Turmoil at Twenty analyzes the run-up to the current crisis and addresses a number of key questions related to vulnerability to the recession, expected recovery, and necessary reforms in the region: · Did the transition from command to market economies, and the period during which this took place, plant the seeds of vulnerability that made transition countries more prone to crisis than other developing countries? · Did the choices made on the road from plan to market shape the ability of crisis-hit countries to recover? What combination of domestic policy reform and international collective action is needed to bring about a recovery and minimize the humanitarian cost of the crisis? · What structural reforms are needed today to address the most binding constraints on growth in a world where capital flows to transition and developing countries are expected to be considerably lower than before the crisis? Turmoil at Twenty will be of interest to policy makers and their advisers, researchers, and students of economics who seek lessons from the current economic crisis, as well as scholars of the transition. Pradeep Mitra and Marcelo Selowsky are former Chief Economists in the World Bank's Europe and Central Asia Region. Juan Zalduendo is Lead Economist in the World Bank's Office of the Chief Economist, Europe and Central Asia Region. ISBN 978-0-8213-8113-7 SKU 18113